PLEASE SCROLL DOWN FOR ARTICLE

This article was downloaded by: [University of Sydney]On: 12 January 2009Access details: Access Details: [subscription number 777157962]Publisher Informa HealthcareInforma Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House,37-41 Mortimer Street, London W1T 3JH, UK

International Journal of Language & Communication DisordersPublication details, including instructions for authors and subscription information:http://www.informaworld.com/smpp/title~content=t713393930

The Peer Attitudes Toward Children who Stutter (PATCS) scale: an evaluationof validity, reliability and the negativity of attitudesMarilyn Langevin ab; Sabina Kleitman c; Ann Packman a; Mark Onslow a

a Australian Stuttering Research Centre, b School of Psychology, The University of Sydney, Sydney, NSW,Australia c Institute for Stuttering Treatment & Research, Faculty of Rehabilitation Medicine, University ofAlberta, Edmonton, AB, Canada

First Published on: 25 September 2008

To cite this Article Langevin, Marilyn, Kleitman, Sabina, Packman, Ann and Onslow, Mark(2008)'The Peer Attitudes Toward Childrenwho Stutter (PATCS) scale: an evaluation of validity, reliability and the negativity of attitudes',International Journal of Language &Communication Disorders,

To link to this Article: DOI: 10.1080/13682820802130533

URL: http://dx.doi.org/10.1080/13682820802130533

Full terms and conditions of use: http://www.informaworld.com/terms-and-conditions-of-access.pdf

This article may be used for research, teaching and private study purposes. Any substantial orsystematic reproduction, re-distribution, re-selling, loan or sub-licensing, systematic supply ordistribution in any form to anyone is expressly forbidden.

The publisher does not give any warranty express or implied or make any representation that the contentswill be complete or accurate or up to date. The accuracy of any instructions, formulae and drug dosesshould be independently verified with primary sources. The publisher shall not be liable for any loss,actions, claims, proceedings, demand or costs or damages whatsoever or howsoever caused arising directlyor indirectly in connection with or arising out of the use of this material.

The Peer Attitudes Toward Children whoStutter (PATCS) scale: an evaluation of validity,reliability and the negativity of attitudes

Marilyn Langevin{{, Sabina Kleitman§, Ann Packman{ and

Mark Onslow{

{Australian Stuttering Research Centre and §School of Psychology, The

University of Sydney, Sydney, NSW, Australia

{Institute for Stuttering Treatment & Research, Faculty of Rehabilitation

Medicine, University of Alberta, Edmonton, AB, Canada

(Received 30 August 2007; accepted 10 April 2008)

Abstract

Background: Persistent calls for school-based education about stutteringnecessitate a better understanding of peer attitudes toward children who stutterand a means to measure outcomes of such educational interventions. Langevinand Hagler in 2004 developed the Peer Attitudes Toward Children who Stutterscale (PATCS) to address these needs and gave preliminary evidence ofreliability and construct validity.Aims: To examine further the psychometric properties of PATCS and toexamine the negativity of attitudes.Methods & Procedures: PATCS was administered to 760 Canadian children ingrades 3–6. Measures included reliability, a confirmatory factor analysis (CFA), aknown groups analysis, convergent validity with the Pro-Victim Scale of Rigbyand Slee, and the negativity of attitudes.Outcomes & Results: PATCS appears to tap a second-order general attitude factorand three first-order factors representing the constructs of Positive SocialDistance (PSD), Social Pressure (SP), and Verbal Interaction (VI). In the knowngroups analysis, participants who had contact with someone who stutters hadhigher scores (more positive attitudes) than those who had not, and girls hadhigher scores than boys. PATCS correlated moderately (0.43, p,0.01) with thePro-Victim scale. Finally, one-fifth (21.7%) of participants had scores that weresomewhat to very negative.Conclusions & Implications: Results provide evidence of the validity and reliabilityof PATCS and confirm the need for school-based education about stuttering.The PSD and SP factors suggest that education include discussions about (1)similarities and differences among children who do and do not stutter in order

International Journal of Language & Communication Disorders

ISSN 1368-2822 print/ISSN 1460-6984 online # 2008 Royal College of Speech & Language Therapistshttp://www.informahealthcare.comDOI: 10.1080/13682820802130533

Address correspondence to: Marilyn Langevin, Institute for Stuttering Treatment & Research, 1500,8215 — 112 Street, Edmonton, AB, Canada T6G 2C8; e-mail: marilyn.langevin@ualberta.ca

INT. J. LANG. COMM. DIS.

2008, iFirst Article, 1–17

Downloaded By: [University of Sydney] At: 22:40 12 January 2009

to increase acceptance, and (2) making personal choices and handling peerpressure in thinking about children who stutter. The VI factor suggests thatopen discussion about stuttering may alleviate frustration experienced bylisteners and provide the opportunity to give strategies for respondingappropriately. Results also suggest that education involve contact with a personwho stutters.

Keywords: Stuttering, attitudes, factor analysis, children.

What this paper addsWhat is already known on this subject?Preliminary psychometric analysis of the Peer Attitudes Toward Children whoStutter scale (PATCS; Langevin and Hagler 2004) indicated that internalconsistency for the total scale was 0.96. A principal components analysissuggested a three-factor structure with factors representing the constructs ofPositive Social Distance (PSD), Social Pressure (SP), and Verbal Interaction(VI). Confirmation of the extant psychometric properties and furtherinvestigation are needed. In addition, given the persistent calls for school-based education about stuttering, it would be useful to examine the negativityof attitudes.

What does this study add?Results provide further evidence of the validity and reliability of PATCS andindicate that attitudes of one-fifth of participants were in the slightly to verynegative range of the response scale. The latter finding supports the persistentcalls for school-based education about stuttering. Replication of the three-firstorder factors of PSD, SP, and VI in the hierarchical model of attitudes thatbest fit the data in this study suggests that these constructs can be used indeveloping educational programmes about stuttering.

