INFORMATION TO USERS - TSpace

359
INFORMATION TO USERS This manuscript has been repcoduced fFom the microfim master. UMI Mms the text diredly from the original or copy submitted. Thus, some thesis and dissertation eopies are in typewriterfBceF while others may be from any type of computer printer. The quality of this reproduction is dependent upon the qualii of the copy submitted. Broken or indisfrtnd print, dored or poor quality illusttations and photographs, pcint bleedthrough, substandard margins, and improper alignment can adversely affect reprodudion. In the unlikely event that the author did not send UMl a complete manuscript and them are missing pages, these will be noted. Also, if unauthorized copwght material had to be removed, a note will indicate the deletion. Oversize materials (e.g., maps, drawings, cham) am reproduced by seething the original, beginning at the upper teft-hand comer and continuing ftom left to right in equal Worn with small overlaps. Photographs induded in the original manuscript have bean reproduced xerographically in this copy. Higher qualii 6" x 9" black and Mite photographic prints am available for any photographs or illustratims appearing in this copy for an addiional charge. Contact UMI diredly to order. Bell & HowaU information ad Learning 300 North Zeeb Road, Ann Arbor, MI 48106-1346 USA 800-52t-0800

Transcript of INFORMATION TO USERS - TSpace

INFORMATION TO USERS

This manuscript has been repcoduced fFom the microfim master. UMI Mms

the text diredly from the original or copy submitted. Thus, some thesis and

dissertation eopies are in typewriterfBceF while others may be from any type of

computer printer.

The quality of this reproduction is dependent upon the qualii of the

copy submitted. Broken or indisfrtnd print, dored or poor quality illusttations

and photographs, pcint bleedthrough, substandard margins, and improper

alignment can adversely affect reprodudion.

In the unlikely event that the author did not send UMl a complete manuscript

and them are missing pages, these will be noted. Also, if unauthorized copwght material had to be removed, a note will indicate the deletion.

Oversize materials (e.g., maps, drawings, cham) am reproduced by

seething the original, beginning at the upper teft-hand comer and continuing

ftom left to right in equal W o r n with small overlaps.

Photographs induded in the original manuscript have bean reproduced

xerographically in this copy. Higher qualii 6" x 9" black and Mite

photographic prints am available for any photographs or illustratims appearing

in this copy for an addiional charge. Contact UMI diredly to order.

Bell & HowaU information a d Learning 300 North Zeeb Road, Ann Arbor, MI 48106-1346 USA

800-52t-0800

A COMPARATIVE STUDY OF HUMAN SERVICE PROFESSIONALS' PERCEPTlONS OF MINIMUM

CHILD PROTECTION ENTRY POINTS FOR NEGLECT: AN ONTARIO PERSPECTIVE

Deborah Goodman

A thesis submitted in conformity with the requirements for the degree of Doctor of Philosophy,

Faculty of Social Work. University of Toronto.

Copyright @ I999 by Deborah Goodman

National Libmy B*m of Canada Biblioth&que nationale du Canada

Acquisitions and Acquisitions et Bibliographic Sewices sew ices bibliograp hiques

395 Wellington Street 395. rue Wellington OaawaON KIA O N 4 OtrawaON KIA O M Canada Canada

The author has granted a non- exclusive licence dowing the National Library of Canada to reproduce, loan, distribute or sell copies of this thesis m microform, paper or electronic formats.

The author retains ownership of the copyright in this thesis. Neither the thesis nor substantial extracts fiom it may be printed or otherwise reproduced without the author's permission.

L'auteur a accorde une licence non exclusive pennettant a la BibIiotheque nationale du Canada de reproduire, preter, distrr'buer ou vendre des copies de cette these sous la forme de microfiche/film, de reproduction sur papier ou sur format eIec tronique .

L'auteur conserve la propriete du droit d'auteur qui protege cette these. Ni la these ni des extraits substantiels de cefle-ci ne doivent &e imprimes ou autrement reproduits sans son autorisation-

ABSTRACT

=A Comparative Study of Human Service Professionals' Perceptions of Minimum Child Protection Entry Points for Neglect:

An Ontario Perspective"

Deborah Goodman, 1999.

For the degree of Doctor of Philosophy, Faculty of Social Work, University of Toronto.

ABSTRACT

&L** In North America, neglect is more prevaient ttIO,. child abuse. In Ontario, neglect is a

significant fador in the leading cause of death for children: unintentional injuries. Yet,

recognition of neglect as a serious danger to children and clear child protection service

(CPS) definitions are lacking. Despite the critical features of neglect and the key role

human service professionals (HSP) hold as referral sources and gatekeepers of CPS,

there has been a paucity of research on what HSP perceive as the minimal point at

which CPS is warranted i.e. the minimum CPS entry point or intervention line (MIL).

The objective of this secondary study was to analyze secondary survey data on 679

Ontario HSP perceptions of the MIL for nine forms of neglect and test whether regional

socioeconomic (SES) variance and agency type infi uenced perceptions. An explanatory

approach was taken and Adaptation-level (AL) theory used to guide understanding

about individual, group and community adaptation to contextual facton, as well as

expand Wolock's (1982) American test of AL theory. Secondary analysis used

nonparametric tests, non-diredional hypotheses and an alpha feveI of pc.05.

The secondary sunrey data were obtained from a convenience sample of six Ontario

CPS agencies that had independently surveyed CPS staff and community HSP

regarding the minimum CPS entry point for 24 Reasons for Service. AII six regions used

the same CPS eligibility tool (Intervention Spectrum) and the same survey method.

Regional SES findings were mixed, as differences in HSP perception of entry points

appear to exist (p<.02), but only with some forms of neglect. Suggested rationale for the

lack of support for AL theory may be this study's large sample unit (region) and the study

effects of more extreme SES polarization in the United States compared to Canada. No

differences between agency type (investigating and noninvestigating agencies) were

found. Overall, the analysis suggests that the construction of neglect is not a static

phenomena within and across agencies and regions and Canadian child welfare may

differ Porn American experiences. Additional research is needed to better understand

which factors, in particufar contextual ones, influence HSP decisions about child neglect

in Canada.

ACKNOWLEDGMENTS

We are each of us angels wHI, only one whg. And we can only fly by embracing each other;

Lucian0 de Crerrcenzo ,q928

This is my opportunity to formally thank and acknowledge the many "angelsn who assisted me in this "flight'.

Heartfdt thanks is given to Professor Robert MacFadden. Throughout my academic career he has always encouraged me to try different and more challenging flights. Quite simply, as a teacher, a mentor, my comprehensive advisor, and now as my thesis supervisor, he personifies professorial excellence. A special thanks is extended to Professor Esme Fuller-Thomson for her statistical wizardry and for helping me to remember to laugh along the way. To my other thesis committee members, Professors' Marion Bogo, Grant Macdonald, and Nico Trocme, 1 thank them for their belief in this study, their perceptive comments, their unending support, and their gentle and patient guidance towards my independent flight.

I would also like to acknowledge the contribution and support of Mary Ballantyne, George Leck (Simwe Children's Aid Society and original authors' of the Spectrum), and Margaret Morrison (Ontario Association of Children's Aid Societies). They encouraged me to join them in their Right. They are pioneers in the field of inviting the communities they serve to be full and contributing partners in €he area of child protection. Additional thanks is given to Mary McConville and the Ontario Association of Children's Aid Societies for authorizing the use of the Intervention Spectrum as the survey tool.

l sincerely thank Bruce, Frontenac (Kingston). Halton, Huron, Peel, and Sudbury Children's Aid Societies' for sharing their survey data with me. l also want to acknowledge the contribution of nearly 700 human service professionals who gave their time and thought to the community survey. The total hours dedicated to just their rating of the Spectrum was over 21 00 hours.

Sincere appreciation goes to my editor, my motheFin-law, and my ftiend, Mary Mahoney. 1 also extend very special thanks to Aida Figueiredo; she showered my children with love and watched over them so carefully during this endeavor.

Finally, to the love of my lile. my husband, Andrew Mahoney, and to the lights of my life, our children, Ian and Kate and Colin Mahoney, thank each of you for your indefatigable support and encouragement throughout this lengthy process. 1 also want to recognize my parents. Captain Frederick Goodman and Patricia Goodman. While their flights ended long ago, their love and influence continues.

I dedicate this thesis to %on qua a, " for want of knowing.

TABLE OF CONTENTS

ABSTRACT ACKNOWLEDGMENTS TABLE OF CONTENTS LIST OF TABLES LIST OF APPENDIXES

CHAPTER 1: FOCUS OF STUDY f ntroduction Substantive fssues Theoretical Framework Survey Instrument - The Spectrum

Development of the Spectrum Eligibilii Tool Design Purpose of Original Survey Purpose of Secondary Analysis Survey tool Design The Minimum Intervention Line: Same, More Inclusive, Less Inclusive Spednrrn Research Study Relationship of Secondary Analysis to Spectrum Research Study Ethics Review Reliability and Validity

Definitions Relevance of Study to Advance Knowledge Analytic Stance

CHAPTER 2: LITERATURE REVIEW Social Work Epistemology

The Dominance of Borrowed Knowiedge Paucity of Theory Assessment

Social Construction of Child Wehre Child Protection Eligibility Instnrments Studies On HSP Decision Making in Child Welfare Research Hypotheses Summary

ii iii iv vi vi

CHAPTER 3: ORIGINAL SURVEY - DESIGN AND METHODOLOGY Original Suwey Study Definition Overview of Original Suwey Design Description of Suwey Procedures Overview of Suwey Tool

Benefit to CPS Agencies in Using Suwey Tool Pretest & Pilot Test of Respondent Information Form: Regions A, 8 & C Sampling Issues - Agencies Selection Sampling Issues - Respondent Selection

Ovmoverage of CPS Respondents Undercoverage of Community HSP Respondents

Data Collection Methods Duplication Summary

CHAPTER 4: SECONDARY ANALYSIS - DESIGN & METHODOLOGY Secondary Analysis Study Design

Ability of Survey Tool to Differentiate MIL: Individual, Agency & Region Level Sampling issues - CPS Agencies & Regions Sampling Issuer - HSP Agencies Methods to Stabilize Secondary Data Coding of Secondary Data Method of Data Analysis

Methods Used to Increase Power of Statistical Tests Determination of Regional SES Rank Rationale for Nonparametric Tests Hypothesis I : Regional SES Groups - Kruskal-Wallis Test Hypothesis 2: Agency Type - Mann-Whitney U Test

Summary

CHAPTER 5: SAMPLE AREAS AND RESPONDENTS Description of Sample Regions

Region A Region B Region C Region D Region E Region F

Regional Characteristics Profiles of HSP Respondents: Regions A, 6 & C

Personal Characteristics Professional Characteristics

Summary

CHAPTER 6: RESULTS Tests of Hypotheses

Hypothesis 1 : Regional SES Groups Hypothesis 2: Agency Type

Summary

CHAPTER 7: DfSCUSSlON & LlMlTATlONS Discussion

Hypothesis I : Regional SES Groups Hypothesis 2: Agency Type

Limitatfons Original Survey Design Survey Instrument Sampling lssues with Original Survey Data Secondary Data lssues Replication lssues

Summary

CHAPTER 8: SUMMARY & CONCLUSIONS Summary Conclusions

LIST OF TABLES Table 1

Table 2

Table 3 Table 4 Table 5 Table 6 Table 7 Table 8

Table 9

Table 20 Table 22 TabCe 72 Table 13

Table I 4 Table 15

Example of Eligibility Tool Design: Ordinal Format with RisklSeverity Categories

Example of Survey Tool Design: Ordinal Format without RisWSeverity Categories

Example of Completed Survey Score Sheets for IA and IIA Example of a Same MIL, a More Inclusive MIL & a Less lnclusive MIL Characteristics of the Six CPS Agencies & Regions Breakdown of Survey Forms Collected by Region HSP Groups by Region Number of Direct ServicelManagement CPS Staff to Number

of CPS Survey Respondents: CPS Participation Rate Ratio of CPS Respondent to Community HSP (CHSP) Respondent

by Region Similarities & Difterences in the Set MIL for 8 CPS Agencies Similarities & Differences in Set Neglect MIL for 8 CPS Agencies Regional Convenience Sample Variation Relative Proportion Between the Regions' Population and the Survey Sample Set Number of Spectrum Descriptors by Neglect Seaon Ranking of Sample Regions by SES Variables

Table 16 Breakdown of Sample Regions into Three SES Groups Table 17 Characteristics of Regions in SES Groups Table 18 Overall HSP Characteristics: Regions A, 6 8 C Table 19 Regional Variations in HSP Personal Characteristics:

Regions A, 8 & C Table 20 Regional Variations in HSP Pmfkssional Characteristics:

Regions A, B & C Table 21 Hypothesis 1 Results: Regional SES Group Variable Table 22 Hypothesis 2 Results: Agency Type Variable

LIST OF APPENDlXES Appendix A Spectrum with riskheverity groupings

- all 24 CPS 'Reasons for Servicen Appendix 0 Spectrum with out risklseverity groupings

-only 9 Neglect CPS 'Reasons for Servicen Appendix C Spectrum Matrix Appendix D Child Protection Services Staff Groups Appendix E Community Human Senrice Profsssional Groups Appendix F 'a siden Personal & Professional Characteristics Appendix F 'b side* Survey Score Sheet Appendix G 'a siden Survey Score Sheet: Example Appendix G 'b siden Common Information Sheet Appendix H Consent Form: CAS Agency Staff Appendix I Consent Form: Community Members Appendix J Definitions Appendix K Community Invitation Letter Appendix L OACAS letter - use of copyrighted material Appendix M University of Toronto, Ethics Review

CHAPTER 1 : FOCUS OF S f UOY

'Child abuse' is dtimatsly what a community, through its mprwsentatives (Le. cMld protection worken), deffnes as

unreasonable and inappropriate child cam. Michael Robin,l99l,p.3

Introductton

Child maltreatment is now recognized as a serious national and international problem.

In fact, compared with other childhood problems, '...child abuse and neglect are

particularly important because they are often directly associated with adverse physical

and mental health consequences in children and families' (National Research Council,

1993, p.6). The national incidence rate f c r child maltreatment in the United States in

1986 was 22.6 per 1,000 children (Sedlak, 1991); In Ontario in 1993, it was estimated at

21 per 1.000 children (Trocme, McPhee, Tam & Hay, 1994). In the last century, the

picture of an abused child carries enormous symbolic strength in North America as it

runs contrary to our broader social values and has resulted in 'child protection" being a

well accepted social concern (Finkelhor, t 994; Robin,?99t). Viewed as a social problem,

public and private agencies such as Child Protection Service (CPS) agencies have been

created and professional intervention, funding and research sanctioned. While research

efforts have begun to establish reasonable parameters regarding incidence, cause,

effects, treatment and prevention of child abuse (Daro, 1988). research knowledge in

many areas in child welfare is still lagging behind practice requirements. In particular,

despite the key role human service professionals (HSP) hold as referral sources and

gatekeepers of CPS, there has been a paucity of research on what HSP perceive as the

minimum point at which CPS is warranted i.e. the CPS entry point or minimum

intentention line.

Understanding the links between perceptions, attitudes, decision making, behaviour,

and the contextual factors is important to know, as these factors may influence HSP who

make chiid maltreatment decisions. While the importance of the attitudes-behaviour field

is evident from the vast literature in the social science fields (Fishbein, 1967; Helson,

1966; Krebs & Schmidt. 1993; Petty & Krosnick, 1995). in the past behaviour has too

often been studied in isolation (Upmeyer & Six, 1989), without regard to the contextual

element The contexiual/environ mental factor has been identified as an important factor

in the etiology of child maltreatment (Belsky, 1993; National Research Councii, 1993;

Wolock, 1982). Yet, recognition of an issue does not always mean there is a focused,

comprehensive research agenda. For example, despite continuing documentation of

professional error in referring and accepting CPS cases (Howitt, 1992; Robin, 1991;

US. Department of Health & Human Services [DHHS], 1991), empirical knowledge

remains limited about the factors that affect HSP perceptions of CPS entry points. One

goaI of this thesis is to advance knowledge about HSP perceptions of child protection

entry points by examining specific contextual factors that may influence perceptions.

Another area that requires an increased research agenda is child neglect (National

Research Council, 1993; Trocme, 1 WZa). As one type of child maltreatment, neglect is

the omission of care of a child by the caretaker that results h harm or substantial risk of

harm to the child. In North America, neglect is poorly defined by law, lacks operational

definitions, and has vague intervention standards (Giovanonni & Becem, 1979).

Variance in neglect is not only evident when intemationai, provincial and interregionai

definitions are examined, but interpretation of neglect can differ intraregionaKy between

different agencies, between HSP groups, and even between individual HSP in the same

agency (Erickson & Egeland, 1996; Lack & Ballantyne, 1994; National Research

Council, 1993; Trocme, 1 992a).

The absence of definitional clarity for child neglect rests in part with the fact it is a

socially constructed concept, affected by factors such as time, place, and culture. In the

Unitad States, neglect is more prevalent than child abuse (DHHS, 1991). In Ontario,

neglect is a significant factor in the leading cause of death for children: unintentional

injuries (Ontario Child Mortality Task Force, 1997). Neglect is also more closely

associated with professionals' level of exposure to neglect and environmental issues

such as poverty (Robin, 1991; Trocme, 1992a). Yet, despite the critical features of

negled. it lacks widespread recognition as a serious danger to children. In order to try

and gain a better understanding about HSP perceptions about child neglect, this thesis

examines where HSP perceive the CPS entry point to be for nine fonns of negled

Adaptation-Level theory (AL) is used to give guidance in thinking about how contextual

or environmental factors may affect HSP perceptions of neglect entry points. AL theory

posits that due to internal and external factors, an individual (or group or community)

actively adapts or adjusts to their environment. The adaptation-level is the 'pooled

effect" of an individual's adaptation to various sources of stimulation- What develops as

a result of adaptation is a reference system or referent point AL theory is tested for

three key reasons. One, AL theory emphasizes that individuals actively adapt to their

situation and it is the pooled effect of a11 park of the environment that influence

responses (Helson, 1964). Two, AL theory has been well tested in a variety of fields, as

welI, this theory can allow examination of individual, group and community levels

(Appley, 1971). Finally, testing AL theory replicates Wolock's (1982) use of AL theory

with CPS workers in New Jersey, although her survey design, methodology, and locale

differ significantly from this test of AL theory.

Using secondary survey data and guided by AL theory and empirical knowledge to date,

three contextual variables were selected to test whether the variables affected where

679 Ontario HSP perceived the minimum intervention line (MIL) for nine forms of

neglect. The independent variables chosen for the secondary analysis are: regional

socioeconomic (SES) variance (Drake & Pandey, 1996; Garbarino 8 Crouter, 1978,

Spearly & Lauderdale. 1982; Wolock, 1982; Zuravin, 1989), agency type (DHHS, 1988;

Zellman and Faller, 1996), and CPS workers' years of experience with CPS cases

(Trocrne, McPhee. Tam & Hay, 1994). While a CPS worker's years of experience with

CPS cases is an important variable to examine, unfortunately there were not enough

inexperienced workers in the sample (n=7) to do the secondary analysis.

The secondary survey data were obtained from a convenience sample of six Ontario

CPS agencies that had independently surveyed CPS staff and various community

groups regarding where the survey respondent perceived the minimum CPS entry point

to be in their community for 24 Reasons for Service. All six CPS agencies asked the

same question, used the same CPS eligibility tool as the survey instrument, and

employed the same survey method. The tool used to collect the original survey data on

24 Reasons for Senrice, is the lntenrention Spectrum (Spectrum). The secondary

analysis only examined the following nine Reasons for Service pertaining to neglect:

Abandonment Lack of Medical Attention, Lack of Supervision, Lack of Appropriate Child

Care Arrangements, Nutritional Neglect, lnadequate Physical Living Conditions,

Inadequate Personal Hygiene, Lack of Household Sanitation, and lnadequate Clothing.

Used as a CPS gatekeeping tool in direct practice situations, the Spectrum assists CPS

workers in making more consistent eligibility decisions at the time of the initial CPS

referral. As an eligibility instrument, the Spectrum has a matrix design. The horizontal

row has four risklseverity categories: most, medium, least and no. The protection

segment of the Spectrum has 24 vertical columns or scales. Each protection scale

dsscri-bes a specific CPS Reason for Service; each scale has a range of descriptors,

listed in an ordinal format. The most risklseverity situations are always described first,

followed by medium riskheverity descriptors, least risklseverity descriptors, and the last

descriptor is always a no risklseverity situation (see Appendix A). The relevant Ontario

child welfare legislation precedes each protection scale.

Used as a tool to survey HSP perceptions of CPS entry points, the Spectrum's

horizontal row (the four risklseverity categories) was removed (see Appendix 8). In the

primary research, the task of the survey respondent for each protection scale was to

read the legislation and the ordinal descriptors, and then place each descriptor under

one of the four risklseverity headings (mosV mediumlleastlno) on the Survey Score

Sheet Essentially, in collecting the original survey data each HSP respondent identified

two types of cases. The first type were the mandated CPS situations; these included alt

descriptors the respondent perceived as falling under the most and medium risklseverity

categories. The second type of cases were the non-CPS cases or community cases,

and included all descriptors that fell into the least and no risklseverity categories. For the

purposes of the secondary analysis, it is this demarcation line between the two types of

cases that is called the CPS entry point or minimum intervention line (MIL). The same

MIL meant the same number of descriptors had been seleded by the HSP respondents

for the mandated CPS situations and for the non-CPS cases. A significantly different

neglect CPS entry point either meant the MIL was more inclusive, where more

descriptors of a less severe nature were included, which may broaden the purview of

CPS, or the MIL was less inclusive, which meant fewer higher severity situations were

selected, which in practice, may narrow the interpretation of the CPS mandate.

The original survey data are ordinal. The secondary analysis of the survey data used

nonparametric tests and an alpha level of pc.05. The decision on the direction of the

hypotheses was made based on the following rationale. If the venue for the original

survey data had been the United States, then directional hypotheses and one-tailed

tests would be appropriate for the secondary analysis based on AL theory and findings

from other American studies. The prediction would be the low SES group would

perceive a less inchsive MIL compared to the high SES group (Wolock, 1982; Zuravin,

'l989) and investigating agencies would perceive a less inclusive MIL than

noninvestigating agencies (Zellman & Faller. 1996). However, because the current

knowledge base about the ecological determinants of child maltreatment and what

factors affect HSP decisions is still evolving (Belsky. 1993), the sample unit (region) in

the analysis of the SES variable is very large, few Canadian studies exist on the issue,

and the fad American and Canadian SES contexts differ considerably, non-diractional

hypotheses and two-tailed tests were selected.

The first chapter identifies the substantive issues that support the secondary analysis,

the theoretical framework is presented, and the development, design and psychometric

properties of the survey tool are noted. How the minimum intervention line (MIL) is

determined and what is meant by a same MIL, a more inclusive MIL or less inclusive MIL

is clarified. Also included are key definitions, the relevance of the study to advance

knowledge, and the analytic stance of this author. The second chapter examines the

literature on CPS eligibility determination instruments and reviews professional surveys

and decision making studies regarding child maltreatment. Analysis is set within a

broader examination of social work epistemology and the social construction of child

welfare. Social constructionist theory (Best, 1989; Parker, 1998) is drawn on to show

how certain facton have affected knowledge development in social work and its sub-

disciplines, like child welfare. The third chapter describes the study design and

methodology used in collecting the original survey data. The fourth chapter presents the

design and methodology used in the secondary analysis. The fifth chapter profiles the

regions and the HSP respondents in the secondary analysis. The sixth chapter presents

the secondary findings and the seventh chapter discusses the results and limitations.

The last chapter provides an overall summary of the secondary research and presents

conclusions.

Substantive Issues

One of the primary responsibilities of the Child Protection Service (CPS) agency is to

receive and investigate reports of child maltreatment. However, the protection of

children from maltreatment should be a community-wide concern (Besharov, 199 1 ).

Human service professionals (HSP) are key players in both child welfare practice and

policy. They deliver child protection services to different client groups, col laboratively

investigate child maltreatment with CPS, interpret CPS legislation, enact and define

CPS policy, and of particular relevance for this study, make referrals to and accept

raferrals for CPS agencies. Yet, knowledge about the HSP views on what cases are

reportable, as defined by legislation, and therefore eligible to receive child protection

services, is still in its formative stages. In the original survey study, CPS eligibility

determination referred to whether a child's situation met legislative requirements to

warrant a protection assessment.

With respect to the actions of HSP, referrals to CPS are crucial because protection of

and services to children can only be provided when child maltreatment is reported. North

American child abuse reporting laws mandate designated professional groups report

suspected child abuse. Despite such reporting requirements, a reporting rate to CPS

that has consistently increased since 1970, and the fact that professionals account for

the majority of child maltreatment reports to CPS agencies (American Humane

Association. 1988; Besham IWf; DHH S, 1991), a number of studies have confirmed

that HSP are not complying with reporting raws (Finkelhor & Zellman, 1991;

Kalichman, Craig, & Follingstad. 1990; Wells, Stein, Fluke & Downing, 1989b).

NationaI American data on the issue, provided by the I988 Study of the National

Incidence and Prevalence of Child Abuse and Nealect N S ) found 40% of the 1,196

professlonab surveyed, admitted that they failed to report child maltreatment

cases to CPS (DHHSJ988). In other words, a substantial number of children were

known by HSP to be rnaItreated, but discretionary or non-reporting to CPS occurred

(DHHS. 1988; Sedlak, 1 987; Zellman 8 Faller, 1996; Zellner, & Antler, 1990).

A range of reasons for discretionary reporting by community HSP exists: a perceived

negative effect for the professional, a belief that the referral would be deleterious to the

child, fear it will be mishandled by CPS, and by far the most frequently cited reasons, a

belief there was either a lack of sufficient evidence to make the report or an impression

the abuse was not serious enough to report (DHHS, 1988; Kalichman & Brosig.1992).

Furthermore, although the preponderance of non-reporting tends to be for mild or less

severe cases of abuse and neglect (DHHS, 1988), Zellman and Faller (1 996) found the

most serious cases of child maltreatment were also not being reported.

Professional error by CPS workers in interpreting the CPS mandate is also well

documented (How& 1992; Leck & Ballantyne, 1994; Robin, 1991). While error and

fallibility are a part of any clinical work, there is considerable variability by CPS

workers in how maltreatment is defined. Factors that inff uence interpretation of the

CPS mandate are workload pressure, inexperienced staff, minimal training, high staff

turnover, and community factors (Leck & Ballantyne. 1994), case characteristics

(Zellman.l992), occupational differences (Giovannoni 8 Becema, 1979), as well as,

gender, race, and socia1 class of participants, including individual beliefs and biases of

CPS investigators regarding child development and discipline (Robin, 1 99 1).

These individual factors take place within an organizational and community context

where formal and informal rules exist for how CPS decision making is handled by

workers (Wells, 1988). The lack of consistency and accuracy in determining CPS

eligibility by relying solely on clinical judgments is illustrated in two studies. One study

examined the screening practices from 12 CPS sites in five states and found, of the

2504 CPS contacts reviewed, '...potentially onequarter or more of the referrals should

have been investigated but were nor (Wells, Fluke, Downing & Brown, 1989a, p.26).

The second study, a review of Rhode Island CPS cases by the Children's Resource

Center (1 993) found:

... between 15% and 25% of the 'high riskm cases are not opened for agency services while many low risk families are carried on caseloads for months or even years (p. 8).

in exploring HSP decision making, most American studies have focused on how various

professional groups define child abuse (Hazard & Rupp,1986) or their attitudes to child

maltreatment (Saunders, 1988) or experimental simulations that offer respondents a

binary choice (Giovannoni & BecerraJ 979; Misener, 1986; Wolock, 1982) or compliance

with reporting laws (DHHS, 1988; Finkelhor et at.. 1991; Kalichman & Brosig, 1992).

While inter-group differences have been found in service philosophy, roles, protocols.

definitions of child abuse, and failureto-report rates (National Research Counci1,1993)

perhaps the most consistent finding from the studies to date is that the definitions of

child maltreatment are vague and fail to delineate standards fw reporting.

Lack of definitional consensus at the practice, policy and research levels, stem from the

child maltreatment laws, which vary by province or state, tend to be imprecise and allow

considerable interpretation (Trocme, l992a). Besharov (1 985, 1991) argues that many

cases of child abuse and neglect are "clear cut", where severe maltreatment or

substantial risk of harm is evident, and so a CPS referral and intervention is inarguably

clear. However, the majority of CPS referrals fall within the more interpretable 'gray" or

'minimum CPS intervention" area. Robin (1 991) highlights the issue:

It is in these types of situations, that decisionmakers rely on their personal interpretations and values, rather than on clearly articulated social guidelines.. .The problem is that while some discretion is reasonable, the extent of ambiguity in existing [child maltreatment1 definitions provides insufficient guidelines for case decision making (p. 6).

It is well established in the literature that clarity regarding the eligibility of cases for CPS

and the determination of minimum CPS intervention levels needed to require a CPS

assessment have not been well addressed in the child welfare field, policy, or research

(Goodman & MacFadden.1996). A review of all fifty state laws and screening policies in

the United States in 1987 found '...screening occurs on several levels and that

'screening' means different things to different people" (Wells, Stein, Fluke &

Downing, l989b, p.47). Similarly, Alter's (1 985) study found "...inconsistency within and

between jurisdictions concerning what constitutes child abuse and neglect '(p. 100).

Despite reported prevalence, the operational definitions and intervention standards for

neglect are neither explicit nor the norm in either CPS practice or policy (frocme, 1 WZa;

Zuravin, 1991). The lack of clear criteria, whether for severe or marginal abuse and

neglect means practice standards vary and existing studies employ an array of different

measures that yield results that conflict or are hard to interpret

Despite the challenges in conducting research on child maltreatment intenrention

standards, it is crucial to know HSP views on child welfare eligibility. Non or

discretionary reporting means the full extent of child maltreatment is neither reflected nor

known. Furthermore, unreported cases undermine the child abuse reporting system and

may reduce funding and policy support for CPS (Zellman et al., 1996). As well, non-

reporting or failure to act on an eligible CPS referral may leave the professional liable.

The greatest concern regarding non or discretionary reporting by HSP is for the children

who are known to HSP to have been or are at risk of being maltreated and who are not

referred to CPS by the HSP or not accepted by CPS. These children are denied

protection they have a right and a need to have.

There are clear substantive reasons for the selection of neglect as the type of child

maltreatment examined in the secondary analysis. One compelling reason is because

neglect is more prevalent than child abuse, yet neglect lacks widespread recognition as

a serious danger to children (DHHS, 1991). In Ontario. for example, neglect is a key

factor in child deaths due to unintentional injuries (Ontario Child Mortality Task Force,

1997). Another reason is because of the strong relationship between family income and

the incidence and severity of child neglect (National Research Council, I 993;

Sedlak.1991; Trocme, 1992a). This link allows the best opportunity to view whether

contextual variables affect HSP perceptions of the MIL Pelton (1994) notes:

There is overwhelming and remarkably consistent evidence - obtained across a variety of definitions and methodofogies and from studies perlormed at differant time periods - that poverty and low income are related strongly to child abuse and neglect and to severity of child maltreatment (p. 16).

Theoretical Framework

The theoretical framework used to guide thinking about how contextual factors influence

HSP decisions is Adaptation-level theory (AL). Harry Helson (1964) first proposed AL

theory in 1938 and argued it be applied quantitatively to explain variations in judgment

about sensory stimuli, like colon and intensity of illumination. Housed under the

umbrella of "cognitive theories", AL theory posits that both internal and external factors

must be considered to understand how thoughts and perceptions influence individual

behaviour.

Since Helson's (1964) initial conception, AL theory has been used by other disciplines

to investigate the attitude-behaviour field, as well it has been extended as a theoretical

frame of reference for the prediction of psychophysiological data. For example,

international testing of AL theory has occurred in such varied fields as stimulus

generalization (Helson & Avant, 1967; Thomas & Jones. 1 962), mathematical model-

building (Restle, 1 968), motivation and affectivity (McClelland, Atkinson. Clark &

Lowel1,1953), learning (Bevan & Adamson, 1960; Capehart & Pease, l968), intelligence

testing (Heim,l955), and social work (Wolack,1982). AL theory has most often been

tested in psychological subdisciplines, like clinical psychology (Campbell &

Brickman, 197 i ; Goldstone, 1 967; Hunt, Campbell & Lewis, 1 957), and social psychology

(Manis, l965,l96?, 1971) where behavioural choice, adaptation, decision making,

judgment theory, and attitude measurement are examined ( all cited in Appley, 1971,

P-9)-

The preponderance of studies to date on attitude and behavioural decisions have been

from the psychology field. While many approaches to studying attitudes exist, most have

ocarned in a laboratory setting (Aiwin,lQQt) not in situ. As well, study focus often has

been limited to either developing scaling methods or examining internal, individual

cognitive factors that affect judgments (Albert, Aschenbrenner, & Schmalhofer, 1989).

Environmental and group behaviour factors have often been absent from examination.

Furthermore, attitudes have been almost exclusively measured by judgments, either of

opinion statements where there is a binary choice, such as with Wdock's (1982) or

Giovannoni and Becerra's (1979) studies or where the attitude object is evaluated on a

Likert scale and assessed for predicting choices (Upmeyer & Six, 1989).

Overall, social work knowledge development to date regarding the attitude-behaviour

relationship has generally been narrow, where attitudes and behaviours are distinct

subjects of inquiry. As well, while field study and inquiry into contextual elements have

occurred (Wolock, 1982; Zellman, 1392), recognition of the contribution of such factors in

child maltreatment by direct practice, policy, and legislative arenas is for the most part,

absent. As Robin (1991) aptly notes:

The degree to which economic and social stresses contribute to child abuse and neglect is acknowledged in research studies, but tends to be discounted or ignored in clinical practice. Consequently, most efforts at responding to the problem have focused on identifying and treating individual abusers and vidims with psychological therapies, rather than remediating the economic and social conditions that allow abuse to occur (p. 4).

The three frrndamentat precepts of AL theory, as suggested by Helson (1964) are: i)

adaptation is a fundamental fact of social, as well as individual behaviour; 2) both

intemal and external factors affect adaptation; and 3) the adaptation-Ievel is a pooled

effect of all internal and external factors. AL theory helps explain how referent systems

develop at individual, group and community levels.

&Adaptationn is clearly a critical concept in AL theory. Rooted in biology, adeptation

generally refers to a species adjustments to conditions under which they must live in

order to survive. Adaptation is also a concept used in many human behaviour theories.

For sxarnple, in Piaget's (1932) theory of cognitive development, in addition to four

major developmental stages, he proposed two in born biologically invariant functions for

all species related to heredity: organkatrbn and adaptation- Adaptation is the process by

which structures of the mind develop over time to achieve a better fit with the

environmental context (Robbins. Chatterjee. & Canda, 1998). In contrast, in social

learning theory, adaptation is viewed through a behavioural lens where, regardless if the

behaviour is adaptive or maladaptive, it is learned due to contextual factors

(Payne, ? 991 ). Similarly, with Germain and Gitterman's (1 980) "life modeln of social work

practice, which is based on ecological systems theory, adaptation is not only the means

but the norm by which an individual responds to interpersonal and environmental

changes. Applying social systems theory to a community context, Noriin and Chess

(1996) argue that adaptation occurs when goals conflict, such as economic expansion

versos protection of the environment Furthermore, a review by Cicchetti & Toth (1995)

of nearly thirty years of developmental research on attachment theory, concludes that a

child forms a bond with an abusive parent because the child's attachment pattarns

proceed adaptively within a maltreating context. As well, recent theories of ethnic

assimilation, acculturation, and bicultural socialization also refer to transitional states of

adaptatron (Robbins et a[., 1998).

Another key concept in AL theory is ahomeostasis" or the 'general law of constancy of

the internal environment" which allows mammals to counteract internai and external

changes so as to achieve a needed steady physiological state. Dempsey (1951) noted,

"...the adjustments of societies to their physical environments and to each other show

the operation of homeostatic principles" (cited in Helson, 1964, p.4041). This concept

is also an integral part of systems theory where the implicit assumption is that stability is

required for the health of the system. As Robbins. Chatterjee and Canda (1998) point

out in their review of system theory:

Homeostasis is necessary to maintain a healthy adaptive balance, and incremental and gradual change is seen as the norm for well-being. Radical change is disruptive to the system and antithetical to well-being (p. 56).

Testing theory is an important toot in knowledge development. Lipsey (1993) aptly

summarizes the function of theory in social work:

KnowFedge, in contrast to facts, requires theory [italics added], that is, a framework of interconnected concepts that gives meaning and explanation to relevant events and supports new insights and problem-solving efforts (p. 10).

Wolock's (3982) test of AL theory involved 1 1 CPS offices in New Jersey and analysis of

184 CPS staff decisions about the severity of CPS cases. The study design involved

ranking each of the i f neighbourhood-based CPS offices on a scale: offices in the

highest SES area to ofices in the lowest SES area. Thirteen socioeconomic (SES)

indicators were used to rank the 11 CPS areas. A trained coder then rated 289 files from

the T I CPS sites for the revel of severity and the I T district CPS offices were then

ranked as to the severity of the child maltreatment cases. The correlation between the

SES ranking of the office and severity of cases sewed by the office were examined.

Previous studies found that rates of child maltreatment were highest and most severe in

the most socially and economically disadvantaged areas (Pelton, 1 994; Seldak, 1991). In

testing AL theory using vignettes. Wolock (1982) hypothesized:

... the point at which a case was perceived [by the CPS worker] as not meeting a minimum level of child care or a parent% problem was regarded as endangering a child's welfare would shift according to the severity of the cases that the workers in an ofke are accustomed to handling. Overall, the result would be B lower set of standards for judging the severity of child abuse and neglect situations in ofices with the more severe caseloads tnen in offices with Iess difficult caseloads [italics added] (p. I 7).

Wolock's (1982) findings supported AL theory. CPS workers in the more disadvantaged

areas who handled more severe caseloads, viewed situations as less serious than CPS

staff in more advantaged sites with less serious caseloads. AL theory advanced

understanding that it is the characteristics or weight of the entire caseload at an agency

that is a critical determinant of the judgments that are made about the severity of a case

and the level of intervention required (VVolock.1982). Although the secondary analysis

tests AL theory and uses it to assist in understanding the influence of contextuat

variables, this research differs significantly from Wolock's in design, methodology, and

locale. Details about these differences will be addressed in Chapter 7 - the Limitation

section of the thesis.

One of the many attractive features in using AL theory is that empirical investigators in

fields far removed from its genesis in sensory stimulation have applied the theory and

advanced knowledge in their fields ( Appley, 1971; Corso, 1971). Using AL theory also

expands WoloMs (1982) test of AL theory. By both broadening and retesting AL theory

it is hoped that the test of AL theory will contribute to the advancement of generalization

and theoretical development, as hypothesis testing sooner or later eliminates

unsuccessful theories (Coubn, 1979).

Another strength of AL theory is that the more recent research on attitudes and

behavioural decisions support AL theory's tenet that both internal and external sources

affect attitude responses (Albert et al., 1989; Alwin, 1993; Krebs & Schmidt, 1993).

Upmeyer and Six (j989) write, The current view is one of similarity between judgment

and behavior and that both internal and external sources affect attitude responses"

(p.15). An additional feature of AL theory is that it explains that individual, group and

community adaptation is an active process where attitudes, values, ways of structuring

experiences, judgments, learning, and interpersonal relations all represent modes of

adaptation to environmental and organismic forces (Appley, 1 971 ; Corso, 1 971). In other

words, the adaptation level is a pooled effect of the adaptation response.

Another advantage in using this theory is it can be applied substantively to judgments or

perceptions of traits, objects, events and phenomena which occur in natural settings.

such as child neglect. And finally, the key concepts of AL theory are congruent with

dominant and influential social work theories. AL theory has a high degree of

consistency with social work values and ethics that recognize human growth,

development, adaptaton and ecological influences (Robbins et a[., 1998).

Suwey instrument - The Spectrum

As noted previously, the primary survey data were collected using a CPS eligibility

instrument, called the intervention Spectrum (Spectrum). In the original survey, the

same survey tool, the same survey method, and the same question was asked of the

survey respondents - Where do you perceive the CPS minimum intenrention line (MIL)

or entry point for 24 Reasons for Service?

Since the substantive focus in the secondary analysis is HSP perceptions of neglect

entry points, some critical questions are whether the original survey data accurately and

reliably captured HSP perceptions of CPS entry points? What is known about the

validity of the survey tool? Can the tool delineate between types of maltreatment and

provide clear descriptions regarding the different forms of neglect and levels of severity?

These are important questions given that '...little progress has been made in

constructing clear, reliable, valid, and useful definitions of child abuse and neglect"

(National Research Council, 1993, p. 57). Furthermore, it is important to understand why

the primary research occurred, which CPS agencies undertook to task of collecting the

HSP survey data, and how does the survey tool delineate intervention standards. Before

answering the aforementioned questions, it is necessary to begin with first

understanding the Spectrum as an eligibility tool.

The following section details the Spectrum's development and design as an eligibility

tool. As welI, the reasons why the individual CPS agencies conducted the primary

survey and what modifications to the eligibility instrument were made to make it a survey

tool, are noted. The purpose for the secondary analysis is given and definitions and

depictions provided regarding what is a same minimum intervention line (MIL), a more

inclusive MIL, and a less indusive MIL An overview on the purpose of the Spectrum

Research Study is given, along with the relationship of the secondary analysis to the

Spectrum Research Study, and the Ethics Review Approval. Finally, the Spectrum's

psychometric properties are examined in order to determine the quality and limitations of

the data used in the secondary analysis.

Development of the Spectrum

Leck and Ballantyne (1994), senior administrators with Simcoe Children's Aid Society

(an Ontario CPS agency), developed the Spectrum in 1991 and implemented it in 1992

to address a number of issues confronting their agency. These issues are, in whole or in

part endemic to most CPS agencies. For example, within and across the agency's

branches the intake workers were inconsistent in determining what was an eligible CPS

case and what was not an eligible CPS case. In other words, the clients' need for

service, as defined by the legislation and by agency policy, did not always determine the

decision to open or not open a particular referral. Inconsistent CPS intake practices

resulted in the community being both frustrated with the agency and unclear about what

constituted an appropriate CPS referral. Furthermore, the senior management had

difficulty in clearly illustrating the types of cases it was servicing in their discussion with

the prouinciaI ministry responsible for child welfare in the province, the Ministry of

Communirj and Social Services (MCSS). Additionally, the agency was in a fiscally deficit

position and needed to know the complete range of CPS services provided to the clients

and to the communify so the CPS agency could make informed decisions about which

services were to be maintained and which would be cut. Finally, the CPS staff knew they

made inconsistent decisions and so they asked their senior management for a tool to

assist them in making better decisions at the intake level (Ballantyne, 1995). Essentially,

Leck and Ballantyne (1 994) designed the Spectrum as a CPS gatekeeping tool to assist

CPS workers in making more consistent eligibility decisions at the time of referral.

Throughout 1992, suggestions to improve the Spectrum's content and use were

collected from the Wo or three other Ontario CPS agencies that were voluntarily using

the Spectrum. These recommendations were incorporated into a revised January 1993

version of the Spectrum. In 1995, the Spectrum was further developed and refined

again, this time nine Ontario CPS agencies participated in refining the eligibility

instrument. The entire range of child protection cases was incorporated into scale format

and clarity was added to some descriptors, language, and method of use. The 1995

Spectrum or Research Version is the version used in the Spectrum Research Study to

test the eligibility tool's reliability and validity (refer to Appendix A).

Eligibility Tool Design

As an eligibility instrument, the Research Spectrum (Spectrum) has a two-dimensional

matrix design (see Appendix C). The horizontal row indudes four risklseverity

categories: most, medium, least and no. VerticalIy there are 17 columns: 14 columns

(with 24 ordinal data scales) are directly related to the mandatory protedion

requirements under the Ontario Child and Farnib Services Act. 1984 (CFSA) - Part llI,

and 3 columns (with 5 nominal data scales) refer to the prevention section of the CFSA - Part 1, as well as other duties (see Appendix A).

Each of the 24 protection scales have a range of descriptors, presented in an ordinal

scale format. The most risklseverity sluations are always described first, followed by

medium risklseverity descriptors, least risklseverity descriptors, and the last descriptor is

always a no risklseverity situation. The minimum CPS intervention line (MIL) or entry

point is between medium n'sklseverity and least n'sklseverity descriptors. The relevant

Ontario CPS legislation precedes each scale. Scales in the Spectrum are identified by

Roman numerals (e-g. 1. 11, Ill, IV); sub-scales are designated alphabetically (e-g. A, 8,

C); and descriptors are noted by numben that range from a minimum of 1 to 4 with

some scales, to a maximum of I to 13 with the Physical Ham scale (e.g. 1,2,3).

The Spectrum is not a risk assessment tool. "Risk categoriesw in the Spectrum do not

infer "risk" as the literature or research arenas define risk, where a risk assessment tool

connotes predictive or future intent (Ward & Wooiverton, 1990). Rather. 'risk'' in the

Spectrum is a practicebased term referring to the severity of the present situation.

Throughout the thesis, in an attempt to ensure the concept of 'severity" is attached to

the Spectrum's use of the word 'risk", the word is presented as riskl~everity~ To date, no

such comparable comprehensive eligibility determination tool has been found within the

North American CPS context (MacFadden. Goodman, Ballan€yne, Leck. McConville, &

Morrison,lSQG). For an example of the eligibility tool design with an ordinal format and

risklseverity categories, see Table 1 -

Table t

Exam~le of Eliaibilitv Tool Desian: Ordinal Format With RisklSeveritv Cateaories

child & F ~ I Y A child is in need of protection where, smrw kt, (e) the cfhld raquims medical treatment to cure, prevent or alleviate physical ISW

(CFW harm or suffering and the chiid's parent or the penon having charge of the child docw not provide, or tefirses or is unavaltable or unabk to consent to, the tmatmenc

MOST RISK

At least one cMld is not r d g medical trentment for an injury, Wess, disability or dental problem. If left sihretion urttreatcd, the condition is lifa threatening, or will result in pmanent impakment, or Is a senous threat to public health.

2, Women tnu ConditionlNo Diaanostic Assessment CMd has an illness or disability that interfwes with normal functioning. With treatment the condition could be comcted or at least controlled, However, without treatment the illness or dbsabilii will w o r n (though it is not life- ttueatsrring).

andior A chikt has had same phyaical qmptoms (e.g., pain or signs of contagious disease) for some time, but no diagnostic aswssmsnt has beem sought by way of a medical w dental axam.