Introduction

There have been persistent calls over many years for educational programmes aboutstuttering (e.g., Langevin et al. 1998, Ruscello et al. 1988, Murphy and Quesal 2002).This has been in response to anecdotal and empirical evidence that children whostutter (CWS) are teased and bullied at school (Mooney and Smith 1995, Langevin et al.1998, Hugh-Jones and Smith 1999), appear to be at greater risk than non-stutteringpeers for being bullied (Blood and Blood 2004), and are rejected more often than theirnon-stuttering peers (Davis et al. 2002). These findings suggest that many CWS suffersocial consequences at school, that some peers hold negative attitudes toward CWS,and that CWS are at risk for long-term psychological harm. Clearly, educationalinterventions that have been called for and are being undertaken are warranted. Thus,there is urgent need for a valid and reliable measure of attitudes to evaluate outcomesof such interventions. There is also need for guiding information regarding theconstruct of peer attitudes toward CWS and the extent to which attitudes are negative.Langevin and Hagler (2004) developed the Peer Attitudes Toward Children whoStutter scale (PATCS) to address these needs. This scale was developed from an initial

2 Marilyn Langevin et al.

Downloaded By: [University of Sydney] At: 22:40 12 January 2009

pool of 116 items written at the third-grade reading level (Fry 1968) to represent (1)the attitudinal domain, which included peer interactions that occur at school, at home,and in public places that are related to academic study, play, and friendships; and (2)the tripartite model of attitude which is comprised of affective, cognitive, andbehavioural intent components (Eagly and Chaiken 1993). Items were developed frominterviews with stuttering and non-stuttering school-age children and drawn fromexisting scales that measured attitudes toward children with disabilities (Langevin andHagler 2004). After pilot testing with 267 children aged 8–13 years in grades 4–6 infour schools in Western Canada, 40 items were retained (table 1). Internal consistency(coefficient alpha) was 0.96. To examine construct validity, a three-factor structure washypothesized on the basis that three factors representing the tripartite model ofattitude would emerge. Principal components analysis with oblimin rotation suggesteda three-factor structure; however, rather than reflecting the tripartite model of attitude,items loaded on three factors defined as Positive Social Distance (PSD), SocialPressure (SP), and Verbal Interaction (VI). The PSD factor reflects general comfort inbeing with CWS (e.g., ‘I would let a kid who stutters hang out with us’) and isconsistent with the construct of social distance. Social distance refers to the degree ofacceptance or comfort that one has toward others, in particular members of ones owngroup (in-group) or another group (out-group) and has long been used to measuresocial acceptance among peoples of differing ethnicity, social class, occupation,religion, etc. (Owen et al. 1977).

SP reflects concern with what others think about CWS (e.g., ‘I would beashamed to be seen with a kid who stutters’) and is consistent with the construct ofsubjective norm that has been shown to be one of three determinants of the intention toperform a behaviour in the theory of planned behaviour (Ajzen and Madden 1986).Subjective norm refers to ‘the perceived social pressure to perform or not toperform the behaviour’ (Ajzen and Madden 1986: 454). Underlying subjective normare beliefs concerned with ‘the likelihood that important referent individuals orgroups would approve or disapprove of performing the behaviour’ (Ajzen andMadden 1986: 455). The importance of the SP factor is highlighted by recentfindings that children’s behavioural intentions to interact with peers with disabilitiespredicted actual behaviour (Roberts and Smith 1999).

The VI factor is characterized by frustration (e.g., ‘I would feel uptight talkingwith a kid who stutters’). This is consistent with anecdotal reports that listenerssometimes report frustration when stuttering interferes with communication andthat school-age peers sometimes do not wait for CWS to finish their sentences.

A known groups analysis, which posits that certain groups of respondents willscore higher on a scale than others (Eagly and Chaiken 1993), further measuredconstruct validity. Drawing from the literature on peer attitudes toward children withdisabilities and other speech disorders (Langevin and Hagler 2004), it washypothesized that (1) children who had had contact with people who stutter wouldhave higher scores than those who had not, (2) girls would have higher scores thanboys, and (3) attitudes would become increasingly positive as grade levels increased.As hypothesized, peers who had had contact with someone who stutters hadstatistically significant higher scores (more positive attitudes) than those who hadnot had contact, and fourth grade students had statistically significant lower scalescores than fifth and sixth grade students. However, differences between boys andgirls and fifth and sixth grade students were not statistically significant. Langevin

Peer attitudes toward children who stutter 3

Downloaded By: [University of Sydney] At: 22:40 12 January 2009

Table 1. Means (standard deviations) for items in PATCS and factor loadings and communalities for Models 1 and 5 in the calibration sample and, inparentheses, Model 5 in the validation sample

Itemnumber Item Mean (SD)

Model 1: Hypothesized first-ordermodel Model 5: Second-order model

*Factor loadingsCommu-nalities

*Factor loadings

CommunalitiesPSD SP VI PSD SP VI

1 I would like having a kid who stutterslive next door to me

2.81 (1.04) 0.554 0.307 0.566 (0.571) 0.320 (0.327)

3 I would walk in the hall with a kidwho stutters

3.26 (1.08) 0.451 0.223 0.420 0.640 (0.641) 0.409 (0.411)

8 I would enjoy doing a class projectwith a kid who stutters

3.39 (1.13) 0.714 0.510 0.748 (0.690) 0.512 (0.475)

11 I would introduce a kid who stuttersto my friends

3.51 (1.18) 0.718 0.516 0.727 (0.640) 0.529 (0.409)

12 Kids who stutter are easy to get alongwith

3.15 (.92) 0.477 0.228 0.474 (0.497) 0.225 (0.247)

15 I would be happy to have a kid whostutters for a friend

3.59 (1.09) 0.771 0.595 0.775 (0.784) 0.601 (0.615)

16 A kid who stutters can be a goodfriend

4.02 (.94) 0.607 0.368 0.604 (0.573) 0.365 (0.329)

18 I would walk home with a kid whostutters

3.55 (1.09) 0.785 0.616 0.565 (0.645) 0.265 (0.159) 0.628 (0.603)

19 I would like to sit beside a kid whostutters

3.24 (1.05) 0.772 0.596 0.783 (0.773) 0.612 (0.598)

21 I would go to the movies with a kidwho stutters

3.42 (1.12) 0.568 0.323 0.653 (0.756) 0.426 (0.571)

25 I would invite a kid who stutters tomy birthday party

3.56 (1.14) 0.782 0.612 0.789 (0.831) 0.623 (0.690)

28 I would like a kid who stutters to talkfor my group in class

3.12 (1.07) 0.660 0.436 0.670 (0.678) 0.449 (0.460)

30 I would let a kid who stutters hangout with us

3.59 (1.11) 0.799 0.638 0.795 (0.786) 0.631 (0.618)

4M

arilynL

angevinet

al.