MEDIUM R1SK 3, Risk of Com~licationdOn-aoina Pain Chifd is not receiving medical care for an mjwyT mess or dental problem that usually should rewive treatment. The chiid's mdMon will probably correct itself even without medicai treatment However, medical treatment now would reduce ridc of cornplicabons, relieve pain, speed healing, or reduce cidt of contagion.

LEAST R1SK 4. Preventfve Care Lacking Them is no drild with untreated medical conditlonsthst could benefit fmm medkat treatment, but preventive medical or dantei care (c.8. immunizations, dentaf checbps) hould be improved.

NO RISK 5. Adeauate Treatment Them is no chsd with untreated hides* illnessesT or disabiIes that could benefit from medical IreatmcntChild is taken for checkup& promptly when symptoms of ilhess appear. C W muWs normal preventive health care for his age.

Note. - I This scab is Lack of Medical Attention (HA). To review all Spectnrm scales used in the eligibility tool,

refer to Appendix A 2 For each pmta-on scale. the first dm-p to r always describes the most Wsevere situation for a

child. The other desm-ptors follow in descending risWverity order to the last descriptor which always describes no njklseuerify or a healthy family situation.

Purpose of Original Survey

Prior to the implementation of the Spectrum as an eligibility tool, a number of CPS

agencies in Ontario wanted to first understand where their various community groups

perceived the CPS entry points to be in their communities. Were entry points the same

or different for HSP groups such as CPS staff, police, or teachers? Where did non-HSP

groups, e.g. foster parents, CPS clerical staff, CPS volunteers, board members from

various community agencies, media representatives, former foster children or the

general public, perceive entry points? Were particular Reasons for Service a problem?

Which Reasons for Service had agreement on intervention points? As a relationship

building endeavor between the CPS agency and its community, the Spectrum survey

process marked the first time the CPS agency had asked the community for their views

on CPS intervention points. Additionally, understanding which community groups and

which Reasons for Service had variance, advanced the ability of the individual CPS

agency to clarify the extent of the CPS mandate.

Some CPS agencies only surveyed CPS staff (e.g. Toronto Catholic CAS). Other CPS

agencies conducted a limited survey involving only some CPS and community

professionals (e.g. Muskoka CAS and Jewish Child & Family Services). However,

seven CPS agencies (Simcoe CAS, Halton CAS, Bruce CAS, Peel CAS, Kingston CAS.

Huron CAS and Sudbury CAS) did extensive community surveys, involving CPS staff,

and numerous community professionals (see Appendix D and E) and public groups.

Unfortunately, Simcae CAS destroyed their survey data. Therefore, between 1993 - 1996, six CPS agencies fram sk different regions in Ontario, independently conducted

an extensive survey of their CPS stafF and various community groups regarding

perceptions of the minimum CPS entry points or MIL for 24 Reasons for Service. The

survey respondents in each region were asked: Where do you perceive the minimum

CPS entry point or MIL to be in your community? All six CPS agencies asked

responclants the same question, used the Spectrum as the survey tool, and the same

survey method. The original surveys were used for the purpose of each individual

agency's analysis of the survey respondents from their region.

Purpose of Secondary Analysis

The intent of the secondary study was to aggregate the original survey data from the

convenience sample of six CPS agencies, and analyze whether specific contextual

variables affbcted HSP perceptions of the MIL for nine forms of neglect Only HSP views

on entry points were examined in the secondary study.

Survey Tool Design

Used as a tool to survey HSP perceptions of CPS entry points, the 1995 Research

Version of the Spectrum was modified. The modification involved removing the

Spectrum's horizontal row, (the four risklseverity categories) in order to eliminate any

possible bias for the survey respondents (see Table 2). In collecting the origins[ survey

data, all 24 Reasons for Service were modified, but since the focus of this thesis is the

secondary anaIysis, only the nine forms of neglect are presented (refer to Appendix 6).

Table 2

Exarn~le of Sunmv Tool Desian: Ordinal Format Without RisldSeveritv Cateaories

IM LACK OFMEDICAL AmNTION 37(2.0

A child ls in need of pmtection where, (e) the child requires medical treatment to cure, prevent or slleviste physical tssr

ham or suffering and the child's parent or the person having charge of the chiCd does not provide, or rehses or is unavailable or unable to consent to, the treatment;

At least one child Es not receiving medical treatment for an injury, illness, dhrabm or dental problem. If left aituab'on untreated, the condition is Me threatening, or will readtin permanent impairment. or Q a seriourr Weat to public ham.

2. Wotsenino ConditiodNo Diaanostic Assessment Child has an iUn- or disability that iMeres with normal fundoring. treatment the condit[on could be corrected or at least controlkd. However. without treatment the Illness or disabUity wHl worsen (though it is not life- threatening).

andlor A child has had some physical symptoms (eg., pain or signs of contagious dise!ase) for some time, but no diagnostic asser#iment has been sought by way of a medical or dentiat warm.

3. Risk of Cornolication~On-aoina Pain Child is not receiving medicat care for an injury, illness or dental problem that uwally &odd r&e treatment. The childk cunditfon wit probably correct k i f even without madicaf ha!ment. However, medical trecrtment now would reduce risk of complications. relieve pain. spaed healing, or reduce risk of contagion.

4. Preventive Cam Lacking There is no child with untreated medicat wnQtions that could beneiit from medcal treatment, but preventive medical or dental care (e.g. imrnunkations, dental checkups) should be improved.

5. Adeauate Treatment There is no child with untreated injuries, dlnc9ws. or dhwbl[itfes that could benefit from medical treatmentchild is taken for cf'1ecfcup6 promptly when symptoms of iUntss appear, Child receives normal preventive health care for his age.

For each of the 24 protection scales, the task of the survey respondent in the primary

research was to first read the Ontario child welfare legislation (CFSA) relevant to that

scale, then read the descriptors in the scale, and then, still maintaining the ordinal

format, place each descriptor number under one of the four risklseverity headings (most/

medium/leasffno) on the Survey Score Sheet (refer to Appendix F "b" side).

Essentially, each HSP survey respondent identified two types of cases. The first type

were the mandated CPS situations; these included all descriptors the respondent

perceived as falling under the most and medium risklseverity categories. The second

type of cases were the non-CPS cases or community cases, and included all descriptors

that fell into the least and no risklseverity categories.

Table 3 illustrates how survey respondent 1120, #213 and #363 scored the Spectrum

survey's first two Reasons for Service: I A (Abandonment) and I1 A (Lack of Medical

Attention) +

Table 3

Exam~le of Com~feted Suwev Score Sheets for IA and flA

Respondent #I20 Spectrum Score Sheet

CASMANDATEDCASES NOT CAS CASES

MOST SECTION SEVERE

MEDIUM S M R I T Y

LEAST NO SEVERE SEVERITY

NOTE. Scales are identified by Roman numerals (e.g. I, 11, Ill, IV); sub-scales are designated alphabetically (e.g. A, B, C); and descriptors are noted by numbers (1,2,3..).

Respondent a 1 3 Spectrum Score Sheet

CASMANDATEDCASES NOT CAS CASES

MOST MEDIUM LEAST NO SECTION SEVERE SEVERITY SEVERE SEVERITY

Respondent #363 Spectrum Score Sheet

CASMANDATEDCASES NOT GAS CASES

MOST MEDIUM SECTION SEVERE SEVERITY

LEAST NO SEVERE SEVERITY

The Minimum Intervention Line: Same, More Inclusive, Less Inclusive

For the purposes of the secondary analysis, it is this demarcation line between the two

types of cases (CAS Mandated Cases and Not CAS Cases) that is called the CPS entry

point or minimum intervention line (MIL). The same MIL means the same number of

descriptors had been selected by the HSP respondents for the mandated CPS situations

and for the non-CPS cases. Referring again to Table 3, IA, respondent #20 and

respondent #363 both had the same MIL for Scale IA - they perceived the CPS entry

point to be between descriptor 3 and 4.

A significantly different neglect CPS entry point can elher mean the MIL was more

inclusive or less inclusive.

A more inclusive MIL is when more descriptors of a less severe nature are included. A

more inclusive MIL may broaden the purview of CPS. Referring again to Table 3, IA,

compared to respondents #20 and #363, respondent #213 indicated a more inclusive

MIL by indicating the CPS entry point to be between descriptor 4 and 5.

A less inclusive MIL means fewer higher severity situations are selected. In pradice, this

may namw the interpretation of the CPS mandate. Referring to Table 3, Scale IIA,

respondent #363 perceived the MIL to be between descriptor 2 and 3, which is a less

inclusive CPS entry point when cornparad to respondent #20 (MIL between 3 and 4) and

respondent #Zt 3. (MIL between 4 and 5). See TaMe 4 for a scale depiction of a same

MIL, a more inclusive MIL and a less inclusive MIL.

Table 4

Ejcarn~(e of a Same MIL. a More Inclusive MIL & a Less Inclusive MIL.

IIA LACK OF MEDEAL ATTENTlON 37(2NE)

A child is in need of protection where, (e) the chifd requires medical treatment to cure, prevent or alleviate physical harm

or suffering and the child's parent or the person having charge of the child do- not provide, or refuses or is unavailable or unable to consent to,

the treatment;

At teast one child is not receiving medical treatment for an injury, illness, dkabifity or dental problem. If left untreated. the condition is We threatening, or will reaudt in permanent impairment, or is a serious threat to public health,

Child has an Rlnass or aisab~v that interferes with normal functiontng. Wfth treatment the condition could be correctad or at least Conbolled. However, without treetment the illness or dbabiltty wilt worsen (though it is not life- threatening).

andlor A child has had some physical symptoms (e.g., pain or s i p of contagious disease) for some time, but no diagnostic awrsc#unent has been sought by way of a medical or dental exam.

3- Risk of Comalications/Onlcroina Pain Child is not receiving rnedlcat cam for an injury, illness or dental problem that usually should receive treatment The c W s condition wNI probably conect itself even without medical treatment. However, medlcaf treatment now would reduce risk of wmplicatfons, relieve pain, speed healing, or reduce risk of contegion.

IIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIII~~ Minimum In terven tian Lhe IIIIIIIIIIIIIIIIIIIII~III~IIIIIIIIII

4. Preventive Care Lacking There is no ctrlld with untreated medical conditions that could benefit from medical freatment, but preventive medid or dental care (e.g. immunbrattons, dentat checkups) should be impmved. - Mom tnciusiite

S. Acieauate Treatment there b no child with untreated injuries. il[netw!s, or cbab~'libles that could benefit from medlcal treatmentchird is taken for checkups promptly when symptoms of iiIn458 appear. Chiid receives normaf preventive haam care for his age-

Spectrum Research Study

In 1994, the Ministry of Community and Social Services (MCSS) provided a research

grant to the Ontario Association of Children's Aid Societies (OACAS) to test the

reliability and validity of the 1995 Research Spectrum. Led by Principal Investigator, Dr.

Robert Macfadden. from the Faculty of Social Work, University of Toronto, the

Spectrum Research Team included myself as Research Assistant and Project Manager,

and representatives from Simwe CAS, Halton CAS, and OACAS.

Five Ontario CASs were part of the four year Spectrum Research Studysample

selection involved a nonprobability procedure called convenience sampling, That said.

the sample did meet a number of predetermined criteria: variance in agency size (small,

medium and large), variance in geographic location (east, west, norVI and central

regions), variance in setting (rural, urban, inner*), and willingness to use the

Spectrum. Over 1000 intra and interreliability ratings were analyzed from the five CASs.

Relations hip of Secondary Analysis to Spectrum Research Study

Prior to the testing of the Spectrum's reliability and validity, three of the five agencies in

the Spectrum Research Study wanted to survey their own communities' perceptions of

CPS entry points. The desire to undertake such a major survey stemmed fram the

positive feedback from other CPS agencies that had previously conducted extensive

community surveys using the Spectrum. 60th the survey process and the outcome

analysis were viewed by the CPS agencies and their communities as beneficial.

The combination of my interest in, and willingness to, aggregate the survey data and

analyze the HSP respondents perceptions of neglect entry points, the potential of the

secondary study to advance knowledge on the eligibility tool, and the promise by the

three CPS agencies in Spectrum Research Study to provide their sunray data, resulted

in the secondary study becoming part of the larger Spectrum Research Study.

The three CPS agencies and their regions from the Spectrum research are referred to in

the secondaly analysis as Region A, Region B and Region C. Initially, only these three

CPS agencies and the HSP from their respective regions, were to be included in the

secondary analysis. However, a decision was made to broaden the secondary analysis

to include a total of six CPS regions. Three other CPS agencies that had previously

conducted the same community survey using the Spectrum, agreed to provide this

researcher with their original survey data so it could be included in the secondary study.

These other three regions are identified in the secondary analysis as Regions D, Region

E and Region F. In sum, the intent of the secondary analysis was to aggregate the

original survey data from Regions A, B, C, D, E, and F and analyze only the secondary

data on HSP perceptions of neglect entry points.

With respect to the secondary study, it was the responsibility of the Spectrum Research

Study to provide Region A, 6 and C with survey materials, such as the Spectrum Survey

tool. Spedrum Score Sheets, consent forms and training overheads. The prototypes for

the score sheets, training overheads and invitation form letters had been developed by

Regions D. E and F. It was my responsibility as Research Assistant for the Spectrum

Research Study and sole researcher for the secondary study, to assist the CPS

agencies in Regions A. 8 and C in the community survey by training CPS staff on how to

conduct the community surveys. Also, I developed an expanded list of personal and

professional questions to ask survey respondents in Region A, 6 and C.(see Appendix F

'aw side). As wetl, I assumed all tasks associated with aggregating and analyzing the

secondary data. Each of the six CPS agencies were responsible for their own sample

selection, covering all costs associated with data cdlection, inviting respondents,

atranging the survey sites, conducting the survey, collecting the survey data, and upon

completion of any analysis for their own purposes, forwarding to me the original survey

forms.

Ethics Review

In February 1996, as part of the Spectrum Research Study, the University of Toronto,

Ethics Review Committee approved this secondary analysis study on human service

professionals (HSP). For specifics, refer to Appendix M. The three regions from the

Spectrum Research Study are identified in the secondary analysis as Regions A, B and

C. All HSP respondents from these regions completed consent forms. In addition to

completing the Spectrum survey (see Appendix F 'b" side), these HSP respondents also

answered a variety of individual questions of a professional and personal nature (see

Appendix F 'a" side). For examples of CPS and community HSP consent forms refer to

Appendix H and I respectively.

The survey data from the survey respondents in the other three regions, Regions D and

E and F, were obtained from CPS agencies that had conducted an extensive community

sutvey prior to the Spectrum Research Study. While the respondents completed the

same survey question with the same survey tool, no written consents were obtained.

However, the CPS agencies ftorn regions 0, E, and F. advised this author that when

completing the survey their respective respondents had consented to their surveys

being used for purposes of analyses. Also, the only respondent information the

secondary analysis uses with the survey participants from regions D, E, and F, are the

survey score, the regional location and job designation. No personal characteristics are

either known or used.

Reliability and Validity

As noted above, the Spectrum is a comprehensive, field developed CPS eligibility tool

intended to improve clarity, accuracy, and consistency of screening decisions. The

Spectrum is not theoretically based but rather it is grounded in the Ontario child welfare

legislation, the CFSA. As a screening tool, the 1995 Research Spectrum attempts to

identify the entire range or 'spectrumw of alI possible CPS Reasons for Service and

capture the complexity of protection situations in 24 CPS Reasons for Service. As

identified earlier, the 24 protection Reasons for Service are divided into four risklseverity

levels. The result is an eligibility instrument that is comprehensive in its breadth and

depth but in practical terns provides HSP with 169 distinct choices (140 under child

protection and 29 under prevention and other services). While the huge number of

options adversely affects reliability, it also contributes to the Spectrum being described

as uoverwhelming" and 'unwieldly".

Preliminary examination of individual CPS workers' rankings from the Spectrum

Research Study finds particular sections heavily favored. One rationale is some workers'

may have psychologically reduced the breadth of the Spectrum to more manageable

terms. Furthermore, descriptors in some sections overlap, such as with Secfion Xlll - Parent Child Conflict and Sectlbn XIV - Child Behaviow Problem. This creates confusion

in use and diminished reliability. As well, Section IX-B and Section XI1 -B are missing

least risWseverity descriptors. Finally, while as a field-based tool it offen relevancy and

practicality for the CPS field, it is not psychometrically designed. While the sheer

number of scales affects reliability, some sections have a multi-dimensional aspect,

which further compromises the tool's value.

Development of the individual scales in the Spectrum varied. Some, such as all nine

neglect scaies, come directly from Magura and Moses (1986) Child Well Being Scales

(CWBS). Other scales, such as Section IN - Physiwi Hem, and Section V - Emotional

Abuse were adapted and modified from the CWBS. Whereas, Section IX(a) - Caretakerr

Hidory of Abuse and Section /X(b)- Camtaker Ability to Protect scales were created

specifically for the Spectrum. Given the variation in scale development end testing, the

Spectrorn Research Study was designed to test the overall reliability and validity of the

t00L

To examine the validity of the tool, the Spectrum Research Study examined whether the

Spectrum measured accurately the concepts in child protection it intended to measure.

Mutchnick and Berg (1996) point out that, "...considerations of validity should precede

reliabil*w because validity examines the instrument of measurement while reliability

addresses the consistency of responses* (p.80). Findings on the Spectrum's validity as

a CPS eligibility tool varied depending on the type of validity assessed.

Face validity (do the scales look like child protection concerns are being measured?)

and content validity (does the Spectrum cover the entire range of child protection

possibilities?) are established on the basis of judgments. Face and content validity were

assessed by an expert panel consisting of two senior Ontario Family Court judges, two

lawyers specializing in family law, and two senior CPS workers from two Ontario CPS

agencies that were neither involved in the research nor using the Spectrum.

Assessment of the Spectrum's legal validity included having the panel indicate where

they individually perceived the minimum CPS intervention levels to be and then

comparing their responses to the Spectrum's intervention lines. The panel also

completed a semi-structured questionnaire and written assessment about whether the

Spectrum's faca and content accurately reflected the CFSA. And finally, the panel

participated in a focus group meeting. The expert panel concluded the Spectrum has

good face and content validity.

Assessment of predictive validity, which is one type of criterion-related validity (does a

Spectrum rating have the ability to predict other criterion that wilt occur in the Mure?) is

not complete. Although initial analysis by MacFadden and his research team (1997)

found:

... there is some evidence to suggest that the Spectrum has criterion validity. When the data are dichotomized into two groups: high severity versus medium, low or no severity, there appears to be some significant difference across these groups. As an example, in the parental capacity area, the higher severity group uses more resources (meant4.2) than the medium, low or no severity group (mean =2.1), e2.21. pc.05 (p.7).

Construct validity (does the Spectrum's individual scales fit theoretical expectations?) is

one of the most diflicult types of validity to assess. The Spectrum's construct validity

was not tested because it is not a theoretically grounded tool. Rather it is a screening

instrument based on child welfare legislation. Testing for concurrent validity (does the

Spectrum's ratings correspond to other similar instruments?) was not possible because

no other similar CPS eligibility instrument has been found in North America.

Reliability of an instrument considers the extent of agreement between raters. In the

Spectrum research, 1062 cases were assessed for intrarater (test-retest) reliability and

interrater (interobserver) reliability. Percentage agreement was used to calculate the two

types of reliability. Generally, for psychometric instruments, if the interrater and the

intrarater responses are in agreement 80% of the time, then it is assumed that the

amount of random error in measurement is not excessive and the instrument is viewed

as having acceptable stability. Yet, as Rubin and Babbie (1997) point out, 'Some would

argue that even 70% agreement would be acceptable* (p.176). While acceptable

reliability was approached but not met, the final reliability figures were aggregated for the

entire tool. Reliability on the specific neglect categories was not specifically tested and

calcu[ated, and is therefore, unknown. However, previous testing by Magura and Moses

(1986) on the CWBS used in the nine Spectrum neglect sections, found the neglect

scales met acceptable standards of reliability and validity.

The Spedrum demonstrates considerable promise as a CPS eligibility framework that is

moving toward standardization. Some types of validity have been established. Reliability

needs to be tested in the next generation of the tool. Testing the reliability and validity of

this CPS eligibility taxonomy and its psychometric properties alone advances child

maltreatment research, for it addresses a key recommendation by the National

Research Council (1 993) that:

... instruments need to be developed, together with associated documentation such as training manuals; interrater reliability studies need to be conducted to document consistency together with studies of construct validity and (in the longer term) predictive validity (p. 63).

Many of the Spectrum's limitations are evident only when it is used in actual practice

settings as an eligibility tool. Used as a survey instrument, confusion and overlap

regarding section and descriptor selection is not an issue as all sections are completed

for the survey.

By design chance, the neglect sections have less of the multi-dimensional nature than

other sections. Unlike other instruments used to assess neglect and abuse, such as the

NationaI Incidence Study maltreatment classification system (DHHS. 1988) or the

Manitoba Risk Assessment System (Reid, Sigurdson, Christianson-Wood, &

Wright,qQ95) the advantage of the Spectrum is that it not onIy defines and describes

the various types of neglect, but it also provides levels of severity and minimum

intervention levels. Additionally, using a too( that is being standardized allows for

replication of the primary study and secondary analysis to occur. The factors that

enhance the quality of the data used in the secondary analysis are the Spectrum's good

face and content validity, the tool's purpose (identifying CPS eligibility and intervention

points), the un l of analysis (individual perceptions of CPS entry points), and the tool

was developed specifically for the Ontario CPS field. The need for Canadian specific

instruments that define and measure child maltreatment is noted by Trocme (1 992a):

With the exception of the Manitoba scales, they are American instruments unlikely to refiect differences between American and Canadian laws, social sewices. and social norms. (p. 72-73).

There are however limitations with using the modified Spectrum as the survey tool. In

addition to the unknown reliability of the specific scales, one drawback with using the

Spectrum as a survey tool is it requires respondents to have very good language

comprehension coupled with an ability to maintain a high level of concentration

throughout the survey. Another limitation with the instrument is the time commitment to

complete the entire survey on 24 Reasons for Service is considerable - nearly three

hours for each respondent, And last, while not an issue in this study. some scales such

as physical abuse and sexual abuse were not treated as ordinal data.

Definitions

The following €wo key definitions are highlighted in this seaon and will be used for the

purpose of the secondary analysis. However. in addition to these two definitions. all

other terms and definitions used in the secondary study, are listed in Appendix J.

Min/,imum lntetvention &hie (MIL): The Child Protection Service (CPS) MIL is where

each survey respondent put the demarcation point or line between the two possible type

of cases: the most and medium risklseverity, mandated CPS situations (CAS mandated

cases) and the least and no risklseverity, ron-CPS community cases (not CAS cases).

CAS mandated cases ere above the MIL. Not CAS cases are below the MIL. In practice,

any cases above the MIL means the CPS agency will intervene in a family by conducting

an intake assessment of that family because the child's situation falls within the most or

medium risWseverity categories in at least one area. The same MIL means the same

number of descriptors had been selected by the survey respondents for the mandated

CAS cases and for the not CAS cases. A more inclusive MIL is when more descriptors

of a less severe nature are included. A less inclusive MIL means fewer higher severity

situations are selected (MIL is also referred to as the ChiM Protection Entry Point).

Neglect Is the presence of certain deficiencies or omissions in caretaker obligations

(usually the parent) that ham the child's psychological andfor physical health or place

the child at risk of ham. If the child were receiving appropriate care, the risk of harm or

actual ham could have been avoided (Ontario Child Mortality Task Force, 1997). Child

neglect coven a range of caretaker behavioun and for this secondary study include:

Abandonment Lack of medical attention Lack of supervision Lack of appropriate child care arrangements Nutritionat neglect l nadequate physical living conditions Inadequate Personal hygiene

* Lack of Household sanitation l nadequate CIothing .

Retevance of Study to Advance Knowledge

The secondary analysis has the potential to be of direct relevance to practitioners, CPS

agencies and their communities, professional groups, policy makers, child welfare

legislators and researchers. From a practke standpoint, analyzing the professional

community's views on their interpretation of where the minimum intervention line is for

difierent f m s of neglect is critical for a number of reasons. As noted previously, HSP

are instrumental in CPS practice and policy. The views of the various professional

constituencies on neglect eligibility are important to know since referrals in child

maltreatment are predominantly provided by HSP groups such as police, doctors, social

workers (external to CAS), community HSP, and teachers. It is crucial to know whether

community HSP have common and accurate perceptions about eligibility for CPS

neglect. Otherwise, referrals may not be made or may be inappropriate, costly in terms

of resources, and may result in professionals antagonism over misconceptions and

unrealistic expectations (Macf adden et al.. 1 996). Clearly it benefits the CPS agency,

the community it serves, the professional groups and policy bodies, and most

importantly, the children that require protection, if interpretation, benchmarks, definitions,

and descriptions of child welfare minimum eligibility have a high level of similarity

(MacFadden et al., 1996; National Research Council, 1993).

The literature identifies that consistent and accurate determination of CPS eligibility has

been a significant problem for the child welfare field (Howitt, 1992; Leck et alJ994;

Zuravin, 1991). The lack of consensus in determining eligibility has been attributed to a

variety of factors; vague definitions, workload pressures, inexperienced staff, minimal

training, high turnover, and community factors (Giovannoni et al., 1979). It is knowledge

advancement about the community or contextual factors that affect HSP decisions that

is of interest in the secondary analysis. Identifying specific contextual factors that do or

do not influence HSP perceptions of neglect provides researchers with suggested paths

for Mure inquiry. At the practice and policy levels, knowledge about contextual factors

that affect decision making combats the ontogenic approach to understanding child

maltreatment

The secondary analysis benefits from the original study's expansion of both the numbers

and type of human service groups previously studied (See Appendix D for list of CPS

staff groups and Appendix E for list of human service professional groups). Data

collected on some respondents includes: years of experience in their field, years of

experience in child welfare, gender, age, marital and parental status, education attained,

and income level (See Appendix F -'a* side for list of Personal & Professional

Characteristics). Aggregating the survey data from the six regions in the secondary

analysis allows for comparison between specific groups and communities to occur. As

the prime referral sources for CPS it is critical to understand at what point and where

the variances and similarities exist for the human service groups regarding their

determination of CPS neglect eligibility. Furthermore, the secondary analysis has the

potential to identify community standards. Analyses within, across, and between various

human service constituencies and their communities can occur. The secondary findings

can identify which human service groups and which regions have the same MIL* which

have less inclusive M k and which have more inclusive MIL, in neglect determination.

The secondary analysis also has the potential to advance knowledge about specific

forms of neglect since nine different forms of neglect are examined. As well, by ranking

communities by their socioeconomic determinants, it will give a context to the

community-level factors that are important in understanding child maltreatment

(Giovannoni. et al.. 1979; National Research Council, 1993; Wolock, 1982). Knowledge

of which factors influence HSP perceptions of neglect entry points where the result is a

less inclusive CPS entry point is important for practical reasons. Such knowledge could

aid the CPS agency, the community and the ministry responsible for child welfare in

anticipating which areas are in need of immediate assistance because contextual

determinants or the region's characteristics place it at future risk of a less inclusive MIL.

In using secondary data from a survey that used an eligibility tool that covers the

mandated range of CPS neglect Reasons for Service, the secondary analysis

addresses the validity problem that, '...the child abuse research community has not

properly explored the realities of child abuse decision making" (Tomison 8 Goddard.

1993, p.1). Also, the use of an eligibility tool that is being standardized allows for

replication of the original survey and the secondary analysis. Replication of a study in

combination with the theory testing is a vital part of the knowledge-building process, in

that replication contributes to generalization and theoretical development

(Coulion, 1979). It is hoped that the secondary analysis will advance understanding

about some of the contextual factors that may influence professionals' perceptions about

child neglect eligibility as well as increase our knowledge about the specific forms of

neglect

Analytic Stance

In identifying the substantive reasons why the secondary analysis needs to occur it is

also important to identify this writer's personal position. This is so the reader may better

understand both the context in which the research was undertaken, as well as the

biases and assumptions that accompany the writer.

As a social worker with over seventeen years experience in the child welfbre field, my

work has included direct front-line intake and family service, assessment and treatment

services, group work, and supervision and management duties. Also included are

research and teaching responsibilities. An ongoing problem I encountered in direct

practice work, and one common for many child protection agencies, is the CPS agency's

often strained and sometimes difficult relationship with its community. More specifically,

there were issues with community professionals who referred cases. My front-line

experience mirrored previous research findings - determination of CPS eligibility was

neither done accurately nor consistently by community professionals or CPS workers.

A mobile CPS intervention line created inconsistency in the refeting of and accepting of

maltreatment referrals. It was apparent this inconsistency resulted in a mutual lack of

trust and confidence about the abilities' of both the CPS worker and the community

HSP. As Sullivan (1994) notes:

This scenario has contributed signficantiy to eroding public confidence in the competence of those charged with responsibility of protecting children when alt that remains of a child-welfare system is the capacity to respond to the worst case scenarios (p. t89).

It is not uncommon for the CPS agency to be described by the community as either

'child snatchers" or 'unresponsive*. Similarly, the CPS agency may spend considerable

time dealing with what they view as 'inappropriateR referrals by community HSP; this can

mean 'reading the riot act" to a professional for not having referred a child protection

case earlier or explaining that the case they are trying to refer does not fall under the

CPS mandate. While over-involvement is a concern, meaning CPS is involved when no

abuse exists (Robin, 1 99 1). it is the under-involvement that is most troublesome: leaving

children to suffer abuse andlor neglect. For example, despite having been referred to

CPS by a HSP from a women's hostel, on June 23,1997 in Toronto, five week old

Jordan Desmond stawed to death in his mother's care,

While professional error in child abuse is not a new phenomenon (Howitt,i992), the

tension between the CPS agency and the community has perhaps been heightened in

the past decade due to retrenchment in the public sector, erosion of funding, and

reduced service mandates (S Jivan, 1994; Whar1,I 995). This is coupled with escalating

maltreatment reports (Lovell & Thompson,i995; Trocme,t991) and continued lack of

clarity regarding definitions and intervention for child maltreatment.

Common sense suggests comprehensive training for community professionals on the

CPS mandate is needed to improve consistency and understanding about what

constitutes an appropriate CPS referral. My direct practice experience did not support

that tact in 1989, as part of my management duties at a CAS in southern Ontario

(seMng a population of nearly 600,000), 1 organized CPS training to nearly 1600 day

care providers. Child care providers (e-g. child care centers, nursery schools, licensed

home day cares, after-school programs) were targeted as they had daily access to

neatly 8,000 children. Despite contact with substantial numbers of children, child

maltreatment ref6rrals were almost negligible from child care providers.

To facilitate better relationships between the reporters and the investigators, it was

agreed that CPS staff and key child care providers would lead the training. Nearly 800

day care providers completed the three hour CPS training on identification of

maltreatment, professional duty to report, and the CPS agency reporting process. While

the training was rated highly by the participants, and improvement of professionaI

relationships occurred in the short-term, at the one year and two year follow-up, child

maltreatment referrals by this HSP group had not increased at all.

This experience left me with many questions: Despite training do different professions in

the same community perceive CPS entry points the same or differently? Do the same

profiessions in different communities view maltreatment the same or differently? Do

different communities have different minimum CPS intenrention lines? If so. is this true

for all CPS Reasons for Service or just some? And which ones? What contextual

fsctors, either regional or agency affect maltreatment decisions? If CPS training alone is

not the answer to improving reporting rates, what is?

My desire to answer these questions, coupled with the inclusion of the secondary

analysis study as part of the Spectrum Research Study, the willingness of six CPS

agencies to provide me with their original survey data, and the confidence that the

survey tool could adequately capture professionals' perceptions of entry points, created

the interest, opportunity and momentum for the secondary analysis.

Drawing on social constructionist theory (Best, 1989; Parker, 1998), the epistemological

position taken is postmodern. k posits we cannot separate the observer from the

observed. A major thesis advanced in this analysis is that professions, with specific

reference to social work, are socially constfucted activities, and since social work

grows out of social, political, economic, and knowledge development contexts, it can

only be understood within the social and cultural context of the participants

(Harfford, 1994; Howitt. 1993; Payne, 1 991). Furthermore, as Robin (1 991) observes:

Child abuse is not a fixed, objective entity, but is essentially a socially constructed mediated conclusion about parent-child relationships that is based on community standards of appropriate and reasonable child care (P-2)-

This postmodem position is translated into a critical stance. A critical stance suggests

that the theory and knowledge that we adopt is done so knowing that they are shaped

and influenced by context. Hartman (1994) argues:

We make use of knowledge and theory but always examine the value assumptions and fhe soc~opolifical and culturai pmcess8s that gave birth to and sustained them ptalics added] (p. 16).

CHAPTER 2: LITERATURE REVIEW

For sock1 workers to adequately understand the determinants of their contwnporary thinking about such issues [social welfare activities], they

must have some appreciation of the histon'cPI and philoMpMEa1 origins. Fmderic G. Reamer, t 993, p. 2

This chapter reviews the studies on HSP and their decision making regarding CPS entry

points for child maltreatment. However, in order to examine the state of this knowledge

base it is first essential to examine how specific factors have influenced the broader

tapestry of social work epistemology and the knowledge development of the profession's

disciplines, such as child maltreatment. Hence, the literature review moves from the

broad to the specific. It is broken down into four parts: social work epistemology, social

construction of child welfare, examination of CPS eligibility tools, and review of the child

maltreatment studies regarding HSP and CPS decision making.

The intent is to be brief, so to ensure the essential elements are examined, but broad

enough so that the contextual forces that shaped and influenced how knowledge has or

has not developed in this particular area of child welfare can be understood. To achieve

both brevtty and breadth requires selectivity. This rn-ter acknowledges biases

accompany both the selection of topics, as well as the interpretation of what they mean

and how important they are.

Social Work Epistemology

From this author's review of the scant social work literature on the philosophy of science

(Orcuff, 1990; Reamer,l993) and the epistemology of social work knowledge (Hudson,

1 985; Payne. 1 991 ; Peile, 1988; Reamer, l994a; Reid, i 994; Wells, 1994) a number of

factors have strongly influenced our current social work knowledge model. Illustrations

of such factors are: conservative political ideology, free market economy, the dominance

of the empirical paradigm, social forces and philosophical beliefs (Goodman, 1991). I

pose that two forces in particular have significantly influenced child welfare knowledge

development at all levels: theoretically, conceptually, empirically, and dinically. One

factor is noted by its dominance and the other is significant because of its almost total

absence. The forces are: the dominance of borrowed knowledge and the paucity of

theoretical assessment.

The Dominance of Borrowed Knowledge

It has often been stated that the fulcrum or common denominator that ties the various

elements of social work together are our core social work values (Levy, 1973). Yet

common values alone do not constitute a profession. A review of the development of

social work into the status of a profession shows social work in the early 1900s was

primarily a charity organization movement that worked with disadvantaged and

disenfranchised people. The profession was rich in practice wisdom and poor in

theoretical and empirical knowledge. Acquiring a professional knowledge base was vital

to social work on two accounts.

First, there was the need for soda1 work to satisfy the principal criterion of a profession:

the existence of an identifiable and specialized body of knowledge that is transmitted

systematically through formal education (Reamer, 1994a). Therefore, in order to address

the theoretical and empirical gaps, while at the same time advance, legitimize, and

justify its status as a 'profession", sociai work bornwed and adapted knowledge from

other soda! science professions, such as psychology, psychiatry, sociology, political

science, and economics.

Second, social work's evolvement into a profession coincided with the emergence of

the 'knowledge sociev. Developing, gathering, and acquiring 'knowledge' was central

to not only all professions, but to society itself. Knowledge confarred power and on this

point Gaventa (1985), in his article, 'The Powerfid, the Powerless, and the Experts:

Knowfedge Sttuggles in an infomation Age" writes:

The power of the knowledge industry is derived not simply from what knowledge is produced and for whom, but also from the growth of new elites who people the knowledge production process ... Who are these knowledge producers? Who benefits from their knowledge? Knowledge production, then is accountable not to the public interest, not to the needs of the powerless who may be affected by it, but to an ideology which serves to justify the superiority of the expert - the ideology of science and objectivity [italics added] (p. 25 -29)

The combination of the onset of the knowkdge era with the outcome of the debate that

existed in the philosophy of science field regarding the logic and parameters of science

as the means of discovery (Orcutt, IQQO), was that knowledge identified with science

was elevated in status, value, and power and thus became dominant knowledge. By

implication, practice wisdom ana common knowledge, the basis of social work

knowledge at that time, was subjugated knowledge. Social work did not have the

luxury of time to cultivate its own knowledge base- The profession had to borrow

legitimate dominant knowledge from other disciplines.

The issue is not the borrowing of knowledge. For knowledge sharing among disciplines,

specifically the social science professions, is accepted and to some extent expected

practice, given the complexity of human study. Rather, it is the preponderance of

borrowed knowledge at the time of social work's genesis into a profession that

profoundly affected both the direction and fod of study, and ultimately, what is defined

and pursued as knowledge in child welfare. Furthermore, not only is knowledge

borrowed from other disciplines, but the preponderance of child welfare knowledge in

Canada is borrowed fmm knowledge developed about the United States experience.

One of the outcomes of the plethora of borrowed knowledge in social work has been a

potpourri of theories, conceptual frameworks, practice methods, treatment techniques,

worker descriptions, and research methodologies (Reamer, 1993). While this approach

is not unique to social work, one concern is the lack of continuous exchange of

theoretical ideas, developments, and methods between disciplines. This point is

underscored when the historical path of adaptation-level (AL) theory is traced.

Subsumed under cognitive consistency theories, AL theory now falls under the

psychology of judgment and measurement of attitude literature. Measuring attittides is a

ftequent activity of social scientists yet there is a significant gap between disciplines

sharing theoretical and technical innovations (Krebs & Schmidt, 1993).

Another outcome has been the emphasis on understanding individual processes in

social work knowiedge, practke and research. Glissonys (1995) analyses of the research

articles published between 1977-1988 in five social work journals: Journal of Social

Work Education. Journal of Social Service Research. Social Service Review, Social

Work, and Social Work Research and Abstracts found:

...(it )he person is used as €he unit of analysis in more than 60% of the reseeah articles published in the 12-year period. The organization is a distant second, used in 14% of the articles, followed by the document (9%) and the family (7%). This is important because it again documents a focus on the individual, showing that there are few t ~ e studies of culleciives (famiIies. grwps, organizations, states, e€c.)[Ralics added] (p. 21 1 ) .

This paradigm for thinking about social problems, like child abuse and neglect, has

meant a primacy on developing knowledge that provides conceptual explanations and

predictions about the individual. Accordingly, child maltreatment knowledge is focused

on understanding child abuse and neglect as solely an individual problem. This

ontogenic approach to knowledge development affects theoretical development, din ical

practice, policy approaches, legislation, and research. As Robin (1 99 1 ) points out

Although there is considerable evidence of a strong relationship between poverty and child abuse and neglect, many practitioners argue that abuse cuts across class lines. While not without an element of truth, such an approach leaves the impression that abuse and neglect is caused primarily by individual psychological factors (p. 4).

Individual and family factors as the dominant model for understanding the occurrence of

child abuse and neglect compromises knowledge development and acquisition, as well

as practice interventions. Ignoring or not even considering community and

environmental factors, social and cultural factors, and protective factors as equally

important elements in the etiology of child maltreatment has limited practice, policy and

research dicectrDons (National Research Council, 1993). Two dominant forces have

significantly influenced our Western cultural practices that emphasizes the correction of

problematic or damaging parental practices as the primary factor for child maltreatment

The first force was the broad, social, political, legal, economic, and philosophical beliefs

of the early nineteenth century. Specifically, the elevated status of the rights and

responsibilities of the individual. This philosophy was demonstrated in a number of key

areas: the English Poor Law ReTom Bill of 1834, the dominance of the freemarket

economy, and the prevailing "laissez-faire" approach of liberalism regarding government

intervention (Hollinger, t 994). Poverty, for example, was viewed as 'a condition that

resulted from the moral infenom of individuals [italics added]* (Reamer, 1993, p. 12).

This is an important point, for it lays down in the early stages of social work knowledge

development in the child abuse field the "explanation" as to why and who is poor.

Theories and practice principles duly followed. It has taken nearly a century of research

for the field to conclude that while there is a strong relationship between poverty and the

incidence and severity of child abuse and neglect, it is a fact not an explanation (Pelton,

1994).

Hence, as part of these sociopolitical and cuItural processes of the times, social work,

along with the allied social science disciplines, and psychiatry were strongly influenced

by prevailing, dominant Western ideology. As Epstein (1994) argues, this led social work

knowledge development to '...emphasize personalities, feelings, and lifestyles as

primary targets of intervention' (p.12). The common view was the problems of the

individual were the responsibility of the individual (Goodman. 1991 ). For social work, the

imp1 ications meant:

Western social work is based on the importance of the individual, with an associated concept of individual rights, and services have developed from a residual policy where they are provided only to those who cannot be helped otherwise (Payne, 1991, p.5).

The need to first bomw and then develop knowledge to explain individual processes

and provide recommended interventions is most evident in clinical social work. The

origins of foundational theories tend to be from other disciplines, such as psychology,

psychiatry, sociology, and medicine. The beliefs and values embedded in the other

professions' theories are reflected in the various theories used in social work: family

systems, egopsychofogy, psychoanalysis, object-relations, cognitive theory, behavioural

theory, social structure, role theory, and communication theory.

The second force that directed the focus of knowledge building towards the individual

was the outcome of the battle that ensued within the fledgling social work profession

during the early 1900s. Social work's historical roots are grounded in two ends of the

human service spectrum. The macro-mezzo end reflects the work of Jane Addams and

the concerns with social justice, social reform, and community development This faction

argued for educated humanism and viewed the practice of social work as an art. The

micro group. led by Mary Richmond, argued for individual well-being, case analysis, and

casework that was grounded in a scientific foundation (Lindsay & Kirk, 1992). Given the

outcome of the debate in the philosophy of science field and the need for social work to

have a uscientifi~* knowledge base, it is not surprising that casework as the venue for

intervention and knowledge development in social work advanced at the same time

group work. community work, and social action regressed as foci for study.

Clearly the strong influence of borrowed knowledge, the growth of empirical knowledge,

and the focus on developing knowledge based on practice with individuals assisted

social work's growth into a profession and the child and family field's development into a

scholarly discipline. However, one ramification is a limited cultivation of social work's

own knowledge base that reflects the profession's "unique value base, conceptual

principles, and practice methods" (Reamer, 1 993, p-w'i) . Another outcome is hindered

theory development about complex social problems such as child abuse, poverty,

powerlessness, racism, unemployment, and gender discrimination (Glisson. 1995;

Maluccio, 1991). Another result is the practice and policy focus in the child welfare field

on ontogenic rationales for child maltreatment when research has consistently shown a

broader ecological model is needed (Drake et al.. 1996; National Research

Council, 1993; Zuravin, 1 989).

Further complicating the issues regarding borrowed knowledge in general, is the fact

that the Canadian child welfare network relies heavily on and is significantly influenced

by the knowledge produced by the American child maltreatment field. The practice is

based on necessity, as a national child welfare research program and peepreviewed

child maltreatment journals are virtually non-existent in Canada. Nevertheless, one

outcome of the practice of borrowing knowledge is Canadian child welfare professionals

may offen know more about the American experiences with child abuse and neglect

than their own context. It seems that embedded in this common practice of borrowing

American knowledge is the assumption that the child welfare knowledge and services

appropriated from the States will translate relatively easily into the Canadian context.

While this may be true with some child welfare practices it is not the case with all

transplanted knowledge and programs. One recent example of possible adverse

ramifications associated with borrowed American knowledge has been the problems with

replicating the economically motivated and politically supported American Intensive

Family Preservation Service model into the Canadian child welfare and child and family

fields (Goodman. 1995). The unsuccessful migration from the States to Canada of these

residually based, timelimited services aimed at high-risk child welfare families came at a

substantial cost of finite child w e b re resources, money and research endeavors.

Despite this author's position, the views in the field remain diverse regarding the

ramifications of borrowed knowledge on social work. Some argue social work is still in

the early stages of forming its own knowledge base (Reamer, 1993) and research

development is at a point of crisis (Austin;i992; Lindsay et a1.,1992). Others, like

Thomas (1980 cited in Orcutt.1985) and Meyer (1992) argue that in the last few

decades, knowledge about the "micron and the "macro" domains of social work have

been expanding; that knowledge has been both increased and strengthened with the

deveIopment and growth of peer reviewed social work journals, comprehensive

curriculum development at the B.S.W.. M.S.W., and doctoral level, and stronger

research agendas. While this may be evident in the States, it is in a more primitive state

of deveIopment in Canada.

The likelihood is that borrowed knowledge will continue to influence the building of social

work knowledge because it is, and will continue to be, part of our foundation knowledge

and teaching. What this thesis posits is that both the amount of knowledge, as well as

the focus of the borrowed knowledge at the time of social work's genesis as a

profession, strongly influenced the direction of what knowledge is developed and

promoted - knowledge about the individual. Furthermore, Canadian child welfare and

child and family field professionals will continue to need to borrow knowledge from the

States. One, because of the breadth and quantity of knowledge produced by the States.

And two, national and provincial networks and infrastructure for knowledge development

in child welfare are just at the beginning stage in Canada. This point will be further

developed when the child abuse field and the knowledge base of HPS professionals and

CPS eligibility determination is examined.

Paucity of Theory Assessment

The second factor is significant because of its near absence in the development of

social work knowledge, and in particular, child welfare knowledge: the paucity of theory

assessment Support for this position is illustrated by the fact that theorylknowfedge

development constituted only 4% of the social wok research articles published

between 'T 983 -2 988 (Glisson, 1995). Although many argue that knowledge developed

through the empiricist framework needs to filled out by qualitative methods, practice

knowledge, and theory (Reamer, l994a; Reid, 1994) the use, assessment. and research

on theory is lacking in social work This absence of theoretical testing is clearly reflected

in the studies produced in the child welfare field regarding HSP decision making and

determinants of child maltreatment, as only a very few indicate a theoretical test

(Bronfenbrenner. Moen & Garbadno, 1984; Tmme, 1992; Woiock, 1982; Zuravin,

1 989).

Why? is it because theory development and testing are not important to knowledge

building? Inquiry into the relevancy of theory for knowledge building begins with the

following questions. What is theory? What are its functions? Is it important in the

development of knowledge? If it is, what effect does the paucity of theory assessment

have on knowledge development in social work?

The literature on the necessary fom, structure, principles. key concepts, and

philosophy of what constitutes ?heow is extensive, and spans the natural and social

science literature, philosophy, and mathematical fields (Payne, 199 1 ; Phillips, 1992).

Entering into this debate is beyond the scope of this thesis. While acknowledging that,

".. as an explanatory, causal story theories are rarely as predse as we wouM like them

to be Italics added]" (Miller, 1987, p. 140, cited in Phillips, tgQ2.p. 133). for the purpose

of this study the foliowing definition of theory will be applied:

A theory is whatever explans empirical fecfs (often, regularities or patterns) of nlativdy obsenrational kinds, through the description of less directly observable phenomena ritalics addedj (Milrer, 1987,p -135. cited in Phillips, t992.p. 136)

In fields such as the naturaE sciences, theory development and assessment is a weli

established, highIy valued part of scientific inquiry. Theory has a critical function and that

is to explah the taws or generalizations generated from ernpinid findings (Phillips,

1992). More specifically, theory can serve a number of functions for social work pradice

and research. Polansky (1986) notes theory can guide practitioners by helping them to

know where to start; theory can direct workers as to what is significant, thereby

organiring their work; and theory provides a historical context as to previous work, as

well as suggest explanations to client's behaviour and motivation (cited in

Grigsby.1993). Simon (1994) points out that ' theory makes possible both the framing

and screening of knowledge for practice" (p. 144). Rothery (1980) persuasively argues

that the "mutual influence of knowledge and theory is one of the best imaginable

examples of real reciprocity, as not only can theory direct the search for knowledge, but

the process of setting a direction and seeing the result will inevitably influence

subsequent choices" (p. 63).

This thesis argues theory is both the driver of practice as well as the guide and

interpreter to research. As Chow (1 996) argues, 'Social constructionists are correct in

pointing out that all observations are theory-dependent" (1 84). Therefore, inquiry into the

theoretical underpinnings of social work interventions is vital as it advances social

work's knowledge base, thereby allowing for continued development as a profession.

Omutt (1 990) writes:

It is imperative that adequate foundational theories support clinical professionar work and important that the social worker contribute to this basic knowledge, as well as to practice theory. Though practice theory may be a priority, knowledge that is fbndamental to practice should not be totally reiegated to other sciences (p. 7).

So, if the role and function of theory is essential in informing and guiding both practice

and research knowledge then why is theory in a subordinate position to empirical

knowledge in the building of social work research knowledge?

Two factors appear to be significant in understanding the reason for the lack of

attention to theory development and testing in social work. The first factor appears

embedded within the social work profession. More specifically, theory, its use, value

and relevance to practice and research evokes a wide range of support and

responses. This range is evident in both the professional literature and in practice

(National Research Council, 1993; Payne, 1991 ; Saltman & Greene, 1993; Simon, 1994;

Thyer, 1994).

A review of the social work literature on attitudes and use of theory in social work finds

that one body of research suggests that practitioners make little use of the human

behaviour theories even though they are a part of the foundation level of teaching

(Franklin, 1986; Kolevzon & Maykrantz, 1982; Thyer, 1994). For example, Kolevzon

and Maykrantz (1982), examined social workers' knowledge of theory, use of theory,

and ability to describe how theory guided them in their practice. They found very few

workers were able to state which theory they used, let alone describe how the theory

directed them in their work. Essentially, even the workers who supported theoretically

based models were not dear on the relationship between theory and practice, nor on

their use of theory in practice. Other studies found that there was a relationship between

a practitioner's theoretTcaI orientation and the year they graduated (Franklin, 1986;

Saltman et al.. 1993). Perhaps in social work, theory dominance and use might have

more to do with contextual influences that result in trends than a theory's causal

explanations and evidential requirements of relevance.

At the other end of the continuum are studies that suggest that theory is both important

to and used by social workers. Recent studies found schools, students, and practitioners

of social work have a high degree of consensus regarding the importance and use of a

multi-theory orientation to human behaviour (Saltman et al., 1993).

This wide variance on the merits of theory is evident in other ways in social work

Examination into the philosophical assumptions, theories, and knowledge base of social

work remains in the early stages (Reamer, 1993). Teaching of theory is often poorly and

inconsistently done, making its value and use questioned for practice needs (Thyer,

1994). Until recently, with the inclusion of feminist theory, the foundational theories have

lacked relevancy for many practitioners, as explanations regarding context or the

experiences of women and minorities have been absent (Callahan, 1993 b). Again,

perhaps at the most basic level, theory's subordinate position in social work reflects the

deep-seated belief that social work is and should be practice based, not theoretically

based. By all appearances, the assumption. 'sound social work practice is based on an

established theoretical orientation" (Saltman et al., 1993, p.90) is not a practice reality-

The second fador contributing to the paucity of theory assessment lies in the broader

arena of social science research. Over the last ten years there has been a shiff in the

emphasis in the social work literature fiom theoretically based models to empiricaUy

based models (Blythe 8 Briar, 1985). Support for this point is found in a study in the

Journal of Consultinu and Clinical Psvcholo~y which found that during the 1960's about

69% of psychotherapy outcome studies had a theoretical rationale but by the 1980's it

had dropped to around 30% (Omer & Dar, 1992). This suggests, of the two scientific

methods used to develop and test knowledge, namely, empirically based models that

are derived from clinical research or theoretically based models that are extracted from

theory, the preponderance of knowledge, at least in the social sciences, is now being

developed from empirical models.

This raises a critical issue and an unresolved debate. If data are collected in an

atheoretical manner, how do researchers decide what concepts to use when cdlecting

data if no theory is tested? Responses are diverse. Rothery (1980). Phillips (1992).

Wtkin (1992), and Chow (1996) argue that knowledge itself is not atheoretical, therefore

data cannot be gathered in an atheoretical manner. On the other hand, Thyer (1994)

and Blythe (1992) take the position that practice guided by empirical data absent of

theoreticat constructs, is not only possible, but may be preferable. Is it possible to

proceed fruitfully without any theory whatsoever? Miller (1959) makes the point, and

one supported by this author, that

Pure empiricism is a delusion. A theorylike process is inevitably involved in drawWIng boundaries around certain parts of the flux of experience to define observable events and in the selection of the events to be observed (cited in Helson, 1964, p.2).

Limited theory assessment in social work can be attributed to a variety of factors: the

dominance of the empirical research agenda, the reliance on knowledge and theories

from other fields, the ambivalence in the profession about whether social work should be

practr*ce based or theoretically based, confusion as to whether theory is or is not a key

element of knowledge building, and the often daunting conceptual and methodological

challenges involved in evaluating theories in the human service fields. Clearly, social

work knowledge is characterized by a paucity of theory development and assessment.

So what are the effects?

At the broad social work profession level, one outcome of limited theoretical assessment

is an enormous premium on knowledge derived ftom empirically based research. For

social work this is problematic since a review of the research literature finds, '...the

methods social workers use reflect little mastery of basic scientific methods, especially

the use of experimental designs and advanced statistics" (Task Force, 1991, p. 43).

Furthermore, if the quality and quantity of knowfedge generated from empirical research

is in question, then so is the relevancy of that knowledge for practitioners. The more

specific effects on the child maltreatment field of the combination of a paucity of

theoretical assessment, a mixed position by the larger social work profession on the

importance of theoretical assessment, and the construction of child maltreatment as a

phenomena due to individual fado=, has meant what limited theoretical testing has

occurred to date has an ontogenic focus, despite the need for comprehensive ecological

theories (National Research Council, 1993).

If theory assessment continues to be devalued in social work, then the ability to either

improve our theories, and ultimately improve and refine practice interventions or rule out

the validity of a theory, remains absent from our knowledge and skill base. For example.

to date, no unified theory of social work practice exists. Variance and diversity are the

key features of pradice interventions, methods, and treatment approaches. Some

suggest diversity in theory and practice are distinctive hallmarks of social work and

multiplicity is a virtue in itself (Orcutt,7990). If variance and diversity are the main

features of practice, then they are key features in theoretical formulations, methods, and

definitions, or in other words, theoretical heterogeneity.

Maas (1979) stated that not only would unification of social work theory be a daunting

task, it might be an impossible one; he argued the likelihood was the continuation of

theories being developed that were specific to discrete areas of practice. In support of

this line of thought, although not writing about the discipline of social work specifically,

but rather society in general, post-modernist theorist J.F. Lyotard (1 984) argued, "... the

needs of clients are no longer determined by grand social theories such as marxism,

liberalism, behaviourism, and psychoanalysis, but are constituted through public policy

and input-output systems of accountability and budget controls" (cited in McBeath &

Webb, 199T. p.751). Recent literature (Orcutt. 1990; Reamer, 1993) suggests the lack

of attention to theory development coupled with a preoccupation with empirical research

has hindered our profession. Perhaps in the end, a unified theoretical framework is not

possible, but if no effort is ever made to develop, examine, research, and test the

possibility, then it will always remain an impossibility.

It is not just that theoreticat testing about different theories is needed. Chow (1996)

strongly argues for the need to condud multiple and qualitatively different studies to

assess the same substantive hypothesis; he points out why this process advances

knowledge development:

In this sense, the advance of knowledge assumes the form of refining, with valid evidential support in each step, what is said in the explanatory theory. This process of refinement is not the accumulation of quantitative information (and definitely not the accumulation of the same type of data) because qualitatively different data from various experiments are kept separate. Instead, it assumes the form of (a) eliminating, or modifying, incorrect theoretical statements, and (b) refining correct theoretical statements by explicating the theoretical properties of the hypothetical mechanisms. In sum, the growth of knowledge is a process of trial and error at the conceptual level (1 16).

In sum, the lack of attention to theory building and assessment forces social work to

rely heavily on theory development from other disciplines. It also means social work is

viewed as a profession that is uncoordinated intellectually and practically. Research has

contributed substantially to the empirical knowledge base of practice, but as it has not

proved to be a dominant factor in the knowledge that guides practice (Reamer, 1993).

SirniIarIy, although knowledge generated from practice has been the dominant form of

knowledge to guide practice, it, as Reid (1994) describes, "...suffers from problems of

bias, over-generalization, and lack of rigorous testing" (p.23). According to Austin

(W98), only 15% of the 863 research-based articles from 13 social work journals

between January 1993 and July 1997 address intervention research - the critical

element in connecting research to the knowledge base for professional practice. It

should not be surprising that a profession that is utilizing limited tools to build knowledge

for practice has in the end limited knOWIedge for practice.

This thesis argues that theory has a key role in explaining, describing, and justifying

what social workers do. In social work research, theoretical testing and refinement

clearly needs to be elevated in importance. Quite simply, social work must reorient itself

to the importance of theory in knowledge building, for theories are developed within the

social construction of social work and are products of the context in which social work

operates (Hartman, I 994; Payne, 1991 ; Witkin, 1992). At this stage of knowledge building

in child maltreatment, the need is to emphasize theories and methods that have a

research base because theoretical testing would advance knowledge in general (Reid,

1993; Wells, 4994; Wells & Biegel, 1992). The need is also to retest and refine

theoretical hypotheses (Chow.1996). And of particular relevance in this secondary

analysis, is the need in the child maltreatment field to examine etiological theories and

models of chiM maltreatment (Drake et aL.1996; Garbarino & Crouter, 1978; National

Research Council, 1 993; Zuravin, 1 989).

The secondary analysis attempts to address the aforementioned knowledge building

needs by testing adaptatioklevel (AL) theory. Other advantages in using AL theory are it

gives guidance in thinking about how contextual factors influence neglect decisions, it

allows examination of individual, community and group levels of adaptation, the theory

has been previously tested and supported in a variety of fields, and this test replicates

-uVdock's (1982) test of AL theory.

Social Construction of Child Welfare

The purpose here is not a detailed review of the historical development of child welfare

in North America. Rather, the intent is to build on and weave in the previous analysis on

social work epistemology by examining the social construction of child welfare. Arguing

that social work is a socially constructed activity influenced by various contextual facton,

means the various knowledge bases, like child welfare, are also socially constructed.

While a review of world history has found child maltreatment to occur in all cultures

across time (Dumbrill, 1998), 'child abuse" as a social construction is a relatively new

perspective (Robin, 1991). It is hoped, that by identifying the beliefs and values that

underpin child maltreatment, the result is a better understanding of the current residual

system and insight into the existing dilemmas and limitations of the literature and

research specific to the secondary analysis.

Like the social work profession, the dawn of organized child welfare in North America

began in the 1850's. As social work evolved into a profession so too did the

professionaliration of child welfare services. This is best understood as part of a larger

developmental process by which the concept of social welfare gradually became

integrated into society and its legal, economic, and public institutions (Testa, 1985).

Child welfare services now form a regular part of social service provision in all Western

welfare states. While often perceived just as, '...a field of social service practice in which

the state, operating through specific statutory law, takes over 'fundions normally canied

out by parents for the childrenr " (McCall. 1990, p.347), another integral part of CPS

service delivery is supporting children remaining with their families.

A brief overview of the development of child welfare as a social service shows that

coinciding with the emergence of the concept of childhood, child welfare reforms in the

latter part of the lgn century were intended to separate children from the capitalist

market situation of adults, combat the internment of dependent children with adult

paupers, and require the state to take more of an enforcing role in the education and

supervision of children (Giovannoni et al., 1979; Macintyre, 1993).

During the first half of the 2om century, child welfare concerns turned to, "...the care of

children who were orphaned, abandoned, or otherwise dependent. including those who

had suffered severe abuse or neglect" (Schuerman & Littell, 1996, p.1). Over the last

fifty yean, child welfare services and knowledge development have increasingly focused

on abuse and neglect. For the most part, CPS agencies concentrate on identifying and

providing services to children who have been harmed or are at risk of ham by their

caretakers. The binary options for CPS service are inhome and out-of-home

alternatives.

Selection of one of the options invokes considerable tension between two values: the

value of protecting children from ham versus the value of respecting and protecting the

integrity of the family. Influencing the decision is the intrinsic belief in our society that

the best place for children is in their family. Other factors that influence both society and

the CPS agency's intervention decision are organizational beliefs and customs,

provincial laws, and available CPS responses (such as staff, resources, or funding). The

default position is to respect the integrity of the family, or in other words, non-

intervention.

Understanding why the task of constructing standardized definitions and clear minimum

CPS intervention lines for child maltreatment is so difficult and such little progress made,

entails closer examination of values deeply entrenched in and offen in conflict with each

other in Western culture. Additionally, the principles embedded in CPS - permanency

planning, reasonable efforts, and least restrictke alternative, are affected by the lack of

a clear political constituency for children in general coupled with the current aggressive

financial agenda of reduced spending and cost-cutting initiatives. These factors.

coupled with values implicit to a conservative ideology: heightened emphasis on

individual rights and responsibilities, distinction between 'deserving and undeserving

poof, and minimal govemment intervention, all support a residual child welfare system

(Armitage, 1993; Goodman, 1995).

The social construction of child welfare adds to the above list the following beliefs and

social values: the best place for children is with their families; caring for children is a

family matter, primarily a female task, unencumbered by income issues or systemic

factors; it is best for families to learn to deal with their problems and not rely on the

state; and government intervention only as a last recourse (Goodman, 1995; Hollinger,

1994; Wharf, 1993). Combined. these philosophical beliefs and values underpin our

current residual child welf&e system (Armitage. 1993; Goodman. 1995).

Situated within a residua! approach, the basic social construction of child abuse and

neglect as a concept in North America has remained constant over the past century.

The first part of the construction encourages the public to identify and sympathize with

the abused child not the abusing parent. The second part is for the public to avoid

responsibility for elher creating, supporting or changing the economic or social

conditions that contribute to maltreatment (Macintrye, 1993; Meyers, 1994; Robin, 1991).

In addition to the increased public exposure over the years to the problem of child

maltreatment, one of the key factors in eliciting public support for 'poor battered

children" is the fact that many adults can personally identify with being an abused or

neglected child. This creates strong anecdotal alliance with the 'psychopathology of the

individual" paradigm and also means terms such as 'child abuse" or 'sexual abusen

generate personal reactions that may usurp intellectual arguments or empirical findings.

While this may contribute to society's strong emotional support for the protection of

children as a shared social value, it further entrenches support for the residual approach

and allows the considerable research evidence that systemic issues contribute to child

maltreatment to be ignored or dismissed.

Contributing further to the complexity of child maltreatment are elements involved in the

second part of the social construction of child abuse and neglect if child maltreatment is

constructed as a psychological problem of the abuser then it is a form of medical

deviance or 'illness* and not a poverty problem. 'Illness" at the individual level can be

€mafed whereas facfors such as poverty or unemployment require major systemic

changes. This 'individual pathology' perspective also allows people to conveniently and

effictively separate themselves from the abusers. As Robin (1 991) points out:

By identitying with the innocence of the abused child, the rest of society is able to protect itself from the insight that they may have something in common with abusive parents,' share to some degree the character flaws of those who abuse children, or have any degree of responsibility for creating or supporting economic and social structures that allow child abuse and neglect to occur (p. 5).

Entangled in this ontogenic paradigm is both the culture of violence and the punitive

approach taken to deviant behaviour in North America. Punishing the individual

'criminalw (with child maltreatment it is those people who severely harm children) allows

society to believe they have addressed the issue by condemning the offender.

Accordingly, this 'psychopathology of perpetrators" paradigm minimizes the importance

of contributing systemic factors and "...conceals the extent to which we are an abusive

society" (Scheper-Hughes and Stein, 1987.p.341).

An additionai factor in the complicated construction of child abuse and neglect is

discrimination against particu tar children in our society. Specifically, within certain

cultures, illegitimate children, female children, adopted children and stepchildren are

seen as less valuable than children who do not fall in these categories (Korbin,l98i).

Williamson's (1976) cross-culturaI survey of gender preferences in the United States

found sons were favored over having daughters. And an examination of studies focusing

on the gender of abused children found females were more severely harmed than males

before the report war made to CPS, and that girls died more often due to abuse than

boys (Madden & Wrench.198t). Additional support for these last two points is found in

two recant studies. Pillitteri, Seidl, Smith and Stanton (4992) examined nurses and the

influences of parent gender, victim gender, and family socioeconomic (SES) level on the

reporting of child maltreatment. They found respondents were significantly less apt to

report abuse if the child was female rather than male and when the family was viewed

as middle and not low or high SES. Additionally, of the eleven children known to the

Ontario CPS network between January 1994 to December 1995 who died as a result of

homicide by a parent or person known to the child, four of the children were males and

seven were females (Ontario Child Mortality Task Force, 1997).

In summary, constructing child maltreatment as a phenomena due to individual

ontogenic factors or individual characteristics means child abuse and neglect remain

nested within traditional North American beliefs and values and a conservative ideology.

To date, the residual approach to child welfare has dominated theory development.

practice, policy, legislation and research (Annitage. 1993; Goodman, 2 995; Lindsay,

1993). Not surprisingly, public, and perf-taps many professionals' beliefs about the

causes of child maltreatment, are in contrast to the mounting evidence that etiological

models of child maltreatment need to focus on a range of often interactive factors:

family income, unemployment, cultural and social values, other environmental risk

factors, as well as child, parental and family characteristics (National Research Council.

1993). Clearly, the child welfare knowledge base fits within the broader epistemology of

social work know ledge development and mirrors the same limitations found previously.

Both reflect the beliefs, values, and ideoIogies of the larger society that has wnstructed

socia! work. child welfare, and child protection.

Child Protection Eligibility Instruments

The development and use of risk assessment models, child well-being scales, and

structured decision making systems is a rapidly growing phenomenon in CPS. For

examples of risk assessment models refer to Washington State Department of Social

and Health Sewices (1995) for the Washington Risk Assessment Model and Reid,

Sigurdson. Christianson-Wood, and Wright (1 995) for the Manitoba Risk Assessment

System. See Magura, Moses, and Jones (1987) for the Child WelCBeing Scales and

Children's Research Center (1995) for the Michigan Department of Social Services

Structured Decision Making System. While the practice and research fields are showing

considerable interest in the use of risk tools, defining the concept of eligibility and

developing instruments for specifically determining the eligibility of CPS referrals are

relatively new areas of need and development in child welfare (Goodman &

MacFadden, 1996).

As noted earlier, eligibility determination in child protection services (CPS) refers to

whether a child's situation meets legislative requirements to warrant a protection

assessment. Notwithstanding that agencies may not confine service delivery strictly to

the legal requirements for eligibility, the legislative mandate does underpin eligibility

determination (MacFadden at al., 1 997).

Determining whether a referral is eligible for CPS service is frequently viewed as just

part of the investigation phase. The concept of specific steps within the investigation

phase appears to still be new. For example, some agencies may open and investigate

most cases to avoid risk of making an error in judgment. While other agencies recognize

that some selectivity is essential given diminished funding, resources, and the number of

false reports (Barone, Adams. & Tooman, 1981; Howitt; 1992; Robin, 1991).

Practice preferences aside, the decision about whether a referral is eligible for CPS

investigation carries considerable resource implications, as well as important

consequences for the children and families involved. Furthermore, as the number of

child protection referrals grows steadily, coupled with significant decreases in funding

and resource availability, CPS agencies are no longer able to tolerate inefficiencies.

There is growing recognition in the field that the initial screening decision about whether

to open a case and proceed with an investigation or whether to reject a referral, is in

fact a critical and distinct decision point. Agencies are incorporating this understanding

into CPS policies and procedures and are now examining their screening practices in

order to give priority to protection over prevention services (Wells et al., 1989b) .

Policy statements about eligibility for CPS services reference the need for the CPS

worker to comply with legislative and policy requirements (Institute for the Prevention of

Child Abuse, 1994; Wells. Fluke, Downing & Brown, 1989a). Yet, to date, screening

decisions have relied solely on clinical judgment; judgment that is known to be

inconsistent and inaccurate. One reason is because CPS definitions remain vague and

imprecise; and two, HSP interpretations of those definitions can vary considerably (Alter,

1905; Children's Resource Center, 1993; Doueck, Engiish, DePanfilis, & Moote, 1993;

Leck et a[., 1994; Wells et al., f 989a). A long standing recommendation in the literature

has been the need for an eligibility instrument to assist CPS workers in the difficult task

of determining eligibility so that consistency and accuracy are increased (Giovanonni et

al., 1979; Led< et al., 1994; National Resource Council, 1993; Wells et ai., 1989b).

Wolodr (1 982) makes this point in her study:

To eliminate the high degree of subjectivity that enters into staff judgments and results in the bias found in this study, the author recommends that child welfare agencies move more rapidly toward the adoption of a structured and systematic model of case decision making..Such an approach would facilitate more uniform and effective decisions at key points in the identification and management of child abuse and neglect (p. 14).

A review of the North American literature on CPS eligibilitylscreening tools finds some

key factors have hindered the development of either a theoretical or clinical 'eligibility"

framework (Doueck et al., 1993).

One significant factor is a wide variance in the underlying philosophical beliefs or

approaches that CPS agencies take regarding screening. A literature review identified

four approaches to the issue. The first is that some CPS agencies accept all referrals

regardless of the level of severity. This 'open-door" approach may be because the

agency has substantial resources or because the agency is the only community sewice

available or because the agency culture leans toward a broad CPS service delivery

mandate. The outcome of this approach is that eligibility is not an issue because it is

neither a concept nor an option. For example, in Ontario, the CPS agencies that take a

broad, prevention approach to refsrrals have consistently voiced their reluctance to

voluntarily use the Spectrum because they view the tool as narrowing their service

options.

The work by Schuerman, Stagner, Johnson, and Mullen (1989) on worker decision-

making reflects the second approach. They found that while the importance of eligibility

as a distinct phase may be stated by an agency, closer examination finds the screening

process does not actually happen. Rather, sole use of clinical judgment occurs. The

third screening method appears to be that some agencies acknowledge the concept of

eligibility and have developed tools andlor guidelines regarding eligibility, but their

actions stem from the premise that 'screening our only refers to bogus or malicious

reports and some low-risk cases (Barone, et al., 1981; Howitt, 1992; Robin, 1991). The

final approach, and one only a few agencies appear to have reached, is that eligibility

assessment is a distinct and critical stage in CPS decision making. While some studies

have identified the general need for a CPS eligibility framework (Wolock, 1982), some

have been more specific, arguing for a relevant gatekeeping instrument for CPS workers

that reflects legislative requirements along with clear, specific criteria (Leck at a1.,7994;

Washington State Dept of Social and Health Services, 1995).

Another key factor that has limited the development of a specific framework to assess

eligibility is that structured risk assessment (RA) tools have been used for screening

purposes. While it is often appealing to do so for many reasons, their use in determining

eligibility is inappropriate. These instruments "...have not yet demonstrated an ability to

differentiate effectively between high- and low- risk cases at the time of the initial report"

(Wells et aLJ989b, p.48 ). Current RA instruments are intended to predict Mum risk of

abuse o r neglect. Consequently. RA tools shift the focus from determining present

eligibility based on current or past events to future predictions (Doueck et a1.,1993).

Even if the CPS agency stated that eligibility determination would be based on future

risk, perhaps thereby justifying the use of RA tools, the ability of such instruments to

accurately predict future maltreatment has been seriously questioned (Doueck et

al., 1993; Seaberg, 1990; Wald & Woolverton, 1990).

There are addlional concerns with RA instruments. For example, implications of using

RA may be profound for clients and CPS workers. If the instrument is to identify and

service only those clients who fall in the 'highest risk" category, than this effectively

rations CPS services, which by implication further entrenches the residual approach.

And, if use of the tool is not to rehabilitate but for risk verification, than the relationship

between the CPS worker and the client is modified. The focus for CPS may be not on

making the right decision but making the defensible one (Sullivan, 1998).

A further factor limiting the development of an eligibility framework is that previous work

on eligibility focused on trying to identify eligible cases by examining common case

outcome measures such as substantiation rates or refenatrreoccumnce rates. Another

method used to determine eligibility was ranking cases based on key criteria such as

source of mkmI, amount of information, and presence of injury. In short, neither using

case outcome status nor case information proved h M u l methods in determining

eligibili (Washington Dept. of Social & Health Services, 1 991).

Finally, another fador limiting development of a screening tool is an absence of a

standard theoretical or clinical framework to assist CPS workers in applying child welfare

[egislation to determine referral eligibility. A North American search for a CPS eligibility

framework found only two: the Intervention Spectrum and the Washington Risk

Assessment Model (WRAM). Both the Spectrum, which is solely an eligibility tool and

the WRAM. which has four 'Screening for Sufficiency" questions as part of the larger

risk assessment model, are undergoing testing for reliability and validity.

Clearly, screening for eligibility is permitted by law or policy but it tends to be a

generalized provision and lacks specific criteria (Besharov, 1 985; Robin, 1991 ; Wells et

a1.,1989a). The lack of standard definitions and clear guidelines coupled with the

dependence on often faulty clinical judgment alone to determine eligibility, the

problematic use of RA tools, plus the absence of any systematic screening framework

ensures CPS eligibility determination will, for some time, continue to be a matter of

professional judgment Judgment that is often inaccurate. The effects of CPS inaccuracy

in determining eligibility means other professions, the general public, and clients will

most likely continue to view CPS workers as incompetent, biased, and unfair in their

decisions to accept or not accept a referral (Alter, 1985). It also means inappropriate

service delivery continues where some children and families who are not eligible for

CPS services will receive them and many children and families who are eligible for CPS

services will not receive them. Perhaps, the most compelling reason for the

development and implementation of an CPS eligibility framework is to ensure that

children who are eligible for and need protection, actually receive protection services.

ldenfification of the factors limiting the development of a screening framework only

partially contributes to our understanding of why eligibility determination in CPS remains

a new concept and prob[ernatic in application, Why, despite continued identification by

the research, academic, and to some extent, practice field, is there a noticeable

absence of standard child maltreatment definitions, expl kit criteria consistent across

jurisdictions, clear intervention lines and supporting instruments? It is obvious such

definitions, criteria, and instruments are needed to assist HSP in the difficult and critical

task of either making or screening CPS referrals.

Referencing the earlier analysis on social work's epistemology and construction of the

phenomena of child welfare in Western culture, I posit the underlying philosophical

beliefs and values of our conservative ideology support both the rationale for and

support of a residual approach. This type of approach supports imprecise neglect

definitions, vague standards for intervention, and defines child maltreatment for the

most part, under a medical model - psychologically deviant paradigm. The focus on

individual pathology is maintained by having individual HSP responsible for identifying

and treating €he individual abuser, by using vague definitions and standards, and by

employing theories and interventions aimed predominantly at individual therapeutic

treatment, not correction of social structures. This ensures CPS eligibility determination

remains nestled within the confines of the child welfare field, the social work profession,

and the larger societal beliefs* Contemporary prad-ces and policy ensure it is incumbent

on the HSP to judge the individual, not to judge the society. As Robin (1991) points out:

Defining child abuse as a problem of 'troubled persons* is a way of udepoliticizing" the problem. The psychologizing of sociaI phenomena draws attention away from the institutional or structural aspects of the problem. The term &child abuseu itself displaces our attention from the context of abuse to the moral and psycho[ogical failings of the individual abuser. By making the individual abuser the ILcausar agent, rather than the scene, becomes the focus of attention (p. 4).

The type of data gathered and method of service delivery in CPS also lends support to

this position. Individual CPS supervisors are responsible for individual CPS workers who

are assigned individual cases. Callahan (1993a) points out that. 'Probably the most

significant outcome of this approach to [CPS] work is that it encourages a sense of

individual responsibility and obscures the impad of social conditions" (p. 77). Likewise,

data are individually and psychologically oriented not systemically focused. Key factors

such as low income, type of housing, community environment, or cultural factors are

neither routinely collected at the practice level nor are they required by law, policy or

funding bodies. Hence, social and economic conditions that foster maltreatment remain

invisible from practice analysis, policy inclusion, legislation, and most theoretical models

of child maltreatment Furthennore, the absence at the case level of ecological data

make it difficult for researchers to aggregate these important etiological factors so to

better understand child maltreatment.

Studies on HSP Decision Making in Child Welfare

Previous analyses suggests that a number of forces have influenced professional

surveys and decision-making studies in CPS. Reamer (1993) argues that in order to

undersfand and appreciate the current context, it is important to first understand the

broader sociopolitical and cultural context in which studies are conducted. The goal of

this chapter's previous sections was to identi@ the often hidden forces that govern

personal and institutional behavioun. Understanding the broader context assists in

better understanding any conceptuaI, design andlor methodological concerns noted in

the studies. The goal here is to provide a concise portrait of the child welfare research

milieu. In other words, the intent is not to diagnose the cultural context as though it were

an illness but to describe it as a network with various layers.

One key force impacting on the child welfare research environment are the sociopolitical

and cultural processes in Western ideology. These processes elevate the rights and

responsibilities of the individual and the dominance of the freemarket economy and thus

support the residual approach to child welfare, as evidenced in the laws, social values

and institutions. It is this set of philosophical, cultural and social values that pervade and

support individual and family life styles and community services, as well as research

endeavors in our society. The critical role politics takes in fostering these often invisible

beliefs and values is underestimated and often denied.

This heightened emphasis on the individual as the focus of study has meant the

preponderance of child maltreatment studies have a medical or pathogenic focus where

the unicausat connection between individual functioning and child maltreatment is

examined. That said, a number of the studies under review have taken the approach of

trying to address the complexity of the phenomena of child abuse and neglect by

examining other etiological factors, such as economic, cuItural, and social elements. For

example. in addition to poverty, neglect has also been empirically linked to other

ecological determinants such as residential transience (Garbarino 8 Crouter, 1978;

Zuravin, l989), unemployment (Steinberg, Catalano & Dooley, t 981). an agency's

philosophical approach and sociodemographic variables e.g. percent change in the

population and population size (Wells, Fluke, Downing & Brown, 1989a). social isolation

(Garbarino et af., 1978). and lack of resources (Craft & Staudt, 1991; Wells et al.,

198Qa; Zuravin, 3989). However, integration of these factors into more complex theories

of child maltreatment is just beginning. Adequate testing of these theories is limited.

Additionally, the relationships among €he different etiological factors is still not well

understood (Zuravin, 1989). The absence of such definitive knowledge means that the

development of a multivariate model of child maltreatment is still in its infancy. An

ecological paradigm for understanding child maltreatment cannot yet counter the

mainstream beliefs and values and the political system that governs the residual

approach. Accordingly, since government sectors are the primary source of child welfare

funding, the direct practice and policy fields of child maltreatment tend to ignore the

findings from the more systemic studies (National Research Council, 1993: Robin.

1991).

The child welfare research field faces additional challenges in that 'child welfaren is a

socialIy constructed problem. This means definitions of child welfare, whether legal or

practice opinion, are vague, interpretive, and change depending on time and place. In

other words, definitions of child welfare are created through collective definition, unlike a

condition such as myopia that objectively exists to be studied. As a result of the

malleable nature of child werare, CPS studies on professional decision making have

been conducted in a context where common language, precise operational definitions

and explicit intervention standards for child maltreatment (in particular neg k t ) are

absent The lack of clear criteria, whether for severe or marginal maltreatment, means

practice standards differ. Also, professional survey studies use many measures and

tools that are not psychometrically sound. Consequently, findings often conflict or are

difficult to interpret Even when maltreatment types are more precisely defined by

operational definitions, the comparability between studies is compromised by wide

variations in the definitions used (Zuravin, 199i).

Survey and professional decision making research is situated within the field of child

maltreatment, which in turn, is part of the overali body of social work research. The

research literature on chiid abuse and neglect has been generally viewed by the

scientific community as having the same characterizations of social work epistemology:

fragmented, decentralized, lacking cohesion and organization, a paucity of theoretical

assessment, and for the most part 'not done well" (National Research Council, 1993).

Analysis of the child maltreatment research finds the majority of studies are cross-

sectional and often fraught with problems such as small sample sizes. Samples are

often selected by case workers and clinicians and the samples are often dominated by

an overrepresentation of poor families, often with multiple problems such as

unemployment, mentat health issues and substance abuse. There tends to be a

genera[ oversight in maltreatment studies regarding incorporating ethnic and cultural

variables- Comparison or control groups are often absent, analysis is either limited or

poor, and there is little theoretical assessment (Glisson, 1995; Lindsay et al., 3992). The

paucity of adequately evaluated instrumentation continues to adversely affect the quality

of studies. What is surprising is not the abundance of instruments, but in using tools the

eagerness with which professionals accept a very thin veneer of science. Issues

concerning the validity of the tools are often transformed into the more concrete and

somewhat iess troubling concern of reliability. The National Research Council's (I 993)

comprehensive review and critical assessment of child maltreatment research literature

states:

In the past decade, significant improvements have occurred in the development of child maltreatment research, but key problems remain in the area of definitions, study designs, and the use of instrumentation (p. 47).

Notwithstanding concerns regarding the quality of the studies, the quantity of research is

uneven within the different categories of maltreatment. For example, the recently

released Child Maltreatment 1996: Reaorts from the States to the National Child Abuse

and Nealect Data Svstem (NCANDS. 19981 found of the almost one million children in

the United States in 1996 who were victims of substantiated or indicated child abuse

and neglect (approximately an 18% increase from 1990) 52% suffered neglect, 24%

from physical abuse, 12% ftom sexual abuse, 6% ftom emotional abuse and 3% from

medical neglect, noting some children suffered more than one type of abuse. Yet, a

greater abundance of studies on physical abuse and sexual abuse exist than research

on neglect And the cwccurrence of different forms of child maltreatment has been

examined only to a limited exfent in addition to many studies on professional decision

making not differentiating between the different types of maltreatment there is a dearth

of literature on the topic of neglect

Finally, there are some roadblocks that Canadian child maltreatment researchers and

pra&*tioners face in generating knowledge and understanding about child maltreatment

in Canada. First, the preponderance of child maltreatment studies are American not

Canadian based. Continued reliance on these findings is a concern as they may not

reflect the Canadian context or our experiences with child maltreatment. Second, in

Canada there is a noticeable absence of a leadership role by the federal government in

the area of child abuse and neglect This lack of national leadership forces child

maltreatment research endeavors to be scattered and essentially the responsibility of

uncoordinated provincial, local, academic and individual efforts.

There has been a vigorous debate in the literature over the past two decades regarding

the lack of clear child maltreatment definitions and intervention standards. This debate

has been the catalyst for many studies. The following review of the literature on

professional decision making attempts to provide more of an overview or meta-analysis

of the findings from the relevant body of research rather than a critical analysis of each

study.

Review of the literature on surveys of professional decision making regarding child

maltreatment finds the inquiry process has taken different approaches. One approach

has been to focus on gathering information regarding different HSP groups perceptions

of child maltreatment by having them define neglect or abuse, by assessing their

attitudes to child maltreatment or by testing knowledge about child maltreatment (Cohen

& SussmanJ975; Gelles, 1977; Giovanonni, 1993; Hanard & Rupp, 1986; King, Baker

& Ludwig, 1993; Nagi. 1977; Pillitteri, SeidI, Smith, & Stanton, 1992; Reidy 8

Hochstadt, 1993; Saunders, 1988; Woolf, Taylor. Melniwe. Andolsek, Du bowitz, De

Vos, 8 Newberger, 1988).

An offshoot of the aforementioned approach to knowledge development has been

numerous studies either regarding HSP groups perceptions' about the 'seriousness" of

an CPS incident or the degree of agreement regarding intenrention or substantiation on

specific maltreatment situations (Alter,1985; Campbell, Hunt & Lewis, 1958; Craft &

Clarkson, 1985; Giovanonni, 1989b; Kelley, 1990; King, Baker & Ludwig, 1993; Lena &

Warkov, 1 978; Magura & Moses. 1986; Misener, 1986; Pillitteri, Seidl, Smith & Stanton,

1992; Rosen, 1981 ; Saunders, 1988; Snyder & Newberger, 1986; Streshinsky,

Billingslelyt & Gurgin, 1966; Wells et al., l98Qa; 1991 ; Trute, Ad kins & MacDonald,

l992a).

Some studies have attempted to examine the contextual determinants of child

makreat ment such as investigating socioeconomic or rurallu rban differences in HSP

decisions (Craft & Staudt, 2991; Giovannoni, 1989b; Pillitteri et al., 1992; Trute et

a1.,1992ar 1992b; Wells et al., 1989a; Wolock, 1982; Zellman, 1992). Other studies

have examined differences in decisions relating to specific worker characteristics such

as gender, age, experience with CPS cases and parenthood status ( Magura et a1.,1986;

Pillitted et a1.,1992; Reidy et a1.,1993; Snyder et a1.,1986; Trute et a1.,1992; Wells et

aL.1989a; Woo(f et a1.,1988). Only a few studies have focused their inquiry to specific

maltreatment categories, such as sexual abuse (Kelley. 1990; Reidy et a1.,1993;

Saunders. 1988; TNte et a[., 1992a. 1992b) or neglect (Alter, 1985; Craft & Staudt,

i 991; Trocme, i992a). And even fewer studies indicate a theoretical rationale (Rosen,

1981; Trocme, 1992a; Wolock, 1982).

Another research direction has been the model building approach. This knowledge

generating method attempts to first describe each step in the CPS decision making

processes and then order it into a prescribed sequence (Baird, Wagner, Caskey, &

Neuenfeldt, 1995; Rossi, Schuerman, & Budde, 1996; Schuerman, Stagner, Johnson, 8

Mullen, 1989). While the knowledge generated by these studies may be helpful to a

specific agency in the study, the wide variance in organizational structures, intake

methods, definitions of maltreatment and procedures, often precludes much

generalizability of these studies findings.

lntra and interprofessional suweys about HSP views of child maltreatment have used

different methodologies in trying to untangle the ambiguity of terms and interdisciplinary

controversy regarding how child abuse and neglect is perceived by HSP groups. Some

examples of various methodologies are: questionnaires or opinion surveys (Cohen &

Sussman, 1975; Gelles. 1977; Hanard & Rupp, 1986; Woolf et al., 1988); vignettes with

choice tasks offering binary options for severity or seriousness level (Giovanonni et

al.,?979; Misener, 1986; Snyder et aI..i986) or an ideal versus actual selection (Rosen,

1981; Streshinsky et a[., 1966; Wolock, 1982) or an evaluation task using standardized

scales (Hazard et al., 1986; Magura and Moses, 1986; Reidy at a13 993; Saunders,

1988; Trute et aIJ992; Zellman, 1992); analogue studies (Alter, 1985) and index

construction (Giovanonni et al., 1979; Trocme, 1992a; Wells et al., 1989a).

Given the absence of definitional standards, it is not surprising the majority of studies

generally found differences within and between HSP groups warding perceptions of

the 'seriousness" of specific incidents of maltreatment or recommended actions to be

taken. Illustrations of intraprofessional studies demonstrating differences within a

professional group are: judges (Campbell et al.,1958); CPS social workers (Craft et

al., 1991 ; Giovannoni, 1989b; Rosen, I981 ; Streshinsky et al., 1966; Wells et al., 1 989a;

Wolock, 1982;); medical residents (WooM et al., 1988); nurses (Pillitteri et al., 1992).

Examples of interprofessional studies finding differences between HSP groups are:

CPS and community social workers, doctors, police (Gelles, 1977); CPS workers,

judges, police and medical personnel (Nag i, I 977); CPS workers, police, pediatricians

and lawyers (Gionanonni et a1.+1979); pediatricians, mental health professionals,

teachers, and college students (Hazard & RuppJ986); hospital social worken, nurses,

psycho~ogists, psychiatrists, and physicians (Snyder et al., 1986); CPS workers, police,

judges, assistant district attorneys, assistant public defanders (Saunders, 1988); CPS

workers, police, and nurses (Kelley,1990); CPS workers, police, and mental health

prufessionals (Trute et a1.,1992a,1992b); primary care physicians, mental health

professionals. child care providers, and education personnel (Zellman, 1992); nurses,

paramedics and emergency medicaI technicians (Kng et al.,I993); and psychologists,

psychiaMsfs, social workers and mentaf health couosekrs (Reidy et a1.. 1 993).

That said, there appears to be consensus among HSP that for the most serious forms of

maltreatment, regardless of type, reports should be made (Dukes & Kean, 1989; Korbin,

1981). Additionally, analyses from a number of studies found professionals can and do

discriminate among the types of maltreatment; the broadest categories being physical

abuse, sexual abuse, neglect and emotional harm (Alter, 1985; Cohen et a[., 1975;

Gelles, 1977; Giovanonni et a[., 1979; King et a/., 1993; Lena & Warkov, 1978; Misener,

1986; Snyder at a1.J 986;). Many studies found that HSP rank the types of maltreatment

by seriousness. For example, cases were viewed as 'more serious" when parental

characteristics, such as willful ham, were tied to the case characteristic of actual

physical injury to the child (Altar. 1985; Craft et a!., 1985; Gelles. 1977; Giovanonni et

at., 1979; Lena et a[., 1978; Magura et al., 1986; Wells et al., 1989a; Zellman, 1992).

Another factor that appears to affect how HSP decide what constitutes child

maltreatment is both the amount and the level of detail of the available information,

When situations are routine, straightforward and simple, then consistency of worker

responses is good. Whereas, the consistency of worker judgments is adversely affected

when situations are complex lack specificity, and require considerable judgment; these

are common characteristics in neglect cases (Nagi. 1977; Wolock ,? 982). Furthermore,

since physical abuse is more easily operationalized than neglect, a corollary finding is

intervention and substantiation is greater in physical abuse than neglect (Gelles. 1 9?7;

Giovanonni, 1989b, t 99q; Giovanonni et a[., 1979; Rosen, 1981; Snyder et aL.1986;

Streshinsky el al., 1966).

The greater weight given to physical injury by HSP at the definitional, intervention and

substantiation levels affects neglect reporting, investigation, and substantiation rates.

Quite simp(y, parental commission of child physical injury is more easily distinguishable

for HSP at major, moderate or even minor levels. Yet with neglect, which is an omission

of care by the caregiver, actual physical harm may only be evident at the major level.

Support for this point is demonstrated in Giovanonnls (1 989b, 1991) study of 1150 CPS

reports and 117 CPS workers @om nine CPS agencies in three American states. She

found CPS workers required indications of more pervasive parental failure in order to

substantiate neglect forms, such as lack of supervision or failure to pmvfde; "simple"

neglect or improper supervision was even less likely to trigger substantiation. The

greater difficulty in defining, intervening and substantiating neglect is a critical issue for

the field, policy, and research arenas to address because neglect is not only the most

prevalent type of child maltreatment in North American, but it is also a significant factor

in the leading cause of death for children: unintentional injuries (Ontario Child Mortality

Task Force,1997).

Findings to date regarding whether worker characteristics, such as, gender, age, role,

education, marital status* parental status or previous experience influence CPS

decisions are mixed and inconclusive. In addiiion to the definitional problems in the

studies, one of the difficulties in drawing definitive conclusions from the surveys is the

intent of the study varied. For example, some studies examined attitude differences

(Reidy et aL.1993; Saunders. 1988; Streshinsky et aI.+i 966), some focused on severity

differences (Giovanonni et al., 1979; Magura et al., 1986; Snyder et al., 1986; Wobck,

1982), and others on investigation or substantiation choices (Trute et a1.J 992; Wells et

a1.,1989a). That noted, a number of studies found no differences in response patterns

for respondent characteristics by age, marital or parental status, previous experience

with or education in CPS (Hazard et al., 1986; Magura et al., 1986; Snyder at al., 1986;

Wolock, 1 982).

A number of studies did find gender affected ratings (Giovanonni et a1.,1979; Reidy et

al., 1 993; Snyder et a1.,1 986; Woolf et al., 1988). The problem with the gender variable is

it is confounded with profession. For example, the fields of nursing, social work, and

general counselors tend to be dominated by women. Yet, in psychiatry, psychology,

police, justice, and to some extent the medical profession, the dominant gender is men.

Therefore, profession is by far the best variable to account for rating differences.

Finally, of particular relevance for this thesis, studies that examined context found it to

be a significant factor in influencing worker decisions (Campbell, et a1.,1958;

Giovanonni, ? 989,t99l; Manis, 1971 ; Wells et a1.,1989a; WoIock,1982; Zellman, 1992).

In particular, Wolock's (1982) study of 184 CPS workers from 1 1 CPS sites in one

northeastern state found the more severe the SES conditions were in a site, the greater

the severity of child maltreatment cases, yet the CPS workers judged the vignettes as

less severe. The reverse held - the more advantaged SES areas had less severe CPS

reports but the CPS workers viewed the simulations as more severe. Findings suggest

that one way the CPS system adjusts to different contexts, such as pressures of more

reports, is to restrict the meaning of abuse and neglect.

Additional support for the significanoe in SES site differences is found in Wells. Fluke,

Downing and Brown (1989a) and a later paper on the same study by Wells, Downing

and Fluke (1991). The study examined CPS intake screening and substantiation

practices. The sample included five states, 12 CPS sites, and a mix of urban/rural

agencies, richlpoor sites, and homogenoustmulticultural populations. When respondent

characteristics were expanded, the only worker variables that significantly related to

the 'decision to investigaten were worker belief that the CPS role in the community was

broad and included intervention if a concern was noted. For example, one site

investigated 97% of contacts whereas five sites screened out over 40% of referrals

received (Wells et at., 1991). Second. "...worken were more likely to investigate if their

decision was not affected by whether resources were available' (Wells et al., 1989a, p.