Downloaded By: [University of Sydney] At: 22:40 12 January 2009

Itemnumber Item Mean (SD)

Model 1: Hypothesized first-ordermodel Model 5: Second-order model

*Factor loadingsCommu-nalities

*Factor loadings

CommunalitiesPSD SP VI PSD SP VI

32 I would enjoy being with a kid whostutters

3.40 (1.07) 0.789 0.622 0.796 (0.829) 0.634 (0.687)

34 I would be best friends with a kidwho stutters

3.18 (1.12) 0.789 0.622 0.795 (0.787) 0.632 (0.619)

36 I would like having a kid who stuttersin my class

3.48 (1.09) 0.756 0.572 0.760 (0.724) 0.577 (0.525)

38 I would sit on the bus with a kid whostutters

3.56 (1.11) 0.779 0.606 0.779 (0.788) 0.606 (0.621)

40 I would play at recess with a kid whostutters

3.59 (1.10) 0.793 0.629 0.790 (0.813) 0.624 (0.661)

2 I would avoid a kid who stutters 3.72 (1.02) 0.641 0.411 0.632 (0.511) 0.400 (0.261)5 Kids who stutter are like normal kids 3.63 (1.22) 0.542 0.294 0.524 (0.453) 0.274 (0.206)6 I would be ashamed to be seen with a

kid who stutters3.80 (1.15) 0.756 0.572 0.740 (0.697) 0.547 (0.486)

7 I would not go bicycle riding with akid who stutters

3.76 (1.11) 0.766 0.587 0.748 (0.640) 0.559 (0.409)

13 I would be embarrassed if a kidstuttered in front of a cashier

3.57 (1.17) 0.650 0.422 0.994 (0.729) 20.311 (20.089) 0.521 (0.422)

14 I would not play inside my house witha kid who stutters

3.78 (1.14) 0.761 0.578 0.793 (0.757) 0.629 (0.573)

17 I would not go to the store with a kidwho stutters

3.79 (1.12) 0.697 0.485 0.802 (0.765) 0.643 (0.586)

26 I would not do homework with a kidwho stutters

3.54 (1.16) 0.676 0.456 0.670 (0.648) 0.449 (0.420)

9 Kids who stutter are weird 3.98 (1.09) 0.703 0.494 0.699 (0.661) 0.488 (0.437)10 I would feel foolish if a kid stuttered

in front of my parents3.73 (1.21) 0.744 0.554 0.752 (0.745) 0.565 (0.555)

Table 1. (Continued.)

Peer

attitudestow

ardchildren

who

stutter5

Downloaded By: [University of Sydney] At: 22:40 12 January 2009

Table 1. (Continued.)

Itemnumber Item Mean (SD)

Model 1: Hypothesized first-ordermodel Model 5: Second-order model

*Factor loadingsCommu-nalities

*Factor loadings

CommunalitiesPSD SP VI PSD SP VI

22 I would feel uptight talking with a kidwho stutters

3.53 (1.10) 0.648 0.420 0.639 (0.665) 0.408 (0.442)

24 I would be frustrated listening to a kidwho stutters

3.15 (1.23) 0.605 0.366 0.571 (0.628) 0.326 (0.394)

29 Listening to a kid who stutters wouldannoy me

3.45 (1.20) 0.733 0.537 0.718 (0.744) 0.516 (0.553)

31 I would not play with a kid whostutters

3.88 (1.03) 0.757 0.573 0.776 (0.693) 0.601 (0.480)

33 Kids who stutter expect too muchhelp to do things

3.31 (1.04) 0.568 0.323 0.553 (0.492) 0.306 (0.242)

35 I would be embarrassed to be with akid who stutters

3.65 (1.17) 0.227 0.572 0.608 0.809 (0.748) 0.654 (0.560)

37 I would not go to the playgroundwith a kid who stutters

3.81 (1.08) 0.716 0.513 0.784 (0.788) 0.615 (0.621)

39 Kids who stutter should not playgames that involve talking

3.88 (1.19) 0.641 0.411 0.645 (0.628) 0.416 (0.394)

4 Kids who stutter do not want to go toparties

3.43 (1.07) 0.326 0.107 6 6 6 6

20 Kids who stutter feel sorry forthemselves

2.88 (1.00) 0.371 0.138 6 6 6 6

23 Kids who stutter should be allowed tomake speeches

3.82 (1.13) 0.519 0.269 6 6 6 6

27 Kids who stutter are friendly 3.67 (.89) 0.357 0.128 6 6 6 6

PSD, positive social distance; SP, social pressure; VI, verbal interaction; 6, item dropped in model respecification.*p,0.01.

6M

arilynL

angevinet

al.

Downloaded By: [University of Sydney] At: 22:40 12 January 2009

and Hagler noted that the lack of a main effect for gender may have resulted from asubsample of grade 4 girls who had unusually low scores.

Langevin and Hagler’s (2004) results gave preliminary evidence of internal consistencyreliability and construct validity; however, further examination of the psychometricproperties of PATCS is required. The specific aims of this study were as follows:

N To re-examine the factor structure of PATCS using a confirmatory factoranalysis (CFA). It was hypothesized that a first-order three-factor structurecomprised of PSD, SP, and VI factors would be replicated. However, becausemodifications are typically made in initial model testing (Benson andBandalos 1992), cross-validation of CFA studies has been suggested (Cliff1983). Therefore, the total sample in this study was randomly split into acalibration sample in which initial model testing was done and a validationsample in which results from initial model testing were cross-validated. Twocross-validation approaches were used. In the first and least stringent level ofcross-validation (Bentler 1980, Bagozzi and Yi 1988) the factor pattern forthe model accepted as best fit to the data in the calibration sample washypothesized in the validation sample. In this method the parameterestimates were free to vary. In the second and more stringent level of cross-validation, a sequential series of invariance analyses were conducted.Invariance analyses test for the appropriateness of the equality constraintsfor sets of parameters across groups.

N To re-examine known groups validity. It was hypothesized that (1)participants who had had contact with people who stutter would havehigher scores than those who had not (contact), (2) girls would have higherscores than boys (gender), and (3) there would be a developmental trend withattitudes becoming increasingly positive as grade levels increased (grade).

N To investigate convergent validity. The Pro-Victim scale (Rigby and Slee1991, 1993), a measure of attitudes toward bullying and victims, was selectedas a measure of convergent validity because CWS are often rejected andbullied. However, because not all CWS are victimized, it was hypothesizedthat the PATCS would correlate moderately with the Pro-Victim scale.

N To re-examine internal consistency and investigate test–retest reliability. Itwas hypothesized that high internal consistency would be replicated and thatscores would be stable over a 1-week test–retest interval.

N To explore the negativity of peer attitudes. Frequencies of mean scores in thenegative range ((2.99) of the response scale for all participants and forparticipants in each grade level were examined.

Methods

Participants

Participants were 760 third- to sixth-grade students (the full sample) from sevenschools in socio-economically advantaged and disadvantaged urban and rural areasin a western Canadian province.1 Participant characteristics for age and gender foreach subsample in the CFA and the investigations of validity, reliability, andnegativity are reported below. All grades were represented in all subsamples exceptthe test–retest subsample as described below. Schools were recruited by sending

Peer attitudes toward children who stutter 7

Downloaded By: [University of Sydney] At: 22:40 12 January 2009

notices of the research project to school boards. Because PATCS was still in itsdevelopmental stage and to prevent possible embarrassment of and/or deleteriousheightened sensitivity toward a specific stuttering child, classrooms in whichteachers indicated that there were no CWS were selected. Measures were completedby all children who were in class at the time the measures were administered. Oneurban school was randomly selected as the test–retest sample.

Measures

PATCS

The 40 item PATCS scale comprised three-factor subscales: PSD, SP, and VI(Langevin and Hagler 2004). Twenty-one items were positively worded and 19 werenegatively worded. A five-point Likert response scale was used (where 15stronglydisagree; and 55strongly agree). All items were summed to derive a total scale scoreand subscale items were summed to derive subscale scores. Negatively worded itemswere inversely coded. Higher scores reflected more positive attitudes.

Pro-Victim scale

The ten-item Pro-Victim scale (Rigby and Slee 1991, 1993) measures attitudestoward bullying (e.g., ‘Kids who get picked on usually deserve it’), support forvictims (e.g., ‘It makes me angry when a kid is picked on without reason’), andrejection of weak kids (e.g., ‘Nobody likes a wimp’). A slight change in terminologywas made to one item to match that used by Canadian children and, consistent withMenesini et al. (2003), a five-point Likert scale was used in place of the original three-point scale. Using a five-point scale allowed increased depth of content validity andthe capturing of more true variance per item (Morey 2003). Internal consistency inthis study was 0.72.

Procedures

The PATCS and Pro-Victim scales were administered during regular class times.Before completing the PATCS, participants were shown the videotape of a 9-year-old boy with moderate stuttering and an 8-year-old girl with severe stuttering thatwas used by Langevin and Hagler (2004).2 Thereafter, they were asked to indicatewhether or not they knew someone who stutters. To control for reading difficulties,items were read aloud. To minimize fatigue, participants were given a break betweenadministrations of the scales. To minimize socially desirable responding, childrenwere told that there were no right or wrong answers and questionnaires werenumerically coded to assure anonymity. To measure test–retest reliability, a secondPATCS was administered 1 week after the first to participants in the test–retestsample. With the exception of asking children if they knew someone who stutters,the same procedures were used. To minimize the potential for inadvertentlyfostering a negative bias toward CWS by the inclusion of negative items in thePATCS, administration of the scales was immediately followed by an interventionprogramme called Teasing and Bullying: Unacceptable Behaviour (TAB) (Langevin2000). TAB aims to improve attitudes toward CWS and improve attitudes toward

8 Marilyn Langevin et al.

Downloaded By: [University of Sydney] At: 22:40 12 January 2009

bullying. It was delivered by teachers who had attended a training workshop. In apilot study of the effectiveness of TAB using the 40-item PATCS reported on inLangevin and Hagler (2004) as a pre- and post-intervention measure, TAB wasshown to improve attitudes toward CWS (Langevin 1997).

Data preparation

In the Langevin and Hagler (2004) study, a principal components analysis (PCA)with orthogonal rotation was used to reduce a set of items into components, a ‘goodtest-building technique when used at an early stage’ (Osterlind 2006: 382). To clarifythe factorial structure of PATCS further, prior to performing the CFA in this study,data from Langevin and Hagler were re-analysed using the exploratory factoranalysis (EFA) maximum likelihood method with promax (oblique) rotation. Theuse of the EFA (rather than PCA) allowed clarification both of the dimensionalstructure of the scale and the assignment of the items to each dimension. Moreover,the oblique rotation is more consistent with the theory of PATCS, as Langevin andHagler suggested that it measured a multidimensional construct in which PSD, SPand VI are not independent but are assumed to measure different, yet relatedaspects of peer attitudes toward CWS.

The re-analysed factor structure was the same as that previously achieved. Thesame three factors representing the constructs of PSD, SP, and VI emerged;however, assignment of nine items to factors was clarified. For example, one item(item 7) moved from cross-loading on PSD and SP to loading only on SP. Only oneof the nine items moved to a completely new factor. The resultant factor loadingpattern for the re-analysed Langevin and Hagler data is shown in table 1 under thecolumn titled ‘Model 1: Hypothesized First-Order Model’. This model was firsthypothesized in the CFA in this study. Factor correlations ranged from 0.70 to 0.78,confirming the need to use the oblique method of rotation.

For each scale less than 5% was missing. These data were imputed using theexpectation maximization procedure (SPSS 12.0; SPSS Inc., Chicago, IL, USA). Theprocedure is widely used for missing data replacement as it results in little or nodiscrepancy with the original covariance matrix (Little and Rubin 1989).

Results

Descriptive statistics

Item means, shown in table 1, ranged from 2.81 to 4.02. All inter-item correlationsexcept one between items 20 and 27, were positive and significantly greater thanzero, ranging from 0.12 to 0.70.3

CFA

To examine the theory of the peer attitudes toward CWS construct proposed byLangevin and Hagler (2004), the structural validity of PATCS was evaluated usingCFA. CFA is widely accepted as a more powerful test of a ‘hypothesized latency inthe data’ (Osterlind 2006: 374) than the exploratory factor analysis. Using CFAallows the (1) evaluation of the correspondence of the data with a model specified

Peer attitudes toward children who stutter 9

Downloaded By: [University of Sydney] At: 22:40 12 January 2009

earlier; (2) examination of the extent to which the proposed factor structure ofPATCS explains its postulated dimensions as compared to alternative models; (3)assessment of the relative fit of competing models, and (4) examination oftheoretically meaningful hierarchical models.