20).

These two findings suggest a CPS agency's philosophical approach needs to be a

variable that is included when examining ruralfurban differences or similar site

variances. And clearly, the resource avaiia bility issue becomes elevated in importance if

CPS investigation rates increase due to a strong resource base in the community. Logic

implies reporting and substantiation rates are influenced too. The corollary is reporting,

investigation and substantiation rates being adversely aff@cted in areas with a limited

resource base. Further support for this point is found in Craft and StaudYs (1997) study

of child

workers

neglect in urban and rural communities. They found that while urban CPS

tended to substantiate neglect more than rural workers. they wondered whether

it was because rural workers had more limited resources. Intuitively, one concludes that

limited resource availability refers to poor areas (low SES) but this may also have

implications for higher SES areas undergoing dramatic changes, where services have

not kept up with rapid population growth.

When case characteristics were analyzed. Wells and her colleagues (1989a; 1991),

found that, as other studies previously noted, actual injury was a significant predictor in

the decision to investigate; of the 2556 CPS cases examined they found 36% of specific

allegations were screened out because no injury was reported. They concluded that

even though neglect constituted the majority of referrals, cases of physical and sexual

abuse took priority - a finding that has deleterious implications for neglect reporting and

investigation. However, when ecological factors were analyzed, site was a critical

variable in screening decisions.

When all case characteristics were accounted for, the importance of site stood out dramatically- When presence of injury, allegation of maltreatment. age of child, source of contact, perpetrator's access to the child (in the home), completeness of information recorded, and presence of other family problems were held constant, the site in which the refenal was made had more impact than any other variebte [italics added] (Wells et atJ989a. p.16).

When screening practices by type of maltreatment by site was examined in greater

detait by Wells and her colleagues, the variance by site was significant For example,

just focusing on neglect and the most common category reported. lack of supe~ision,

one site screened in those reports 100% of the time yet another site accepted them only

11% ofthe time; similar findings were evident in other sites where failurn to provide was

screened in 70% of the time in one area but only 5% of those referrals accepted at

another site (Wells et af.. 1 99 1).

Finally, when Web and associates (1989a) examined the influence of socio-

demographic variables they found percent change in the population, the size of the

populatrbn, the number of families with children under 18 years, and median household

income, negatively associated with screening-in rates. This means when the percent

change in the population was high or when the site had more families with children

under 7% yeam old - the odds of being screened in decreased. Their interpretation of

the overall findings was that some agencies were clearly not investigating valid

maltreatment referrals. Also, screening practices varied considerably by site and

reflected, "...the contacts encountered by the agencies represent those that the

community of reporters has teamed that CPS will accept In other words, the community

knows from previous experience which cases are more likely to be acceptedn (Wells et

al., l989a, p. 22).

Giovanonni's (1 989,199 1) inquiry into unsubstantiated reports of child maltreatment

included a survey of 11 50 maltreatment reports from nine CPS agencies in three states

and interviews with 7 1 7 CPS workers on their views on unsubstantiated reports.

RuraVurban differences by maltreatment types were included in the analysis.

Giovanonni (1991) found when '...maltreatment was verified but CPS intenrention was

not wananted" (p. 601, nearly 75% of the workers interviewed indicated over 60% of

those cases not serviced by CPS needed concrete services such as food, housing,

money, health care and legal assistance. Since it is these aforementioned "concrete'

services that are most often required in neglect referrals, the reasonable conclusion is

that neglect referrals are frequently screened out. Another important finding was that

workers identified that agency policy and lack of community and agency resources were

the most common barriers to secvicing neglect cases. Additional analysis of reports

found urban sites had a higher proportion of physical and sexual abuse cases, which

were viewed as 'more serious" and had higher substantiation rates. The rural areas,

which were found to have more neglect type cases, viewed neglect cases as "less

serious', and, as found with previous studies, these cases had lower substantiation

reports. Furthermore, substantiation of neglect only improved if both lack of supervision

and failurn to provide were found together (Giovanonni, 1989). Similar to Wells, Fluke,

Downing and Brown's (1 989a) study. Giovanonni. (1 989) finds the "severity" benchmark

and evidential requirements are higher for neglect referrals than for abuse reports.

Combined, these findings raise the question of resource availability as an important

factor to consider in neglect

The sampling limitations with the studies by Wells and her colleagues (1989a, 1991) and

Giovanonni (1989,1991), preclude generalizability. However, due to the level of the

examination. the similarity in the conclusions regarding the importance of context, and

the limited number of studies examining context, weight is given to the findings.

fellman's (1992) national survey of 1196 HSP fram 15 states also examined urbanlrural

differences and SES variation. While SES was a significant predictor of vignette

outcomes, what is interesting is the higher status families were judged more severely on

the neglect vignettes and lower status families were viewed more unfavorably with

physicaI and sexual abuse. One rationale suggested was that somehow '. . .middleclass

parents are seen as behaving in a way that is more violative of expectations of them,

and thus behaviour is rated as more serious" (Zellman, 1992, p.70). If this is true, this

too has implications for understanding rating differences by profession, by maltreatment

type and by location.

Context is very important in understanding worker decisions. Review of the literature on

ecological studies on child maltreatment finds further support for the argument that in

addition to high poverty, a community's socioeconomic characteristics such as

unemployment rate, vacant houses and residential transience, are associated with

neglect (Drake & Pandey, 7996; Garbarino & Crouter, 1978; Spearly & Lauderdale,

I 982; Steinberg, Catalano & Dooley, 1 981 ; Zuravin, 1989).

Therefore, given the strong link between neglect and sitefSES variance, SES will be one

of the independent variables examined in the secondary analysis. The six regions will be

divided into three SES groups: low SES group (indudes the most disadvantaged regions

as determined by SES indicators), medbrn SES group, and high SES group (includes

the most advantaged regions as determined by SES indicators). A two-way test of

significance is employed and the alpha revel is pc.05.

The rationale behind the selection of a nondirectional hypothesis needs to be noted. In

testing AL theory and examining the SES variable, if the original survey and secondary

analysis had been conducted in the United States, then given the theoretical and

empirical evidence to date (Wolock,l982), coupled with the more extreme SES

polarization in America, a directional hypothesis is appropriate. where the prediction is

that the low SES group will be less inclusive than the high SES group. However, a two-

tailed test is used for the analysis of the secondary Ontario survey data because the unit

of analysis is very large (region), the Canadian SES context differs considerably from

the United States, and the empirical evidence is still emerging regarding the impact of

the SES variable on neglect As Zuravin (1 989) points out, '...despite many studies of

co-variation between community characteristics and child maltreatment rates, the

knowledge base is sparse" (p. 1 06).

In identifying the importance of context as a key variable in understanding HSP

perceptions of neglect entry points, the next question is what occurs within a particular

context to set or further entrench perceptions? A number of studies found exposure a

key component in influencing perceptions and attitudes. For example, Campbell, Hunt

and Lewis (1958) found in a series of vocabulary definitions provided to judges that the

judges who had been primarily exposed to 'high pathology" definitions rated mid-scale

items as indicating "less eccentricity" than judges who had been predominantly exposed

to "low pathology' definitions. Manis (1971) in his examination of the social

consequences of context efbcts, conducted four experiments with subjects who had to

either describe or decode descriptions o f emotions. He found real difficulties in

communication occurred because persons exposed to one antex€ gave verbal

descriptions of stimuli different from descriptions provided by persons exposed to

di irent contextual stimuli-

With regards to testing agency exposure to neglect, since profession has been

established as the best variable to account for worker characteristic rating differences,

the secondary analysis will investigate the influence of profession on neglect eligibility by

comparing agency type, specifically, investigatory to noninvestigatory agencies.

Support for this variable is found in Zellman and Falleh (1 996) comparison of the I986

American National Incidence Study (NIS) findings with 1980 data; they found that

noninvestigatory agencies, such as schools, hospitals and community and social service

agencies, reported significantly more child maltreatment reports compared to

investigatory agencies, such as police and CPS workers. Again, if the primary study and

the secondary analysis took place in the United States, given AL theory and the

empirical findings to date from American data, a directional hypothesis would be

selected. Again, a nondirectional hypothesis is used because the Canadian context

differs considerably from the American context, no Canadian data exists regarding HSP

perceptions of neglect entry points by agency type, and there are ambiguities with some

of the American data. As Zellman and Faller (1986) note, '...the meaning of the

differences in these proportions across agency types is not entirely clear" (p.364).

It was the intent in the secondary anaIysis to also examine the influence of individual

exposure to neglect by examining a CPS workers' yean of experience with CPS

cases. The rationale behind examining just CPS workers' experience with CPS cases

and not all HSP in the sample is because the issue to be examined is accumulated

exposure to and experience with CPS cases. This can best be tested with CPS workers.

It is generally accepted at the practice level that CPS workers with less than two years

experience are not fully trained child welfare workers. Using 'practke judgment' as the

criteria for distinguishing experienced CPS workers from non-axperienced workers, then

for the purpose of the analysis, 'less experienced" included CPS workers with two years

or less experience with CPS cases; "more experienced" included CPS workers with

more than three yean experience with CPS cases. Theoretical and substantive

evidence (Trocme et al., 1994) supports the examination of this independent variable in

the secondary analysis. However, while the sample size for the "more experienced"

CPS workers was ninety-nine (n=99), the sample size for the 'less experienced" CPS -

workers was only seven @I=?). There was not enough "less experienced" workers in the

sample to do the analysis.

Research Hypotheses

The research hypotheses are:

Hypothesis 1 - Regional Socioeconomic (SES) Group

HI - Human Service Professionals (HSP) in the three SES groups (low,

medium and high) will perceive the minimum CPS entry point differently

for the nine forms of neglect

Hlo - There will be no difference in where HSP in the three SES groups

perceive the minimum CPS entry point for the nine forms of neglect.

Hypothesis 2 - Agency Type

H2 - Investigatory agencies (CPS & police) will perceive the minimum CPS

entry point differently for the nine forms of neglect than HSP in non-

investigatory agencies (all other HSP).

HZ0 - There be no difference where HSP from investigatory agencies and

non-investigatory agencies perceive the minimum CPS entry point for the

nine €oms of neglect

The alpha level is p (-05 and based on the rationale previously noted, two-tailed tests of

significance are used.

Summary

This examination has taken a broad brush approach to understanding the contextual

facton, intellectual traditions, philosophical assumptions, social construction, political

elements, and debates that have influenced child maltreatment epistemology, and more

specifically, child neglect knowledge and research. Overall, the weaving of the broad

tapestry of social work research knowledge, in particular, child maltreatment

epistemology, has not been even in its breadth or its brilliance. The one dimensional

focus on the individual is limiting when there is a need for a multi-causal, ecological

approach. Poor use and little use of theory assessment has hindered the extent to which

research knowledge guides social work practice.

Since child neglect is a socially constructed phenomenon, the issues surrounding the

definition of neglect are essentially value issues (Giovannoni et al., 1979). For example

one issue that creates considerable conflict for professionals, the public and society, is

whether preserving the family is more important than protecting the child? Societal

values, both conflicting and agreed upon, are played out in all areas in the child

maltreatment field; for in the end, clinical action, policy and research are as much a way

of thinking as a set of methods.

This secondary study attempts to address a number of areas that have been identified in

the literature as either knowledge gaps or issues that have hindered knowledge

development about the etiology of child neglect

First, in using adaptation-level (AL) theory to guide thin king about how contextual factors

influence HSP decisions, this secondary research addresses the need to include

theoretical testing in studying a problem. Also, the need for theoretical refinement is

addressed by replicating Wolock's (1 982) test of AL theory.

Second, despite growing literature identifying context (social, economic, community and

professional facton) as an important element in child maltreatment, North American

social work knowledge development has had a narrow focus - understanding the

individual. Societal beliefs and values regarding the responsibilities of the individual, and

the primacy of the individual over community, influence the type of knowledge developed

and accepted by the public, practice and policy levels. Whether borrowed from other

professions or generated within social work, knowledge on understanding individual

factors as the cause of child maltreatment has meant examination of environmental and

group behaviour as possible influences for understanding chiid neglect, is subjugated in

importance. Therefore, this secondary study tries to address the need to examine more

complex models of child maltreatment and etiological theories by using a theory that

allows context to be examined. The specific contextual variables tested as to whether

they influence HSP perceptions of entry points, are: regional socioeconomic variance

and agency type.

Third, it is crucial to know HSP views on CPS eligibility because protection and sewices

to children can only be provided when child maltreatment is reported. Yet, knowledge is

far from definitive about how, why or where diierent HSP groups perceive CPS entry

points. Neither the child welfare laws nor practice wisdom are explicit in stating what the

minimum level of unacceptable care of a child is to warrant child protection services.

This secondary study is an attempt to try and begin to address the knowledge gap

regarding where are the HSP perceptions of the CPS entry points.

Fourth, despite neglect being the most prevalent form of ham to children and having

profound health consequences for children, the preponderance of maltreatment studies

to date have focused on child abuse. Research has largely overlooked child neglect.

Nine forms of neglect are the focus of the secondary examination.

Fifth, the issue that confounds the majority of maltreatment studies is the absence of

standardized definitions. Even though standard definitions logically precede discussion

of incidence, etiology, treatment, research, and policy, as well as the issue to be

investigated, HSP perceptions of the minimum CPS entry point decisions for neglect

This secondary research attempts to address the need for clearer, standardized neglect

definitions by analyzing survey data that used a survey tool that was Ontario based and

designed to assist in CPS eligibility determination. Additionally, the tool has good face

and content validity. A firrther benefit in analyzing data colIected from a tool that is being

standardized is that replication of the original survey can occur, so that future secondary

research can compare findings to this secondary study.

Sixth, while challenging the residual approach and the univariate model of child

maltreafment continues to falI to research and academic arenas. there is a need for

collaborative prackes between the field and research. This secondary study

demonstrates that the practice field can be a full and contributing partner in the area of

child welfare research, in that the field collected the original survey data. As well, the

research sector can be of benefit to the practice community. For example, the original

survey data were generated by individual CPS agencies for their own use, in that the

CPS agency wanted to do their own analysis andlor wanted to benefit from the process

of conducting the survey with their community. While there was no attempt or ability by

the CPS agencies to aggregate the six regions' survey data, the CPS agencies willingly

shared their survey data with this researcher so that secondary analysis of all six regions

survey data could occur.

Finally, this secondary study attempts to address the ovenhelming need for Canadian

child maltreatment research. In order to try and understand child abuse and neglect

etiology as it exists within the Canadian context and experience, research is needed at

the national, provincial and local levels. This thesis is an example of taking primary

survey research done at the local level, and then elevating the examination to the

provincial level by aggregating the local data and conducting secondary analysis.

CHAPTER 3: ORIGINAL SURVEY - DESIGN & METHODOLOGY

Suwey research is probably the best method available to the social sckntist interested irt collecting orfginal data

for describing a populatton toe large to obsewe directly. A Rubin & E Babbie, 1997, p.346

Since the secondary analysis is vulnerable to all limitations associated with secondary

data, it is imperative to understand the strengths and limitations inherent in the data

generated from the original survey design and methodology. This chapter presents the

original survey study definition and research design, the process involved in pretesting

and pilot testing the survey forms, and the sampling and data collection methodology

used in the primary research. The sampling section details the agency and respondent

sample selection methods. Sampling bias and potential biases are noted.

Definition of Original Suwey Study

The rationale for presenting six individual survey data sets as one, called the "original

sunrey study", is because all six CPS agencies were Ontario based, all conducted the

same extensive community survey, all used the same survey tool, all asked the same

survey question, all used the same unit of analysis, all collected non-identifying

respondent information, all completed the survey within the same three year period, and

all were willing to submit their original data for secondary analysis.

Ovenriew of Original Suwey Design

The original survey data came from a convenience sample of six individual CPS

agencies, that had, independent of one another, conducted a survey on their

communlies' perceptions of CPS entry points (community suwey). In particular, these

CPS agencies wanted to obtain their professionalsr views on CPS intervention

standards. The eligibility tool, the Spectnrm was modified for use as the survey tool.

A survey method was selected by the six individual CPS agencies as the most

appropriate method to collect CPS staff and community perceptions of CPS entry points,

because large numbers of community and CPS respondents were to be surveyed and

the survey tool (Spectrum) allowed perceptions of entry points to be easily analyzed.

The unit of analysis is the individual. Examples of CPS staff surveyed: front-line,

supervisors, senior management, executive director, and clerical. Illustrations of

community members surveyad: HSP, CPS foster parents, CPS volunteers. CPS and

community board members, media, general public, former foster children, and clerical

staff from community agencies. For the purposes of the secondary analysis onIy the

data from the CPS direct service staff and community HSP groups were analyzed.

All six CPS agencies (known in the convenience sample as Regions A, B, C, D, E and

F) used the same survey design. A face-to-face, group survey technique was assessed

by the CPS agencies to be the most reliable and cost-effective approach for collecb-ng

the original survey data. As well, this method had a high respondent completion and

return rate. Survey participation by the individual respondents was based on informed

consent and protection of confidentiality. Respondents were not asked to give their

names in the survey.

Also, all six CPS agencies used a modified version of the Spectrum as the survey tool

(see Table 2). All six CPS agencies examined all the CPS Reasons for Service. All six

CPS agencies asked the same survey question of the community respondent Where do

you perceive the minimum intervention line (Mlt) or entry point to be in your community

for each of the following 24 Reasons for Senrice? All six CPS agencies collected their

survey data between 1993-1 996.

Each of the six individual CPS boardlmanagement groups sanctioned the community

survey and approved both funding and time for the CPS agency and staff to do the

community survey. The sampling fmme was constructed by each CPS agency. Sunrey

participation by CPS staff was required, but for community respondents and associated

agencies it was more haphazard, as respondents self-selected. For details on the

specific characteristics of the six CPS agencies and their regions in the sample, see

Table 5. Rationale behind the determination of each region's SES status is provided

later in this thesis-

Through the combined experiences of the CPS agencies that had previously conducted

a community survey (Regions D, E and F), an array of survey materials had been

developed and refined. These materials were made available for Regions A, B and C to

use. Illustrations of the material are: community invitation letten, survey forms, training

material, lists of HSP groups to invite, a standardized training script, overheads, and

even flyers.

Table 5

RegionA Region B Region C Region D w o n E Region F

Ontario regional west east north west central central locale

population 59.065 129,089 161,210 65,268 732,798 323,236 1991 "

% of pop. change: 5.5% 12% 5.7% 10.9% 23.8% 9 5.4% T 986-3 991 *

Oh immigrant pop. 8.6% 13.6% 8.1% 8.0 36.7% 23.0%

tvpeof ama rural nrmUurban nrraVurban rural urbanfrural urban/mrat

culture English English English English multi- multi- French cultural cuftuml Native

size of CPS small medium large small large medium agency

regional SES high medium low high medium high status

Total # 55 163 f 45 104 284 124 surveyed (N=875)

# HSP surveyed 36 T I ? 128 87 214 97 (N679)

Note. * Census data, Statistics Canada, 2991

Description of Survey Procedures

As noted above, distribution of the original survey to the respondents in the six regions

was achieved through faca to face, group administered meetings in community settings.

It was the responsibility of the CPS agency to arrange the dates, times, venues and

RSVP arrangements for the community and CPS staff surveys.

The size of the group meetings ranged from 5 to 50 respondents at a time. CPS

agencies used both CAS office and community venues to conduct the survey. All six

CPS agencies designated one person as the projectiresearch liaison. The liaison's

responsibility included providing the Introduction/training part of the survey, attending all

CPS staff and community survey sessions, answering questions, monitoring the data

collection part of the survey process, and collecting the completed sunray score sheets

and Spectrum survey tool. Only one survey per respondent was accepted.

The group method was selected because a three hour time frame was required to

complete the entire survey. This included a 40 to 60 minute lntroductionltraining period

for survey respondents. The time variance was due to whether it was a CPS group

familiar with the Children's Aid Society (CAS) mandate, the purpose and language of

the tool or a community HSP group, where knowledge ranged widely regarding the CAS,

the Ontario child welfare legislation, and the Spectrum survey tool.

The introduction section included a brief review of the history of the development of the

Spectrum, details into the organiration, structure and intent of the tool, the purpose of

the survey, examples of the forms, an overview of the study, potential use of the data,

and the consent process. The Introduction section concluded with a training segment on

how to complete the survey and score the MIL or CPS entry points. Respondents were

encouraged to ask questions before the survey commenced. In addition to completing

the Spectrum survey, respondents from Regions A. 6 and C completed Individual HSP

Variables (see Appendix F 'a" and 'b' side). While survey participants in Regions D, E,

and F, also completed the Spectrum survey, only the information on the locale,

respondent designation (CPS, HSP, non-HSP), and profession of the respondent were

collected. It took approximately two hours to complete the actual sunrey. Upon

completion of the survey, respondents handed in their completed survey score sheet

and the survey tool.

The Introduction period allowed for the use of a standardized script, visuaI aids and

provided the survey respondents with the opportunity to ask questions or clarify any

issues regarding the survey. This method also allowed information regarding the

purpose of the survey and the details on how to complete the survey to be relayed in a

consistent manner. Compared to a mail or phone suwey method, the on-site group

method contributed to a higher completion and return rate. The group format also a

allowed the CPS agencies to monitor the survey proceedings so to ensure respondents

did not discuss or copy their responses.

Overview of Suwey Tool

As noted previously, the Spectrum is an Ontario developed, field-based, CPS eligibility

tool that is moving toward standardization. The I997 version of the Spectrum is now

implemented province-wide. As an eligibility instrument, the Spectrum is not theoretically

or empirically based, but it is grounded in the Ontario child welfare legislation and other

sources of authority, such as provincial ministry standards on investigation of child

maltreatment, the Ontario Association of Children Aid Societies (OACAS) accreditation

standards, and best field practice wisdom (see Appendix A).

As outlined in Chapter 1 - Eligibility Tool Design, the 1995 Research Spectrum has a

two-dimensional matrix design (see Appendix C). The horizontal row has four categories

of risklseverity - most, medium, least and no. The minimum CPS intervention line (MIL)

or CPS entry point is between the medium and least risklseverity descriptors. The

vertical column presents all 24 CPS protection or mandated Reasons for Service along

with five non-mandated Reasons for Service, such as adoption, request for counseling

and administrative services. For each Reason for Service, the child welfare legislation

precedes the scale for that Reason for Service category. There is at least one descriptor

for each of the four risldseverity categories.

As identified in Chapter 1 - Survey Tool Design, the eligibility instrument was modified

for use as a survey tool. The Spectrum's risklseverity categories were removed to

eliminate any possible bias or suggestion to the respondents where the MIL is. (To view

the modified neglec€ sections, see Appendix B).

The purpose of the community study was to target the prof8ssionals' views on eligibility

determination. Each respondent was to indicate where they perceived the MIL to be for

their community for each Reason for Service. The modified Spectrum was the survey

tool used to capture those perceptions of CPS entry points. In summarizing the steps to

complete the survey, respondents first referenced the CFSA and read the accompanying

protection scale and then placed each descriptor number under one of the four

ris klseverity headings (most/ mediu rnileastlno) on the Survey Score Sheet (refer to

Appendix F 'bn side). Each HSP survey respondent identified Wo types of cases. The

first type were the mandated CPS situations; these included all descriptors the

respondent perceived as falling under the most and medium riskheverity categories.

The second type of cases were the non-CPS cases or community cases, and included

all descriptors that fell into the least and no risklseverity categories. For more specific

details on survey tool design and use, refer to Chapter 1 - Survey instrument section.

Benefit to CPS Agencies in Using Survey Tool

The CPS agencies that conducted a community survey were able, for the first time, to

assess CPS entry points. As wel, the process of conducting the survey appeared to

enhance community and CPS relations. Additional benefits to CPS agencies in using the

Spectrum as a survey too[ were the language in the Spectrum was familiar to the CPS

respondents and OACAS authorized the we of the copyrighted Spectrum as the survey

fool. Furthermore, previous use of the Spectrum as a survey instrument found the group

format method had a completion rate of over ninety percent

Despite the actual and perceived survey benefits, the ovewhelmingly positive feedback

from the agencies and communities that have completed a community survey, and the

fact that by 1997, over half of the CPS agencies in the province were implementing the

Spectrum as an eligibility tool, onty a few CPS agencies in Ontario have elected to use

the Spectrum to survey their community regarding perceptions of entry points. The

surveys, whether limited or extensive, have only been done in conjunction with CPS

agency implementation of the Spectrum. While a number of CPS agencies saw the

community survey as valuable research endeavor, the associated costs and time

required to do the survey, were prohibitive for many agencies.

Pretest & Pilot Test of Respondent Infomation Form: Regions A, B & C

Respondents from all six regions completed the same Spectrum survey score sheet.

Regions D, E and F completed their community surveys prior to Regions A, B and C.

Regions D, E and F collected minimal respondent information: region, CPS agency, and

profession (e.g. CPS worker, CPS manager, community HSP worker, community HSP

manager).

Regions A, B and C however. asked their survey respondents extensive professional

and personal information (see Appendix F 'a" side). As part of my duties as Research

Assistant for the Spectrum Research Study and the secondary analysis researcher, I

developed the respondent information fom. On the CPS agencies behalf. I conducted a

pretest and pilot test of the form and entire survey process prior to the CPS agencies in

the above regions commencing their formal survey.

The pretest occurred in late 1995 and involved a small sample of HSP respondents (two

community social workers, two early childcare educators, one CPS staff person), who

were not HSP respondents in any of the six sample regions. The purpose was to pretest

the 1995 survey tool along with the profassional questions that accompanied the survey

and establish the length of time it took to complete the survey with the new questions

added. Based on the feedback from the pretest group, the professional questions were

revised and a pilot test of the complete survey process, including the Introductory

section, was done in March 1996. The pilot test involved a branch of a Toronto CPS

agency; the agency was not part of the original survey data. A total of six CPS social

work staff and eleven community HSP (two teachers, two police, one lawyer, two

religious counsellors, two community social workers and two other HSP were surveyed.

Based on analysis from the pilot test, the professional questions for the survey

instrument were again refined and the survey instrument finaliied for use by Regions A,

B, and C.

Information collected on the participants from Region A. B and C included respondent3

survey location, local CPS, profession, place of work (e-g. hospital, CPS agency.

community, court), position, last professional degree attained, highest level of education,

number of years living in the community, number of years of experience in the

respondent3 chosen field, number of years of working inkith the child welfare field.

respondent3 assessment of their knowledge' of the child welfare legislation, estimated

number of actual cases where child abuse andlor neglect was involved, gender, age,

marital and parental status, annuaal income and finally, clientele SES group. All

respondents from Region At B and C completed consent forms (see Appendix H and I).

Sampling Issues - Agency Selection

Each of the six CPS agencies constructed their own sampling frame for agency

selection. While the overall goal was to encourage participation by HSP groups, each of

the six individual CPS agencies defined for themselves, which community groups to

invite. Both HSP and non-HSP groups were invited to attend the Spectrum rating and

each agency or group wold send as many staff as they wanted to the community survey

(see Appendix K). Although no master list exists noting all HSP groups in each region,

key community HSP agencies were targeted by the CPS agencies to attend the

Spectrum survey. Assisted by the experience from the CPS agencies in Regions D, E.

and F, the CPS agencies Region A, B and C were provided with a suggested list of

CPS staff and community HSP groups to invite (see Appendix D and E).

A btaI of 875 Spectrum survey forms were completed by respondents from the six CPS

regions. The breakdown is follows: Region A (n=55), Region 0 (n=163), Region C

(n=145), Region O (n=f 04). Region E (n=284), Region E (n=124). Only 2% of the survey

forms had critIcaI data missing, such as job designation or the survey form was

incomplete (n =f 7). Non-HSP groups that attended the survey were clerical staff, foster

parents, board members, and students (n = 155), as well as the general public and

media groups (n = 24). HSP groups (n= 679) were responsible for 79% of the total

surveys, and groups included CPS, police, teachers, and various other community HSP

agencies, such as hospital sock1 workers. See Tabla 6 for breakdown by region on all

original survey forms collected.

Table 6

Breakdown of Survev Forms Collected bv Reaion

REGIONS HSP A 8 D E F TOTAL

CPS 19 34 53 21 85 42 254

CHSP 17 83 75 66 1 29 55 425

nonHSP 18 37 I 6 I 5 67 26 179

Missing data 1 9 4 2 3 I 17

TOTAL 55 163 145 104 284 1 24 875

The remainder of the discussion on the original survey design and methodology will

focus on analyzing the strengths and limitations associated with the original HSP survey

data. Rationale for this decision is based on the fact that the six CPS agencies targeted

HSP groups to participate in the survey, plus the focus of the CPS agencies' original

analyses was the professionals' perceptions of CPS. As well, the secondary study

analyzes only the HSP respondents' perceptions of the MIL

Sampling lssues - Respondent Selection

The unit of analysis in the original survey was the individual. HSP were targeted for

participation in the survey. The intent of the six sample agencies was to be inclusive and

broad, not exclusive in the invitations to community HSP (refer to Appendix K). While

there is no master list of all HSP in each region and some HSP individuals were unique

to particular regions (a-g. Native elders), generally, HSP who worked with children

andlot families were targeted to attend the surveys.

The final sample of 679 HSP participants included CPS direct service and management

staff (n=254) and community HSP (n= 425). All CPS groups targeted for the survey by

the CPS agencies participated (see Appendix D). The breakdown for the total original

sample of CPS respondents: CPS direct service staff (n = 197), CPS supervisors and

senior managers (n = 56) and CPS executive directors (n = I), for a total CPS cohort

sample from six regions of n = 254. The community HSP sample included for example,

police, teachers, family doctorslpediatricians. nurses, day care personnel, and

counsellors in child and family clinics.

The list of community HSP the CPS agencies targeted for the survey was extensive

(see Appendix E). Almost all community HSP groups targeted did participate.

Illustrations of such HSP groups participating in the original survey are: social workers

who work in community settings such as hospitals, farnib clinics, counseling agencies

(n=65), managers and supervisors of community agencies (n=102), community HSP

who dassified themselves as therapists, mediators or counsellors (35), religious

counsellors (n=3), caseworkers/caseaides (n=66), a doctor (n=i), nurses (n=9),

medical-other includes mid-wives, emergency response team members and nursing

assistants (n=5), lawyers (n=i I), a judge (n=l), HSP who work in the juvenile or adult

correction field (n=9), police (n=16), teachers, vice-principals and principals (n=30),

preschool, nursery school, day care and child home-care staff (n=65). See Table 7 for

the complete breakdown of all HSP groups in the original survey data by region.

Table 7

HSP Grou~s bv Reaion

REGIONS HSP A B C D E F Total CPS-HSP Direct Service 1 5 Management 4 7 14 4 25

SociaC worker Management Therapist Religious Counsellor

Caseworker Physician N u m Med. Other Lawyer Judge Corrections Police Teacher PreschooE HSP-Other

Overall

Note- 'CPS- direct - incfudes aIE those staff in intake, famiiy ,children service positions and other direct

setvice positions 'Medicat -Other examples are Emergency Response HSP, midwives and nursing assistants 'HSFWhel are identified only as Community HSP

Overcoverage of CPS Respondents

In the design of the original survey, it was known that sampling bias would occur in the

form of overcoverage of CPS respondents. One reason was because the CPS agencies

had the responsibility for arranging the survey sessions, which created a sort of 'home-

court advantage" and the likelihood of a greater turnout by CPS respondents. Another

reason was the survey topic - perceptions of child protection intervention - which

attracted CPS staff to participate. Furthermore, the directive from the six host CPS

agencies boardlmanagement group was for an 'all hands" participation in the Spectrum

survey. That sanction was reflected when the actual number of CPS staff at the time of

the survey (only direct service staff, supervisors, senior management, and executive

director complement) are compared to the Spectrum survey CPS participation rate.

Impressive by most survey response expectations, the CPS staff participation ranged

from a low of 70.8% in Region B to a high of 87.6% in Region E; see Table 8.

Table 8

Number of Direct SewicelManaaement CPS Staff to Number of CPS Suwev Respondents: C PS Particbation Rate

Rwion A Reahon 8 Reaion C Reaion D Rwion E Reaion F # CPS 23 48 70 25 97 50 Staff

#r CPS 19 34 53 2 T 85 42 Survey Respondents

Participation 82.6% 70.8% 75.7% 04% 87-6s 84% Rate

-- - - - - --

Note. - CPS staff onIy refer to front-fine workers, supetvisors, management and executive director.

The response rate for community HSP was not estimated by the CPS agencies in the

original survey design because the actual ratio of CPS staff to community HSP by region

was not known. However, based on participant breakdown from previous Spectrum

surveys and agencies' time and budget available for data collection, the planned

sample size was to minimaIIy achieve a ratio of 40% CPS to 60% community HSP

(CHSP) respondents for each region. See Table 9 for ratio of CPS to community HSP

respondent by region.

Table 9

Ratio of CPS Res~ondent to Communitv HSPICHSP) Res~ondent bv Recrion

HSP Region A Reuion B Recrion C Reaion D Reaion E Reqion F Total # CPS 29 34 53 21 85 42 254 % 53% 29% 41 % 24% 40% 43% 37%

# CffSP 27 83 75 66 129 55 425 % 47% 74 % 59% 76% 60% 57% 63%

Total # 36 117 128 87 214 97 679

Three individual regions (Region C, E and F), as well as all six regions combined,

achieved the approximate 40=CPS:606HSP split. Two regions surpassed the minimum

ratio with a 30:70 split Region 6 and Region D. Region A's sample ratio was reversed

with a 55-CPS:45-CHSP split The minimum CPS:CHSP ratio was achieved in five of

the six regions. As noted earlier, the original survey data set consist of a total of 875

surveys collected by the six CPS agencies and the survey compLetion rate was 98%.

Each agency completed their own independent analysis on their respective data.

Undercoverage of Community HSP Respondents

The haphazard, self-selection method for sampling community HSP respondents

resulted in undercoverage of community HSP groups, such as police, teachers, doctors

and nurses. In sum, self-selection by the community HSP occurred at the individual and

group level, resulting in sampling bias and undercoverage with some key HSP groups. A

potential bias with the community respondents, is they may have sel'seleded because

they have an overly positive or negative view of the CPS agency. There is the

possibility that the community HSP sample is composed of suwey participants with

extreme views of the CPS agency, and these views may affect how they determined

eligibility.

While undercoverage of key community HSP respondents did occur in the original

survey data, the undercoveraga did not occur because of lack of effort by the CPS

agencies to elicit survey attendance by groups, such as teachers and police. For

example, to encourage participation from both key groups and key individual HSP,

Region 0 had invited all three police departments: provincial, municipal and military;

Region A offered to provide individual survey sessions just for school teachers at a time

and locale of their convenience. Additionally, to try and reduce sampling bias, a number

of options were used by the CPS agencies. One approach was through the letters of

invitation that were sent to key community HSP groups (see Appendix K). For example,

some agencies, like child and family agencies, court clinics or social services, received

general invitations. Where possible, the CPS agencies directly invited individuals, such

as psychologists, therapists, family court lawyers, famiiy dodon and pediatricians. Other

community agencies received a general invitation, along with invitations to specific

individuals at the agency. One example, was the Chief of Pediatrics at the community

hospital received a specific invitation.

Wrth the key HSP groups (police and teachers), protocol dictated that the invitation be

sent to the chief of police or school board. This fact may have adversely affected the

attendance of key individuals from those groups that the CPS agency wanted to target,

such as front-line youth bureau police officers or elementary and high school teachers

and principals.

A potential bias identified in the design stage of the original surveys of Regions A, B,

and C, was that the community HSP sample might be restricted to agencies above a

certain threshold size. Based on the rationale that both police and schools are large

community service providers, key CPS referral sources, and together, account for 40

percent of CPS referrals (Garbe, 1998), it was predicted that more participants would be

sent ftom the law enforcement and educational settings. Despite the formal invitations

and creative informal networking by the individual CPS agency agencies, with offers to

arrange times or survey sites convenient to these key HSP groups, the teachen and

police remained significantly underrepresented in the original study.

Data Collection Methods

AII data were collected between 1993 to 1996. The same Spectrum survey form was

used for all six agencies (refer to Appendix F 'b" side). Previous mention has been

made regarding the host CPS agency arranging the multi-site community based survey

settings. Identical training overheads, information handouts, consent and sunrey forms

were used by Regions A, B and C. These materials were based on the previous

Spectrum survey forms and handouts from Region D, E and F surveys. Additionally, the

project liaisons from Regions A, 6 and C were trained by this researcher on the use of

the forms and method of survey presentation, as well as provided with a standard script

and format of presentation for the infmation/training period preceding the survey.

Duplication

Duplication could occur if the same HSP completed the survey for different regions or

completed more than one survey for the same region. Due to respondent anonymity it is

impossible to ascertain whet her duplication of the original survey forms occurred.

However, the potential for multiplicity was addressed in the Introductory phase of the

survey when participants were asked if they had already completed such a survey.

Additionally, duplication was not assessed as serious problem as the sunrey's three hour

time requirement significantly diminishes the likelihood of this bias occurring.

Summary

Assessment of the quality of the data from the original study is based on the overall

source of error from the elements included in the total survey design. These elements

include the original research design, the size and representativeness of the sample, the

techniques used for collecting the data, and viewed as the major source of error in

survey estimates - the design of the survey questions (Czaja & Blair. 1996).

Essentially, a good research question has to relate to the measure used. In the original

survey it is: Where is do you [the HSPJ perceive the minimum CPS intervention line to

be in your community for this [neglect] Reason for Service? The quality of the data is

dependent on the quality of the instrument used to answer the research question. The

survey tool that was used is moving toward standardization. Furthermore, it is unique in

that it is designed, for Ontario use, as an eligibility determination tool, it is accompanied

by the legislation it is grounded in, and the tool provides a range of CPS situations from

extreme severe to that of a healthy family.

It is the range of examples or descriptors that gives the HSP a more complete picture of

the maltreatment situation. It is this 'bigger picture" that then allows the HSP to more

accurately indicate where they perceive the CPS minimum interwention line (MIL) to be.

The strength of the original survey study is the combination of a tool with strong face

validity and a good question coupled with a large interregional sample of HSP.

Limitations in the overall study clearly exist, but fw the most part lie with the

methodology. This is an important point when evaluating the worth of the original

research, for Fowler (I 998) notes:

Thus, ftom the perspective of total survey design, investing in the design and evaluation of questions is a best b y , one of the endeavors that is most likely to yield results in the fonn of better, more emf-ftee data (p.343).

CHAPTER 4: SECONDARY ANALYSIS - DESIGN & METHODOLOGY

Secondary analysis k a special case of existing statistics; it is the reanalysis of previously colkcted sunray data or other information

W. Lawrence Neuman, 1981, p.280.

This chapter outlines the design and focus of the secondary analysis. Methodological

tasks included analyzing the ability of the survey tool to capture individual, group and

regional differences. Also, the six CPS agencies in the regional convenience sample are

assessed on an number of dimensions, and secondary data problems associated with

the regional, agency and respondent sampling, are detailed. The methodology for

dividing the six regions into three socioeconomic (SES) groups is outlined. As well, the

methods used to code the original data and stabilize the secondary data are presented.

The nonparametric tests and data analysis methods are also reported in this chapter.

Secondary Analysis Study Design

The goal of the secondary analysis was to aggregate the original survey data from a

convenience sample of six CPS regions and analyze whether regional SES variance or

agency type affected 679 Ontario HSP perceptions of the MIL for nine forms of neglect.

The original survey data and secondary analysis used the same unit of analysis - the

individual. Also, the research question in the secondary analysis was - Where did you

(the HSP) perceive the MIL for nine forms of neglect? is directly related to the research

question in the primary survey - Where do you (the sunrey respondent) perceive the MIL

for 24 CPS Reasons for Service in your community? Hence, the secondary data are

appropriate for the research question.

An explanatory approach was taken and Adaptation-level (AL) theory used to guide

thinking about how contextual factors influence professionals' child maltreatment

decisions plus a test of AL theory allowed Wolock's (I 982) test to be replicated.

Ability of Suwey Tool to Differentiate MIL: Individual, Agency & Region Level

A critical issue concerning the quality of the secondary data was whether the survey tool

could accurately capture and delineate between different MIL at the individual, group

and regional level. The survey instrument's sensitivity to individual and intrarregional

group differences was established through examination of the analyses from both the

Spectrum Research Study expert panel and the already completed community survey

forms from another CPS agency not involved in the original survey study. Differences

were found in individual and HSP community groups identification of the MIL.

Establishing interregional variance was accomplished by examining the set MIL for each

of the 25 Reasons for Service from the eight of the nine CPS agencies using the

Spectrum. The set MIL is the actual MIL or entry point the CPS agency uses in

determining eligibility in the actual practice setting. The set MIL may or may not differ

from where a HSP or community member perceives or wants the MIL to be.

In 1995, use of the Spectrum was voluntary and Ontario CPS agencies set their own

intervention lines. In setting the MIL some CPS agencies used a community process

(eg. Peel. Hafton, Simcoe), some set the MIL soleIy through an internal process (e.g.

Samia, Haldimand-Norfolk. Muskoka), and some, h adopting the tool from another

CPS agency, simply applied the other agency's set MIL (e-g. York. Perth. St. Thomas-

Elgin). Since the set MIL is a matter of public record, the following eight CPS agencies

set MIL were analyzed:

HALDIMAND/ NORFOLK HALTON MUSKOKA PEEL PERTH SARNlA SIMCOE ST. THOMAS I ELGIN

All eight Spectrum's had the same 25 Reasons for Service. Two agencies added

unique sections which were not included in the analysis; Muskoka had a separate

section on Housing and Haldimand had a unique section on Child Sexual Play. Broad

analysis found that for 12 of the 25 Reasons for Service all eight CPS agencies had the

same set MIL. One level difference existed in the MIL for 9 of the 25 Reasons for

Service. Two level differences were evident in 4 of the 25 Reasons for Service. See

Table 10.

Table 10

Level of Agreement 25 Reasons fbr Sewice % of Agmement All agree - . 12 Reasons for Service - .) 48%

Wthin 1 Level 1 - 9 Reasons for Service - - 36%

Wrttrin 2 Levers - - 4 Reasons for Se"ce - - 16% - OveralC 25 Reasons for Secvice - 900% -

Contrary to the many field, community and public perceptions that variance among CPS

agencies entry points abounds, the analysis found a great deal of similarity in the

agencies' set MIL For example, in nearly half of the 25 Reasons for Service all eight

agencies had the same set MIL. When one level differences were taken into account,

the agencies agreed on the set MIL in 84% of the Reasons for Service. Furthermore, in

eight of the nine Reasons for Service with one level variance, it tended to be only one or

hnro agencies that differad. Not surprisingly, the four sections viewed as the most

vaguely written in the CFSA had two levels of variance: 1)Emotional Abuse,

2)Unattended Mental, Emotional and Developmental Condition, 3)ParentlChild Conflict,

and 4)Child Misconduct.

A more detailed analysis identified which agencies differed on which Reasons for

Service. Since the focus of the secondary analysis is neglect, only the nine Spectrum

sections that relate to neglect will be highlighted. Closer examination of the neglect

sections found that they accounted for seven of the nine (78%) Reasons for Service with

one level difference. While the explanation for the variance at the agencyfregional level

was not evident, the analysis confirmed that CPS agencies did have different MIL for

neglect and that the survey tool (the Spectrum) could capture and delineate interregional

differences in the MIL for the neglect. See TaMe 11.

Table 11

Similarities & Differences in Set Nealect MIL for 8 CPS Aaencies

Spectnrm Section # Agencies with # Agencies with MIL Ssme MlL Dii%rent MIL mom incfusiim or /ess indusive c

la- Abandonment 8 0 no differences in MIL

Ila- Medical Attention 8 0 no differences in MIL

Vlla- Supervision 7 b- Child Care 7

VILla-Nutrition 6 b-Living Condition 6 c-H yg iene 5 d-Home Sanitation 7 *Clothing 7

more inclusive MIL > more inclusive MIL >

more inclusive MIL > more inclusive MIL > less inclusive MIL < less inclusive MIL c less inclusive MIL c

Sampling Issues - CPS Agencies & Regions

The secondary analysis aggregated the original survey data so that the regional

examination could occur, therefore regional sampling issues need to be noted. At the

regional level, all six sample areas met the most basic secondary analysis requirement - that the secondary data be Ontario based. As identified earlier, at the regional level this

was e convenience sample. A probability or random sample method was not feasible for

a number of reasons. Even though a number of the 55 CPS agencies in the province

were implementing the Spedrurn, most CPS agencies were either not interested in or

not able to conduct a community survey because of the considerable time and costs

assodated with the survey- Furthermore, only a fw CPS agencies did extensive

suoleys of their community. In order to have adequate sample sizes to conduct the

secondary analysis at the regional and agency type level, the CPS agencies had to

have adequate sample sizes (e-g. N > 50).

In sum, the regional criteria for the secondary analysis were that the CPS agency had

conducted an extensive community survey using the Spectrum, had adequate suwey

sample sires, had their original data, were willing to share their secondary data, and had

obtained either written or verbal consent from their survey respondents to use the data

for purposes of analysis. Consequently, the potential sample group of CPS agencies

and their regions was limited to a very small convenience sample of six CPS regions.

And although the regional convenience sample was not a provincially representative

sample, as metropolitan Toronto, relig ious-based CPS agencies and native agencies

were absent, by chance, the regionaI sample had variation on a number of criteria (see

Table 12.

Table 12

Reaional Convenience Sam~le Variation

Cn'tenii Region

Variance in Agency Size: SmalI (10 - 39 full time staff) - RegionA Region D Medium (40 - 79 full time staff) - Region B Region F Large (80 + full time s€a€f) - Region C Region E

Variance in GeographidRegional Location: East Region - West Region - North Region - Central Region -

Variance in Setting: R u ~ - RurallUrban Mix -

Cultural Variations English - English, French 8 Native - M ~ l t i ~ ~ l t ~ ~ l -

Region B RegionA Region D Region C Region E Region F

RegionA Region D Region 6 Region E RegionC Region F

Region A Region D Region B Region C Region E Region F

In addition to the limitation associated using a convenience sample for regional

sel6Cfi*on, further analysis finds that in using the secondary data, regional population sue

and agency size did not directly comelate in each situation. In particular, Region F.

which geographically is a relatively small region in central Ontario, has a population of

over 300,000 with a dense urban core but has a medium size CPS agency with 40 to 79

full time staff. While Region C, has a large CPS agency with over 80 staff that senre

English, French and Native sectors, but serves a population of less than 200.000 in

northern Ontario. Additionally, Region C includes an island whose population is Native

and eight HSP survey forms are included in the Region C secondary data set

As well, the relative proportion between each region's sample size and the regional

population size was not always reflected in the secondary data. For example, Region E.

representing 50% of the total sample population, had the largest data set (31%). While

Region A, the region with the smallest population in the total sample (4%) contributed

the smallest data set (5%). While the other four regions respondent sample contributions

were somewhat clumped together (Region B = 17%. Region C = 19%, Region D = 13%

end Region E = 14%) the relative proportion between the sample set contribution and

region's population was not equal; see Table 13.