There were 751 participants in the CFA (age56–12 years, mean59.8 years,gender5404 boys and 347 girls).4 In the calibration and validation samples therewere 375 (age57–12 years, mean59.7 years, gender5192 boys and 183 girls) and376 (age56–12 years, mean59.8 years, gender5212 boys and 164 girls) participants,respectively. The CFA was conducted using the maximum likelihood method fromthe AMOS programme (Arbuckle and Wothke 1999). Chi-square (x2) is one of themost commonly used fit indexes. Small values relative to the degrees of freedomindicate statistically non-significant differences between the actual and the impliedmatrixes, signalling no discrepancy between the hypothesized model and the data.However, this statistic is sensitive to sample size. Thus, following current practice weused the root-mean-square error of approximation (RMSEA) and its 90%confidence interval (CI) to assess approximate goodness of model fit in thepopulation. Values lower than or close to 0.06 suggest good fit (Hu and Bentler1999) and a narrow confidence interval suggests good precision (MacCullum et al.1996). We also report the relative likelihood ratio of x2 to degrees of freedom (x2/d.f.) statistic. Values less than 3 are considered to indicate good fit (Kline 1998)when the sample size is large. In addition, the goodness of fit index (GFI) was usedto reflect a relative amount of covariance accounted for by the model. Values greaterthan 0.90 suggest good fit (Kline 1998). Finally, the Tucker–Lewis index (TLI) andcomparative fit index (CFI) incremental fit indexes that have been shown to berelatively independent of sample size were used (Marsh et al. 1988, Fan et al. 1999).Values greater than 0.90 are required; however, Hu and Bentler (1999) suggest a cut-off value of close to 0.95. The difference in the Chi-square test (Dx2) was used toevaluate alternative nested models. A statistically significant value indicates that aless parsimonious model fits significantly better that the initial model (Kline 1998).

Results of the CFA for all models tested in the calibration and validation samplesare shown in table 2. As would be expected when samples are large, x2 wassignificant for all models. Model 1, comprising 40 items and three first-order factors,had acceptable RMSEA and x2/d.f. However, GFI, TLI, and CFI were belowacceptable levels. Also, three items had low communalities (0.11–0.27) indicatingthat they shared little variance with the extracted factors. Although there are nostrict cut-off values, the general rule of thumb is that communalities lower than 0.30are problematic and indicate that relevant items do not share the adequate amountof variance with the extracted factors, possibly measuring a different construct orsimply being problematic items. These items were deleted;5 hence, all subsequentmodels comprised 36 items. Model 2, the 36-item first-order three-factor model,was then hypothesized. The three factors represented the constructs of PSD, SP, andVI. Thereafter two error covariances were allowed: items 24 and 29, and items 5 and12,6 resulting in Model 3. Model 4 was a first-order four-factor model. This modelwas exploratory and was hypothesized on the basis of the modification indicesrather than an a priori theoretical model since no theoretical model of peer attitudestoward CWS could be found in the literature other than the work of Langevin andHagler (2004). The fourth factor comprised six items that previously loaded on PSD.Although three items (8, 11 and 15) appeared to tap into a friendship or intimaterelations factor, as a group the six items did not appear to be sufficiently distinctive

10 Marilyn Langevin et al.

Downloaded By: [University of Sydney] At: 22:40 12 January 2009

from the remaining items on the PSD factor. Although as shown in table 2, Model 4had good fit, Model 3 was accepted as a theoretically better fit to the data; the PSDfactor reflects general comfort in being with CWS, the SP factor reflects concernwith what others think about CWS, and the VI factor is characterized by frustrationand annoyance that can be experienced by listeners. Although GFI for Model 3 didnot exceed 0.90, the remaining fit indices were acceptable. Correlations amongfactors were 0.83 and 0.88 between PSD and SP and VI, respectively, and 0.93between SP and VI. Given such strong correlations between factors, a hierarchicalmodel with one second-order factor was hypothesized. Thus, Model 5, a second-order factor model comprising a general attitude factor and the three first-orderfactors of PSD, SP, and VI was posited. This second-order model was hypothesizedon the basis of the high correlations among factors rather than an a priori theoreticalmodel of a hierarchical factor structure. Fit indices suggested that Model 5 was aplausible fit to the data in the calibration sample. Factor loadings of the first-orderfactors on the second-order factor were 0.87 for PSD, 0.91 for SP, and 0.99 for VI.

Cross-validation

Table 2 shows that Model 5 yielded fit indices in the validation sample that were similarto the calibration sample and invariance tests indicated equality of the parametersconstrained across samples to the level of first- and second-order factor loadings andfirst-order factor disturbances (x2

121552639, x2/d.f.52.17; GFI50.83; TL50.91;CFI50.91, RMSEA50.040, its 90% CI: 0.037–0.042). Replicating the strongcorrelations among the first order factors revealed in the analysis in the calibrationsample, factor loadings of the first-order factors on the second-order factor were 0.85for PSD, 0.93 for SP, and 0.98 for VI. Overall, these results suggested that Model 5

Table 2. Goodness-of-fit indices for first- and second-order models of PATCS

Model x2*

Degrees offreedom

(d.f.) x2/d.f. Dx2 Dd.f. GFI TLI CFI RMSEA (90% CI)

Calibration sample (n5375)Model 1a 1887 734 2.6 0.78 0.86 0.87 0.065 (0.061–0.068)Model 2b,c 1587 588 2.7 0.79 0.88 0.88 0.067 (0.063–0.071)Model 3b,c 1259 587 2.2 328* 1 0.83 0.92 0.92 0.055 (0.051–0.060)Model 4c,d 1139 584 2.1 120* 3 0.85 0.93 0.94 0.050 (0.046–0.055)Model 5b,c 1259 587 2.2 0.83 0.92 0.92 0.055 (0.051–0.060)

Validation sample (n5376)Model 5b,c 1342 587 2.3 0.83 0.90 0.91 0.059 (0.054–0.063)

Pooled sample (n5751)Model 5b,c 1791 587 3.1 0.87 0.92 0.93 0.052 (0.050–0.055)

GFI, goodness-of-fit index; TLI, Tucker-Lewis index; CFI, comparative fit index; RMSEA, root meansquare of approximation; CI, confidence interval.aForty items.bThree first-order factors.cThirty-six items.dFour first-order factors.*p,0.001.

Peer attitudes toward children who stutter 11

Downloaded By: [University of Sydney] At: 22:40 12 January 2009

provided an adequate fit to the data across the calibration and validation samples. Thus,the samples could be pooled for subsequent analyses. As shown in table 2, a final test ofModel 5 in the pooled sample resulted in fit indices that were acceptable. Item loadingsand communalities for Model 5 are reported in table 1.

Known groups analysis

There were 734 participants in the known groups analysis (age56–12 years, mean59.8years, gender5395 boys and 339 girls).7 Participants who did or did not know someonewho stutters were 241 (32.8%) and 493 (67.2%), respectively. Results of a three-wayANOVA of contact (two levels), gender (two levels), and grade (four levels) revealedmain effects for contact [F(1, 718)518.84, p,0.001; partial g250.026] and gender [F(1,718)514.36, p,0.001; partial g250.020]. As hypothesized, children who had contactwith someone who stutters had higher mean scores (mean53.78; SD50.76) than thosewho did not (mean53.43; SD50.72) and girls had higher mean scores (mean53.65;SD50.67) than boys (mean53.45; SD50.81). No main effects were found for gradeand there were no statistically significant interactions.

Convergent validity

There were 732 participants in the convergent validity analysis (age56–12 years,mean59.8 years, 393 boys and 338 girls).8 Mean scores for the PATCS and Pro-Victim scales were 3.54 (SD50.75; range51.06–5.00) and 4.04 (SD50.58;range51.0–5.0), respectively. As expected, a moderate positive correlation(Spearman’s rho) of 0.43 (p,0.01) was obtained.

Internal consistency

Internal consistency for the total scale was 0.97. These results meet the suggestedcriteria of greater than 0.70 for group comparisons and greater than 0.90 for usewith individuals (Nunnally and Bernstein 1994). For the PSD, SP, and VI subscalesinternal consistency was 0.95, 0.89, and 0.88, respectively. Although these resultsmeet the suggested criterion for group comparisons, internal consistency for the SPand VI scales fall slightly short of the suggested criterion for use with individuals.9

As a further indication of internal consistency, item-total score correlations rangedfrom 0.47 to 0.76 for the total scale, and from 0.48 to 0.79, from 0.46 to 0.72, andfrom 0.50 to 0.71 for PSD, SP, and VI subscales, respectively. These item-total scorecorrelations exceed the cut-off of 0.30 suggested by Nunnally and Bernstein (1994).

Test–retest reliability

There were 48 participants in the test–retest sample (grades53–5, age56–10 years,mean58.7; gender528 boys and 20 girls; contact with someone who stutters: 18 (38.3%)knew someone who stutters, 29 (61.7%) did not). Although the range for grade and age,respectively, fell one and two years short of that for the full sample, demographics for thetest–retest sample were similar to the full sample.10 The 1-week interval test–retestreliability (Pearson correlation) was 0.88 (p,0.01) for the total scale. Correlations for the

12 Marilyn Langevin et al.

Downloaded By: [University of Sydney] At: 22:40 12 January 2009

PSD, SP, and VI subscales were 0.84, 0.80, and 0.74 (p,0.01), respectively. Streiner andNorman (2003) suggest that a correlation of 0.75 is a minimal requirement.

Negativity of attitudes

Results indicate that: (1) 21.7% of participants had mean scores that were in thesomewhat negative (2.00–2.99) to very negative (1.00–1.99) range of the scale; and(2) participants in grade 4 had the largest proportion (24.7%) of negative scores.

Discussion

The purpose of this study was to examine further the psychometric properties ofPATCS and to investigate the negativity of attitudes toward CWS. Findings suggestthe following:

N PATCS taps a general peer attitude toward CWS factor as well as the specificconstructs of PSD, SP, and VI.

N Internal consistency for the total PATCS and its subscales exceed therecommended minimum for group use. However, although internalconsistency for the SP and VI subscales fall just short of the recommendedminimum of 0.90 for individual use, there appears to be no consensus onsuch a criterion.

N PATCS scores were stable over a 1-week test–retest period.

N PATCS differentiated children in the expected directions when children weregrouped on the basis of contact with someone who stutters and gender.

N A statistically significant moderate correlation with the Pro-Victim scale gaveevidence of convergent validity.

The small effect size for the difference between boys and girls in this study isconsistent with that generally found in other investigations of attitudes toward thosewith physical or intellectual disabilities (Nowicki and Sandieson 2002). The lack of amain effect for grade is inconsistent with the findings of Langevin and Hagler(2004). However, this result is consistent with a recent meta-analysis in which noconclusive evidence for a developmental age trend in attitudes toward persons withphysical or intellectual disabilities was found (Nowicki and Sandieson 2002). One-fifth (21.7%) of participants had mean scores that were somewhat to very negative.These results are consistent with those of McDougall et al. (2004) in which 21% ofhigh-school students had slightly below neutral to very negative attitudes towardpeers with disabilities.

Implications for educational programmes

Results of this study provide empirical support for the persistent calls for school-based educational programmes about stuttering and give guiding information for thedevelopment of such programmes. Replication of the first-order factors of PSD, SP,and VI suggests that these constructs can be used in designing school-basedinterventions. The PSD factor, which reflects general comfort in being with CWS,suggests that discussions might address acceptance of CWS, in particular, similaritiesand differences among children who do and do not stutter. Levy et al. (2005) found

Peer attitudes toward children who stutter 13

Downloaded By: [University of Sydney] At: 22:40 12 January 2009

that a combined similarities–uniqueness message was more effective in increasingsocial tolerance than similarities only or uniqueness only messages.

The SP factor, which reflects concern about what others think about CWS, isconsistent with the construct of subjective norm that has been shown to be one ofthree determinants of the intention to perform a behaviour in the theory of plannedbehaviour (Ajzen and Madden 1986). This factor suggests that educationalinterventions might include discussions about making personal choices about waysin which to interact with CWS and about handling peer pressure.

The VI factor, which reflects frustration and annoyance that can surface inverbal interactions with CWS, had the highest loading (0.98) on the general attitudefactor suggesting that it made the strongest contribution to the general attitudefactor. This is not surprising given that the defining feature of stuttering is itsinterference with smooth and effective verbal interaction. This factor suggests thatopen discussion about the interference of stuttering on communication may validatethe frustration that can be experienced by listeners as well as the CWS and providethe opportunity to give strategies for being respectful and helpful listeners. Findingsalso suggest that arranging contact with someone who stutters may be an importantcomponent of intervention programmes.

Limitations and further research

A potential limitation is that the videotape used in this study did not portray a childwith mild stuttering. However, it is likely that results will be generalizable to childrenwith mild stuttering, given that Langevin et al. (1998) found that children with mildstuttering were teased. Notwithstanding, further research is needed to determine ifthe findings of this study will be replicated when referent children include those whostutter mildly. In addition, future research could investigate whether or not findingswill be replicated when referent children also include those who are older than the8–9-year-old children used in videotape in Langevin and Hagler (2004) and in thisstudy. Recall that the participants in grade 4 who were closest in age to the childrenin the video had the highest proportion of negative scores.

Further research is also needed to re-examine test–retest reliability, examinediscriminant validity, and determine if a negative item method effect exists.Although test–retest reliability was satisfactory in this study, results should beviewed with caution because the size of the sample in the test–retest group was lessthan 10% of the full sample and children in grade 6 were not represented. Thus, alarger and more representative test–retest sample is needed.