Table 13

Relative Pro~ortion Between the Reaions' Po~ulation and the Suwev Sample Set

REGlONS Po~ulation A B C 0 E F Total $991 59,065 7 29,089 161,210 65,268 732,798 31 3,136 1,460,566

%ofthe 4% 9% 11% 4.5% 50% 21.5% 100% total sample population

% of €he total sample of HSP 5% 17% 19% 13% 31% 15% 100% resmndents

Note Source= 1993 Census of Canada, (q) 1993 OMan'o population was 9,840,000 (2) percentages Rave been rounded to the nearest whole number

Sampling issues - HSP Agencies

Referencing the sampling issues identified in the previous chapter, secondary analysis

of the data found additional limitations with the data. The intent in the originaI survey

was for each survey respondent, in particular the HSP participant, to clearly identify

their job designation, work locale and profession but considerable professional

information is absent and thus the secondary analysis is limited. For example, in order

to analyze the MIL by work locale and by other professional variables like job position,

then the secondary analysis must be able to distinguish whether a nurse worked in a

hospital setting, a community home care sewice, or was attached to a school.

Accordingly, to examine whether CPS positionlrole affects eligibility determination, the

secondary data have to identify whether the worker holds an key intake position or an

adoption-support job.

It was disappointing that most of the survey participants from Region D, E and F had to

be coded in the secondary analysis under the catch-all tern 'community HSP" or "CPS

- unknown" as specific titles were not requested on the suwey form; respondents were

asked to indicate whether they were community HSP or CPS staff. If the community

HSP specified they were a social worker they were coded accordingly. However, if they

indicated they were a community front-line worker but they did not indicate their

profession they were coded as a caseworkerfcaseaide. The lack of specificity with

respondents' job and profession categories ultimately limited analysis and relevance of

the study as data could only be dichotomized into CPS-staff and community HSP.

All CPS groups targeted for the survey by the CPS agencies participated (see Appendix

D). The breakdown of the secondary data of CPS respondents: CPS direct service staff

(n = 197). CPS supervisors and senior managers (n = 56) and CPS executive directors

(n = I), for a total CPS cohort sample from six regions of N = 254. Problems were

encountered in coding the forms by CPS job type. Surveys from the CPS staff in

Regions D, E and F usually indicated 'CAS-social worker" or 'CAS-staff. Whether the

respondent worked in an intake, family service or children's service position was not

noted and therefore unknown to the researcher and thus absent from secondary

analysis. Hence, the preponderance of CPS forms from those regions had to be coded

"CPS unknown". Also, in Region F, there is the possibility one CPS management

respondent is coded as a community HSP participant (a falsepositive); more

specifically, one of the 24 community HSP management in Region F may be a CPS

management respondent.

The list of community HSP the CPS agencies targeted in the original survey was

extensive (see Appendix E). Almost all community HSP groups targeted did participate.

The breakdown of the HSP groups in the secondary data are: social workers in

community settings, such as hospitals, family clinics, counseling agencies (n=65),

managers and supervisors of community agencies (n=.i02), community HSP who

classified themselves as therapists, mediators or counsellors (35), religious counsellors

(n=3), caseworkerslcaseaides (n=66), a doctor (n= t ), nurses (n=9), medical-other

includes midwives, emergency response team members and nursing assistants (n=5),

lawyers (n=ll), a judge (n=l), HSP who work in the juvenile or adult correction field

(n=9), police (n=16), teachers, vice-principals and principals (n=30), preschool, nursery

school, day care and child home-care staff (n=65) and finally, the HSP-other category

(n=65). Community HSP were placed in the 'HSPother" category when the respondent

only indicated they were a HSP in the community. Most of those participants coded

'HSPother" are respondents from Region D, E and F. Examples of HSP groups not

included in the secondary data set are psychiatrists and teaching assistants.

Another problem with the secondary data stems from limitations associated with the

haphazard sampling of the community HSP. This assessment is based on the fact that

in Ontario, professional referrals account for about 65% of all referrals to CPS (Trocme,

1991). Using Peel Children's Aid Society's (CAS) analysis of their referral sources from

1987-1996, community HSP referral breakdown is as follows: schools (34%), medical

(1 5%), police (9%) and all other social agencies combined (7%) (Garbe, 1998).

Using Peel CAS's community HSP referral percentages as the base to compare the

secondary data to, a representative sample of the four key community HSP groups

would be police (n=58), schools (n=223), medical (n=97) and all other social agencies

(n*7). However, analysis of the actual sample of the four key community HSP groups

shows a highly unrepresentative sample, with police (n =I 6), schools (n = 30), medical

(n = ?5) and agencies (n = 364). It is evident in the examination of the secondary

data that the community HSP sample of key HSP groups, is highly unrepresentative,

thus poses a limitation with the secondary analysis.

Methods to Stabilize Secondary Data

In order to stabilize the secondary data, two methods were employed, one direct and

one indirect The direct method was by having a large sample size (N=679). Chow

(1996) notes, 'A function of increasing the sample size is to stabilize the data" (p.122).

Another method that helped to stabilize the data was ensuring the CPS liaisons had

extensive training on the survey task before the data were collected.

Coding of Secondary Data

All 875 data f o n s received by the secondary study undewent a process of evaluation

in which they were assessed for confomity to the study's ordinal measurement. A code

book was created to classify the secondary data. Prior to data entry, all forms were

reviewed by an independent data entry person and this author. AlL 875 survey forms

were coded and the data entered into the computer. The data were analyzed using

Statistical Package for Social Sciences (SPSS).

Post data entry, in addition to the process of ensuring the data were clean, this

researcher randomly sampled every tenth form to ensure data were accurately coded

and data entered. However, as the secondary analysis is focused on examining only

HSP perceptions of €he MIL, the non-HSP surveys and data-missing forms were

excluded from the secondary study's sample data and analysis.

As noted previously, the survey has ordinal data and the respondent task was to rank

order the descriptors for each scale into four risklseverity categories: most I medium/

least i no. Determination can be made that a descriptor has a higher rank than another

descriptor but with ordinal data it is not possible to ascertain exactly how much higher

that descriptor is than another. The minimum CPS intervention line (MIL) is set between

medium risklseverity and least risWseverity descriptors. Each of the nine neglect

sections has a total of four to six individual descriptors. See Table 14 for a breakdown of

the number of descriptors for each of the nine forms of neglect studied.

Table 14

Number of Soectrum Descriptors bv Nealect Section

Neglect Section Number of Descriptors

ZA - Abandonment 2A - Lack of Medical Attention ?A - Lack of Supervision 78 - Inappropriate Child Care

Arrangements 8A - Lack of Nutrition 8B - Physical Living Conditions 8C - Personal Hygiene 80 - Household Sanitation 8E - Clothing

On the survey fon, respondents placed the descriptors they perceived as examples of

the CPS mandate under the most or medium risklseverity heading. Then, they marked

the descriptors they view as non-mandated CPS sluations under the least or no

riskheverity heading.

Coding each of the 679 €oms in SPSS entailed assigning all 'most" descriptors a '1"; all

"medium' descriptors a 'T; 'leasf' descriptors a '3" and; "nd descriptors a "4' (see

Table 4 for coding examples). Designation of all respondents minimum CPS intervention

line or 'break point" data was between '2" and '3". Essentially, the data were

dichotomized into: CPS-mandated ('I" and '2") and non-CPS cases C3" and n4n).

Analysis involved examining each HSP 'break poinf' on each of the nine neglect forms

by two independent variables: regional SES variance and agency type.

Method of Data Analysis

The unit of analysis for the secondary analysis is the individual HSP. Adaptation-level

(AL) theory was used to guide thinking about how contextual factors influence HSP

decisions. Taking an explanatory approach, the secondary analysis employs the nulC

hypothesis significance-test procedure (NHSTP). Chow's (1996) thorough examination

of the limitations of NHSTP finds that while it is relevant only to the statistical conclusion

validity and not intended to be used to interpret data at any level of discussion other

then the statistical level, nevertheless, NHSTP is a particularly useful method with

research on theory corroboration because. '...NHSTP provides a welldefined rational

basis for deciding whether or not research data are due to chance influencesn (p.175).

The issue of whether the secondary findings are due to chance is a primary concern for

all researchers. I have taken the approach of setting up a rejection level of an alpha

level of p.c 0.05. This means the null hypothesis (Ho) will be rejected if a finding is less

than 0.05. This level of significance, or alpha level, represents the probability that

approximately one time in twenty or 5 percent of the time, I will make the mistake of

saying that significant differences exist between the groups tested, when in fact, no

differences exist. The intent of this low rejection level is to reduce the likelihood of

alsely rejecting the null hypothesis (type I error). In other words, I am more concerned

with avoiding a type 1 e m (reject null hypothesis when it is true) rather than a type !I

error (fail to reject the null hypothesis when it is actually false). Beta is the probability of

making a type I1 error. Reducing the beta error increases the power of a statistical test

where power is equal to 1 minus the beta error (1 - 25) or in other words, the power of a

test is the ability of a test to correctly reject the null hypothesis (Pett. 1997; Runyon &

Haber, 1976; Spring hali. 1 997).

Methods Used to Increase Power of Statistical Tests

Increasing the power of the statistical tests used is important because the likelihood of

committing a type I1 error is reduced. Since power varies as a function of various

statistical and substantive criteria, the secondary analysis tries to increase power in a

number of ways. First, large sample sizes are used (e.g. N >30 per group) in both

hypothesis 1 and hypothesis 2 (N=679). Using large sample sizes increases the power

of the test and the ability of the test to detect statistical significance at large, medium

and small effict sizes (Pett, 1997). The trade-off or disadvantage of large sample sizes

is that the likelihood of detecting statistical significance increases, where the result may

f3e a finding that is statistically significant, but may not be substantively or clinically

important

A specific limitation with the sample sizes in ' secondary study, is that while the overall

sample sizes are large, the actual sample sizes of the groups tested in each hypothesis

are not equal. For example, in hypothesis 1 the sample sizes between the three grwps

are not equal and range from the high SES group where n= 220, the medium SES

group where n = 331 and the low SES group where n = 128. In hypothesis 2, the sample

size with investigating agencies is n = 270, while with noninvestigating agencies n =

409. The greater the imbalance between the sizes of the sample groups tested, the

greater the adverse effect on the power of the test. Because the sample sizes are

unequal in the hypotheses tested, the calculation of the harmonic mean is required.

Essentially, the harmonic mean calculation takes unequal sample sizes and makes them

equal by indicating how much the smaller sample size needs to be increased in order to

ensure adequate power of the test Using Cohen's (1988) method to calculate the

harmonic mean (2 x n l x n2 I N), finds the sample sizes in hypotheses 1 and 2, both

have an adequate ability to detect medium and large effect sizes (power is greater than

the standard acceptable level of 80%). However, there is less confidence that the

sample sizes in hypotheses 1 and 2 can adequately detect a small effect size (power is

less than 80%).

The second method used to increase power was by choosing an alpha level of pc 0.05.

rather than a more stringent alpha level of pc 0.01. Selection of p c.05 means the

likelihood of falsely rejecting the null hypothesis is reduced (type L error), and this

increases the power of the tests used in this study.

Third, by choosing to use both the appropriate type of test (nonparametric) for the type

of data (ordinal) and by selecting the appropriate statistical test for each hypothesis,

power increases. In other words, the data being anaIyzed met the assumptions of the

tests used. Meeting the test's assumptions assists in the ability of the test to control for a

type I error (Pett, 1997). Other options that increase a tesfs power, such as one-tail

tests, were not avai table to use because the hypotheses are nondirectional (Spring hall.

1 997).

Determination of Regional SES Rank

Establishing each region's SES status was done by following the same steps Wolock

(1982) used to rank the 11 CPS regions in New Jersey by 13 SES indicators.

Essentially, the six regions were ranked from 1 (low SES) through to 6 (high SES) on

each of the following SES regional variables: unemployment rate, percentage of lone

parent families and incidence of low income (rationale behind the selection of the three

SES variables is noted later in this chapter). The three ranks scores were summed and

each region then received an accumulative score; based on similar groupings of the

accumulated scare, the regions were then divided into a High SES Group, Medium SES

Group and Low SES Group. The scores of the regions in the High SES Group are:

Region F (16), Region A (IS) and Region D (14). The scores of the regions in the

Medium SES Group are: Region E (8) and Region B (7). Finally, the score of the region

in the Low SES Group is: Region C (3). See Table 15.

Region SES Indikators Overall Score

UnempIoy. % Loon llncdence of Rate t5+ Parent Low income

Region A 6.1% 2.3% 6.1% RANK 5 5 5 15

Region 8 7.7% 3.6% 10.1% RANK 3 2 2

Region C 8 -6% 3.8% I 1.7% RANK ? i 7

Region D 6.7% 2.2% 7.7% RANK 4 6 4

Region E 7.8% 3.0% 8.7% RANK 2 3 3

Region F 6.0% 2.5% 5.3% RANK 6 4 6 16 Note. - Source: 1991 Census, Statistics Canada Final Rank : most disadvantaged or lowest SES region has a low number i.e. 3. The most advantaged or highest SES regions have high numbers i.e. 14,15, f6.

The rationale behind the selection of the SES indicators used in the hypothesis 1,

needs to be noted, since one of the goals of this researcher was to try and replicate

WolocKs (I 982) study. The secondary analysis employs three SES indicators:

unemployment rate, percentage of lone parent families, and incidence of low income.

Yet, Wolock used the fol[owing SES indicators to assess New Jersey neighbourhood

SES levels: 1) density rate (population expansion rate), 2) percentage of femaleheaded

households with children under eighteen years of age, 3) percentage of femaleheaded

households with children under eighteen years of age, 4) percentage of families below

the poverty level, 5) percentage of illegitimate births, 6) rate of infant mortality, 7)

venereal disease rate, 8) rates of identified drug cases, 9) percentage of non-white

individuals, 10) suicide rate, 17) percentage of individuals with eight years or less of

education, 12) percentage of families on General Assistance or Family Benefits, and

13) support services available in the community.

So why only use three SES indicators when Wolock selected thirteen indicators? Fitst.

no set formula for selection of SES indicators exists. Each indicator is just one piece of a

very large jigsaw puzzie: the social and economic context. Second, previous use,

profassional assessment, personal preference mixed with indicator reliability, accuracy

and timeliness, all influence choice. Selection is then further modified because some

indicators ere quite accurate but not necessarily related to the area studied; and some,

while relevant, are very crude in their ability to portray or capture the specific social or

economic element it is intended to measure. Furthermore, context influences selection.

For example, the same SES indicator may differ considerably by American or Canadian

standards; variance may be attributed to conceptual or definitional differences, sampling

limitations, measurement bias, timing of data collection, method of preparation, or the

data collection is specific to federal, provindaVstate or region (Grant. 1992).

Examination of the SES indicators in Wol&s (1982) study found most do not translate

well into Canadian text Some statistics Wolock used are not collected, are very difficult

b obtain, are unreliable or are not available a the regional leveI in Ontario. Examples of

these are: percentage of illegitimate births, venereal disease rate, rates of identified drug

cases, percentage of non-white individuals and suidde rate. Hence, selection of the

three SES indicators for this study was based on three criteria. One, on trying to attain

some similarity to Wolock's (1982) study. Two, since neglect is linked to low income and

poverty is not distributed randomly, to use indicators that reflect a region's level of

economic deprivation. And three, to use indicators that are regularly and systematically

collected in all Ontario regions and are proven to be significant as an indicator of poverty

as well as reliable as a statistic. (Garbarino, 1992; Grant, 1992; Ross, Shillington &

Lochhead, f994). Based on the above rationale, the indicators selected for secondary

analysis are: unemployment rate, percentage of lone parent families, and incidence of

low income. For the breakdown of the six regions into three SES groups see Table 16.

Table 16

Breakdown of Sam~le Reaions into Three SES Groups

SES Regions # of Respondents Group in Region N

High SES Region A 36 Region D 87 Region F 97

Medium SES Region B 117 Region E 224

Low SES Region C 128 128

TOTAL 679 679

Rationale for Nonparametric Tests

As noted above, the characteristics of the data determine which statistical test(s) are to

be used. For this secondary survey study, the scale of measurement of the dependent

variable (HSP perception of the minimum CPS neglect intervention line) is ordinal and

so nonparametric tests of significance, which are specific to the rank nature of the

ordinal data are employed.

Since the data are rank ordered, the advantage of the nonparametric tests is a focus on

the rank ordering of scores. Additionally, these tests have few assumptions concerning

the population's distribution, they can be used for small sample sizes, and are not

influenced by atypical data like outlien (Pett, 1997). While nonparametric statistics are

typically less powerful than parametric tests, this is only true if the assumptions of the

parametric test have not been violated. Again, for comparison purposes only, the

equivalent parametric test will accompany the nonparametric test. However, all findings

will be based on the nonparametric results. As noted previously, the hypotheses are

nondirectional and an alpha level of 0.05 was used.

Hypothesis 1 : Regional SES Groups - Kruskal-Wallis Test

Analysis of the three SES groups' perceptions of the MIL, employed the Kruskal-Wallis

(&W) one-way ANOVA by ranks test The advantage of the nonparametric K-W test is

it ranks the values of the regions instead of just noting that they are above or below the

median. Additionally, this test does not require equal sample sizes. The limitation of the

K-W test, is that similar to its parametric counterpart, the one-way ANOVA, this is an

omnibus test of significance, where significance simply indicates there are differences

among the groups but does not indicate where the differences are among the groups.

If the K-W finding is significant (p <.05), the appropriate nonparametric post hoc test is

the Mann-Whitney U pairwise comparisons to test for K population medians. The post

hoc test determines which groups are significantly different from one another.

Hypothesis 2: Agency Type - Mann-Whitney U Test

In hypothesis 2, where two independent group medians are tested, the appropriate

nonparametric test to use is the Mann-Whitney U. Again, the parametric counterpart, the

t-test, accompanies the Mann-Whitney U test Like the independent t-test, the Mann-

Whitney U compares measures of central tendency between two independent groups.

However, the Mann-Whitney U uses medians for comparison. The advantages in using

the Mann-Whitney U is that it is sensitive to the central tendency of the scores, it has

excellent asymptotic efficiency, and the power efficiency approaches 95.5% as N

increases, which makes the Mann-Whitney U nearly as powerful as the independent t-

test in being able to correctly reject the null hypothesis (Pett, 1997).

Summary

The strengths of this secondary analysis (i-e. appropriate research question, test of AL

theory, comparable u n l of analysis to original survey data, appropriate statistical tests

employed) are balanced against the use of secondary survey data, which make the

study vulnerable to all limitations associated with secondary data-

CHAPTER 5: SAMPLE REGIONS & RESPONDENTS

Life is the ut of drawing sufficient conclusions from insufficient evidence.

Samuel Butler, Wn century poet.

This chapter describes the six CPS regions in the convenience sample and highlights

some of the unique regional characten'stics, such as locale, geography, regional size,

population growth, and the cultural makeup of the region. Regions are also profiled

within their respective SES group. Possible implications regarding regional differences

within the SES groups are noted, such as with rates of maltreatment substantiation and

neglect allegations. Description of respondents' professional and personal

characteristics were limited to the 28T HSP survey respondents from Region A (n=36),

Region 8 (n=l17) and Region C (n=128) and their similarities and differences are noted.

All figures are based on 1991 Census data gathered from Statistics Canada.

Description of Sample Regions

Region A

Region A is predominantly a rural area with a rich agricultural base in south-western

Ontario. Situated along side one of the Great Lakes, this region is composed of

numerous small towns and villages, The majority of the population are white, English

speaking; essentially a homogeneous population with a regional population growth of

small proportions (5.5%). Of the six in the sample, this region has one of the lowest

unemployment rates (6. t%), percentage of [one parent families (2.3%) and incidence of

low income ( 6.1 96). Region A is ranked in the high SES group for this study.

Region 0

Region 8 is situated in south-eastern Ontario. While it is a region that is predominantly

mra! with a strong tourism component, it is dominated by large, university-based city in

the southern part of the region. Another unique feature is has a long history as an active

military site. For the most part, residents are white and English speaking. Population

growth is moderate at 12%. This region has an unemployment rate of 7.7% and lone

parent families make up 3.6% of the population, the incidence of low income is 10.1 %.

Region 0 is placed in the medium SES group.

Region C

This region is situated in the southern most sector in northern Ontario. Situated on the

shores between Lake Huron and Lake Superior, the land is primarily undeveloped with

one main city being the urban centre. A mix of English. French and Native populations

reside in the region. In addlion to serving ~e regional mainland, the CPS agency also

sewes the Largest freshwater island in the world, where the residents are predominantly

Native. While forestry is a key industry, agriculture is limited and marginal. Population

growth can be described as small at 5.7%. This region is distinguished in the sample as

having the highest unemployment rate (8.6%), the highest percentage of lone parent

families (3.8%) and the highest incidence of low income (1 1 -7%). Region C is the only

region in the low SES group.

Region D

This region is directly north of Region A Both regions border the eastern shores of Lake

Huron in the south-western sector of Ontario. Region D is also a rural area punctuated

with numerous small towns. Much of the land is used for pasture or cropland. The

homogeneous population is mostly white, English speaking and agriculturally focused.

Compared to Region A, this region's population growth is more moderate at 10.9% and

this appears to be because the region is viewed as an attractive retirement community.

Like its neighbor to the south, this region has a similar unemployment rate ( 6.7%), a

similar low number of lone parent families (2.2%). but a somewhat higher incidence of

low income (7.7%). Region D is in the high SES group,

Region E

With over 700.000 residents, Region E has by far the largest population in the sample.

Geographically this is a large region and it is situated in the intensely developed Golden

Horseshoe area in south-central Ontario. The region has three distinct areas. The

southern segment is heavily developed and urbanized and has Lake Ontario to the

south. The central section lies north of the TransCanada Highway and while it is a mix

of urban and rural, it is rapidly being urbanized. The northern part of this region is very

rural and consists of small towns. Region E has the distinction of being one of the

fastest growing communities both in Ontario and in Canada. This regions is a preferred

locale for new immigrants to Canada and as such, is distinctly multicultural. This region's

population increased from 260,000 to more than 800,000 in two decades; just beween

1986 to 1991 the population increased 23.8%. One ramification of the mushrooming

population has been a struggle for all services to keep up with population growth. This

region has a unemployment rate of 7.8%, percentage of lone parents of 3.0% and the

incidence of low income at 8.7%. Region E is in the medium SES group.

Region F

Wth the residents numbering over 300.000, this south-central Ontario region's

population is the second largest in the sample. Region F lies adjacent to the western

borders of Region E. The preponderance of Region F's population is situated in the

southern part of the region in a city that has many adjacent bedroom communities. The

southern area is characterized as heavily developed and urbanized. The northern

section of Region F is heavily wooded, rural and replete with conservation areas, park

lands and hiking trails. Similar to its eastern neighbor, a significant proportion of Region

F's overall population growth (15.4%) is comprised of new immigrants (23%). Once

characterized as a predominantly white, English speaking area, Region F is now

described as multicultural. The low unemployment rate (6.0%), low incidence of low

income (5.3%) and small percentage of lone parent families place this region in the high

SES group.

Regional C hamcteristict

Broad analysis of the sample regions characteristics finds some regions naturally group

together because of the similarities on variables such as regional locale, cultural

makeup, population growth and immigrant population percentage. For example, Region

A and Region D are quite similar in that they are both located in south-western Ontario,

both have small CPS agencies, both are rural, and both are defined as having English

speaking, homogeneous populations. As well, these two regions have very similar rates

on unemployment, percentage of lone parent families and incidence of low income.

Both Region A and Region D are in the high SES group.

Region E and Region F are matched on characteristics such as locale (south-central

Ontario), rapid population growth, and having a multicultural population. Moreover, the

population change in Region E and F is marked by a significant influx of immigrants and

has resulted in strained sewice delivery in both regions. However, despite the

similarities between the two regions, Region E is in the medium SES group and Region

F is in the high SES group.

Closer scrutiny of the regions within each of the three SES groups finds regional

characteristics within the high and medium SES groups are more dissimilar than similar.

In addition to the variation in the characteristics noted above, there is another important

distinction: regional variation in maltreatment substantiation and neglect allegation rates

exists (Trocme, 1994). A possible implication for the secondary analysis is that the

accumuiation of significant differences within the medium and high SES groups may

adversely affect the ability to reasonably test HSP perceptions of neglect entry points by

regional SES variation. Table 17 presents the regional variations between and within the

SES groups.

Table ?7

Characteristics of Reaions in SES Grou~s

SES Group Regions

Reaion D Reaion E Hiah SES l n 3 1 CPS region Population size' Oh population change* Urban [rural Predominant cukure % immigrants* CPS agency sue Regional maltreatment rates" Regional neglect rates*

Medium SES in =2) CPS region Population size* % population change* Urban I rural Predominant cukure % immigrants CPS agency sue Regional maltreatment rates* Regional neglect rates*

Reaion A south-west 59,065 5.5% rural white 8.6% small 34% 25%

Reaion B sout h-east 129,089 12% rural some urban white 43.6% medium t 8% 24Oh

~outh-west 65,268 10.9% nrral white 8.0% small 34% 25%

Low SES h=lI Region C CPS region Population ske* % population change* Urban /rural Predominant culture % immigrants CPS agency sire Regional maltreatment ratas- Regional neglect rates*

northern 161.220 5.7% rural English. French 8 Native 8.1 % large 32% 39%

south-central 313,136 I 5.4% urban some rural multicultural 23% medium 27% 28%

Reclion D south-central 732,798 23 -8% urban some rural multicultural 36.8% large 27% 28%

NOTE = data fram Statistics Canada, 7991 census

*= data from Trocme, (1994) Ontario Incidence Study of Reported Child Abuse and Neglect

Profiles of HSP Respondents: Regions A, B (L C

Examination of the respondents personal information, such as age, gender, marital and

parental status, is limited to HSP participants from Regions A, 8 and C (N=281). Details

on these survey participants' prof8ssional characteristics include: education, income,

years living in the community, years of experience in the HSP chosen field, self-

assessment of the respondent's knowledge of the Ontario child welfare legislation, years

of experience with child maltreatment cases, and an estimate of the total number of child

maltreatment cases the HSP has experienced. An attempt was made to gather client

SES status information but it was evident when reviewing the data there were problems

with the quality of the responses, and so those results are not reported. See Table 18

for overall findings.

Personal Characteristics

The majority of the survey respondents are from helping professions such as social

work, CPS, and general counseling services. At the practice level, women dominate

these fields and so it is not surprising that females make up the preponderance of the

participants in this study: out of 277 respondents, 203 or 73% of the HSP were women

and 74 or 27% were men. Overall, nearly half (47%) of the respondents were over the

age of fotty-one (n=124), threequarters of the sample were in a relationship where they

were either married (69%) or were living common-law (8%), and the majority were

parents (77%). For the most part, these survey respondents were long-term residents of

their communities, with 84% residing in their region for more than six years. In fact, over

two-fifths or 44% of all respondents (n=t21) indicated they had lived in their

communities more than twenty yean. For breakdown of respondents personal

characteristics by region see Table 19.

Professional Characteristics

Regarding professional characteristics, the respondents were well educated, with sixty-

nine percent of the 281 respondents having at least an undergraduate degree (n =191).

Overall, between survey participants with BSW and MSW degrees, the ratio was

approximately ?O:3 0. These respondents also appeared to have considerable

experience in their chosen fields. More specifically, over half of the respondents had

more than 11 years of experience in their profession. Two-thirds of respondents stated

that their knowledge of the CFSA was either moderate (36%) or considerable (30%).

While only 44 or 16% of survey participants indicated substantial familiarity with the

Ontario child welfare legislation, 12 or 4% said they had no knowledge of the CFSA and

38 or 14% indicated they had some knowledge of child protection laws. Yet, for the most

part these HSP indicated they had considerable practice experience with child

maltreatment, as nearly half of all respondents (n=45) indicated they had 'more than 51

child maltreatment cases" during their professional career.

Overall. 71% of those HSP surveyed as to their annual income (n=187), indicated they

made between $10,000 to $49,999 a year. Over half of all the HSP respondents (57%)

indicated their salary range was between $30tOO0 to $49.999 per year. While 29% of the

participants made over $50.000 (n=78), most of those respondents (n=62) said they

made $50.000 to $69.999. In other words. there seems to be a 70:30 split be-n

those respondents who made less than $50,000 and those who made more than

$50,000. Although regional variation of HSP salaries is not known, review of the study's

overall front-line to supervisorlmanager ratio reflects that same ratio. Thus, the income

breakdown appears to reasonably reffed the salary differences bebeen fiont-line HSP

and management HSP respondents. See Table 20 for breakdown of respondents

professional characteristics.

A more detailed analysis finds unique regional differences. For example, regarding

gender and age; Region A differs considerably from Region B and Region C, in that

over half of Region A respondents are men and two-thirds of Region A survey

participants are under forty years of age. Whereas, Region C, women made up four-

fifths of the sample, and 45% of those respondents were over the age of forty. Region 6

appears distinct from the other regions in that over fifty percent of the respondents in

Region B are over the age of forty-one and are well educated. The value of higher

education is evident in that threequarters of the 114 sample participants in Region 6

have at least an undergraduate degree, compared to 67% for all regions combined.

Furthermore, where the other two regions have an average BSW-MSW ratio of 75:25,

Region B is unique in that the ratio of 8SW:MSW is 50:50. The higher level of social

work education for Region B respondents may be due to the fact that Region 0 is the

only region in the sample that has a university. See Table 15 for breakdown of the

personal characteristics of respondents from Region A, Region B and Region C. See

Table 16 for breakdown of the professional characteristics of respondents from Region

A, Region 6 and Region C.

Table f 8

Overall HSP Characteristics: Reaions A B & C

Chsracteristics N Column Percentage (=287)

Gender (n=2?7) %male 203 73% male 74 27%

Age [range is 23-63] (n=264) under 30 35 30 to 40 105 40 pius 1 24

Marital Status (n=268) married common law divorced single widowed

Parental Status (n=277) children no children

Education (n=275) BSWIMSW 88 BA 76 Master -other 23 Ph.DA4.D- 4 College 65 Other HSP degree 13 High School 6

BSW MSW

Years Living in Community [range 1 -561 (n=278) 1-5 years 44 6-10 yean 51 1 1-19 years 62 20+ years 121

Years in Chosen Field [range 1-38] (n=278) 1-5 years 45 6-1 0 years 77 11-19 yean 77 20+ years 79

CFSA Knowledge (n=277) none some moderate considerable substantial

Number of Maltreatment Cases (n=262) never ?S 1-5 cases 29 6-20 cases 54 21-50 cases 43 57 + cases 1 17

Note. Column percentages have been rounded to the nearest whole percentage.

Table t 9

Reaional Variations in HSP Personal Characteristics: Reaions A. B 8 C

Personal Region A Region B Region C Characfen'sdics ('36) (N=1 f 7) (W728) [column percentages in italics]

Gender (n=277) female male

Age [range is 23-63] (n-264) under 30 30 to 40 40 plus

Marital Status (n=268) married common law divorced single widowed

Parental Status (n=277) children no children

Years Living in Community t -5 years 6-10 years 1 ?-I 9 years 20+ years

Note. Column percentages have been rounded to the nearest number

Table 20

R ~ i o n d Variations in HSP Professional Characteristics: Reaions A. B and C

Pm&ssional Region A Region 8 Region C Chefacfetisfr*~~ (N=36) (N=117) (N=128) [column percantages in italica

Education (n=275) BSWMSW BA Master Ph*D/M*D* Col Iege Other HSP degree High School

Social w o k BSW : MSW

Years in Chosen Field 1-5 years 6-10 years 1 ? -19 years 20+ years

CFSA Knowledge none some moderate considerable substantial

Number of Maltreatment Cases never 1-5 cases 6-20 2T-50 cases 57 +cases

- - - - - - pp -

Note - Column percentages have beerr rounded to the nearest whole percentage,

Summary

The problems associated with a convenience sample at the regional level are evident in

this secondary analysis. While there is a clear rationale for dividing the six sample

regions into the three SES groups, closer examination reveals considerable differences

exist between the regions grouped within the high and medium SES groups. It is not

known whether these differences may affect the ability of the secondary analysis to

adequately test HSP perceptions of neglect entry points by regional SES variance.

The overall personal profile of the HSP respondents from Region A, B and C is: married

with children and long-term residents of their communities. Professionally, these HSP

are mature, well-educated professionals. Over two-thirds (n=191) have at least an

undergraduate degree (69%) and of those, 56 or 20% also hold a master's or doctorate

degree. Generally, the secondary analysis found respondents did not indicate great

familiarity with the Ontario child welfare legislation, as 'moderaten knowledge on the

CFSA was most frequently checked. Yet, for the most part these HSP indicated they

were very experienced with child maltreatment, as nearly half of all respondents

selected 'more than 51 child maltreatment casesa during their professional career.

CHAPTER 6: RESULTS

The great tragedy of Science - the slaying of a beautiful hypothesis by an ugly fact,

Thomas Huxley - Victorian sage

Statistics is a language that can speak where other tongues are mute.

Paul D. L d y , 1989, p.173.

This chapter presents the results from the secondary analysis. As noted earlier, both the

non parametric and parametic statistics are presented. However, all findings are based

on the nonparametric statistic. It is interesting to note that in each case, the parametric

and nonparamettic results paralleled.

In the first hypothesis (HI), it was expected that there would be differences between the

SES groups perceptions of the MIL for the nine forms of neglect. The overall findings

from the secondary analysis are mixed and inconclusive, Only four Reasons for Service

had p<-05. Moreover, for eight of the nine neglect forms, the results from the secondary

analysis are opposite to what would be expected of a test of Adaptation Level (AL)

theory if American data were tested.

The nonparametric, Kruskal-Wallis test found no significant differences between the

three SES groups of pc.05, with five of the nine neglect Reasons for Sewice. Since the

nonparametric test did identify four neglect Reasons for Service with significant

differences of pc.05, the Mann-Whitney U paiwise comparison post hoc test indicated

where the significant diierences were between the three SES groups. With three

Reasons for Service, the regions in high SES group had a less inclusive MIL (a lower

mean rank score) than the low SES group (a higher mean rank score). Again, this is

opposite of what would be expected from a test of AL theory and previous findings from

other American studies. The Mann-Whitney U post hoc test showed only one Reason for

Service was in support of AL theory, where the low SES group had a less inclusive MIL

(a lower mean rank score) than the high SES group (a higher mean rank score). While

it is disappointing not to be able to advance AL theory, it may be a fair test of AL theory

was confounded because of the large sample unit (the region) coupled with the fact that

the Canadian context and social safety net differ considerably from the United States.

Despite the mixed results ftom the hypothesis 1 analysis, given three neglect Reasons

for Service had significant differences of pc.004 (indicating the differences are not due

to chance alone), this suggests the results do support other studies that posit eligibility

determination by HSP is affected by context.

The analysis of agency type found no significant differences in where investigating

agencies and noninvestigating agencies perceived the MIL for the nine foms of neglect;

this result differs ffom American studies.

Overall, the secondary analysis suggests that the construction of neglect is not a static

phenomena within and across agencies and regions. Furthermore, the findings suggest

that Canadian child welfare may differ from American experiences. Additional research

is needed to better understand which factors, in particular contextual ones, inffuence

HSP decisions about child neglect in Canada.

Tests of Hypotheses

Hypothesis I: Regional SES Groups

Based on Adaptation-level theory and empirical findings to date, hypothesis I is:

H1 - Human Sewice Professionals (HSP) in the three SES groups (low, medium and high) will perceive the minimum CPS entry point differently for the nine forms of neglect

Hlo - There wilt be no difference in where HSP in the three SES groups perceive the minimum CPS entry point for the nine forms of neglect.

As a nonparametric test, the Kruskal-Wallis test establishes whether there is

significance of pc.05 among the groups, by ranking the data and calculating the average

sum of the ranks for each group. If the average sum of the ranks for each group is the

same then the independent groups have similar underlying populations and the null

hypothesis is true. M the discrepancy between the average ranks for the independent

groups is sufficiently large (p<.OS), the null hypothesis is rejected. For comparison

purposes only, the equivalent parametric test, the Analysis of Variance (ANOVA), is

presented along side the Kruska l-Walis test in Table 21 .

If an alpha level of less than -05 is found with the Kruskal-Wallis test then the Mann-

Whitney U paiMlise comparison post hoc test is used to analyze the three groups and

establish which group or groups have significant diirences of pc.05. The three

possible combination of SES groups are:

low SES gmup (I) to the high SES group (3);

low SES group (I) to the medium SES group (2);

medium SES group (2) to the high SES group (3).

To calculate the Mann-Whitney U test for each of the above SES group combinations,

the data on the MIL for each pair of SES groups are first combined and the scores on

the dependent variable (HSP perceptions of the MIL) are ranked from lowest to highest

without regard to their position on the independent variable. Then, the independent

groups in each pair (e.g. low SES group and high SES group) are separated out and the

sum of the assigned ranks for each independent group is calculated. If the sum of the

ranks is similar (e.g. see Table 21 - 88, the low SES group mean rank is 178.91 and

the high SES group mean rank is 170.33) it means the two SES groups come from the

same population, ps.05 and the null hypothesis is We. However, if the sum of the ranks

for one of the independent groups is very different (e-g. see Table 21 - 2A, the [ow

SES group mean rank is 196.20 and the high SES group mean rank is 159.65), pc.05,

and this is the evidence to suspect the two samples did not come from the same

population.

The post hoc test results are also presented in Table 21. Both the alpha level and the

mean rank scores of the two SES groups being compared, are shown. The mean rank

is presented instead of the sum of the ranks to allow comparisons of ranks even

if the groups are not equal in size (Pett, t997). The relationship between the mean

rank score and the MIL is this: when examining the post hoc findings for each SES pair.

if pc.05 then the higher mean rank score means the HSP in that SES group

perceived a more inclusive MIL. Conversely, the lower mean rank score means the

HSP in that SES group perceived a Iess inclusive MIL. However, the findings do not

indicate beween which specific descriptors the MIL is. Only that the MIL is perceived as

being the same or more inclusive or less inclusive.

All HSP (N=679) in the six sample regions are included in the test of hypothesis 1. As

noted earlier, the six regions are divided into three socioeconomic (SES) groups.

High SES group Medium SES group Low SES group Region A (n = 36) Region 6 (n = 1 17) Region C (n = 128) Region D (n = 87) Region E (n = 214) Region F (n = 97)

TOTAL n =220 TOTAL n = 331 TOTAL n =128

Significance between the three SES groups was not found in the following five neglect

categories: (IA) -Abandonment (p = .9046), (8A) - Nutrition (p r.2033). (8C) - Personal

Hygiene (p = .0988), (8D) - Household Sanitation (p = .4943), and (8E) - Lack of

Clothing (p = -2050). Table 21 displays the results of the secondary analysis.

There was significance of pc.05 in four forms of neglect (24 - Lack of Medical Attention

(p = .0001), (?A) - Lack of Supervision (p = .0001), (7B) - Lack of Appropriate Child

Care Arrangements (p = . O W ) and (8B) - Physical Living Conditions (p = -0285). The

Mann Whitney U post hoc test analyzed the mean ranks of the three pain of SES

groupswhere p c.05 and found:

2A - Lack of Medical Attention

Between Group I (low SES) and Group 3 (high SES) p = -0001. The mean rank of

the for Group I is 196.20 and the mean rank for Group 3 is $59.65. The HSP in the

low SES group perceived a more inclusive MIL than the HSP in the high SES

group.

Between Group 1 (low SES) and Group 2 (medium SES) p = -0001. The mean rank

for Group 1 is 259.90 and the mean rank for Group 2 is 217.24. The HSP in the low

SES group perceived a more inclusive MIL than the HSP in the medium SES

group*

There was no significant difference between the mean ranks of the scores between

Group 2 (medium SES) and Group 3 (high SES). where p = .6414.

7A - Supervision of Child

Between Group I (low SES) and Group 3 (high SES) p = .0001. The mean rank for

Group I is 206.81 and the mean rank for Group 3 is 154.82. The HSP in the low

SES group perceived a more inclusive MIL than the HSP in the high SES

group,

Between Group 1 (low SES) and Group 2 (medium SES) p = .0002. The mean rank

for Group 1 is 263.68 and the mean rank for Group 2 is 216.98. The HSP in the low

SES group perceived a more inclusive MIL than the HSP in the medium SES

group.

Between Group 2 (medium SES) and Group 3 (high SES) p = .0276. The mean

rank for Group 2 is 286.72 and the mean rank for Group 3 is 258.54. The HSP in the

medium SES group percalved a more inclusive MIL than the HSP in the high

SES group.

76 - Child Care Armngements

Between Group I (low SES) and Group 3 (high SES) p = .0172. The mean rank of

the scares for Group 1 is 157.1 1 and the mean rank for Group 3 is 181 -62. The HSP

in the low SES group perceived a less inclusive MIL than the HSP in the high

SES group.

There was no significant difference between the mean ranks of the scores between

Group 1 (low SES) and Group 2 (medium SES) p = .9140.

Between Group 2 (medium SES) and Group 3 (high SES) p = -0017. The mean rank

of the scores for Group 2 is 258.27 and the mean rank for Group 3 is 298.10. The

HSP in the medium SES group perceived a less inclusive MIL than the HSP in

the high SES group.

8B - Physical Living Conditions

There was no significant difference between the mean ranks of the scores between

Group 1 (low SES) and Group 3 (high SES) p = .3871.

There was no significant difference between the mean ranks of the scores between

Group 1 (low SES) and Group 2 (medium SES) p = .1408.

Between Group 2 (medium SES) and Group 3 (high SES) p = .0113. The mean rank

for Group 2 is 287.53 and the mean rank for Group 3 is 255.98. The HSP in the

high SES group perceived a more inclusive MIL than the HSP in the medium

SES group.

Table 22

Hv~othesis f Results: Regional SES G m u ~ Variable

Spectrum N Knrskal ANOVA PostHoc Test- N i e c t Wallis M m MMney U Mean Sections pc.05 pc.05 pc.05 Rank

-

[lAl Abandonment 677 .9046 -799 No two groups are significantly different

[q ~ a c k of 676 .0001* -0001 .00Olm Group 1 [low SESj 196.20A Medical Attention Group 3 [high SES] 159.65

,000f- Group 1 [low SES] 259.90A Group 2 [med SESj 21724

,6414 Group 2 [med SES] 277.56 Group 3 [high SES] 27238

FA1 Supervision of Child

678 .0001* .0001 ,0001"* Group 3 [low SESj 206.81 Group 3 [high SES] 154.82

,0002- Group 1 [low SES] 263.6aA Group 2 [med SESj 216.98

,0276 Group 2 [med SES] 286.72A Group 3 [high SES] 258.54

[7BJ Child Care 675 ,0094. -022 -0172 Gmupf( lowSE~ 157.1f Arrangements Group3[highSES] 181.62&

,9140 Group 1 [tow SESI 229.02 Group 2 [med SES] 230.38

.OOt7- Group 2 [med SES] 258.27 Group 3 [high SESI 298.1 QA

Spect~m N Kiuskal ANOVA Post Hoe Test - NegI8Ct Wallis Mann W n e y U Mean Sections pc.05 pc.05 pt05 Rank

[8Al Lack of Nutrition 677 .2033 .I 82 No two groups are significantly d'Hferent,

[eel Physical 677 .0285* -004 .3871 Group 1 [low SES] 178.91 Living Group 3 [high SES] 170.33 Conditions

,1408 Group .I [low SES] 217.t6 Group 2 [med SES] 234.96

.0113 Group 2 [med SESj 287.53A Group 3 [high SES] 255.98

[8CI Personal Hygiene 677 .0988 .I33 No two groups are significantly different

1801 Household 677 ,4943 .476 No two groups are significantly different. Sanitation

[8Ej Lack of Clothing 674 ,2050 .I92 No two groups are significantly different.

NOTE. * pc.05 for KruskaCWallis (nonparametric) test - pe.002 for Mann-Whitney U (nonparametric) pairwise comparison post hoc test A indicates SES group had a significantly more inclusive MIL W.052,where more descriptors of a less

severe nature were included,

Hypothesis 2: Agency Type

Based on theoretical and empincat findings to date, hypothesis 2 is:

H2 - Investigatory agencies (CPS & police) will perceive the minimum CPS entry point differently for the nine forms of neglect than HSP in non- investigatory agencies (all other HSP).

HZ0 - There will be no difference where HSP from investigatory agencies and non-investigatory agencies perceive the minimum CPS entry point for the nine forms of neglect.

Again, all 679 HSP respondents were included in the secondary analysis. The data were

dichotomized into two groups: (1) noninvestigatory agencies (n=409) and (2)

investigatory agencies (n=270 ). With the investigatory group, CPS workers accounted

for 94% of the sample (n=254) while police represented only 6% of the sample (n=16).

Two-tailed tests of significance were used with pc.05. Since two independent samples

are being tested and the data are ordinal the nonparametric test, the Mann Whitney U is

used. Significance between the two groups (noninvestigating and investigating) was not

found in any of the nine neglect forms.

The results are as follows: (IA) - Abandonment (p = .1802), (2A) - Lack of Medical

Attention (p = .I W8), (?A) - Lack of Supervision (p = .2861), (78) - Inappropriate Child

CareArrangements(p=.4994),(8A)-LackofNutrition(p=.0845). (88)-Physical

Living Conditions (p= .2272), (8C) - Personal Hygiene (p = .6t23).(8D) - Household

Sanitation (p = .0759), and (8E) - Lack of Clothing (p = -8592). See Table 22 for

overview of results-

Table 22

Hv~othesis 2 Results: Aaencv Tvm Variable

Specthm Neglect sections

N Mann t-test Whitney U

679 pc.05 pc.05

[l A] Abandonment

[2q Lack of Medical Attention

[7AI Supervision of Child

[76] Child Care Arrangements

[8A] Lack of Nutrition

[8B] Physical Living Conditions

[8C] Personal Hygiene

[8D] Household Sanitation

[BE] Lack of Clothing

Note. - 1 = noninvew-gating agencies (all other HSP groups except CPS and Police). 2= investigating agencies (CPS and Police).