Discriminant validity could be examined by investigating systematic variance thatis unrelated to the construct of peer attitudes toward CWS. For example, PATCSscores could be correlated with measures of personality and social desirability.Streiner and Norman (2003) note that social desirability bias is a common measureof discriminant validity. Socially desirable responding occurs when respondents tendto agree to items that they ‘believe reflect socially desirable attitudes in order toshow themselves in a better light’ (Oppenheim 1992: 181).

It is commonly suggested that negative items be included in a scale to disruptacquiescent responding, a tendency to agree to items regardless of the content(Mueller 1986). However, further research is needed to determine if inclusion ofnegative items results in a negative item method effect (Marsh 1986).11

14 Marilyn Langevin et al.

Downloaded By: [University of Sydney] At: 22:40 12 January 2009

Finally, research is needed to address three further important matters: (1) tocross-validate these CFA findings by comparing the current CFA solution toalternative models, including a simple structure rotation model, performed onequivalent or larger size samples; (2) to standardize the scale with a sufficiently largeand representative sample of school-age children with subgroups selected accordingto specific criteria; and (3) to explore the possibility of a fourth factor capturingintimate–non-intimate relationships with CWS. The possibility that a fourth factormay be present tapping into social comfort with close intimate friendships is ofinterest in light of recent findings of a two factor structure of social distance inadults. In a factor analysis of social distance items measuring social comfort ofAfrican-American undergraduate students, the in-group, with White Americanundergraduate students, the out-group, Rollock and Vrana (2005) found an intimatefactor (e.g., ‘Happy to have a white person as someone I would date’) and a non-intimate factor (e.g., ‘Happy to have a white person on a sports team with me’). Asthe authors indicated, these results suggested that participants made a cleardistinction between intimate and non-intimate relations with the out-group.Although in-group preferences have been found in children (Bennett et al. 1998), noinvestigations of factorial structure of social distance measures in children could befound.

In conclusion, the findings of this study provide needed information about thenature of peer attitudes toward CWS and the extent to which they are negative.Although further research is needed to complement the psychometric analysesundertaken in this study, results provide further evidence of the reliability andvalidity of PATCS.

Acknowledgements

This research was supported by grants from an anonymous donor, the Elks andRoyal Purple of Canada, the University of Sydney, the Australian Department ofEducation, Training and Youth Affairs International Postgraduate ResearchScholarship and Postgraduate Award, and an Alberta Heritage Foundation forMedical Research Studentship Award. The authors thank the children whoparticipated in this study; and the reviewers for their constructive comments.

Notes

1. Participants were part of a larger intervention study; only the pre-intervention questionnaires areutilized in this study.

2. The PATCS was always administered first because video players were not consistently available forwhole class periods. The mean Pro-Victim score of 4.04 in this study is comparable with Menesiniet al.’s (2003) graphed means of approximately 4.17 and 4.25 on two subscales in their eleven-itemversion of the scale. They did not provide means for the third subscale or the total scale. Thissuggests that the inability to counterbalance administration of scales did not appear to affect thePro-Victim scores in this study unduly. For a comprehensive description of: (1) the selection of theCWS for the video-tape, (2) the children’s stuttering behaviours, and (3) a probe of potentiallystigmatizing characteristics other than stuttering, see Langevin and Hagler (2004). Other thanstuttering, the CWS in the video looked like ‘normal’ kids to participants in the probe.

3. Available from the first author.4. Nine questionnaires were excluded: four because they were not completed according to

instruction, four because the participants indicated they had a stutter on the PATCS, which could

Peer attitudes toward children who stutter 15

Downloaded By: [University of Sydney] At: 22:40 12 January 2009

not be verified, and one because the participant told the first author that she had a stutter, forwhich a referral was made.

5. These items were conceptually different from the others as they required participants: (1) to inferinternal states (items 4 and 20), (2) to make judgements about personal characteristics of CWS(item 27), or (3) to make judgements about how others should treat CWS (item 23).

6. This method, albeit not universally accepted, conceptually implies that the random measurementerror from relevant items have ‘something’ in common that is not represented in the model (Kline1998). The covariances were allowed because it is believed that items 24 and 29 tap into similaremotions, and items 5 and 12 tap into similar peer conceptions of CWS.

7. Seventeen PATCS were removed from the full sample due to inconsistencies or missing dataregarding the questions about contact with someone who stutters.

8. Because participants’ attendance in the classes varied at times due to other commitments (e.g.,individual tutoring), 20 participants did not complete the Pro-Victim scale.

9. For a discussion about acceptable levels of reliability when test scores are used for makingdecisions about individuals, see Streiner and Norman (2003). They suggest that suchrecommendations are tenuous since reliability is dependent on sample size.

10. Similarity of demographics is reflected in the following: 53.8% and 58.3% of participants wereboys in the full and test–retest sample, respectively; four of the five grades and five of the sevenages in the full sample were represented in the test–retest sample; and 32.5% and 38.3% of theparticipants in the full and the test–retest samples, respectively, had had contact with someonewho stutters.

11. Post-hoc analysis of a two-factor model comprised of a positive item factor (19 items) and anegative item factor (17 items) indicated that such a model was not an adequate fit to the data(x2

59351465, x2/d.f.52.47; GFI50.80; TL50.89; CFI50.90, RMSEA50.063 [CI50.059–0.067]).

References

AJZEN, I. and MADDEN, T. J., 1986, Prediction of goal-directed behavior: attitudes, intentions, andperceived behavioral control. Journal of Experimental Social Psychology, 22, 453–474.

ARBUCKLE, J. L. and WOTHKE, W., 1999, Amos 4.0 User’s Guide (Chicago, IL: SPSS Inc.).BAGOZZI, R. P. and YI, Y., 1988, On the evaluation of structural equation models. Journal of the Academy of

Marketing Science, 16, 74–94.BENNETT, M., LYONS, E., SANI, F. and BARRETT, M., 1998, Children’s subjective identification with the

group and in-group favoritism. Developmental Psychology, 34, 902–909.BENSON, J. and BANDALOS, D. L., 1992, Second-order confirmatory factor analysis of the Reactions to Tests

Scale with cross-validation. Multivariate Behavioral Research, 17, 459–487.BENTLER, P. M., 1980, Multivariate analysis with latent variables: causal modeling. Annual Review of

Psychology, 31, 419–456.BLOOD, G. W. and BLOOD, I. M., 2004, Bullying in adolescents who stutter: communicative competence

and self-esteem. Contemporary Issues in Communication Science and Disorders, 31, 69–79.CLIFF, N., 1983, Some cautions concerning the application of causal modeling methods. Multivariate

Behavioral Research, 18, 115–126.DAVIS, S., HOWELL, P. and COOKE, F., 2002, Sociodynamic relationships between children who stutter and

their non-stuttering classmates. Journal of Child Psychology and Psychiatry, 43, 939–947.EAGLY, A. H. and CHAIKEN, S., 1993, The Psychology of Attitudes (New York, NY: Harcourt Brace College

Publ.).FAN, X., THOMPSON, B. and WANG, L., 1999, Effects of sample size, estimation methods, and model

specification on structural equation modeling fit indexes. Structural Equation Modeling, 6, 56–83.FRY, E., 1968, A readability formula that saves time. Journal of Reading, 11, 513–515, 575–578.HU, L. and BENTLER, P. M., 1999, Cutoff criteria for fit indexes in covariance structure analysis:

conventional criteria versus new alternatives. Structural Equation Modeling, 6, 1–55.HUGH-JONES, S. and SMITH, P. K., 1999, Self-reports of short- and long-term effects of bullying on

children who stammer. British Journal of Educational Psychology, 69, 141–158.KLINE, R. B., 1998, Structural Equation Modeling (New York, NY: Guilford).LANGEVIN, M., 1997, Peer teasing project. In E. Healey and H. F. M. Peters (eds), Proceedings of the 2nd

World Congress on Fluency Disorders (Nijmegen: Nijmegen University Press), pp. 169–171.

16 Marilyn Langevin et al.

Downloaded By: [University of Sydney] At: 22:40 12 January 2009

LANGEVIN, M., 2000, Teasing and Bullying: Unacceptable Behaviour (Edmonton, AB: Institute for StutteringTreatment & Research).

LANGEVIN, M., BORTNICK, K., HAMMER, T. and WIEBE, E., 1998, Teasing/bullying experienced by childrenwho stutter: toward development of a questionnaire. Contemporary Issues in Communication Scienceand Disorders, 25, 12–24.

LANGEVIN, M. and HAGLER, P., 2004, Development of a scale to measure peer attitudes toward childrenwho stutter. In A. K. Bothe (ed.), Evidence-based Treatment of Stuttering: Empirical Bases and ClinicalApplications (Mahwah, NJ: Lawrence Erlbaum Associates), pp. 139–171.

LEVY, S. R., WEST, T. L., BIGLER, R. S., KARAFANTIS, D. M., RAMIREZ, L. and VELILLA, E., 2005, Messagesabout uniqueness and similarities of people: impact on U.S. Black and Latino youth. AppliedDevelopmental Psychology, 16, 714–733.

LITTLE, R. J. A. and RUBIN, D. B., 1989, The analysis of social science data with missing values. SociologicalMethods and Research, 18, 292–326.

MACCULLUM, R. C., BROWNE, M. W. and SUGAWARA, H. M., 1996, Power analysis and determination ofsample size for covariance structure modeling. Psychological Methods, 1, 130–149.

MARSH, H. W., 1986, Negative item bias in rating scales for preadolescent children: a cognitive–developmental phenomenon. Developmental Psychology, 22, 37–49.

MARSH, H. W., BALLA, J. R. and MCDONALD, R. P., 1988, Goodness-of-fit indexes in confirmatory factoranalysis: the effect of sample size. Psychological Bulletin, 103, 391–410.

MCDOUGALL, J., DEWIT, D. J., KING, G., MILLER, L. T. and KILLIP, S., 2004, High school-aged youths’attitudes toward their peers with disabilities: the role of school and student interpersonal factors.International Journal of Disability, Development and Education, 51, 287–313.

MENESINI, E., CODECASA, E., BENELLI, B. and COWIE, H., 2003, Enhancing children’s responsibility to takeaction against bullying: evaluation of a befriending intervention in Italian Middle Schools.Aggressive Behavior, 29, 1–14.

MOONEY, S. and SMITH, P. K., 1995, Bullying and the child who stammers. British Journal of SpecialEducation, 22, 24–27.

MOREY, L. C., 2003, Measuring personality and psychopathology. In J. A. Schinka, W. F. Velicer andI. B. Weiner (eds), Handbook of Psychology, Vol. 2 (Hoboken, NJ: Wiley), pp. 377–455.

MUELLER, D. J., 1986, Measuring Social Attitudes (New York, NY: Teachers College Press, ColumbiaUniversity).

MURPHY, W. P. and QUESAL, R. W., 2002, Strategies for addressing bullying with the school-age child whostutters. Seminars in Speech and Language, 23, 205–212.

NOWICKI, E. A. and SANDIESON, R., 2002, A meta-analysis of school-age children’s attitudes towardspersons with physical or intellectual disabilities. International Journal of Disability, Development andEducation, 49, 243–265.

NUNNALLY, J. C. and BERNSTEIN, I. H., 1994, Psychometric Theory (New York, NY: McGraw-Hill).OPPENHEIM, A. N., 1992, Questionnaire Design, Interviewing and Attitude Measurement (New York, NY: Pinter).OSTERLIND, S. J., 2006, Modern Measurement. Theory, Principles, and Applications of Mental Appraisal (Upper

Saddle River, NJ: Pearson/Merrill Prentice-Hall).OWEN, C. A., EISNER, H. C. and MCFAUL, T. R., 1977, A half-century of social distance research: national

replication of the Bogardus’ studies. Social Distance Research, 66, 80–98.RIGBY, K. and SLEE, P. T., 1991, Bullying among Australian school children: reported behaviour and

attitudes toward victims. Journal of Social Psychology, 13, 615–627.RIGBY, K. and SLEE, P. T., 1993, The Peer Relations Questionnaire (Adelaide, SA: Institute of Social Research,

University of South Australia).ROBERTS, C. M. and SMITH, P. R., 1999, Attitudes and behavior of children toward peers with disabilities.

International Journal of Disability, Development and Education, 46, 35–50.ROLLOCK, D. and VRANA, S. R., 2005, Ethnic social comfort I: Construct validity through social distance

measurement. Journal of Black Psychology, 31, 386–417.RUSCELLO, D. M., LASS, N. J. and BROWN, J., 1988, College students’ perceptions of stutterers. National

Student Speech Language Hearing Association Journal, 16, 115–120.STREINER, D. L. and NORMAN, G. R., 2003, Health Measurement Scales: A Practical Guide to Their Development

and Use, 3rd edn (New York, NY: Oxford University Press).

Peer attitudes toward children who stutter 17

Downloaded By: [University of Sydney] At: 22:40 12 January 2009