Summary

Overall, the findings from hypothesis I vary considerably by each Reason for Service.

No differences of p< 0.05 were found between the SES groups in the following neglect

forms: Abandonment (p = .9046), Lack of Nutrition (p = .2033), Personal Hygiene (p =

.O988), Household Sanitation (p = .4943), and Lack of Clothing (p = -2050); a trend

toward statistical significance is noted with Personal Hygiene (p =.0988). While findings

of pc 0.05 were found with Lack of Media( Attention (p=.0001), Lack of Supervision

(p=.000 I), Inappropriate Child Care Arrangements (p=.0044), and Inadequate Physical

Living Conditions (p=.0285), the results are mixed and inconclusive. Post hoe testing

indicates the differences between the groups.

The results from hypothesis 2 indicate that there are no significant differences between

where investigating and noninvestigating agencies perceive the minimum intervention

line for the nine neglect forms. The null hypothesis is accepted. although two statistical

trends are noted: Lack of Nutrition (p= -0845) and HousehoId Sanitation (p =.0759).

The following chapter attempts to shed some light on possible reasons for these mixed

findings.

CHAPTER 7: DISCUSSION & LlMtTATlONS

me knowlredp that you can have is inexhauwble, and wltrt is inexhaustibie k Ibenevohnt, Tlie knowledge Ulst you cannot have

h of the n d d k of birUI mddrerth, and of our Wum &Wny and the purposes of God. Hen, Wen, & no knowlrealge, but illusions that twfriict M o r n and limit hope.

Accept the mystery behind knowledge: It is not darkness but shadow. Northmp Frye

In this chapter, the findings from the secondary analysis are discussed and speculations

are presented regarding a possible rationale for the mixed results. As well, the

limitations with this secondary study are noted, and the minor and major problems with

the original survey data are detailed and analyzed.

Faced with such mixed results, the possibility that the findings reflect an underlying

problem with theory andlor the tool andlor the methodology in both the original study and

the secondary analysis must be considered. Acceptance of Adaptation-level (AL) theory

has occurred theoretically and empirically. It has been well established in the science

literature that organisms, (in this study, HSP), adapt to their environments. Although AL

theory has been well tested in a variety of fields it is possible that AL theory is untenable

and the cause of the varied findings. Concerning the survey tool used in collecting the

original survey data, limitations do exist and have been outlined previously. Issues with

the Spectrum as a survey instrument include residual underpinnings, complex design,

multidimensionaI categories, and a high respondent burden. However, an expert panel

assessed this tool as having strong face and content validity. As well, previous testing

by Magura and Moses (1986,1987) established acceptable reliability of the neglect

sections found in the Spectrum. Furthermore, the Spectrum has consistentIy

demonstrated an ability to capture regional, agency and individual differences.

Nevertheless, the problems inherent wi# the survey tool predude ruling it out as the

cause of the varied findings. Another possibility is that limitations identified with the

original survey methodology, such as non-random sampling, over and undercoverage of

key HSP grwps and sampling bias or the large sample unit used in the secondary

analysis, adversely affected the results. The effects of the methodological issues as a

factor in the mixed and varied findings cannot be discounted.

Discussion

In the following section 1 try to resist superficial conclusions. I clearly recognize the

following discussion does not preclude the possibility the results are based on some

combination of a Rawed tool or other limitations associated with the secondary data.

Rather, while the comments on the results, in particular those in hypothesis 1, are only

suggestive, they are generated by weaving together a closer examination of some

contextual realities with the original and secondary survey data, other research findings,

and the position consistently argued in this thesis: child neglect is a socially constructed

phenomenon. The interlocking of these threads of argument allows possible reasons to

be presented and speculates that there may be some logic in the findings.

Hypothesis 1 : Regional SES Groups

Initial examination finds the results inconsistent and inconclusive, and many more

questions are raised than answered. Of the nine neglect forms studied, the following five

were found to have no significant difference of pc.05 between the groups:

Abandonment, Lack of Nutrition, Personal Hygiene, Household Sanitation and Lack of

Clothing.

In the other four neglect Reasons for Service, significant differences were found: Lack of

Medical Attention (p=.0001), Lack of Supervision (p=.000 I), inappropriate Child Care

Arrangements (p=+OO44), and Inadequate Physical Living Conditions (p=.OZ85). Findings

of pc.001 make it unlikely the results are due to chance. However, when significant

differences of p c.05 were found, the differences were not always in the same direction.

For example, in two of the four neglect forms where significance was found in regional

SES variance, the findings were: the HSP in the low SES group perceived a more

indusive MIL than HSP in the high SES group. This suggests that HSP in the low SES

group may intervene sooner with CPS referrals designated Lack of Medical Attention

(p=.0001) and Lack of Supervision (p=.0001), compared to the high SES group. This

finding is opposite of AL theory and previous empirical results from American studies.

Furthermore, differences were not always between the same SES groups. For example.

with the neglect form. Lack of Supervision, differences were found between all three

SES groups, but with Lack of Medical Attention, differences were only found betwean

the low and the high SES groups and between the low and the medium SES groups.

Therefore, one question is why are there significant dmerences between SES groups

with these two brms of neglect where the HSP in the low SES group perceived a more

indusive MIL than those HSP in the high SES group? Another question is why with

PhysicaI Living Conditions, were significant dirences only found between the medium

SES group and the high SES group? And why was no significant difference found

between the SES groups with the neglect forms that are most representative of society's

conceptual view of neglect: Abandonment, Lack of Nutrition, Personal Hygiene,

Household Sanitation and Lack of Clothing? Finally, why was Inappropriate Child Care

Arrangements (pr.0044) the only neglect form where the finding supparted AL theory

and o€her American studies. In other words, why did HSP in the low SES group perceive

a less inclusive MIL or CPS entry point than the HSP in the high SES group?

Closer examination of the realities of the Canadian context in the convenience sample of

six regions, suggest that in testing the SES variable at the broad regional level, the

sample regions may in fact be more similar than dissimilar. Compared to Wolock's

(1982) study where SES variance between neighbourhoods was examined, this

secondary analysis examined large, geographic regions, where all SES groups are

represented, and where the majority of the HSP serve clients from all SES groups.

Wolock's examination of the influence of the SES variable may have been more focused

because she used a smaller sample unit (neighbouhood), had clear SES distinctions

between the high and low SES neighbourhoods, and the tendency was for CPS

workers to serve specific SES groups.

Further support for the suggestion that the Ontario HSP respondents in the original

survey data may have had relatively simiIar SES contexts, comes fmm a more detailed

look at the actual spread between the highest SES region and the lowest SES region on

the SES indicators, unempEoyment mte and petcentage of lone parent femiies. The

differences are not profound- For example. Region C, (the low SES region) had an

unemployment rate of 8.6% versus the Region F (the highest SES region)

unemployment rate of 6.0%. Compare both rates to the current unemployment rate of

some upper-northern Ontario regions of more than 20%, and the low SES Ontario

region moves into an advantaged position. Similarly, with the lone parent families

indicator, Region C had a population of 161,210 in 1991, of which 3.8% or 6200 were

lone parent families; compared to Region F with a population of 313,165 with 2.5% or

7800 lone parent families. So, not only is the range between 2.5% and 3.8% not

substantial compared to a region where lone parent families constitute onequarter of

the total population, but on absolute numbers alone, the high SES Region F has 25%

more lone parent families than the low SES Region C. Trying to unequivocally establish

how poor is a poor region and how rich is a rich region is not a simple endeavor.

That said, when the incidence of low income statistic is examined across each region.

there are significant differences between the low and high SES regions: 1 'I .7% for

Region C versus 5.3% for Region F. However, while lt has been well established in the

literature that neglect is disproportionately reported among the poor. it, '...is not simply

concentrated among the poor, but among the poorest of the poor" (National Research

Council, 1993, p.9). While the number of low income families residing in Region C

constitutes nearly twelve percent of the region's population, the question that is raised is

what is "poof? Is poor in New Jersey the same as poor in Ontario Region C? The

answer is no. In a comparison of relative rates of poverty in ten countries, Canada

included, the United States ranked dead last (Ross, Shillington and Lochhead, 1994).

While the goal of both Wolock's (1982) research and this secondary analysis was to

examine professionals' perceptions of what constitutes a CPS case in their community

by studying the influence of the SES factor, the fact is, Canadian and American contexts

and soda1 safety nets, differ considerably. This point is of particular relevance for this

examination of neglect, as neglect is associated with low income (DHHS, 1991;

Gaharino, 1992; Pelton, 1994) and the poverty levels in the United States are broader,

deeper and more profound than in Canada (Korbin, 1992).

Notwithstanding population differences, three main factors contribute to this fact. First,

poverty is more clearly tied to race in the United States with nearly half of African-

American children experiencing poverty, often of a persistent nature (Korbin, 1992).

Wolock's (1982) research locale is New Jersey - a state with a high residency leveI of

Ahr'can-Americans. Since a significant proportion of African-Americans reside in inner-

cities, the geographic concentration of economically marginal families means '... a

correlation between low income and the risk for child maltreatment on both the individual

and community level [italics added]" (Garbarino, 1992, p.228) Second. Canada's social

safety net is more comprehensive than its southern neighbour, and includes universal

medical coverage, which does not exist in United States. At the very minimum, without

basic health care children are in jeopardy of medical neglect (OEC0,1990). Third,

homelessnass. violence and drug problems are reaching epidemic proportions in the

United States and these problems are associated with economic deprivation which in

turn is linked to neglect (Garbarino, IWZ; Jenkins & Bell. 1997; Osofsky, 1997).

Korbin (1992) further argues that poverty is both urban and rural, and the experiences

and effects are not monolithic, and the impact '...varies depending on the nature of the

poor neighborhoodn (p.214). The corollary is the experience and effects of poverty vary

depending on the county. Simply put, poor neighbouhoods in Ontario are not the same

as poor neighbourhoods in New Jersey. Furthermore, definitions of poverty are not the

same between the two countries. Canada's definition of poverty allows for higher

income levels to qualify for low income status plus universal health care exists for all

Canadians (Ross et al., 1 994).

In sum, the ability of the secondary analysis to adequately and rigorously test whether

regional SES variance affects HSP perceptions of neglect entry points, may have been

severely compromised before analysis even commenced. One reason is due to the more

extreme socioeconomic polarization in United States, which may mean Wolock's (1 982)

American study is better able to find significant differences between CPS workers

perceptions of severity of CPS cases because more extreme SES situations exist.

Another reason is that Wolock's sample unit was the neighbourhood. More specifically,

neighbourhoods with homogeneous SES groups. This secondary analysis used as the

sample unit, large, geographic regions, that include all SES groups. Additionally, the

unit of anaiysis in Wolock's study is the individual CPS worker. Whereas, the unit of

analysis in the secondary analysis is the individual HSP. Furthermore, the CPS workers

in the American study. tended to serve distinct SES groups, whereas the HSP in this

secondary analysis, serve ail types of SES groups. FinalIy, the influence of the Canadian

social s a w net coupled with a non-random sample of regions with relatively limited

diffarences between each of the SES indicators, resulted in a final sample in the

secondary analysis, that may be more similar than different on the SES variable.

Arguing the sample regions may actually have greater similarity than difference on the

SES fador leaves the question as to how to explain the significant differences between

the regional SES groups, especially where the HSP in the low SES group perceived a

more inclusive MIL for Lack of Medical Attention and Supenrision of Child than the high

SES group. If poverty differences between the regions are not substantial than how are

the differences that were found explained? Furthermore, why was significance found

with some and not other neglect forms? While it is clearly beyond the ability of the

secondary analysis to answer these questions, other research offers fntitful suggestions

regarding other factors to consider and possible interpretations regarding the results in

the secondary analysis.

Of the six regions in the convenience sample, four regions, two in the medium SES

group (Regions 8 and E) and two in the high SES group (Region 0 and F) had

population growth percentages of over 'TO%. Two regions in particular had dramatic

population surges. In the five years between 1986 and 1991, the population change for

Region E (medium SES group) was 23.8%. Similarly, Region F (high SES group) had a

population change of 15.446. During the same time period, Region C (low SES) had a

population change of only 5.746. in other words, the HSP in Region E and F experienced

significant population growth, there was a heavy burden on regional services to keep up

with the acute population growth, and CPS referrals mushroomed (Garbe, 1998). Yet,

the HSP in Region C (low SES group) experienced minimal population growth and had

relatively stable resource availability. Guided by AL theory, I posit that perhaps the HSP

in the regions with the significant population growth adapted to the increased population

and CPS referrals and limited resource availability by restricting CPS referrals to only

the most severe cases. The logic in this rationale suggests that having adapted to the

more chaotic and stressed context, the HSP in regions in the high and medium SES

groups will perceive a less inclusive MIL than HSP in the low SES region. This

speculation as to how the HSP might have adapted to these very different contexts, may

help understand the results with Lack of Medical Attention and Supervision of Child,

where the high SES group perceived a less inclusive MIL than the low SES group.

Support for this speculation is found in the study of Wells and her colleagues (1989a),

where resome availabiiity was found to be a factor in a CPS worker's decision to

investigate. Workers were more likely to investigate if the decision was not affected by

whether resources were available. Additionally, percent change in the population and

population size were also found as negatively associated with screening rates.

Additional support for the importance of resource availability and burgeoning population

growth (which correlates into associated increase in maltreatment reports) as key

facton in HSP determination of maltreatment, is found in the studies of Craft and Staudt

(1991). Giovannoni, (1989). as well as Wolock (1982). The studies argue that in the

context of limifed resources and more maltreatment reports. CPS systems adjust to this

situation by restricting the meaning of abuse and neglect Furthermore, Wells, Fluke.

Downing and Brown (1 98Qa) conclude this 'restrWionn is experienced by the community

HSP and they adapt by knowing which cases CPS will most likely accept Again, it is

suggested, that perhaps the CPS workers and community HSP in the medium SES

group (Region E) and the high SES group (Region F) adapted to the dramatic

population surges by having a less inclusive MIL for particular neglect Reasons for

Service. Since supervision problems account for 40% of substantiated or confinned

neglect cases (Trocme et a1.,4994), weight is given to this speculation.

Another possible rationale for the findings with Lack of Medical Attention and Lack of

Supervision is a position presented to me by two senior CPS staff. They posit the

perceptions of neglect entry points are narrower in the advantaged regions on particular

fonns of neglect because knowledge, awareness and services are absent, since

professional experience with these cases is lacking. In the words of a front-line worker

from an Ontario CPS region with high levels of child poverty, &The rich areas don't see it

f ~ o v e M "-

Support for the argument that knowledge and awareness are factors in the construction

of child maltreatment is best demonstrated in a review of the sexual abuse statistics

over a fifteen year period from one Ontario Children's Aid Society. In 1977 only 8 sexual

abuse allegations in the entire year were noted in a region of nearly 400.000, yet by

W92, that figure rose to 359 sexual abuse allegations in a region of over 700,000

(Garbe, 1998). If knowledge, awareness and recognition are key factors in the

professional construction of sexual abuse as a maltreatment form (McPhee, 1998) than

1 is reasonable to conclude that the same is true in the social construction of neglect As

well. some empirical evidence exists that suggests regional variations do occur with

neglect allegations. For example, analysis from the Ontario Incidence Study found that

compared to the other east, west and central regions, northern Ontario, (where the low

SES region is located), had the highest rate of neglect allegations at 39% (Trocrne,

1994). Is it poverty or another factor, such as professional awareness that explains

northern Ontario's higher incidence of neglect allegations? Unfortunately, neither

Trocmers (1994) Ontario Incidence Study nor this secondary analysis can answer that

question.

Two findings in the SES analysis were unique and may just be anomalies. One was with

the neglect form, Physical Living Condition, where the post hoc test found the high SES

group had a less inclusive MIL than the medium SES group (p=.0113). The other

unique finding was with Inappropriate Child Care Arrangements (p=.0044). This was the

only neglect finding in support of AL theory and the results from American studies,

where the HSP in the low SES group perceived a less inclusive MIL than the HSP in the

high SES group.

Neither result can be explained by the aforementioned rationale. One possible

conclusion: mixing regions together that have varied characteristics and placing them

into distinct SES groups, predudes any sort of reasonable answer. However, supported

by empirical findings from other previous studies (Wells et al., 1989a), a factor that may

bear consideration related to these findings, as well as a factor to consider with future

research, is examination of an agency's philosophical approach to child maltreatment

There is no evidence to support that a broad or narrow philosophical approach is

constant across all Reasons for Service within the same agency. On the contrary, earlier

analysis of the nine CPS agencies MIL found the approach to interpreting neglect

eligibility was flexible within and across agencies. While the secondary analysis has no

means of assessing what the agencies' philosophical approach is, or whether it affected

the results, it is a factor worth considering in Mure research. What is clear is that the

construction of child neglect may differ by country, by province, by region, by agency

and even by different categories of neglect within agencies.

The discussion so far has focused on trying to understand possible reasons for why the

SES group, or regions in the SES groups, perceived the MIL differently. The other

question is why are there no differences between the SES groups with five forms of

neglect: Abandonment. Nutrition. Personal Hygiene. Household Sanitation and Lack of

Clothing.

Again, definitive conclusions are beyond the scope of this secondary study, but Swiffs

(1990) analysis of knowledge produwon about child neglect suggests that in the

construction of child neglect as a social category, certain neglect concepts are shaped

and agreed upon. In particular, Swift argues that agreement at the societal level occurs

with those farms of neglect that characterize the "chronic dirt and disorder" concepts of

neglect

1 suggest that the five forms of neglect where no differences between SES groups were

found are the negled forms that essentially reflect the 'dirt and disorder" elements of

neglect: Abandonment, Nutrition, Personal Hygiene, Household Sanitation and Lack of

Clothing. More specifically, it may be with these forms that the individual, the agency

work and the larger set of social values generally agree on what constiitUtes neglect in

other words, in the construction of neglect, these five forms are clearly categorized and

easily understood as "cases of neglect". Hence, no significant differences are found

regarding the entry points.

Hypothesis 2: Agency Type

The H2 hypothesis is investigatory agencies (CPS and police) will perceive a different

MIL for neglect than noninvestigatory agencies (all other HSP). The null hypothesis was

accepted as the secondary analysis found no significant differences between where

HSP in investigating agencies (n=270) and HS P in non investigating agencies (n=409)

perceived the MIL. This result suggests that a disjoint does not exist between where

investigating agencies and noninvestigating agencies perceive neglect entry points for

CPS. It is important to note that this finding differs from previous American studies that

found differences between investigating agencies and noninvestigating agencies

(DHHS, 1988).

The results from hypothesis 2 must be interpreted with caution given the problems

associated with the original survey data. More specifically, the sampling bias, low

representation from key communi€y HSP groups, and the srnalI sample of police in the

primary survey data. Therefore, other Canadian studies were reviewed to see whether

their findings differed or supported the result from the secondary analysis.

Notwithstanding that a dearth of Canadian studies exist on the topic of Canadian HSP

views and attitudes on chiid maltreatment, findings from three studies directly and

indirectly support the results from hypothesis 2.

Trute, Adkins and MacDonald (1994) assessed rural Manitoba child welfare workers

(n=35), police officers (n=80) and community mental health staff (n=35) on three issues

regarding professional attitudes to sexual abuse. In two areas, treatment/punishrnent

and perpetrator identity, significance of pc.Ol between the three groups attitudes to

sexual abuse was found. In support of this secondary finding, Trute and his colleagues

found all three HSP groups viewed the seriousness the same, where "...[the scores]

reflected a similarity of view and were all at the high end of the scale" (p.190).

Further support for the finding that differences may not exist in where investigating

agencies and noninvestigating agencies perceive neglect entry points is found in

Troaners (1 WZa) work in developing a Child Neglect Index (CNI). He used an Ontario

CPS sample to test the development and use of the tool. Trocme found that across

professions, similar definitions of neglect were perceived in a similar way, and therefore

similar definitions could be constructed:

The panelists came from different professional backgrounds and were seasoned practitioners unlikely to be swayed by group pressure. The interdisciplinary panel consensus and the high level of interrater agreement documented in the field test further support studies that have found that professionals define and assess child neglect in similar ways (p.149).

The finding from the secondary analysis that there are no significant differences

between investigating and noninvestigating agencies perceptions of the MIL raises an

interesting consideration when this finding is combined with a related result in the

Ontario Incidence Study (01s). In examining the relationship between the source of

refsrral and investigation outcome in OIS, the researches found the source of referral

was not related to investigation outcome (Trocme et al., 1994). It is interesting to note

that while this secondary study's result differs from American research, the

aforementioned OIS finding also differed from comparative American studies (Trocme et

al., 1994). The fact that the OIS and this study are Canadian based, and more

specifically Ontario studies, and both answer no to two questions when American

research answers yes, adds weight to the position that Canada's HSP experience with

child welfare may differ from the United States.

The Manitoba study, the Child Neglect Index research, the OIS study, and the findings

from this secondary analysis, make for a very limited package of information on the

perceptions, attitudes and decisions of Canadian HSP about child maltreatment.

However, combined, these results do suggest that further exploration is needed as to

exactly which Canadian and American experiences with child mslltreatment differ, and

m y -

Limitations

The interpretation of results must take into account limitations associated with various

components of the secondary study such as the original survey design and methodology

and the design of the secondary analysis study.

Original Survey Design

The research design used by each of the six CPS agencies to collect the original survey

data used in the secondary analysis, was a survey method. More specifically, a face-to-

face, group survey method.

As the research method used to colfect the original survey data, surveys are an easy

method for examining large samples on single or multiple topics. Surveys also allow for

descriptive and explanatory analyses using numerous variables. A limitation with the

original research design, and by association the secondary analysis, is that in general,

sunreys inherently have a difficulty in addressing the context and are prone to

ecological fallacy, which is, '...drawing conclusions about individuals solely on the

obsenration of groups* (Rubin & Babbie, 1997, p. 11 9). For the secondary analysis, the

artificiality of the original survey data is a limitation. While the secondary examination

can identify where the HSP peneived the minimum CPS intervention line at the time the

survey was conducted, the analysis is limited in that it can neither state this is where the

HSP will actually draw the entry point in real life situations nor can it take into account

contextual changes that occurred during or after the completion ofthe survey that might

affect the: views of the HSP.

Future examination of HSP perceptions of CPS intervention levels will most likely need

to continue to employ a survey design. Suggested improvements to the research

technique in collecting primary data, are adding a mixed methodology approach by

including indepth intenriews with key informants, as well as focus groups in conjunction

with the survey format Administering the same survey to the same respondents over a

period of time may address some of the primary and secondary study concerns

regarding artificiality and absence of context. Keeping the examination focused on

specific forms of maltreatment is recommended.

Sunrey instrument

Limitations with the survey instrument used to collect the original data have been

previously outlined in Chapter 1 of this thesis. Weaknesses with the tool include the

reliability of the specific scales have not been determined, some scales have multi-

dimensional descriptors, the philosophical nature of the tool is towards a residual model,

and the survey instrument has an ontogenic focus, which means contextual

determinants are essentially absent as factors in child maltreatment.

Limitations from the participants' perspective are primarily associated with the high

competency level required to complete the survey. The survey requires that respondents

have high literacy, comprehension and conceptual abilities coupled with good

concentration skills. Specifically, the amount of material to be read and complexity of the

task coupled within the three hour compIetion time results in a heavy respondent

burden. The three hours needed to do the survey could be significantly reduced in Mure

studies by having respondents only complete sections on a specific type of

maltreatment, such as just the neglect scales.

Additionally, the complexity of the design of the survey instrument is viewed as a

limitation and this may have adversely affected the data analyzed in the secondary

study. For example, in other surveys most closed questions have two parts: the

statement of the question and the response. The Spectrum's closed questions have

multiple parts. For each Reason for Service respondents had to first read and interpret

the legislation, then understand the respective maltreatment definitions provided, read

all the descriptors, select the minimum intervention line (MIL), and finally, correctly code

the decisions on the survey scare sheet Furthermore, there may have been additional

limitations with the method of using the tool because there may have been some tool

effects associated with having the respondent first read the child welfare legislation

before interpreting the CPS entry point It is possible that by reading the child welfare

legislation just prior to the selection of the MIL, that HSP respondents' responses were

influenced. While problems associated with the complexity of the tool cannot be ignored,

the issues are balanced against the need to provide the participants with the infomation

needed to make an informed response.

Another limitation associated with a complex survey design is the respondent task was

complex. Generally, demanding and complex respondent requirements are not a

prsferred survey method. implications for the original survey data may be respondent

fatigue, which could affect the scoring of the scales near the end of the survey. Whether

respondent fatigue affected the secondary data by influencing HSP decisions is not

known. By chance, the neglect scales are situated near the beginning and middle of the

survey. A possible strength of the complex survey task is that may more accurately

mirror the complexity of both the question (Where is the MIL intervention line for neglect

Reasons for Service?) and the phenomenon under study (neglect). Pehaps the more

challenging process coupled with Vle provision of the entire range of neglect and non-

neglect contexts required the HSP to think longer and harder about their views on

neglect entry points, so their responses may more accurately reflect where they

perceived the MIL in their community. Other surveys on professional decision making

have attempted to understand variance in maltreatment intervention by offering a limited

option - an eitherlor decision or binary selection. The survey tool moved away from the

limitations associated with binary choice by offering a survey format that was most

certainly complex, and not without limitations, but overall, provided the HSP respondent

with richer details about neglect contexts, specifically identified each neglect area, and

provided the entire range of options for each category.

Limitations with Mure use of the survey tool are the cost. time and organizational

requirements associated with the face-to-face survey. These resource issues pose

significant problems for CPS agencies (especially multi-branch , in ner-city CPS agencies

with large numbers of HSP), and researchers. The resource requirements associated

with the group survey fbnnat are balanced against the quality of the data collected and

respondentsr high satisfaction and completion rates; rates that would not have been

achieved through administering a mail or phone survey.

Sampling Issues with Original Survey Data

A known limitation with not using random sampling is the generalitability of the findings

are adversely aff8Cfed. A limitation with the sampling conducted by the six individual

CPS agencies was representativeness did not occur at the group or individual level as

CPS staff were sanctioned to participate and community HSP respondents self-selected,

The effects of sampling bias in the secondary analysis, resulted in overcoverage of

CPS staff and undercoverage of key community HSP groups, such as police, teachers,

dodon and nurses. As well, it is possible that the limited number of HSP from the

medical community, specifically doctors and nurses, may have had an adverse affect on

the secondary results in hypothesis 1: 8A - Lack of Nutrition. This specific form of

neglect often requires a medical diagnosis to facilitate CPS internention. Additionally, in

hypothesis 2, the small numbers of police in the investigatory group (n =16) may have

negatively affected the test between investigatory agencies and noninvest igatory

agencies.

Furthermore, sampling bias limited the secondary analysis because not all members of

the defined HSP population were included in the original survey. This happened either

because the host CPS agency did not invite the HSP group, such as with mid-wives and

emergency response teams. or, as in the case with psychiatrists and teaching

assistants. self sele*on occurred and HSP representatives fram those groups did not

attend the survey. In sum, undercoverage and overcoverage of HSP groups limits the

breadth and depth of the secondary analysis.

Another limitation associated with the sampling of the original data, which by definition

affects the secondary study, is the composition of the community HSP sample. Their

views of CPS prior to the survey are not known. It is possible that the HSP in the

community HSP sample may have had strong positive andlor negative views of the CPS

agency, and this may have biased survey responses. Future research can address this

issue by including a question that asks respondents about their views on CPS.

Further limitations with the original data stem from absent professional information.

Specifically, Regions D, E and F only gathered data on the CPS location and respondent

job. Sometimes, the workplace was identified. Secondary analysis was limited and the

relevance of the study was adversely affected as respondents could only be

dichotomized into two categories: CPS and community HSP.

Lack of job specificity occurred with all regions. For Regions A, 0 and C this was due in

part to a design flaw in the survey form coupled with a hck of clarity in the Introduction

section of the survey regarding not specifying with respondents how detailed the

'position" information was to be. For Regions D, E and F, respondents were not asked

to provided this information. Absent and vague information on the original survey,

resulted in limitations for the secondary analysis. The problem is best illustrated with

CPS respondents. Many HSP respondents in all six regions, indicated their position

was a CPS social worker. The specific details as to whether it was an intake position, a

family service or other type of CPS position were usually not provided. In the secondary

analysis, these respondents were coded as %PS-social worker unknown". The lack of

specificity in the respondents' positions limits the ability of the secondary study to

aggregate the professional information data and differentiate the neglect entry points by

professions and job positions.

Specific personal and professional HSP respondent information was collected from HSP

respondents in Regions A, B, and C. Personal and professional characteristics included:

level of education, age, gender, marital and parental status, level of income, number of

years in the community, number of years of experience in their profession, number of

years of experience with child abuse and neglect cases, level of knowledge with the

CFSA, number of child abuse and neglect cases. However, this information was

completely absent from the survey data from Regions D, E and F. The absence of this

respondent infonation limited the secondary study's ability to assess the impact of

personal variables on neglect eligibility determination. As well, because only the

secondary data from Regions A, 8 and C were available for testing whether CPS

workers years of experience with CPS cases affected perceptions of neglect entry

points, the hypothesis had to be dropped from the secondary study due to insufficient

numbers in the sample of inexperienced CPS workers.

Secondary Data Issues

The secondary study used a non-random, convenience sample of six CPS regions for

selection of the regional sample. Representation at the regional level did not occur as

native, religious and large, inner-city agencies were not included in the convenience

sample. Since representation was not possible, generalikability of the secondary findings

is precluded.

A critical look at the statistical power finds additional limitations with the secondary

study. It has been well established by power analysts that a result that does not reach

significance may be due to an absence of power because the small sample size is too

small (Chow, 1996). The reverse is also possible. the result may be significant because

the sampke used is too large or a number of tests were done on the same data (Cohen,

1988). As identified earlier, the power of the tests were increased by using large sample

sues, an alpha level of 0.05, the appropriate statistical tests for the ordinal data, and

employing a tool with good face and content validity. While the aforementioned factors

to increase power were in effect, the sample sizes in the two hypotheses were unequal.

The result, is that for hypothesis 1 and hypothesis 2, the size of the samples allowed for

large and medium effect sizes to be detected with confidence, but there is not enough

power to detect small effect sizes with confidence. Since significant findings were only

found in hypothesis 1, the possibility that the results were caused by conducting multiple

tests on the same data, which increases the risk of a type 1 error, must be considered.

However, given some of the significant findings in hypothesis 1 were at pc.007 or less,

and the power was sufficient to detect large and medium effect sizes, it is unlikely the

results were due to chance.

Replication Issues

One of the goals of the secondary analysis was to replicate Wolock's (1982) test of

adaptation-level (AL) theory. The reality is this secondary study differs from Wolock's

(1982) on a number of important dimensions. Methodological differences include this

secondary study being broader in design by having community HSP surveyed in addition

to CPS workers. Wolock examined a number of maltreatment types, whereas this

examination is more focused, as only forms of neglect were tested.

Important differences exist between this Canadian analysis and Wolock's (1 982)

American study. One difference is that Wolock's sample unl was the neighbourhood.

Specifically, poor New Jersey neighbourhoods (low SES) were compared to rich New

Jersey neighbourhoods (high SES), where CPS workers primarily serviced just their

assigned neighbouhood. In contrast, the sample unit for this Ontario secondary study is

large geographic regions, where each region includes numerous communities, distinct

neighbourfmods, all SES groups, and most HSP serve the entire region. Also. Woiock

examined just CPS workers' views, whereas this analysis fccuses on HSP perceptions.

Hence, the three key differences between Wolock's (1982) study and this analysis are

the sample unit anaLyzed (smalt neighbourhoods compared to large regions), the focus

of the study (CPS worker compared to all HSP), and the CPSIHSP SES experience

within the sample unit (homogeneous SES group compared to the entire range of SES

groups). These factors may have limited the ability of the secondary analysis to birfy

test AL theory because the HSP across the regions in this Ontario study may have

adapted to fairly similar experiences. From a research perspective, the fact that

Canadian HSP serve a broad range of SES groups may have limited the testing of A 1

theory. However, at the practice level, perhaps the outcome of having HSP serve all

SES groups may be beneficial. Including all SES levels for HSP service delivery might

mean the Canadian HSP provides better service as the adverse effects of always

serving a specific SES dient group. such associated with senring low incornelpoverty

clients, are diluted.

Another limitation identified previously, is that akhough the six Ontario CPS regions were

ranked into three SES groups by three SES indicators, the actual spread between the

rich SES regions and the poor SES region on unemployment rate and percentage of

lone parent families was not signifcant. This suggests that on the SES variable the

regions may be more similar than dissimilar. This factor may have confounded the ability

of the study to find real poverty differences between the low SES region and the high

SES regions, and thus adequately test the SES variable.

Summary

Limitations associated with the tool and the original survey methodology demand all the

secondary results be treated cautiously. More specifically, despite the findings from the

secondary analysis that some neglect entry points varied by regional SES variance, the

limitations associated with the secondary data predude a conclusion in hypothesis I that

poverty did or did not cause the variations. What is evident is that perceptions of neglect

entry points varied by region and by neglect form. The post hoc testing clearly indicates

a further inquiry into the diffbrences is needed.

The results from hypothesis 2 suggest that investigating agencies and noninvestigating

agencies do not perceive the CPS entry point for neglect differently. This result

contradicts American findings. Supported by research from other Canadian studies, the

secondary analysis suggests that the Canadian experience with child maltreatment may

differ on a number of dimensions ftom the American context. If this is the case,

Canadian child maltreatment professionals are faced with a number of problems. One,

empirical knowledge about child maltreatment in Canada is based for the most part on

American research findings. Two, only a few Canadian studies have illuminated the

issue to date. Three. Canadian HSP do not know which areas or issues or experiences

differ or which are the same. And finally, there is no comprehensive national or

provincial research agenda in Canada to gather information about differences and

similarities between American and Canadian child maltreatment experiences.

A summary of the strengths of the secondary study include an attempt to advance

knowledge about specific contextual determinants of child neglect by testing AL theory,

the use of the same unl of analysis as the original suwey data (the individual), an

appropriate research question (where did the HSP respondent perceive the MIL for

neglect?), the we of secondary data that were colteded using a tool that is being

standardized end has demonstrated face and content validity, and a focus on better

understanding the forms of neglect.

As demonstrated in the discussion section, there are many 'perhapsn and few ' for sure"

conclusions regarding neglect it is an intensely complicated type of child maltreatment.

The road to knowledge development and understanding of the phenomena of child

neglect in Canada, is long and bumpy and replete with challenges. For knowledge to be

gained about neglect in the Canadian context, many, many more studies are needed.

Theories need to be tested, studies need to be replicated and the knowledge honed in a

variety of ways for answers to move from the 'perhaps" to the 'for suren column. For

example, less than two years ago, one sheep was cloned after about four hundred

attempts. Now, sheep, mice and cows are being cloned with apparent ease. The point

is that even with sound methodology, nearly four hundred failed efforts preceded

success. Such is the way of knowledge development.

CHAPTER 8: SUMMARY & CONCLUSIONS

G b n that third-party rdbnakc, made mostty by pmfessionals, account for over 75% of famtlies involved with child weffire agencies,

the basis on which these decisions to report are made merits careful examination, Nico Trocme, 11991,p.72

Summary

This secondary study attempted to advance child maltreatment knowledge on a number

of levels. First, at the practice level, because protection and sewices to children can

only occur when maltreatment is reported, and since Human Service Professionals

(HSP) are key players in referring and accepting Child Protection Service (CPS)

referrals, and as the majority of referrals come from community HSP, it is crucial to

know where HSP perceive the minimum CPS entry point. Therefore, the objective of this

secondary research study was to aggregate originaI survey data from 679 Ontario HSP

regarding their perceptions of the MIL for nine forms of neglect and analyze the

secondary data by regional socioeconomic (SES) variance and agency type. An

explanatory approach was taken and Adaptation-level theory used to expand Wolock's

(1982) American test, as well as guide this study's understanding about individual, group

and community adaptation to contextual factors. Secondary analysis used

nonparamettic tests, non-directional hypotheses and an alpha level of pc.05.

Second. as a child maltreatment form, neglect is the most prevalent type of child

maltreatment it is also a significant factor in the leading cause of death in children - unintentional injuries. Yet. negkt is poorly defined by law, has vague intervention

standards, and for the most part, this malt mabent category, has been neglected by the

research field. Therefore, the secondary analysis tried to advance knowledge about this

socially constructed phenomenon by foarsing the secondary study solely on neglect and

by examining nine different forms of neglect

Third, the broad analysis of social work epistemology, the examination of the forces

underpinning the social construction of child maltreatment, and a review of the

substantive findings to date, found that despite mounting empirical evidence regarding

the role contextual variables have in child maltreatment, for the most part, knowledge

development has an ontogenic focus. Swift (1990) argues the discourse of child neglect

is blinded by this narrow focus: the problems of individual parents who experience

economic and social deprivation. Therefore, the secondary anelysis attempted to

develop knowledge about contextual factors involved in child neglect by examining

whether specific contextual variables affected HSP perceptions' of neglect entry points.

The variables tested are: regional socioeconomic (SES) variance and agency type.

Unfortunately, there were not enough inexperienced CPS workers in the sample to

analyze the variable CPS workers years of experience with CPS cases.

Fourth, despite the need for theory testing and theory refinement in the building of

knowledge, there is a paucity of theoretical testing in social work research. It was

important to me as a researcher to ensure theoretical testing was included in the

secondary study. Adaptation-level (AL) theory was used to guide thinking about how

contextual variables influence decisions. Furthermore, replicating Wdock's (1982) test

of AL theory, provided another opportunity to try and advance theoretical knowledge. A

well accepted research method on theory comboration - the null-hypothesis

significancetest procedure (NHSTP), was used in the secondary study to test AL

theory.

Finally, the preponderance of Canadian knowledge about child abuse and neglect is

borrowed from American sources. Far too little is known about the national, as well as

spedfc provincial experiences with child maltreatment. The secondary analysis wanted

to advance knowledge about Canadian experiences with child neglect by examining

contextual factors that may influence professionals determination of neglect

With all the aforementioned goals in mind, the secondary analysis took an explanatory

approach in analyzing survey data from 679 HSP from six Ontario regions, regarding

where each HSP perceived the minimum entry point for nine forms of neglect.

Ada ptation-level theory was applied and the variables, regional SES variance and

agency type, were analyzed. Nonparametric statistics were employed with an alpha

level of p c.05. The nondirectional research hypotheses posited these two variables

would affect perceptions of the entry point for neglect.

The findings from hypothesis 1 (regional SES variance) were mixed and incondusive.

Unfortunately due to the limitations previously noted, the secondary analysis cannot

definitively answer whether SES variance affects HSP perceptions' of neglect entry

points- That said. the results where p t O O l do suggest regional contexts need to be

more closely examined as a fador in the construction of neglect.

Post hoc testing indicates further inquiry is needed to identify and understand the

regional contextual fadors that influence how specific forms of neglect are determined.

The secondary study speculates, that in addition to poverty, other possible contextual

factors may need to be considered in Mure research endeavors. Examples of factors

that may affkct maltreatment decisions are an agency's philosophical approach to child

welfare, regional resource availability, population surges, and profassional knowledge

and awareness regarding specific types of maltreatment.

The results frcm hypothesis 2 (agency type) supported the null hypothesis. In other

words, no differences existed between where HSP in investigating agencies and HSP in

noninvestigating agencies perceived the CPS entry point for all nine forms of neglect.

This finding is contrary to American studies. Coupled with a few empirical findings from

other Canadian studies, this tentative result suggests that the Canadian experience with

child maltreatment may differ from the American one. The specifics, as well as the

ramifications of this possibility, need to be researched and identified and a body of

Canadian knowledge developed.

There are a number of limitations with the secondary study. Foremost is the fact that

secondary data were used and therefore, the research is vulnerable to all limitations

associated with secondary data. Minor limitations include the design and methodology

used in cdIdng the original survey dab. More specifically, the absence of professional

and persona1 information on the HSP in Regions D,E and F; the lack of specifcity

regarding the HSP position within an agency; and sampling issues, such as the different

respondent selection methods (sanctioned for the CPS staff which resulted in

overcoverage and self-selected for the community HSP which created undercoverage)

and small sample sires in some of the key community HSP groups such as police,

teachers and family docton. Additional limitations associated with the original data stem

from the possibility that the community HSP who did respond, did so because they were

biased toward extreme views towards the CPS agency. Limitations associated with the

tool are a heavy respondent burden, untested reliability and complex design. Minor

limitations related specifically to the secondary analysis are the unequal sample sizes in

hypothesis 1 and 2, which adversely affected the ability to detect small effect sizes.

Major limitations with the secondary study include the sample unit used (large

geographic regions instead of small neighbourhoods) and the fact most HSP in this

study serve all SES groups, not a specific SES group. These factors compromised the

ability to adequately test whether HSP perceptions were affected by regional SES

differences. Two direct consequences ftom using a convenience sample for regional

selection is the inability to generalize the secondary findings and the lack of clear SES

distinction between the regions.

While taking into account all the limitations associated with the secondary study, the one

key question is: was knowledge advanced? How confident am I in the quality of the

knowledge produced? Most certainly the combination of some or e l study limitations

hindered the full potential of the secondary analysis to adequately test the independent

variables using AL theory. However, I posit that knowledge was advanced, albeit in just

a few, but nonetheless. important areas.

The hypothesis t results add to the current body of knowledge from ecological studies

that posit regional variation is a factor in the construction of neglect. I am comfortable

with this condusion despite the multiple tests on the data (which increase the likelihood

a significant result will be found) because with a few forms of neglect significance of

pc.001 was found. This suggests the results are not due to chance alone. This finding

directs Mure research to strongly consider regional factors in understanding HSP

perceptions of neglect.

Second, this secondary study advanced knowledge about neglect by identifying that

HSP within regions perceived the CPS entry point differently for different forms of

neglect Again, this position is based on when p c.001, making it unlikely the findings

are due to chance. One speculation is that for those forms of neglect where p c.001,

that these neglect forms may be more efastic in their construction than the neglect forms

where p >.05. In other words, with some forms there was clear agreement by the HSP

about the entry point, and yet with other forms of neglect there was significant variance.

Future research needs to advance our understanding about how and why each neglect

form is constructed. As well, future research needs to examine what factors affect or

do not affect construction of each neglect form at the different levels: individual. agency,

regional, provincial, national and even international.

While knowledge generated by hypothesis 2 is limited, the fact that the results are

contrary to American findings advances the argument for the need for Canadian

research to improve understanding about child maltreatment in Canada.

FinalIy, as a researcher, I gained considerable knowledge about the need to ensure that

research is rigorous. Weighed against all the "right" things that were done - the

inclusion of theoretical testing, a detailed examination of the substantive issues, the

completion of an extensive literature review, providing dear rationale for the importance

of the research questions, ensuring the units of analysis and the research question were

appropriate for secondary analysis, ensuring the appropriate statistical tests were

employed, and understanding the strengths and limitations of the survey tool - are the

limitations of the study. Specifically, a regional convenience sample, a large regional

sample unit, and unequal sample sizes. The limitations associated with the secondary

data and with the secondary study itself, mean the final results are presented with

caution not confidence. I have learned invaluable lessons from this experience regarding

the benefits to knowledge development when research is rigorous.

ConcCusions

The need to better understand HSP perceptions of minimum child protection entry points

is not diminished by the limitations of this secondary study. Assessing HSP perceptions

as to what constitutes a child protection case remains an important and viable research

agenda. HSP are, and wilt continue to be, key players in the detection and acceptance

of child maltreatment referrals. Where HSP perceive the minimum intervention line for

CPS cases is critical to know for incidence rates of child abuse and neglect are affected

by detection and by reporting.

Clearty, inappropriate child protection senrice to children and families that are beyond

the CPS mandate is a concern because it means CPS intrudes into their lives

unnecessarily, and limited CPS resources are provided to children and families not

intended for. However, the far greater concern is for children, who are either not referred

to CPS and should be, or not accepted for CPS interventions, and need to be. The

ramifcations for these children could mean not only continued neglect and abuse, and

long term mental and physical health consequences, but death.

The research challenge in better understanding this complex phenomenon called child

neglect, is disentangling the interaction of the factors that impact on HSP eligibility

decisions: personal, professional and contextual. While suweys will remain useful for

studying large samples of HSP, a multi-method approach is needed. Other research

methods that may assist in illuminating the issue are: key informant interviews, focus

groups, and comparing vignette decisions to actuaI case situations. Wedded to the

research on child neglect is an examination of how society, regions, agencies and

individuals construct the different foms of child neglect

Perhaps what is most evident to me in the completion of this secondary research study

is that the factors involved in the construction of child maltreatment in Canada need to

be examined from the perspective of the Canadian context Continued reliance on

American knowledge to explain the field of child maltreatment in Canada is a concern.

American knowledge about neglect is knowledge that may or may not be relevant to the

Canadian context What is the effect on practice and research when Canadian HSP

know more about the characteristics of child neglect in the United States than about this

prevalent child maltreatment form in Canada? Clearly, Canadian child maltreatment

research needs to occur, as well as research comparing the two countries experiences.

Furthermore, despite considerable empirical evidence supporting social and contextual

issues as significant factors in neglect, the continued policy and practice focus remains

on individual, ontogenic factors as the sole cause of neglect. This position limits our

understanding of the issue and compromises our service to maltreated children and their

families. Responsibility rests with the child maltreatment research community to continue

to lobby for recognition and inclusion of social and contextual factors in policy, practice

and research arenas. Quite clearly, community, regional, provincial and national

contexts are important factors in the construction of child neglect in Canada. Finally, in

order to refine our knowledge there is the need to test and build upon current theory.

There are many ways of developing and testing knowledge. And all need to be

employed to examine the broad area of child maltreatment, as well as the specific forms

of it, such as neglect.

In the end, analysis alone wilI not lead to solutions. Political leadership from the federal

and provincial governments plus the cooperation from the Local CPS agencies is aIso

needed. The absence of collaborative efforts to conduct and fund national, provincial

and local research endeavors, and to establish clear standards, common definitions and

consistent intervention points for child maltreatment, in particular for the more nebulous

yet ubiquitous form of 1 - neglect, means child abuse and neglect will continue to be

experienced and the adverse effects of it felt by significant numbers of our children long

into the next millennium,

Appendix A

Ontario-Child- . . . . . . . . . . - . - -. Welfare . . . . : * .- - .

CHILO PROTECTION ENTRY POINT.. .................................................. 4

Vll ABUSURlSKOFABUSE4NADEQUAtESUPERVlSlON ................................ 33 A) SupanrirbnofChiIdmn ..................................................... 34 8) Asnngemnt Ar Child Cur .................................................. 36

Vlll ABUSURJSKOFABUSE4NADEQUATECHllOCAFIE ................................. 38 A) ~ m n l ~ m t ............................................................. 39

..................... ............................ 6) PfiyricrE M g Conditions .. 41 C) PemMHygirw ......................................................... 43

........................................... 0) HouseholdSanitrtion ........ .., 45 E) Clothing ................................................................. 41

Xf UNAITENDED MENTAL, EMOTIONAL. DEVUOPMEIWAL, CONOll'ION ................... 60 A) P u r n f r C R e a p n s e ~ C h W s ~ ........................................ 61

....................................... XII CH IDUNOERl2COMM~S~ IOUSACIT 64 A) P8cenW Response D C W s Wmvim ........................................ SS

................................................... REU€JESIFOR COUNSELUNG 74 ............................................ A) wl(u.Cou- Sewha 74

(I) ( h . ~ ~ h r r ~ ~ t t h . c h ~ p u m t b d ~ o t b u n m i l r b k l o a x o r c & e h i r o t hr CUttOdirl mhb owr thm eblId and h n not made d rq~#b . proviaton &r orthe childs& un ud w, =(A. rhld h in 8 midanChI phwwntmd the pumt mfuru or i. unrblo or unwining to mum th. e t ~ w s a m urd c-

C h i l d h u h r d O n a i v r o n b r a n ~ ~ o f ( i n r b y a p . n o n w h o d ~ n o t nomdly are lk kim But parent did not kwe rbcup(ly. Parent maintained some contact d u f h $ ~ 8 b m c a PlnnthuJwRyrntunudO~1umec8fedngorisexpecWdb nturnshonly..

(e) tfae child nquims madial tmmornw cum, p m m t or rlkviate phymial ham or au(hring md tho childa* purnt ortha pmon having cham oftha child doas not -vide, or mftmos at l. unni l rbk ot unrbk 0 wIUantt0, the tm!!M*

~ h ~ J w y r n q u h p m m p t m s d ~ r t t b n t i o n v ~ o n m e m a n y ~ t g . . bngbona mcbn#, intoml injufies lhkd degree (mortmmm) bum% brain or spinal cord injuy; eye injuy; deep nrounds or puncbrm that could r#ut k ayamnk inIbcltkn. C

b) E%cnsh or lnappropri.1. Phyakat Fom Wad, lbul(lng in Modentdy Sorkla injury

Prime aretaker did not haw knawkdge of this m o t did not rHaw it to occur,

Prime areWcer did not h8v8 Fmowiedge of this mdbr did not rlbw it to occur.

(Agency intervention hem lior infomation sharing would be opened under "Request lbr CounselIhgW.)

No RISK

It . -m Chifd sustained no abusive m a 1 activity.

(O) than b aub.1.lrml h L (M the child MI muthr amotionaI ham of tiw LM deat- in C chu8e 37(2M() and the cklld's prmt or the prron having chug. of the chtld d o a not p H * , of nfuur or i8 unmibbk or utubk to conrrrrtto, 8owkas or tnrbnant to p m n t ma ham;

PC)NSE TO C w O w e CC)NDITlON

Mnftions of Child'. Emobionrl Conditions

(a) ChiU at &tiom b k to &If md Othon ChlWs rmo(iorpl condition (Le. aggnrjon, anxiety, drpn#ion or wimdnml) places him in situations whrrs he h 8t 88rious rbk to himself or others fie- child hss attempted suicide, child return ta shaths whem shehe has h n beam or mpad* chM has swbusly injured another P-W

(b) Child u ~ b k to Fundion in One or Mom Major Rolos Child h8s an emotional condition @e. aggression, anxiety, depression or withdfawal) that makes himlh.r unable lbnctbn at all in one or mon Mjor roles (q. in school, with ftiends. or u 8 bm*ly mmber),

For situatiorrswhem theemothul condition rppun to be a m mult of adult conflict in the home see Section X (6) - Pmnt Pfobkm -AdultConIlia

No dWcd symptom of mrlnutdtbn or dehydnbion, although child may have become very hungry or thinty.

A 9 I, A-S A€ h a result afthe cnad Wing ddhmtely I o c W out oraxpslkd Crom the home, although the

pannt is in 8 poaffin b admit the chiId otto make m appropriate ahmat8 amngement:

child sustains 8 severe injucy or illness tJmt usu8Ily quires hospitalization.

child b seriously vidinbd (usrubd, kidnapped, ek).

Indudes .y ckild who has no #h place to go (mfltivs/frisndlnaighbout) or who is not old rnaugh or aprbk rnough b go them.

indudes m y chiid who has been out wwml hours or mom in veey bad weather. or who is 00 young b - stmas sably- C

ChildisdenhdrartOhbofhefhamrorarprl#~kDmr H e o t s h e h a d r o ~ to go (nhtive, Mend. neighbout) and is old enough or apobk enough to go them

0 ChW conlined (P cwm Rr extensive period of tbn (depending on the age of the child)* C W con- k my amped or drk mnkwm (egg. clowf bin. shed) br my pwiodofthne. Child not rllowsd outside Ilwr weekor mom.

a Any 8amory deptfvrtkn or p h w m ~ t in frightenhg sitmWnt Child b hrmaurd, W. or bound Ibr r dry or mom

There ir potenW Ibr physic& mdlaremotionrl ham, but the child has not yet sustained m injury rr a mutt of om of the Iblbwhg aebjw:

Child conltnd 0 roa JI day and night or not allowed O go outride Ibt several m- child's movements m phyricrlly tartritbd by harnessing, tying, binding, ek., but hr no mom than metal houn.

Child h nuw drPb.nMy c d n e d , lied, or bound in r y my as r means of punishment

Child b plrd h a v 8 ~ threatening shutbn (ag. hefd out of window, held O W rolldihg m(v. drfib81M8@ rilawsd 0 wander where potential fbt injury is high, eR).

No acbml injury or harm ocwn, though child my have bean frightened.

Tknrb are ruch that, ifanid out, physical oremotional ham to the chY could msuk

ImMed would bethnrLI of physkal abuse, deprivetion of- orwater, sexual abuse, etc.

Them hPI bnn no rtlampt to a r r y out such th-.

No d m and sprdlk themats of abwb or harm am made.

Pamnttgmrdii says they "Ibrl warwhelmed by the chW, "might hurt child", "faat child might have m ltCidentC, wget so mad at chiM they don't know what might happen", etc

indifhdthrwltt are of a qW&y which b8d the listener to bekve them is a danger of mjuy or neglect to the child. b m p l r r indudc ributkns invoking persons with a history of mentat health probCem or ovewhdmed prnnts with v a y smal! children,

VII ABUSHRISK OF ABUSE -INADEQUATE

A child who b to hndk basic needs (e.0. eating. Wet, avoiding accidents) b bR abm with M hadequate al tmath angiver (eg. another young child, adutt invalid). The pmnt does not nbrm M r e the chitd% mads beeom rateL The chi@ is at least ~~o(kn8l ly dbtnught or hungry, and may have had m 8ccident, but no injury muI&d.

md lor

In both as8s hospit8lbtbn b mtordinrrPy required (k mrdierl nuom (although dJ# may be temponnly horpiCJiud lbr his pmtdm) .

Then m one or two obvCourly hrurdous conditions of signfieant nature in the home. ChM has surtrinrd r physial tnjuly muiring medical m m n t 8s 8 nwlt

Child does not h8ve any plrcs of residence or the limiiy b experiencing acu!e shelter problems (e~. no heak elactrfcify, etc.), Thk m y indude a fom~@ l i i g in other than tndiinal ddential housing (eg. a m tents).

Chitd has sustained r physical injury andlor ailment u r mutt

7hem am me or two hazardous conditions of a significlnt nature in the home but child has not sustained inju y u a muk

4. a a u M a m E

Them me no obvious hazardous conditions m the homer,

Complaints about child's hygkrw kM b n n made by @tern (school. ek). Pews will not pby-WithchUd,

Cupot tllc, wab, dews, bathroom fktwa ur bywad with 8mc~tod dM, dab-, IbodwuQI; humulor~fumsteprominmt. RJekdurtMd@l 'U8e~~rythine. b l t s n m r m d b ~ l y o f ~ ~ throughwt T'mdjimkpikd upmdfy~ontheIkorrothrtitbdHticolto getaround. DkW u, not washed; frmAy rrtr Mdhty dbW, of doeant use them Pofbh8bklbodt lbund rpdkd; spobd ibds not dirvrrded. Heavy rodant hhr0ltbn; awphO nmJn lurn %lun ornf. F u n i C y ~ ~ o r r d i l t y ~ , o r o n l E r n ~ ~ ~ w l t h d i r t m d m i I -

As r m u k of thb poor #nitattom, at lust one child b phyticrlly ill ( age htestinaf disorder, b o d ~ ) ~ ~ ~ ) c r J ~ t

Carpet tile, walls, dwn, brthroom lbQnrr us ltyrnd with rn~l~lsted di* dakb, bodwmbs; h u r r m n w m i m r ~ w r r d s ~ Thickdu8tmd:g~eortrvvythingc Home mlb ovawhdmhg@ throughout Trash and junk pikd up m$ laywad orr the floor sa that it it dMcdt to get amund, D ~ r r r n o t w r u ) N d ; ~ # c r ~ d & t y ~ , o r ~ ' t u s e t h e m kfi8h.M. Ibodr mnd rpdkd; 8poad bods not disardd- H u v y r o d m t ~ * ~ v r n n i n h n n " b m ~ r . F ~ r l w p . o n ~ ~ , o r o n Prm blwkwi(hdirtand8oiL

Unt#y, dusty, mhotdkt bubui#* CuprtndOI immrnypr( ic l r ra fmmdmrpaad,~ lvrptron#b*mb& nnly-• T-, rhrhnr. -.nvry dur(y: cobwebs b cornan. Strk, odom. Olrbrge not lraptin proper receptacle. Home €8 notp#od up: mi- m JI ovrr (but no "phC of tfash). Dktydb)NIIlyuwndCwm;wunadrtnightotne%tdry. Growdm hy dl mund (butpubh8bbAodsgemnlly~nted). Eiorn anplnO Hrmh. but few in number, 8ppruinO minty at night (no rats). WUk. windowst doom, br(kmom (Mums m spo!bd. stained, rtrukd with dbt C

a ~ d n n m d o r d r r l y ; n o v 8 n n i n ; ~ n l i r m r r ~ a C I ~ t m d ~ m p t ~ n m d ~ r r m d l d ( b u t r o r n r I i ~ Umad. ppet

renp,myk-)- Ruguhrdmlhg (no mon hn Wm hyerofdurton mb).

a Plw~mlt bD W U ~ odo~n. a Home is ordrc(y (but mtkks kr daily living rmy be mund: new8prp.n. boobB

rrtr not hung up). e D b h e s w u t r r d ~ r t k r r t p u t h l i n k ~ ~ m r r l , * G ~ ~ ~ *

Chsd is irc#ng mr ny buic and u8entbl i(mr of clothing.

Them m ro hw clothes. or so hwr of the right kkdr of clothes, that child b sometimes u- to pwbm n o m l and ~ t a # l r y actkith (going outdoon, to school, etc.)*

Cmhes am worn and of!^ obviouNy mended, but not ripped, tom, or dMy, I

Same clothes fnay not M properly (dms length wrong. unable to button).

A thlW b in nnd of p-n whom,

a) thm 1.8 *uk(mnthl h k that tho child will sufht amotknml ham of the kind d.rcdbd Jn

Pamt has sewed him or. her datiomhip wim this petson. or maintains only a Gmitad rmwomllip.

A child b In n o d of pmtacth whem,

No RISK

The canditbn my be a r a k g corn pain. dimmfbrt, stms, or krt of tiam during me child's i h ~ ~ ~ u A o ( h m 0 r m b m i n o t d j ~ t r i n n W M u w i m t h e t h i i d O

8cwmWab hbn or hor.

Thechld wfbm R#n 8 mMl, motbd ordeva@man!aI concTttkrr n defined in (M) 8 b . th8t Wnotnmdird. eaWwrfoudy implitlh.chYGs dr~rlopmrnt n d the k.eh968 p w r n t a r p n o n ~ ~ o f ( I r c W 6 # r n o t p r o v i d . ~ m ~ o r b u r u n i b b ( a o r u r u b k ! e c u M m t b o t h o s e ~ ~ ~

Adequate ulprnririan i8 mcmsay b pnvsnt r recurrence. The child's pmnf or person having charge ofthe chiid does not provide adequate aupenrbbn ibr the ehibd,

T?m p r n n t p m h s proper and tbmly urpsnririon of#m child's retivitiat inside md outside ofthe home.

Due to confkt betwaen the p ~ n t and child. the child ad in danger of aperotion fbm the frrrdry-

O(hrrMly mmkn tend to m i d contact with child (or chDd tmds to avoid contact with them)*

Vm y Sarious Mbeanduet (Vkknt or Felonious Bohrviour):

(1) th. ehad has baon rbur-IHd, th. thlch purnt has dkd or h unavaikbla to axmclcbe h b or hrr custodtrl rl~hg war tha chUd and h u not nudo rdrqurt. prwbkn bt the child% an a d wsbdy, ot the tM# b in 8 mid.ntW plrcmrnt md (ha prnnt fmfbr or is unrbh or unwilling to nrulnr tho child's un and c-;

Child has bwn rb~pt ly deserted o t abandoned by hi parent ot guardian. Them is no C n d m that ths pamnt intends to mWm.

Chi# has ban abandoned In a residential pfacament Parent mfiises to mums caring fbr me child

) (he ehlld n q u b m d k I tmotmant (o cum, p m r n t or rllrvlrtl physkrl ham or rWerlng and tha childt& parant or the panon having ch- of the cMW doas not provide, or nfuses or b unrvrilrbte or unrbk ta mmrnt to, the t tubnae

VII - I ~ Q U A ~ E SUPERVISION

A chitd ir in n o d of probction wkm,

(8) the chlfd has rulhnd pkyrkrl )urn, inflcWd by the penon having ckug. of tha chAd ot uuod by thrt pmon'8 Wlum (o a m and pfwlde for or supowbe and pmtact tho child WwUCbly;

(b) ( h e & 8umnt i r I fbk (M tke chW 8-t phyllerl ham iMkbd ot alllad 88 dacrlkd in ckma s7(2)(r);

(c) !ha chUd ha8 boon rrrturlb mohstod or semally r rp lohd~ by tka panan having charge of th. chNd at by mothat panon We# the prnom having charpa of the child knom or should know d the pauibEll(y of 8axu8l nrdntatlon or amxwl axploitrtion and hlb to p- the thud;

(d) !ham k r aubrtrntirt rbkthrt thr chNd wilt be saually molostad or soxurlfy axplobd u dncrlkd in ckrao W(2Mc);

(t) the chlld has su2hmd emotfanal hrm, dmnonstmtmd by swan, 0 rrukly, m dw-h, (UI)withdnwrl, or (!v) 8 e K d ~ ~ ot r g g m i w khrviout, and the chIWs purnt or the pmon having charge of the child doas not pmvlde, or mfurr of b unrvlikbk or unrbk to eomont to, marvku ot traalmnt to nmody or rlCwIrt8 the ham;

(a) t h m k i subtarrtkt rkk that the child will rulht amotianal ham of tha kind dtreribad in clrure 37(2)(t) and th. child's prnnt or the prmn hrving chwge of the child doas not pCMI/Ckf or mu or fs unnnihbk or un8bh to consant to, r e w b or tna!!nont to pmvmnt the ham:

Any p m m h.vbrg charge o h &l#, lus than 16 y u m of age, must mrk. rarrombk provisimn Ibr t h . ~ ~ 8 ~ p e ~ b h . n b c u l r , a n 8 ~ r f n g t h s ~ ~ b ~ ~ ~ h r ~ Thepenoninthrfge

child b kR, md the competency of lhe chiid and/or amgiwr in meting basic needs (eg. eating, DiWhg and obtaining kdp in annrgencks).

When the child is less than 10 y a m of age, the onus of rrbblbking that provision f& the chiWs supenrbian md urn md ulbty was msonabk under the d r e u e , rsrtr with the parent or amtafmr.

The child has km improprrly supervised by the prmt k a msuk tha child has bean injured, requiring medicat treatment, or h u been victimized (mobted, etc.),

No child has k r n physially injurad, however;

Parent exwcWs We w ~ ~ n over a younger child, either inside or outside the home. The chDd m y hM been lknd pfaying at home wim o b m that could hurt him The thld my have bwn lbund pl8ymg h unrdb dfwms!anm outside (e.g. in t(naf in a dump. or with older stnngem). P m t okn does not know whefe one af the younger children is. ChlW wvdrn 0 ~nfrmilivafNS md rametimes m d s strangets hetp to m m h o r n In gem* ibmily's you-childnn are given br&o muh mponribitii Artheirown rrkty.

Pamt uurrlCy has no idea what child k doing md m k no attempt to find out Child is known to be W i .

Pannt has d i i k y supenrising the younger child- Parent tends to kove younger chiM unohrvd md does not always know what he ir doing, but do know the child's whereabouts. C W h &n ggetting htoa things that he shouldn't. Sometimas the child b found rngaghg in rough me Hwmmr, no chW has aver been injured u a result ofthis- Pamt would be able to respond to emergency rituatbn in appropriate time hrne-

Parmt provides proper and timely supofvision of child's aclSvilies inside and outside of the horn*

Parent knmm child's wharwbwts and actWes, whom he is with, and when he raturn. Defirlite l imb am set on child's activbs-

A child who is u ~ b k to k u d k basic needs (eg. a m , toilrt, avoidkg accidents) is kft dorm w[m m inadeqw rltsrnatim angiver {eg. mno#tr young child, adult Wid) . The pannt doe8 not m!um bebm the child's needs becorn acute, and during that time an accident aecumd causing some injury to the child.

A cksd wko b unable to handle bask needs (60. rrting, Wet, avoiding accidents) is Ieft alone with m inadequate aItsmrdCvs caregiver (eg. mother young child. adult invalid). The parent does not mtum bebn the child's need8 become acute. The child is at least rmOfiOWly drstnughtor hungry, md m8y rrPy hnn an accident but no injury resultad.

A very young child L never left done or with m inadequate alternative eongiver when the p a n t goes out But an older chad rbfe to hnd dbr hiiWrowmes doas not know wham his pmnt ir at nightot when he or- will return. The a i d would be abk to get help in m emergency if necessary.

Pamnt makes safe and rpprop- s u m chifd a m rmngermnts when needed (including IWyMbg and ovamight mngements). OK ChW is old enough ra !hat he does no? nonWJy mqub. amngemnts ibr subs!itute chtM am*

the child fin 8ul l .d phyricrl ham, MieW by the panon having charge of the child or a u u d by that pornan's hnum to can m d provide fot or supawbo and protact tha child adq~8t@fy: than b ruI#trntbC rbk thrt the chifd will 8-r physical harm inflictrd or truamd u d u e d a d In ehmm 37(2)(r); tho chiH has brrn rrxwlly mohtad or saxually exploitad, by tfw panon havCn@ charga of tho child at by mnothot porn whan tha w o n having charge of the child knows or 8hould know of #a pomlbility of raxurl m o ~ t i o n or 8axuaI exploitation md hik to pmtrct the child; tham b a aubrtrndlrl rCIk that the child will k sexually Mastad or rrxurtly oxptoM u dactibod in e k m 37(2)(c); tho chlld has 8ufbnd mmotional ham, darnomtmbd by swam, (0 IIlljrtY, (il) dapnuion, (ni) Wmdnwrl, or 0 roW4oatmtthre or rggnrrhn khrviour, and the chUdes prrmt or the panocl having charge of the child doas not provide, or mfutas or b unnnilrble or unabk to conrrnt ta, sowicas or tnatmnt to nmedy or aUevirt. the ham; them b 8 substantial rbk that the chnd will ruffat emotional ham of the kind dmcdbad in elrumo 3?(2)(t) md ma child'# prnnt or the panon hrving charge of the child doas not pmvido, or mhmm o r b mmmilrble at unabk to emmnt to, urvic# or tnrtmorrt to pmvant the ham;

and hr

Chiid is su(hring from dinial symptom of malnutribion or dehydmtion mquiring hospHllhrtbn (q. nm-organk Wun to thrive).

Child suffWs from wnw clinical symptoms of malnutrition, dehydrstbn, or fkmd poisoning. Madid attention mdlot r e h a b i l i i dkt nquired-

Mub m imgubr and often not prepared at all, but usually one meal a day is pmvided.

Child may have been seen scrounging br bod.

Child is generally geIthg enough 0 eat, but pmpamd mrJI an often nutritionally unbalanced and skippud.

Chi# gets bod h i l f at home or suppbments his dirt outside home.

Child is pmided with mgulor and ample meals that usually met basic nutribjonal raquiremrrtr.

Examples of Hazardous Conditions:

Then m one or two obviously hazardous conditions of r sfgnbnt nature in the home. Chad h u sustained r physiaf injury requiting medial treatment as a mult

mdlot

Child does not have any pbce of fesidenca or the fPmflj b experiencing acute shelter pmblems (eg. no h u t ekcbicity, ek). This my indude a kmily liMng in other than t m d i i ~ l raridentirl homing (eg. cars tents).

ChiW has sustainad 8 physical injury and/or ailment as r msuk

Rm are an o r t w hazardous condib;w of a significant nature in the home but child has notsus?ainad injuryas a muk

This may hdda r frmily r i p in other than ~~ residential housing (eg. cam, ten&)

4 aaGMmhm There are no obvious hazardous conditions in the home.

ChDd b not wfhffng fknt any ibsses dw to poor prrsoml hygkct8 but condkions are such that thy could s a r illness

Complaints about chius hygiene have been mrde by others (school, etc.). Pews WJ1 not ph)wRtkchi

C.cp8C tib, wW8, doon, bathroom IbUttfes are layefed wit?! encrusted dirt, debris, ibod w8stms; humn or animal waste prominent Thidc dutt8nd g m m cmtrvrrythfng. Home rmrul ov-ming)y of u r l ~ s p o i i a g e thmughout Trash md junk pikd up and byefed on the fbor to that it is dHlia~lt b gat around. DQhrr am not washad; fomily eat8 offdifty dishes, or doesn't use them. Parkhable fbds bund spoiled; spo~hd bods not d i i a d . Hnvy d e n t MhsWo~ -ping vemuh haw %ken wet frmi sleep on dkty mWeme8, or on linens black with dirt and soil,

Household exhibib mny of tha wnditkns described be-

Carpee Wq walis, doom, br!hroom (bduns m layered with encrusted dirt, deb&. bad wastes; humn or mimal waste prominent T l W dwt md gmme amt everything. Home smlb werwhsImingly of u r inWspo ikge throughout Truh md junk pikd up and l8ymd on the float 80 that it is dimcult to get around. Dishem are not was- lbmity a8b off dirty dilhu, or doesn't use them. Pmri8h.M. Ibods bund spoikd; rpoikd bods not dbcrrded- H a m rodent inbsWm; moping vermin h8ve "taken wet Family Jwpr on dkty -. or on lium black with dirt a d rdL

No chiid ir ill as 8 nruk but conditbns 8fe such that thay m at high h k Lr becorning ill.

Household exhibib mury ofthe conditions desdibed k k

Untidy, dusty, minor dirt buildup. C r p l C M d ( 8 , h n n m * y p r ( l c b r o f ~ r b m d m ~ , ~ p t r a m t i m b ~ nnly-, T m , Wfves. objeds am veydur(y; cohek in cornea Sbk, stutfy odoun, GmMge not kept in prow ncsptackC Home h not picked up: things am rW om (but no "p%rW of trash). Dirty d m lay uound how: -had at night or next dry, 0- by rll uound (but perbhable foods genemlly mMge~otad)~ San cmphg vennh, but hw in number, appuarbrg mainly at night (no ntr). VWb, windam, doom, brttrmorn IbQm are spotted. stained. streaked witk die

a O.mnlfy dmn md ofdam no vermin; clun linens. c u p . t m d t i l , r m p t ~ n m d ~ l d h r d ~ ~ a d ( b ~ r o r r w ~ n t t h ~ . p p u -91mryb--w-

a RegrJudWng (no man than thin layer of dust on bbkr). Pt rHnt to neutmt odwm. Horn is orderly (but ub'cl# fbr d8ify living may be amround; newspap.m. books, coats not hung up). ~ r m n s d o r r t I n 8 t p u t i n r i n k ~ ~ m s r l ,

8 G ~ p f o p 8 d y 8 m 8 d .

Child is W n g mny bask and essential items of ckdhing.

Then are so kw dothe$, or so kw of the right kinds of clothes, that child b sometim unabk to peform mmul and necessary ;ilctMti# (going outdoom to school, eR).

Child is M n g mrny basic and essential bms of dothing.

Them are so hw dothe& or ro few of the right kinds of clothes, that child is sometimes unable to perfom normal and necessary activitb (going outdoom. to school. ek).

Cto(h- m romrlilnr insumcient to protect child trom the dements. but no inju y has yet occum*

Chiid b &sing orte or hwo essentiaI clothing item, or some essential items are in such bad condition that timy ahouidnt be worn,

Chi# mty be impropeafy d n w d lbr some of his n o m l and necessary activities. (8.g school). but he b abk to partidpab or attend.

ChM has all .rrmntiai clothing hems and emugh chmggas of clothes to be neat and dean.

Clothes need not be new, but are in good condition and fit adequat8ly.

Cloths are consistent with tima, p h , md weather- For example, child has seasonally appropriate clothes (nirrwsu, gloves, IigWight items).

Appendix C

MOST RlSW MEDlUM RlSW LEAST RlSW NO RISK1 SECTION SEVERE SEVERrrV SEVERrrY SEVERE CFSA (3) - PROTECTION

A - ~ n y ~ r ' c e t Maith cam f, 2 3 4 5 IIf Physial lRhk of Phyricsl Harm

A - Physic81 AirnSC1plrne ' f , 2 , 3 , 4 , 5 6 , 7 , 8 , 9 , t O t3,12 I 3 N SQxwURirk of &%d Ham\

A - Ahu~ivs Sexual Ad* 1, 2, 3, 4, 5, 6, 7, 8, 9, to I 1 V EmotionrURisk of Emotional Abuse

A - Parantaf Response to ChWs Emotional Condition 1, 2, 3, 4, 5, 6,

VI AbudRi~k af Abuu - C-l T-nt A - Deprivation o f FooaVWater 1, 2, 3, 8 - Delibetate "Lockr'ng O N 1, 2, 3, C - Phy&c8S Confinement 1, 2, 3, D - 7'hmt of Abuse 1, 2, 3,

VII Abuse lRisk of Abuse - Inadequate Supewislon A - Sups~3ion of Childmn 1 , 2, 3, 4, 5 I B - Unngemnt lbr Child Cam 1, 2, 3. 4, 5, 6

Vlll &u&k of Abuu - Inrdaquatm Child an A - NuititioMMtt 1, 2, 3, 4, B - PhysiiI Living Conditions 1, 2, C - Personal Hygiene 1, 2, D - Musehold Sanitation t, 2, 3, E - Clothing 1, 2,

IX Risk of A ~ U U - Ability to Prdrct

A - Pafentat Response to Child's ConcTrtion 1,

XI[ Child < f 2 Cammittad Sariour Act A - Patenhi Respon~e to ChiM's 6ehavrbr 2* Prnnt Chiid Conflict A - Chimmiiy Rdations +I,

XIV Child Bahavior Pmbkm A - Chihit's Misconduct I . 2. 3. 4* 5, 6

CFSA (2) - PREVENTlON a CFSA (7) - ADOPTION 8 OTHER SERVICES XV R q m t for Counding

A - Child Wefim CounsUrg Se&s : 1, 2, 3, 4 XVI RequestfotAdoption S.nrfcas

A -AdoPQcon SeNEcC&S: 1, 2, 3, 4, 5, 6, f 8 -Ado~b~~-bisdosum: 4, 2. 3, 4

XVlC ~ w s t t o r A u k f r n c e A - OthsrCASRsq~- iAssr 's tsm: 1, 2, 3, 4, 5, 6, f, 8, 9 B - C O ~ ~ U * m~sts ~ s s r ' ~ b l l ~ 8 : r, 2, & 4, s

Appendix D

Child Protection SWf Groups

CPS social work staff: intake, family services, children's services

(e.g . adoption, group home, foster care, specialized foster care,

emergency receiving, independent living, institution)

CPS Supervisorsimanagers

CPS Senior management/ Executive Directors

Appendix E

Community Human Service Prof-lonal Groups

Police: Municipal, Ontario Provincial Police, Military School teachers/administrators/teaching assistants: Public, Catholic, Jewish, Private Local Ministry of Community & Social Services (MCSS) office Doctors, pediatricians, nurses (hospital, clinics, home-visiting, VON, Community Public Health), mid-wives Probation Housing Family Court lawyers, judges, Court Clinic, Witness Assistance Program staff Day care personnel: private, non-profit, home, licensed Sociat workers: hospitals, family service associations, education, Children's Mental Health Centers, Mental Health Clinics, other community social workers Women's hostel staff External Child Abuse Committee members Hospital Child Abuse Review members Executive Directors, Supervisors and Staff of Community Service Agencies e.g. John Howard Society, Elizabeth Fry Society, Drug & Alcohol Counseling Services, Adult Protective Services, Big Sisters, Big Brothers. Family Counseling Centers, Community Living, Disabled, Intellectually Handicap etc. Community treatment people e.g. child care workers, psychologists, psychiatrists, private practice social workers, group workers, youth workers etc.

Appendix F rags side] Personal & Professional Characteristics

DATE: TOWNICrCY.

LOCAL CAS=

PROFESSION:

Place of Work

Position Last Pcu&ssional Degree Atfained

#Years Living In Cumnt Community

1, What b the number of year@) of experfence in your chosen field? ( 1 urrr [.I*

2 What is the number of year@) of working in or with the child weltare field? 1 Y-r c a

3. What is your assessment of vour own knowledge about the current child welfare legislation, the 1984, Child & Family Sewices Act 3

( ) non.llMe ( ) mockrate ( 1 some ( ) comidanbh

How many cases have ygg had where the child was abused andlor neglected and involved the Children's Aid?

( ) never ( ) 6 - 2 0 - m ( ) S t + c r u s ( )t.bCaru ( ) 21-50c88as

What is your gender 3 ( ) Famk (

What is your age? ( )Yean

What is your marital status? ( ) single ( ) divorced ( ) commn4aw

Do you have children? ( 1 Yes ( 1 NO

What is your highest kvel of education? ( ) Less th.n high school ( ) High school [or s o m postseeonday1 ( ) Community Colkg.

Appendix F [ *bn side]

Spectnrm Score Sheet

CASMANDATEDCASES

MOST MEDIUM SECTION SEVERE SWERlTY

VII A B

NOT CAS CASES

Appendix G [Ua" side1

Spectrum Score Sheet: Example

CAS MANDATED CASES MOST MEDIUM

SECTION SEVERE SEVERlTY 1 A

NOT CAS CASES NO

S€vERlTY

EXAMPLE

Appendix G r b D side]

Common information Sheet

ONTARIO ASSOClATlON OF CHILDREN'S AID SOClETlES LETTERHEAD

lNTERVENTlON SPECTRUM RESEXRCH PROJECT

The Infervention Speclm is an assessment hgrnework developed to assist child protection pmfessionak to make reliable and accurate judgments about whether a case is appropriate for child protection services. It is an eligibility or screening tool to aid child protBCfi*on staff to make informed decisions at the first point of contact with a case.

This research project will test the reliability and validity of the Intervention Spectrum with five children's aid societies (CASs) within Ontario. It will also examine human service workers' perceptions of child welfare entry points and where they would like the entry points to be. The project is funded by the Ontario Ministry of Community and Social Services and coordinated by the Ontario Association of Children's Aid Societies. The Principal Investigator is Professor Robert MacFadden, Faculty of Social Work. University of Toronto and Deborah Goodman, Project Manager and Doctoral Candidate, Faculty of Social Work. University of Toronto. Other project team members include: Mary Baltantyne. Service Diredor and George Leck, Executive Director, Simcoe Children's Aid Society; Margaret Morrison, Service Director, Halton Children's Aid Society, and Mary McConville, Executive Director, Ontario Association of Child rents Aid Societies.

The next page contains a CONSENT FORM for your review. It explains what is requested of you as a participant in this research and outlines the conditions under which the research is conducted. If you have any questions please call Professor Robert MacFadden at 41W8-5818 or Deborah Goodman at 416-604-0753.

Thank you for your consideration.

Appendix H

Consent Form: CAS Agency Staff

ONTARIO ASSOCfATlON OF CHILDREN'S AID SOCIETIES LETTERHEAD

lNfERYENTlON SPECTRUM E S H R C H PROJECT

You are being asked to attend an orientation as background to understanding the intervention Spectm and then, while referencing the Child and Family Services Act (CFSA), indicate your perceptions about where you think the current child protection intervention levels or entry points are in in your community. If you would like the minimum intervention levels to be different in your community you are asked to circle what you wwld like the child protection entry point to be. This involves reviewing a series of scales which reflect different child protection problems and levels of severity. We recognize that these are your perceptions and opinions and that you may be more or less familiar with the CFSA legislation.

The information you are asked to provide includes both personal and professional general background data. No names are requested. You am free not to participate or to withdraw at any time without any repercussions to you. No personally identifying information will be published from this research. There are no known risks to you in participating. Your cooperation will assist in refining the Speclrwm for your community and the child we(fare network, and help in better understanding human service workers' perceptions of child protection entry points.

I agree to participate.

Name:

Date:

Appendix I

Consent Fonn: Community Members

ONTARSO ASSOCfATlON OF CHILDREN'S AID SOCIETIES LETTERHEAD

1NTERVENlldN SPECTRUM RESEARCH PROJECT

You are being asked to attend an orientation session to the intervention Spectrum and then, while referencing the Child and Family Services Act (CFSA), indicate your perceptions about where you think the currant child protection intervention levels or entry points are in your community. If you would like the minimum intervention levels to be different in your community you are asked to circle what you would like the child protection entry point to be. This involves reviewing a series of scales that reflect different child protection problems and levels of severity. We recognize that these are your perceptions and opinions and that you may be more or less familiar with the CFSA leg is lati on.

The information you are asked to provide includes both personal and professional general background data. No names are requested. You are free not to participate or to withdraw at any time without any repercussions to you. No personally identifying information will be published from this research. There are no known risks to you in participating. Your cooperation will assist in refining the Spectmm for your community and the child welfare network, and help in a better understanding of human service worken' perceptions of child protection entry points.

I agree to participate.

Name:

Appendix J Definitions

Caretaker: The use of the word 'caretaker" within the Spectrum refers to three categories of caretaker. One, the primary caretaker - the mother, father, parent exercising access contact stepparent, or adult with a custody and control order for the child. Two, the assr'gned aretaker - the day care worker, baby-sitter, a family member providing temporary substitute care, or a partner of the parent with no legal relationship to the child. And three, the assumed caretaker - the teacher, the recreational leader etc.

Child Abuse versus Harm: Under the CFSA the definition of child abuse has two components and both are essential to the definition; these are 'harm" and 'caretaker". Harm done to a child is child abuse, only if the harm is as a result of something done or not done by the child's caretaker. Reference CFSA, Section 72(1) and 79(2).

Children's Aid Society (CAS): Refers to one of the 55 agencies in the province of Ontario approved to carry out the Ontario child welfare legislation mandate and provide sewices as defined by the CFSA.

Child and Family Sewices Act, 1984 (CFSA): Refers to the most recent Ontario child webre legislation, which provides and defines the mandate of the Children's Aid Societies to protect and serve children and their families.

Child Protection Sewices (CPS): Refers to child welfare services provided by a CAS under Part III (Child Protection) of the CFSA (CFSA, 55. (3)(9) ). Child protection services is also the part that the Spectrum Sections I through to XIV is based on.

Community: For the purpose of this study, community will be defined as an inclusive form of social organization that is territorially based and through which most people satisfy their common needs and desires (Norlin & Chess, 1997, p.348).

Eligibility Determination: Refers to whether a child's situation meets legislative requirements to warrant a Child Protection Services (CPS) assessment.

Human Service Worker: Any community professional in the human service field such as social workers, teachen, police, day care providers, lawyer, CPS worker. Refer to Appendix D and E for examples.

Minimum Intenrention Level: The Child Protection Service (CPS) MIL is where each survey respondent put the demarcation point or line between the two possible type of cases: the most and medium risklseverity. mandated CPS situations (CAS mandated cases) and the least and no riskheverify, non-CPS community cases (not CAS cases). CAS mandated cases are above the MIL Not CAS cases are below the MIL. In practice. any cases above the MIL means the CPS agency wilI intervene in a family by conducting an intake assessment of that famiIy because the child's situation falls within the most or medium risklseverity categories in at least one area. The same MIL means the same number of descriptors had been setected by the sunrey respondents for the mandated CAS cases and for the not CAS cases. A more inclusive MIL is when more descriptors of a less severe nature are included. A less inclusive MIL means fewer higher severity situations are selecteded Also re%rrecf to as the Child Prote&*on Entry Point

Negfecf: Is the presence of certain deficiencies or omissions in carstaker obligations (usually the parent) that harm the child's psychological andlor physical health or place the child at risk of h a m %the child were receiving appropriate care the risk of ham or actual ham could have been avoided (Ontario Child Mortali€y Task Force, 1997). Child neglect covers a range of caretaker behaviors that for this secondary study include::

Abandonment Lack of medical attention Lack of supervision Lack of appropriate child care arrangements Nutritional neglect lnadequate physical living conditions lnadequate Personal hygiene Lack of Household sanitation lnadequate Clothing.

RkWSeverity Categories:

Most RisklSeverity Category: As defined by CFSA, Part Ill (mandatory protection), and therefore actionable by the CAS, where the child is in urgent need of child protection services given that:

The child has already suffered physical or sexual harm at the hands of the person having charge of the child or because of that person's failure to care and provide for or supervise and protect the child adequately andlor The child is at substantial risk of suffering physical ot sexual harm as above or of suffering irreversible psychofogical damage and the child is in imminent danger if intervention is not immediate andlor The child has been orphaned aneUar The chikt has been completely abandoned andlor The family dynamics ate such - that abandonment of the child is imminent if intervention is not immediate.

Medium RisklSeverity Category: As defined by the CFSA, Part Ill ( mandatory protection), and therefore actionable by the CAS, where the child is in need of child pmtedon services, but the need is not as urgent as the ' Most Risk" cases given that

The child is at substantial risk of suffering physical or sexual harm as above or of suffering irreversible psychological damage but the child is not in imminent danger so intervention does not have to be immediate anNor The chiid is at substantial risk of being abandoned but is not in immediate danger of abandonment.

Least RisUSeverity Category: As defined by CFSA. Part II (prevention), the child or family could benefit from intervention but the intervention is not necessary for the physicaf andlor psychological safety of the child or the integrity of the family (related to ihe abandonment of the child).

No RisffSeverity Catego y: 1 he family is healthy in its response to the physical and psychologicat needs of the child.

Appendix K Community invitation Letter

Date

Spectrum Liaison Children's Aid Society

Ontario*

Attention:

Dear Colleague:

The Children's Aid Society of is requesting your assistance in completing a project currently under way within the agency. This project involves the refinement of a tool which will provide for a consistent determination of which families within our community should receive child welfare services.

Given that you are a community service provider with whom we collaborate regularly. we would very much like to have your input in assisting the Society with the determination of the level of this agency's intervention. Your involvement in this exercise if critical to enable us to develop a very dear indication of what the community of expects of its Children's Aid Society.

We are inviting community professionals to join us at one of a series of meetings at which we will be seeking opinions regarding this decision making tool. We wish to invite you and your staff and Board members to join us. We would welcome the involvement of as many staff as you can encourage to attend.

Yours sincerely.

Children's Aid Society.

Appendix L

(I) ontarb association of children k aid societies

OUTLINE OF RESEARCH

THE

FOR ETHICS REYIEW

INTERVENTION SPEcTEt'UM

The Iatemntion Spectnrm is a project M e d by the Minktry of S o d and Commdty Services through a grant to th Ontario Agociation of CbiIdrcn's Aid Socktics. The primrry purpose of this t a c h b test the reliability and nlidity of a new h e w o r k (ir, l7xe Inramrrion Spccbwn) deigned to assist ia dctcduhg ciigibiIity fir child protection sarrlces Tht project will olso a r p h the paaptbus of community profksionsls related to their views concerning child maltreatment

The project is murrgcd by a research team which Wudcs: Profkssor Rob MacFaddeg Faculty of Saciol Work, Univemity of Toronto (416) 978-5818 as the Principal Iwatigator. Debonh Goo- doctoral candidate, Faculty of Social W o e as the pja ma~gcr; George Led& Executive Dimtor d Muy m t y u e , Dinctor of Sewkc, Simca CMWents Aid Sacicry; hkgarct Morrirou, Dkector of Service, IIalton QiI&enrs Aid Sociay; and Muy McComrille, Executive Director of the Ontario Associatioa of ChiIdrenrs Aid Societia. The project rIso &as an Advisory Committee which meets at criticai j~~tmues in the resear&

Eligibility detcmhation h cud protection is hqueotfy viewed as one extensin imtstigatioa phase rad the 110th of- stags within this initiaI bnnsligatioa is often overlooked. YLet tht iaitiaI detaminubn ofwhether to proceed with ongoing hvdpt ioa it nilierl rad a distinct decision pomt. Some agencies m y open and investigate all cases to amid the rkk of making an arot in judgment. Other agencies rccognk that some selectivity b essential @en dkWIhg rao~rccr and the number o f & k reports (Boront. Adam &Toomm_ 1981).

The S m k pot rirt instnrment aac is it based on social s&uce research or e m p i t l a l 5 d h p hntialIy it is a taxoaomy or range of masons 59r child protection s e r n D ~ grouodcd in child w e k e k@lotion (Le, the Child and F d y SetvtCccs Aa) and nted by h e 1 ofscvedty. It answcn arch questloas as, "ShouId thir case be opened?' and 'b this case eligiblt 6Dr c W protection sewices as dcfincd under the kgjsltioa?" Although rpccific a t e g o b and definitions m y k somewhat unique to Oawio. the concept of this eliplbility instrument and process of development and use should be of in!- to the h a d child w e b network in N o d Amcrica. To-date, no such compnhcnsh eligibility hncwork &as ken hund within the Nonh American chiki w e b context,

The dcvcbpment and we of risk assessment models ( e . ~ Washington State Department of So& Services, 199% Reid, Sugurdson, Chriab~n-Wood B .Wright, 1995) , child weII-bbg sak (e-g., Ma- Moses and Jon- 1 9 0 and suumed decision-making systems (e+ Cht'fdrea's Rcscuch Centre, 1995) is a npidy growing phenomenon in cud ptotcetive services (BS), Howeverw d&ning both the concept of e-ility and dcvclopiag b e n t s Lr detcrminisg eligibility of CPS r & d s rre rrlativcfy new areas ofneed inEhild w c k

As the number of child protection r&& grows steadily, coupled with s i p i k m t decreases in 6mddbg and resource avaiIabilay* the 8 S network is no Ionger able to t o h t e inefficiencies. Tlre decbion to accept a report b not only the &st step but a critical stage Q agency compliance with legisktive requitexnents lad in the ahcation of scarce cesoutces. CPS agencies are now needii to cxamiLlt their smeenipg practices to prbrithe semtces rad emphosist protection over pmntive senices (Wells, Stein, Fluke & Do* 1989).

Policy statements conccrniPg eiigi&iliry Br CPS services re& to the wed fix the CPS worker to compIy with the kghlative and poky requitememf (Wells et aL, 1989; Institute 6Df the Prcvmtlbn of Qild hurt, 1W)- Uag clinicoljkdgment ahar to determine CPS eiig'b'tlity b beta problematic hot taro reasons.

A S C C I ) ~ ~ concern at the agency, policy a d feghlative levels k the absence of standards and nEtetiP about what kcton to use in detmhhg eligibi(ay @,ecLr et. aL, 1995) and "... iacomirtency withb and between jlukdiaiom coaccnhg what coustituta child a b w and neglect " (AItcr, 1985, p. 100). Support-hr thk position is b u d m a 1987 miew of bcvs and srrcenbg policies in my US. stua and the Dimkt of Columbia by Wells, Stein, FIulte & Doming (1989, p-47' Th authors state that "...screening occun on several IeveIs and the 'screeningt m a n s dU3kmt things to d i f f i n t people*.

An instnrmcnt ir cCtarIy d t d that could Ysirt BS walkas in the diEdt task of determining eligibility of rrfixnb so that consistency and accuracy arc incteased hide b r n the opcratiopnI d W t b that w a e bund in implementing risk assessment systems, a review of the North American litttattue on CPS elgibilayf%crccning took hds some key ktors have hmdned the deveIopment of eitfvr a theoretial or c!inicaI eligibility bmewotk (Aim, 1985).

One si@amt kctor that has W e d the devcIopmeat of a specSc hmcwork to uscss eligibility is that there is a wide variance in the underiyipe philosophiaI belie6 or approaches that CPS a g ~ * e s take regarding eligibility. A mkw of the titeranare i d d e s dDur approaches to the Some CPS agencies accept rlI reS?rrplq nga~IIess of Ievel of severity- These agencies may &a= substantial resources or an the only cumunity KNia available. Consequently, eligibility b not an igue because it is either not a concept or an option. A second group of agencies, whik acbowledgiug the coacept of eligibility, approach e-iliry &om the premise that screeniug out r&rs only to bogus or malicious nports and some bw &k cases won% eti aL, 1981). A third group recognize that the decision to accept a case is sipikm, but that analysis does not extend past this acknowledgment (Schuemaq Stagm & Johnson, 1989). The lart group vim ewility assessment as a diffiM awl critical stage in CPS decirion-mafcing and that it requires an iastnrment that rr&ns kghlativt along with dear . ria to assist workers with the c i i ~ n k task of sc2zk (ka a. a L , * l E 1 O 1 l Smte Deparfment of Socisl and Hdth Saviecq 1995).

bt rruons, the use of these rlJt assesmeat took bot eligibility determination is

&ivefy bm~etll IOW ~ I C cus at the time of initirl npoi (web tt. PL,

~ o n ~ ~ 1 t ~ 1 1 t o t ~ ~ r n m r t o d a u e p d i c t i o n s @ o P e c t e t , r l , 1993).Akv the abw of such bfnmmts to accmttIy predict hare maItrea-t has km seriously ~ucstr*oaect (Maid & Woolvatort, 1990; Saber& 1990; Doucek e t al, 1993).

A third b a n h to the dcvcbpmat of eligibility h m m o c k ir that previous work on eligibility has b o d e m oa common outcomes such as substantiation and

A 5 d &dot limiting eligibility ~ n v o r k dmbpmcnt is that t h n does not appear to exist t thcorctihI or clinical hwwork to vsbt child protection workers in applying CPS legishtiori to dctembe nfkmi eligibility. A North American search f i r a B S eligibility hamwork bund only two: the Intecvention Spc*nrm and the Washington Risk Assessment Model (WRAM). The WRAM has 6Dur 'Screen@ fbr SuScitncy" questious as part of a larger rbk assessment model and Wre the Specrmm is cunentiy being tested Br rrliability and validity (Whshington State, Department of Social and Health Services, 1991).

EYELOPMENT OF THE SPEcrrrny

The Intenrention Spectrum was &st devebped by Simeoc Chitdrcn's Aid Society. In 1990, the ag-s managmeat team identiEd a number of issues. There was incoIlSiStency h a r b W e deemed eligsk and ineligible cases. The cIients' needs bDr

as d&cd by the Icgiilation and agency poky did not ahrrays detamine the decision to open or not open a putieulu ast. I n c o e e n t intake practices t d t c d in the community being hasrated with the agcncy and uncIear on what was an appropriate child proteCtr*on rekmL Agency SUE thcmstk were asking Lr a too1 to assist thcm in making better decisiom at the intake h L A h , in dircussions with the firndhg Ministry, the agency had diEculty in clear& ilhxmatbg the typ& of cases it was Lastly, being in a ddcit budget situation, the agency needed to bow the complete specrmm of scroicfs provided to mrLe bf6nncd decisions regarding which semtces were to be msiataincd and which w d d be cut,

emotional development coaditbn; child unda 12 commitmi raiolls act: pent/cbild

scaDn of the CEA- Put II and iacludc other duties such as: nquat fix wuwIIing; * adoption semi=; sad dmrmmrtive semices.

Determining the minimum intervention kvcI was an important step. The agency decided that the intamition poi& would be betwen the wmediumw and "Ieastn severity catego&, given agency reso- mupled with CFSA ~ u i r e m e s t ~ , Cws Wing into the "lcastn or "no" severity catcgocies wouM, if quirrd, k rrdcmd out to comrnu&y agencies. To

The agency asked alt of its sutS board members aud o m UO comruaity people which incIuded aster parents, whmtctls, community prodcssionaIs, cIkw and c&ens to give input as to the iatcnmion points on the Spearurn. The data &om t f k process were aadjmd and there was consensus oa 137 out ofthe 143 descriptorr In other w o r e these groups hugely a p e d on which descriptors fit into which categories which detetmines whcn a case ir eligible fit sarvicc After discussion concer&g the 6 remaining doaiptors, mne of these needed to be moved mozc thn one a t e up or down (kt& and BaUantyne, 1994).

A k a one-momh pibt testing paiod, the agency started to use the Spcarum in February 1992 and has experienced two &ions since !hat paiod Early imprrssions from intake workers are &vonbS Wdipo a marked hnptovtmcllt in assessing cases invohring considerable mccmhy. Commi ty dcedback has noted m bcrease in the consistency and cki ty of rrssons'bt accepting or wt accepting a case. Commmity refkols aIso seem to be more appropriate. As of November, 1994, 1646 (8) of the 51 agencies responding to a navy indicated they were using the Specoum. Thiny-three percent (17) were either i m p h d a g or p h m h g to use the Spenr~m, 37% (19) were interested in the Spectrum and another 14% (7) had no phns to w i+

The Specmm bas the potential to d e c t amunity standards and increase common underskdings among podntoluIs within the corn&. Thir study is concerned with asking community pmfkssionrl what they view to be the cumnt standards nhtcd to minimum intcmation lcveb ia child proteaion and ,in their opinion, where the minimum mtemntion 1mk sboufcf be set. How acMlte are thir cumat perceptions and how muck change, if any, b proposed? How satMed/ibsat~ed we communir/ professionals with -m child protection stanub? In what or- would they nise the miniinurn h n e ~ i o a imk Ih what meas wodd they lower these? TI& study will explore accomadathn theory as related to W d pmtectioa standards. Auxmodation theory postulates that people with oagoing expoam to more serious sifilatbas accommodate to this md nise their perceptiom as to what constitutes a &us situation. In thip study, do workas who pactice in settiags with more severe child protection probkms tend to set

The pupose of this research is to test the nlhbility and validity of the Interntion Spectrw~ Acomrenience sampte of& chr76tcn's aid societies hss been scItcttd based on w i l h p s to use the Spcenum and the wiay that these agencies wouId of6rr in testing the Specmn. These agencies reflect r range of shes Oprsc, mediirm, smsn), locations (d, urban, i n n e t a auci multiCUItutal casetoads,

btratater rehbility sPill k tested by hv iug erch worker rate the same case twice using the Spectrum, &st at intake and pOphS about 1-2 weeks ipter. Mixmation 6.om the case at intake (witbout the rat@) will k Scaled and opened by the same worker Lr the second rating after some time has p u t e d The worker arill use ody this o w rnateriol to rate the case a sccoad timt, Whik rhir approach can be a&xed by rrcalIeaiao of the fint rating it is hoped that the busy nature of CPS wortas will make this less likely.

Intenater reIiabiliry wilI bolvt Emwing a second worfu assess the ase, as recorded by the intake w o r k - Thir second w o t h will aka use the Spectrum and k unaware of tht &st w~rket's rat- h t k these appmacha to assessing reliability hPve ken used in othet studies in &Id ptcehcon (eg, Ttocm~ 1994).

W o r k partkiption in tbir study ir vohmrvy. The study will be descrikd to the sociaL tuortcn by agcoy pmoml who have b e a ~ oriented to the rrseuch by the pmjea team. Worlms in each of the 6ve agencies wilt k ulad to micar r consent hm (see Appendk

There are severa1 approaches to validity assessmat which will be employed in this study. From the penpcctivc of content wlidicyp an expen p a d will be dDrmed of two judges, two lawyers and two child protection workers who will assess the Spccmrm in terms of how cIoscly it reflects the CFSAlegisiatioa. This wiH be done in two ways-

SeveraI names of judges, lawyers and child protection workers with knowledge and expertise in child w e h e kgislatioa will be provided to the project by the Ontario Association of Children's Aid Sotietic* Initially, the potential contacted by phone by the ptincipaL hmstiptor or the project sWyt If they a p e to participate a document will be indDrmuion on the Specaum. what is being ruked ofthem, a consent Lrm, ahag With a copy of the Rcscuch Vmion of the Speantm. They will be asked to read and sign the consent fb- if they arc willing tnd then to review the Spatrum and comment in a written ~eport on the extent to which it Meets the Iegirltion. Whrt does it include &om the haqisfation? Whu is not included (See Appendix for this material)?

After the written reports are received ad analyzed, a bow group will be held with these experts to obtain additional kedback to assist in rchement of the Specmm~ Thir fiocus group wilI meet to discus the perceptions within the written submigions by the expert group and wilI explore way3 the Specman might be rehed to benet achieve its objectives- Wah wtitten pcmission &om each participant, the 3CSSiou will be audio-recorded and t m s c r b d with no Eddfjhg names As noted in the consent firm fir mrt group /-

in Appendix A, participants ue h6khdraw at any time or not pacticipate in any of the ptocers if they wish. The audio-upes will be stored in a locked cabinet and emcd one yerr a f h the end of the project No ind'nridual will be identified in any publication that ICSUIts Born the mearch.

The study ar9 explore thepmdiczk volidzy of the Speanrm. It is hypothesized that a ~ s w h h E @ h u ~ r a t ~ o n t h c S p c c t r u m willbeassociatedwith~crtarrsouree use and more negath events SpccibDy, the research agencies will monitor the bolbwhg variabk fbr a maximum of six months &om the point of assessment using the Spectmm: how soon rrsources arc &ncd withia the whether the case ir opened to FhmiQfChikken M c e q the numkr of internal and e s a n r l resources assigned to the arc, the numkr of agency w o r k &ours spent on the case and the number of phcements withinthearebytppc -

Two othr scrks wiN be used to assess concunrnt Holidiry. The W-on Risk Assmment MbdtI (WRAM ) has a screening for suEcieocy section which b used to rsscst eligibility fir child pmtectioo SCIViCes. In this model, hur airafr and dated questbk are compkted If the mpom is posithe to three out of these bPur, the case P oped. Thk screekg tool has accepmbk reliability and wiK be wd as a compafison bacrumcnt fit the Spe*rum. A propnbn of the workers ascssing cases using the

Ckam~gent wlidiry wiII be cxpfored by having afthe research agencies mriw a pool of urliet cases which had ken uwed using tkir aPditionrl approach and compare this with the tahg using r Spemum.

Tluce of the five research agencies have eIecled to incorporate a sigdicaut staE and cammMity process nlrted to introducing the Specrnmt These agencies aril1 orient all stptfand conduct commm*ty matin&p where community profasioarls and Iay pasoas will receive an orientation to the Spechnrm, and its use. As part of this study, both sta.€€ and p r o f k s i d -ammunity members wiU be asked to participate in idea-g what they perceive to be cuncnt minimum intcnmtioa levels fix their child protection agency and *what change thae should be in thie Imfs based on their experience and pet-qtians (Sec Cbn&nt ~orm, StaErnd Community Profcssiioarb, App&dix A).

Each participant wilI use a Child Protection httwatioa Score Sheet and rate a series of vignettes which d e c t a nnge d p m b 1 ~ and I-& of severity in child protection. They axe asked to indicate rht minimum intewention Iml arm11.y and then circle the fmI at which they believe, these ~ * c e s should be afCctCd

As p u t d the info~tiongathctcd, wpondents wilI be ukcd basic background data such as pmfcssiou. position, yeus exped- gender, age, ethnicity, years working in nnnnt coramdty, cumnt perceived IeoeI of LnowIcdge &out the CFS& m b a of &me or negIect cases they bnn had atpasurr to.

Vdidity a a s m c u t will utilize procedures such as comIatious, t-tests, and analysis of wiauce, where appwate. Ourlitatiw respoa#s MI dso be rcponcd and quanti&d, in same cases

The she of the finrl sample has not yet been dctemined It is estimated that the pool of five rrseuch agencies wiII yield 660 contacts per month which r d t irr use of the Spectnnn. This would provide about 3960 assessments within a six month period. This number would a ripificant test of the Spectrum on most if not all Spectrum categories. It would aIso enswe uposurr to a wide wiety of cases to assess the Spcctntrn's utiIity for much of Ontario.

REFERENCES

Adarns, G.R. & Schvaneveldt, J.D. (1 985). Understandina Research Methods. New York, Longman Inc.

Albert, D., Aschenbrenner, M., & Schmalhofer, F. (1 989). Cognitive Choice Proc~sses and the Attitude-Behavior Relation. In A. Upmeyer (Ed.) Attitudes and Behavioral Decisions (Ed.) NY: Springer-Verlag, 61-99.

Alter. C. (1 985). Decision-making facton in cases of child neglect Child Welfare. 64 (2). 99-1 11.

Alwin, D.F. (1993). Attitude development in adulthood: The role of generational and life-cyle factors. in D. Krebs & P. Schmidt (Eds.) New Directions in Attitude Measurement New Y o k Walter de Gruyter, 61-93.

American Humane Association (1988). Hiahliahts of Official Child Abuse and Neglect Rewrtina. 1986. Denver, CO: Author.

App1ey.M.H. (1971). Adaptation-Level Theow: A Svmwsium (Ed.) New York. Academia Press.

Armitage, A. (1993). The policy and legislative context in B. Warf (Ed.). Rethinkina Child Welfare in Canada. Toronto, Ont McClelfand &Stewart Inc., 37-63.

Austin, David (1998). Prwress Reoat on the Development of Research Resources in Social Work Paper (draft) presented at the International Conference on Research for Sodal Work, Miami, FLA

Baird, C., Wagner, 0.. Caskey, R., & Neuendeldt, D. (1995). The Michiaan De~artment of Social Secvices Structured Decision Makina Svstem: An evaluation of its imwd on child ~rot8CZI~on services. Madison. WI: Children's Resource Center.

Ballantyne, M. (I 995). Intervention Spectrum: Overview of Its Creation. Unpublished paper, Bame: Ont, Simcoe Children's Aid Society.

Barnhorst, 0.. & Walter, B. (1991). Child prowon legislation in Canada. In N. Bala, J. Hornick, & R Vogl (Eds.). Canadian Child Welfare Law: Children. Families. and the State. Toronto: Thompson Educational Publishing, 24-32.

Barone, N.. Adams, W., (L Tooman, P. (1981). The screening unit: An experimentat approach to chifd protective service. Child Wetfare- 60(3), f 98-204.

Belsky, J. (1 993). Etiology af child maltreatment A developmenfal-ecologicaI analysis. PsvcholoaicaI BuMetin. $ I 4 (3). 41 3-434.

Besharov, D.J. (1985). Doing something about child abuse: The need to namw the grounds for state intervention. Haward Joumai of Law and Public Policv. 8,539-589.

Besharov, D.J.(t 991). Child abuse and neglect reporting and investigation: Policy guidelines for decision making. In M. Robin (Ed), Assessina Child Maltreatment Remrts: The Problems of False Alleaations. New York NY: Haworth Press. 35-50.

Best, J. (1989). lmaaes of issues: Twkina contem~orarv social ~roblems IEdJ* New York, NY: Aldine de Gnryter.

Blythe, B. (1992). Should Undergraduate and Graduate Social Work Students be Taught to Conduct Empirically-Based Practice? Yesf Journal of Social Work Education 28 (3),260263. -

Brythe, B. J., 8 Briar, S. (1985). Developing Empirically Based Models of Practice. Social Work. Nov-Dec., 483 - 488.

Briere, J., Berliner. L.. Bulkley, J., Jenny, C., & Reid, T. (1996). The APSAC Handbook on Child Maltreatment Thousand Oaks, CA: Sage.

Brown, C.W.,& Ghiselli, D. (1959). General Aims and Methods of Science. In L. Gorlow & W. Katlovsky (Eds.) Readinas in the Psvcholaov of Adiustment NY: McGraw- Hill, 7435.

Ca llahan, M. (1 993a). The administrative and practice context: Perspectives from the front line. In B. Wharf (Ed.), Rethinkina Child Welfare in Canada, Toronto: McCelland & Stewart Inc., 64-97.

Callahan, M. (1993b). Feminist Approaches: Women Recreate Child Welfare. In B. Wharf (Ed.), Rethinking Child Welfare in Canada. Toronto: McCelland & Stewart Inc., 772 -209.

Campbell. D.T., Hunt, W.A., & Lewis, N.A. (1958). The relative susceptibility of two-rating scaIes to disturbances resulting from shifts in stimulus contexts. Journal of Amlied Psvchofoav- 42,213-217.

Children's Research Center ( 4 993) A New Ao~roach to Child Protection Madison WI: National Coundi on Crime and Delinquency.

C hildrenss Research Center (I 995). The Michiaan De~artment of Social Services Decision Makina Svstem: An Eduation of Its [moact on Child Protection

Servt*ces Madison, WI: National CounciI on Crime and Delinquency.

Chow. S.L (1996). Sfatisticak Shnifcanc~'. Rationale. Validitv and Utilitv. Thousand Oaks, CAr Sage Publications.

Cicchetti, D., & Toth, S.L. (1995). Child maltreafment and attachment organization: Implications for intervention. In S. Goldberg, R. Muir & J. Kerr (Eds.) Attachment Theow: Social. Develo~mental and Clinical Pers~ectivas. Hillsdale. NJ: Analytic Press, 279-308.

Cohen, J. (1988). Statistical Power Analvsis for the Behavioral Sciences (2"- edition). Hillsdale, NJ: Lawrence Erlbaum,

Cohen, S., & Sussman, k (1975). The incidence of child abuse in the US. Chiid Welbre. 432-443

Corso, J. F. (1 Wl). Adaptation-level theory and psychophysical scaling. In M.H. Appley (Ed). Ada~tation-Level Theow: A Svm~osium (Ed.) New York. Academia Press, 27-48,

Coulton, C. (1979). The need for replication in social work research. Social Work Research 8 Abstracts. l(2).

Craft, J.L & Clarkson. C.D. (1985). Case disposition recommendations of attorneys and social workers in child abuse investigations. Child Abuse 8 Nealect. 9, 165-1 74.

Craft. J.L. & Staudt. M.M. (1991). Reporting and founding of child neglect in urban and rural communities. Child Welfare* 70 (3). 359-370.

Czaja, R., & Blair, J. (1996). Designinn Survew: A Guide to Decisions and Procedures. Thousand Oaks, CA: Pine Forge Press.

Daro, D. (1988). Confrontino Child Abuse: Research for Effective Proaram Desian. New York Free Press.

Doueck. H., Bronson, D., & Levine, M. (1992). Evaluating risk assessment implementation in child protection: issues lor consideration. Child Abuse & Nealect. 16, 637-646.

Doueck, H., English, Dot DePanfiiis, D., & Moote, G. (1993). Decision-making in child protective services: A cumparison of selected risk-assessment systems. Child WeIfam 72(5), 441-452

Downing, J.. Wells, S., & Fluke, J. (1990)- Gatekeeping in child protective services: A survey of screening policies. Child Welfare- 64(4), 357-369.

Drake. B., 8 Pandey. S. (1996)- Understanding the relationship between neighborhood poverty and specific types of child maltreatment Child Abuse and Nalecf. 20 (I i), t 003 -1 018.

Dukes. R.. & Kean, R. (1989). An experimental study of gender and situation in the perception and reportage of child abuse. Child Abuse & Nealect 13.354 -360.

Dumbrill, (3. (1998). The Historv of Child Welfare. Introductory Child Welfare Training Module 101 for Ontario Association for Children's Aid Societies.

Epstein, L (1994). The Therapeutic Idea in Contemporary Society. In A* Chambon & A Irving (Eds.) Essavs on Postmodemism and Social Work. Toronto: Canadian Scholars' Press, 5 -18.

Erickson, M.F. & Egeland, B. (1996). Child neglect In J. Briere, L. Berliner, J. Bulkley, C. Jenny, 8 T. Reid (Eds.) The APSAC Handbook on Child Maltreatment* Thousand Oaks, CA: Sage, 4-20.

Finkelhor, D. (1 984). Child Sexual Abuse: New Theow and Research. New York: Free Press.

Finkelhor, D (1994). The 'backlash' and the Mure of child protection advocacy. In J.EoBo Meyers (Ed.) Child Protection Under Fire: The Backlash. Thousand Oaks, CA: Sage PuMications, 1-16.

Finkelhor, D. (1996). Introduction. In. J. Briere, L. Berliner, J. Bulkley, C. Jenny, & T. Reid (Eds), The APSAC Handbook on Child Maltreatment. Thousand Oaks, CA: Sage, ix-xiii.

Finkelhor, D., & Zellman (1991). Flexible reporting options for skilled child abuse professionals. Child Abuse & Neglect. 15,335441.

Fishbein, M. (1967). Readinas in Attitude Theorv and Measurement (Ed.)., NY: John Wiley & Sons hc.

Fowler,Jr. F.J. (1998). Design and evaluation of survey questions. In L. Bickman and D. Rog (Eds.). Handbook of Amlied Social Research Methods. Thousand Oaks. Ck- Sage Publications, 343-374.

Franklin. D.L. (1986). Does client social class aflect dinical judgment? Social Casework. 67. (7), 424-432.

Gerban'no, J. (1992)- The meaning of poverty to children. American Behavioral Scientist- 35.(3), 220-237.

Gatbarino, J & Crouter, A. (1978). Defining the community context for parent- child relations: The correlates of child maltreatment. Child Develooment. 49,604-61 6.

Garbe, D. (1998). The Children's Aid Societv of the Repion of Peer: Twentv vear child abuse report. Children's Aid Society of the Region of Peel. Brampton. ON.

Gaventa. John (1985) The powerful, the powerless, and the experts: Knowledge struggles in an information age. In P. Park, M. Brydon-Miller, 6. Hall, & T. Jackson (Eds.), Voices of Change. Toronto: OlSE Press, 21-40.

Gelles, R.J. (t973). ChiM abuse psychopathology: a sociological critique and reformulation. American Journal of Ortho~svchiatrv. 43.61 1-621.

Gelles R. J. (1977). Final Remrt in Definino and Classifvina Cases of Child Abuse in Rhode Island. Kingston, RI: University of Rhode Island.

Gemrain, C. 0. & Gitterrnan, A. (1 980). The Life Model of Social Work Practice. New Yok Columbia University Press.

Giovannoni, J. (1989a). Definitional issues in child maltreatment. in D. Cicchetti & V. Carlson (Eds.) Child Maltreatment Cambridge: Cambridge University Press.

Giovannoni, J, (1989b). Substantiated and unsubstantiated reports of child maltreatment. Children & Youth Services Review 11, 299-318.

Giovannoni, J . (I 99 I). Unsubstantiated reports: Perspectives of child protection workers. In M. Robin (Ed), Assessina Child Maltreatment Rewrts: The Problems of False Alieaations. New York NY: Haworth Press, 51 -62.

Giovannoni, J. & Becerra, R. (1979). Definina Child Abuse. New York: Free Press.

Glisson, C. (1995). The state of the art of social work research: Implications for mental health. Research on Social Work Practice. 5(2), 205-222.

Goldstein, H. (1986). A cognkivehumanistic approach to the 'hard to reach client'. Socia[ Casework. 67 (I), 27-36.

Goodman, D. (1995). A Critics[ Examination of the Theoretical Foundations, Assum~tions, Goals, Interventions. and Outcomes of Intensive Family Preservation Proarams. Unpublished comprehensive paper, Faculty of Social Work, University of Toconto.

Goodman. D. & MacFadden, R.J. (1986). The Intervention Spectrum: a comprehensive tool for determining eligibility for child waIfare intake. Ontario Association of Children's Aid Societies JournaL 40, (1 ),23-26.

Grant, J. (1992). -of Toronto, ON: Univentty of Toronto Press..

Grigsby, K (1993). Theories that Guide Intensive Family Preservation Services: A Second Look. In S. Maybanks 8 M .Bryce (Eds.), Home-Based Sewkes for Children and Families: Policv. Practice 8 Research. Springfield, IL: Charles C. Thomas, 248-259.

Hallett. C. & Birchall, E. (1992). Coordination and Child Protection: A Review of the Literature. Edinburgh: HMSO.

Hanard, A, 8 Rupp, G. (j986). A note of the knowledge and attitudes of pruf&ssionaI groups toward child abuse. Journal of Communitv Psvcholoav. 14.21 9-223.

Heineman-Pieper, M. (f989). The Heuristk Paradigm: A Unifying and Comprehensive Approach to Social Work Research. Smith Colfeae Studies in Social Work. 60,8 -34.

Helson, Harry. (1 964). Ada~ation-Lave1 Theorv: An Experimental and Svstematic roach to Behavior. NY: Harper 8 Row.

Howitt, D. (1992). Child Abuse Errors: When Good Intentions Go Wrona. New B~nswick, NJ: Rutgen University Press.

Hudson, W.W. (1985). Lanauaae. Thouuht and Reason on Social Work Pradice and Education. Unpublished paper presented at Practitioners as Evaluators of Direct Practice: A Working Conference for Social Work Educators, Practitioners and Researchers, School of Social Work. University of Washington, Seattle.

Institute for the Prevention of Child Abuse. (1994). Child Abuse Standards: Self- tnstnrctional Trainina Manual, Toronto, Ont Institute for the Prevention of Child Abuse.

Jenkins, E., & Bell, C. (1997). Exposure and response to community violence among children and adolescents. In J. Osofosky (Ed). Children in a Violent Sociew New York: Guilford Press, 9 -31.

ffilichman,S., & Brosig, C. (1992). Mandatory child abuse reporting laws: Issues and implications for policy. Law and Policv. 14, (2-3), ISH68.

Ka1ichman.S.. Craig, M., & Follingstad. D. ($990). Mental health profkssionals and suspected cases of child abuse: An investigation of factors infiuencing reporting. Communitv Mental Health Journal. 24,4351.

Kelley. S. J. (1990). Responsibilities and management strategies in child sexual abuse: A comparison of child protective workers, nurses, and police officers. Child Welfare, 6, (f ), 4-51.

King, B., Baker, M., & Ludwigt S. (1993). Reporting of chi[d abuse by prehospital personnet. Prehosdtal8 Disaster Medidne. 8 (1),6788.

Kolevzoo. MS., & Maykrant, J. (1982). Theoretical orientation and clinical practice: Uniformity versus eclecticism? Social Senrice Review* 56, (1). 120-1 29.

Korbin, J.E. (198t). Child Abuse and Nealect: Cross Cultural PemPectives. Berkeley, CA: University of California Press.

Korbin, J.E. (1992). Child poverty in the United States. American Behavioral Scientist. 35,(3), 213-219.

Krebs,D., & Schmidt, P. (1993). New Directions in Attitude Measurement (Eds.) New Yo* Walter de Gruyter.

Leck, G., & Ballantyne, M. (1994). Spectrum - gatekeeping in child welfare intake. OACAS Journal. 38. (1).

Lena, H.F. & Warkov, S. (1978). Occupational perceptions of the causes and consequences of child abuselnegled Medical Anthro~olwv. 2 (I).

Levy, C.S. (1973). The value base of social work. Journal of Social Work Education. 9,3442-

Lincoln. Y., & Guba, E. (1985). Naturalistic Inauirv. Beverly Hills, CA- Sage.

Lindsay, D. (1993). The Welfare of Children. New York, NY: Oxford University Press.

Lindsay, D., & Kirk. S.A. (1 992). The continuing crisis in social work research: Conundrum or solvable proMern? An Essay Review. Journal of Social Work Education. 3 370-382. A

Lipsey. M.W. (1993). Theory as method: Small theories of treatment In L.B. Sechrest & A.G. Scott (Eds.). Understandina Causes and Generalkina About Them. San Francisco: Jossey-Bass, 5-38.

Lovell. M & Thompson. AH. (1995). improving the organiration and delivery of child welfare services: Themes, policy implications, and research agenda. in J. Hudson 8 B. Galaway (Eds.) Child Welfare in Canada: Research and Poky Im~lications. Toronto: Thomson Educational Pub.

MacFadden,R.J.,Go~dman~D.~Ballan~ne,M..M., & Morrison, M. (I 996). Eliaibilitv Determination in Child Protective Senrices: The intervention SDectrum. Unpublished paper.

MacFadden,R.J.,Goodman.D.,Ballantyne,M..M., & Morrison, M. (I 997). Preliminarv Findinas: Intervention Spectrum Research Summarv. Decern ber 8. 1997. Unpublished report.

Macintyre, E. (1993). The historical context of child welfare in Canada. In B. Wharfs (Ed.), Rethinkina Child Welfare in Canada. Toronto: McClelland & Stewart Inc. 23-36.

Madden, M E , & Wrench, D. F. (1981). Significant findings in child abuse. Victimoloav. 2J2), 196-224.

Magura, S., 8 Moses, B. (1 986). Outcome Measures for Child Welfare Services: Theorv and Amlications. Washington, DC: Child Welfare League of America.

Magura, S., Moses, B., & Jones, M A ((1987). assess in^ Risk and Measuring Chanae in Families: The Familv Risk Scales. Washington, D.C: Child Welfare League of America.

Maluccio, AN. (1991). The optimism of policy choices in child welfare. American Journal of Ortho~svchiatw. 61, (4), 606-609.

Manis, M. (1971). Context eflects in communication: Determinants of verbal output and referential decoding. In M.H. Appley (Ed.) Adaotation-Level Theow: A S- New Yo* Academia Press, 237-256.

Mass, H.S. (1 979). Assessing Family and Child Welfare Practice. Social Work, 24 365 - 372. A

McBeath,G.B., & Webb, S.A. (1991). Social work, modernity and post modemity. Sociolwical Perspectives. 38, 744-762.

McCall. M.L (1990). An analysis of responsibilities in child welfare systems. Canadian Journal of Familv Law. 8.

McPhee, 0. (1998). The Child Protection System: Oraanizational Reswnses to Child Sexual Abuse and the Sodal Construction of Social Problems. Unpublished doctoral dissertation, Faculty of Social Work, University of Toronto.

McKittflckf C. (1987). Toward a nursing definition of child maltreatment using seriousness vignettes. Advances in Nursinrr Science. 8 (4), 1-14.

Meyer, C. (TQ92). Sodal work assessment: Is there an empirical base? Research 9n Social Work Pradice. 2 .(3). 297-305.

Meyen, J.E.B. (1994). Child Protection Under Fire: The Backlash Ed.1 Thousand Oaks, CA: Sage Publications.

Misener, T.R (t986). Toward a nursing definition of child maltreatment using seriousness vignettes. Advances in Nursina Science. 8. (4), 1-14.

Mullen, E. J.. (1 985). MethodologicaI Dilemmas in Social Work Research. Social Work Research and Abstracts. July. 12 -20.

Mutchnick, R. J., & Berg. 0. (f 996). Research Methods for the Social Sciences: Pfactice and Amlications. Needham Heights, MASS: Allyn & Bacon.

Nagi, S . L (1977). Child Maltreatment in the United States. New York: Columbia University Press.

National Research Council (1993). Understanding Child Abuse and Nealect. Washington, DC: National Academy Press.

Notlin, J.M. & Chess, W A (7997). Human Behavior and the Social Environment 3rd. Edition. Boston, MA: AIlyn 8 Bacon.

Nyden, P., Figert, A, Shibley, M., & Burows, 0. (1997). Buildina Communitv: Social Science in Adion. Thousand 0aks.Ck Pine Forge Press.

Omer, H., & Dar, R (1992). Changing trends in three decades of psychotherapy research: The flight from theory into pragmatics. Journal of Consultino and Clinical PSVC~O~O~V, 60,88*93.

Ontario Child Mortality Task Force (1997). Final Re- JoumaL of Ontario Association of Children's Aid Societies - S~ecial Edition. July.

Orcutt, B. (1990). Sdence and lnauiw in Socia[ Work Practice. New York, NY: Columbia University Press.

Organization lor Economic Co-operation and Development [OECDI (1 990). Lone Parent Families: The Economic Challenae. OECD, Social Policy Studies. WB.

Osofosky, J. (1997). Child and youth violence: An overview of the issues. In J. Osofosky (Ed). Children in a Violent Society. New York: Guilford Press, 3-8.

O1toole, R. . Turbett, P., & Nalepka, C. (1983). Theories, professional knowledge and diagnosis of child abuse. In D. Finkelhor (Ed.) The Dark Side of Families. CA: Sage Publications.

Parker, 1. (f998). Social Constructionism. Discourse and Realism. Thousand Oaks, CA: Sage Publications.

Parker, R. (1989). Themes and variations. In B.C. Kahon (Ed.) Child Care: Research. Policv. and Practice. London: Hodden & Staughton.

Payne, M (1991). Modem Social Work Theow: A Critical introduction. Chicago, ILL Lyceum Books, Inc.

Peel Fair Share Task Force (1995). Peel and Families Deserve Their Fair Share. United Way of Peel Region.

Peiie, C. (1988). Research paradigms in soda1 work From stalemate to creative synthesis. Social Service Review 62, (I ) , 1 -1 9.

Pelton, L.H. (1990). Resolving the crisis in child welfare. Public Welfare. 48 (4) 2 9-25.

Pelton, LH. (1993). Enabling public child welfare agendas to promote family preservation. Social Work. 38.49 1-493.

Pelton, LH. (1 994). Is poverty a key contributor to child maltreatment? Yes! In E. GambriIl& T. Stein (Eds.) Controversial Issues in Child Welfare. Needham Heights, MA: Nlyn 8 Bacon.

P ett, M.A. (1 997). Nonwrametric Statistics for Health Care Research:Statistics for Small Sam~les and Unusual Distributions. Thousand Oaks, CA: Sage Publications.

Petty, R.E., & Krosnick, J A ((1995). Attitude Strenaths: Antecedents and Conseauences (Eds.). Mahway, NJ: Lawrence Erlbaum Associates, Publishers.

Phillips, C.C. (1992). Theories and laws. The Social Science Bestiaw. N.Y. Pergamon Press., 1 2-37.

Piaget, J. (1932). The Moral Judament of the Child. New Yak: HarcouR

Pilliffen', A., Seidl, A, Smith, C., 8 Stanton, M. (1992). Parent gender, victim gender, and family soci08~0nornic level influences on the potential reporting by nurses of physical child abuse. Issues in Com~rehensive Pediatric Nunina ? 5. (4), 239-47.

Polansky, No, Doroff, C., Kramer, L, hess, D., & Polland, L (1978). Public opinion and intervention in child neglect Social Work Research and Abstracts. 3,114 5.

Pratkanis, A., Breckler, S.J.. Oreenwakl, AG. (1989). Attitude, Structure and Function. Hillsdale, NJ: Lawrence Erlbaum Associates. Pub.

Reamer, F.G. (1992). The Piace of Empiricism in Social Work. Journal of Social Work Education. 28, (3), 257-259.

Reamer, F. G. (1993). The Philoso~hical Foundations of Sodal Work. New York : Columbia University Press.

Reamer, F.G. (1994a). The Evolution of Social Work KnowIedge. In F.G. Reamer (Ed.) The Foundations of Social Work Kitowiedae. NY: Columbia Univ. Press, 1 -12.

Reamer, F.G. (1994b). Sodal Work Values and Ethics. In F.G. Reamer (Ed.) The Foundations of Social Work Knowledae. New York: NY, Columbia University Press, - 195 -230.

Reid, G., Sigurdson, E., Christianson-Wood,J., 8 Wright, k (1995). Basic Issues Conceminci the Assessment of Risk in Child Welfare Work. Winnipeg: MAN., University of Manitoba.

Reid, W.J., (1991). Reframina the E~istemoloaical Debate. School of Social Welfare, University of Albany, State University of New York.

Reid, W.J. (1994). Research and Evaluation. in F.G. Reamer (Ed.), The Foundations of Social Work Knowledae. New York: NY, Columbia University Press, 122-1 50,

Reidy, T., & Hochsiadt, N. (1993). Attribution of blame in incest cases: A comparison of mental health professionals. Child Abuse and Nealect. 17,371381.

Robin. M (1982). Sheltering arms: The roots of child protection. In E. Newberger (Ed.), Child Abuse. Boston: Little, Brown.

Robin, M. (1991). The social construction of child abuse and 'false allegations". In M. Robin (Ed.), Assessina Child Maltreatment Remrts: The Problems of False Alleaations- New Yok NY: Haworth Press.

Robbins, S.P., Chatterjee, P., & Canda, €. R. (1998). Contemooraw Human Behavior T heorvr A Critical Pecsmctive for Social Work. Boston, Mk Allyn & Bacon.

Rosen, H. (1981). How workers use cues to determine child abuse. Social Work Research and Abstracts. 17(4), 27-33.

Ross, D., Shillingtan, E.R., 8 Lochhead, C. (1994). The Canadian Fact Book on Poverty. O€bwa. Canadian Council on Social Development.

Rossi, P., Schuerman. J.. 8 Budde, S+ (1996). Understandina Child Maltreatment Decisions And Those Who Make Them. Chicago: Chapin Halt Centre for Children, University of Chicago, Discussion Paper.

Rothery, M A (W8O). The Contribution of Science to Social Work Practice Theow and Education: lm~lications for the Develo~ment of ahfi~irica~lv Grounded KnowIedae Base. Unpublished doctoral paper. Faculty of SociaI Work. University of Toronto.

Rubin, A, 8 Babbie, E. (1 997). Research Methods for Social Work. 3". Edition. Pacific Grove, CA: BmkslCole Publishing Co.

Runyon, R.P. & Haber, A ( 1976). Fundamentals of Behavioral Statistics. 3rd. Edlion, Don Mills, ONT: Addison-Wesley Publishing.

Russell, M.N., (1990). Clinical Social Work: Research and Practice. Newbury Park. CA: Sage.

Saltman, J.E., & Greene. R.R. (1993). Social workers' perceived knowledge and use of human behavior theory. Journal of Social Work Education. 29, (I), 88-98.

Saunders, E., (1988). A comparative study of attitudes toward child sexual abuse among social work and judicial system professionals. Child Abuse and Nealect. 12,83- 90.

Scheper-Hughes, N., & Stein, H. (1987). Child abuse and the unconscious in American popular culture. In. N. Scheper-Hughes & H. Stein (Eds.), Child Survival. Dordrecht D. Reidel Publishing Co.

Schuerman, J.R., & Littell, J. (1995). Problems and Prosoects in Societv's Resmnse to Abuse and Nealect Chapin Hall Center for Children. University of Chicago.

Schuerman, J.R., Rzepnicki, T., Uttell, J., & Chak, A. (1993). Evaluation of the Illinois Familv First Placement Prevention Pmram: Final R ~ D o R Chapin Hall Centre for Children at the University of Chicago.

Schuerman, J.R., Stagner, M., Johnson, P., & Mullen, E. (1989). Child Abuse and Nedect Dedsion-makina in Cook Countv. Chicago: Chapin Hall Centre for Children, UniversRy of Chicago, Discussion Paper.

Scott, 0. (1989). Meaning construction and socia[ work practic8- Social Service Review. 37,39-St

Seaberg, J.R. (1990) Child well being: A easible concept? Social Work. 46, (8), 267-272

Sedlak, A (1991). Studv of National Incidence and Prevalence of Child Abuse and NeaIectiS88 (Revised Remrt). Rockville, MD: W8staI

Shecif. M.. & Wood-Sherif, C. (1 987). The Own Categories Procedures in Attitude Research In M. Fishbein (Ed.) Readings in Attitude Theorv and Measurement (MJ, NY: John Wiley 8 Sons Inc.. 190-1 98.

Simon, B-L. ($994). Are theories for practice necessary? Yes! Journal of Social Work Education. 30, (2), t 44447.

Snyder, J., & Newberger M. (1986). Consensus and difference among hospital prof8ssionals in evaluating child maltreatment Violence and Victims. 1. (2), 125-1 39.

Sommer. B., & Sommer, R. (1991). A Practical Guide to Behavioral Research: Tools and Techninues. New York: Oxford University Press.

Spearly, J., & Lauderdale, M (1982). Community characteristics and ethnicity in the prediction of child maltreatment rates. Child Abuse and Nealect. 7, 91-105.

Sprinthall, R.C. (1 997). Basic Statistical Analvsis. 5*. Edition. Need ham Heights, W Allyn 8 Bacon.

Steinberg, L., Catalano, R., & Dooley, D. (1981). Economic antecedents of child abuse and neglect Child Develooment. 52,975-985.

Streshinsky, N., Billingsley, A*, & Gurgin, V. (1966). A study of social work practice in protective services: Irs not what you know it's where you work. Child Weifbrew 45.444450.

Streufert, S. (1987). Decision making: Research and theory challenges for applied social psychology. Journal of Amlied Social Psvcholo~v. 17. (7), 609-621.

Sullivan, R. (1994). Should child welfare workers be required to pass periodic competency tests? Yesl In E. Gambrill 8 T. Stein (Ms.), Controversial Issues in Child WeMam. Needham Heights, MA: Allyn & Bacon.

Sullivan, R. (1998). Risk Assessment Tools: What we should know! January 27,1998, Guest Speaker at Faculty of Social Work, University of Toronto.

Swiftt K (9988). Knowledae About NeaIect : A Critical Review of the Literature. Working Papers in Social Welfare in Canada, #23. Faculty of Social Work, University of Toronto,

Swift* K (1990). Cmtina ffiowledae: A Studv of the Production of Knowledae pbout Child Nealect Unpublished doctoral dissectation, Faculty of Soda1 W o k University of Toronto.

Task Force on Social Work Research (1991). Buildirnr Social Work Knowledae for Effective Services and Policies: A Plan for Research Development Austin, TX: University of Texas School of Social Work.

Testa, M. (1985). Child Welfare in Socioloaical and Historical Perswctives LDiscussion paper #KM] . Chapin Hall Center for Children, University of Chicago.

Thomas, E. J. (1985). The Validity of Design and Development and Related Concepts in Developmental Research. Social Work Research and Abstracts. 21. (2). 50- 55-

Thyer. B. (1 989). Exploring Epistemologies: The Debate Continues. Journal of Social Work Education. 25, (2) 174-1 76.

Thyer, B. (1994). Are Theories for Practice Necessary? No! Journal of Social Work Education. 30. (2). 148 -1 52.

Tomisan, AM., 8 Goddard, C. (1993). A Failure of the Child Abuse Research Community: Why Case Management Has Not Been Properly Investigated in SRU. Department of Social Work, Monash University, Melbourne, Australia.

Tower, C.C. (1996). Understandim Child Abuse and Nealect. 3rd. Edition. Needham Heights, MA: Allyn 8 Bacon.

Trocme. N. (1991). Child welfare services. In R. Bamhorst & L. Johnson (Eds.), The State of the Child in Ontario. Toronto: Oxford University Press. 63-9I. -

Trocme, N. (1992a). Develo~ment of an Emrt-based Child Nealect Index: Ma king Social Work Practice Knowledae &licit Unpublished doctoral dissertation. Faculty of Sodaf Work, University of Toronto.

Trocme, N. (1992b). Estimating the scope of child abuse and neglect in Ontario: A guide to understanding child maltreatment statistics. Research Connection. Toronto: The lns€itute for the Prevention of Child Abuse, 2-1 1.

Tmme, N., McPhee, D., Tam., KK., & Hay, T. (1 994). Ontario Incidence Studv of Re~orted Child Abuse & Nalect Toronto: Institute fbr the Prevention of Child Abuse-

Trocme. N.. Tam. KK., McPhee. D. (1995). Correlates of substantiation of maltreatment in child welf&e investigations. In J. Hudson & 6. Galaway (Eds.) Child Welfare in Canada: Research and Policv Im~lications Toronto: Thomson Educational Pub.

Tmte, B., Adkins, E, & MacDonald, G. (lS92ah Professional attitudes regarding the sexual abuse of children: Comparing police, child webre and community mental health- Child Abuse & Nealed. 16,359-368.

Tnrte, 6.. Adkins. E-, & MacDonald, G. (1992b). Coordinatim Child Sexual Abuse Services in Rural Communities. Toronto: University of Toronto Press.

Upmeyer. A., & Six, B. (I 989). Strategies for exploring attitudes and behavior. In A. Upmeyer (Ed.) Attitudes and Behavioral Decisions (Ed.1 NY: Springer-erlag, 1-1 8.

U.S. Department of Health and Human Services [DHHSJ (1988). Studv Findinas: Studv of National Incidence of Child Abuse and Ne~lect Washington. DC: Author.

US. Department of Health and Human Services [DHHS] (1991). National Incidence and Prevalence of Child Abuse and Nedect: 1988 - Revised Re~oct. Washington, DC: Author.

Vogl, R. (1991). Initial involvement: reporting abuse and neglect. In N. Bala. J. Homidr, 8 R. Vogl (Eds.) Canadian Child Welfare Law: Children. Families and the State. Toronto: f hompson Educational Publishing, 33-54. -

Wald, M., 8 Wwlverton. M. (1990). Risk assessment: The emperor's new clothes? Child Welfare. 64, (6), 483-51 1.

Washington State, Department of Social & Health Services. (1991). lm~act of lnvestiaations: - 0utcomes for Child Protective Services Cases Receiving Differential Levels of Service. Olympia, WA: Children's Services Research Project

Washington State, Department of Social 8 Health Sewices. (1995). Risk Factor Matrix Guide. Oiympia, WA: Children's Sewices Research Project.

Wells, K. (1994). A reorientation to knowledge development in Family Preservation Services: A proposal. Child Welfare. 73, (5), 475488.

Wells, S. (1988). Factors infl uendng the response of child protective services workers to reports of abuse and neglect In G. Hotaling, D. Finklehor, J, Kirkpatrick, & M. Straus (Eds.), Co~ina with Familv Violence. Newbury Park, CA, Sage Pub.

Wells, S., Downing, J., & Fluke, J. (1991). Responding to reports of child abuse and neglect In M. Robin (Ed), Assessincl Child Maltreatment Reoorts: The Problems of False Alleaations. New York NY: Haworth Press, 63-72.

Wells, S., Fluke, J., Downing, J., 8 Brown. H. (1989a). Executive Summaw - reenina in CMd Protective Services. American Association for Protecting Children -

American Humane Society.

Welk, S-, Stein, T., Fluke, J., 8 Downing. J. (1989b). Screening in child protective se~ces . Social Work. I, 45-48.

Wharf, B. (1993). Rethinkina Child Welfare in Canada (Editor). Toronto: McCelIand & Stewart Inc.

Wharf,, B. (1995a). Organizing and delivering child welfare sewices: The contributions of research. In J. Hudson & B. Galaway (Eds.) Child Webre in Canada: Research and Policv Implications, Toronto: Thomson Educational Pub.

Wharf, B. (1995b).Toward a new vision for child welfare in Canada. Child Welfare 3.820-839.

Wiebush, R. (Nov. 23.1992). Memorandum Re: CPS Screening Criteria. Madison. WI: National Council on Crime and Delinquency.

Wrtkin, S.L (1992)- Should Empirically-Based Practice be Taught in BSW and MSW Programs? No! Journal of Social Work Education. 28, (3), 265-269.

Wolock, 1. (1982). Community characteristics and staff judgments in child abuse and neglect cases. SociaI Work Researcfr and Abstracts* 2, 9-15.

Woolf, A*. Taylor, L,, Melnicoe,L., Ando1sek.K. Dubowitz, H., De Vos. E.. & Newberger,E. (f988). What residents know about child abuse: lmplications of a survey of knowledge and attitudes. American Journal of Disease of Children. 14.668-673.

Zellman. G. (1992). The impact of case characteristics on child abuse reporting decisions. Child Abuse & Neolect. 16, 57-74.

Zellman, G., Faller, K (1996). Reporting of child maltreatment. In J. Briere,, L. Berliner. J. Bulkley, C. Jenny. & T. Reid (Eds.) The APSAC Handbook on Child Maltreatment. Thousand Oaks, CA: Sage, 359-381.

Zellner, G., & Antler, S. (1990). Mandated reporters and CPS: A study in frustration. Public Welfare. 48. ( I ) , 30-37.

Zumvin. S. J. (1986). Residential density and urban child maltreatment. Journal of Familv Violence. 1, (4), 307322.

Zuravin, S. (1989). The ecology of child abuse and neglect Review of the liirature and presentation of date. Violence and Victims. 4, (2). 101-1 20.

Zuravin, S. J. (199 1). Research definitions of child physical abuse and neglect Current problems. In R Stan & D. Wolf& (Eds.) The EftWs of Child Abuse and Nealect: Issues and Research. New Yo& Guibrd Press.

Zwavin, S. J.. & Taylor, R. (1 987). The ecology of child maltreatment Idenwing and characterizing highrisk neighbofioads. Child Welfare. 66, (6) 497-506.