www.elsevier.com/locate/im

Information & Management 44 (2007) 142–153

Advancement, voluntary turnover and women in IT:

A cognitive study of work–family conflict

Deborah J. Armstrong a,*, Cynthia K. Riemenschneider a, Myria W. Allen b,Margaret F. Reid c

a Information Systems Department, Sam M. Walton College of Business, University of Arkansas, Fayetteville, AR 72701, USAb Department of Communication, Fulbright College of Arts and Science, University of Arkansas, Fayetteville, AR 72701, USAc Department of Political Science, Fulbright College of Arts and Science, University of Arkansas, Fayetteville, AR 72701, USA

Received 27 March 2005; received in revised form 27 October 2006; accepted 7 November 2006

Available online 2 January 2007

Abstract

We used quality of work life theory and the causal mapping method to evoke the concepts and linkages of women’s cognitions

about work–family conflict in order to better understand the issues contributing to advancement barriers and voluntary turnover of

women in IT. The major concepts (Managing Family Responsibilities, Work Stress, Work Schedule Flexibility, and Job Qualities)

were found to not only impact each other but also were key factors influencing women’s advancement opportunities and voluntary

turnover. Organizations may use these insights to mitigate voluntary turnover and increase workforce diversity by addressing female

IT professionals’ concerns regarding work–family conflict issues.

# 2006 Elsevier B.V. All rights reserved.

Keywords: Gender and information technology; Work–family conflict; Glass ceiling; Turnover; Cognition; Causal mapping

1. Introduction

Studies since the late 1970s have suggested that

information technology (IT) professionals exhibit char-

acteristics that differ from those in other professions. In a

frequently cited study, Couger and Zawacki [19]

surveyed more than 1600 data processing employees.

Their findings indicated that these individuals showed a

significantly higher need for challenging work but a

significantly lower need for social interaction. More

recently, Wynekoop and Walz [78] compared personality

characteristics of IT professionals to population norms

* Corresponding author. Tel.: +1 479 575 6158;

fax: +1 479 575 4168.

E-mail address: darmstrong@walton.uark.edu (D.J. Armstrong).

0378-7206/$ – see front matter # 2006 Elsevier B.V. All rights reserved.

doi:10.1016/j.im.2006.11.005

and found that IT professionals were more ambitious,

logical, and conservative. Research has found that

members of the IT workforce possess unique attitudes,

interests, sense of identity, and work consciousness [57].

Furthermore, the work setting faced by IT profes-

sionals has some unique features. The IT environment is

characterized by work practices that often include long

hours, late nights, after-hour meetings, on-call duty, and

a continual state of ‘rush’ or crisis [1]. It is characterized

by rapid technological change, with new technical skills

appearing and others becoming obsolete.

IT is also different from other traditionally male-

dominated occupational fields like accounting and

medicine where female participation is rising, whereas

the number of women in the IT field is declining [5,10].

This has triggered discussions on how to retain IT

employees with particular focus on how workplace

D.J. Armstrong et al. / Information & Management 44 (2007) 142–153 143

conditions may impose barriers to women’s advance-

ment (often known as the glass ceiling) and lead them to

leave their employer [42]. It is widely recognized that

female participation in the computing disciplines needs

to be encouraged [71] not only in terms of gender

balance, but also as a means of meeting current and

future demands for qualified employees.

Our study examined female IT professionals’

cognitions about balancing work and family issues

within the context of advancement and voluntary

turnover. By focusing on how women in the IT field

balance work and family we can further assist gender

sensitive research and funding organizations in their

search for ways to increase diversity and retain non-

traditional workers. Our study was initiated to answer

two questions:

(a) H

ow do women in IT perceive the interaction of

work and family responsibilities? and

(b) I

f there is an interaction, how do these women see it

affecting advancement and voluntary turnover?

2. Previous research

2.1. Gender in the information technology

workplace

Prior research on gender in the IT workplace has

centered on four themes:

(a) G

ender differences in the use of technology, such as

the perception and frequency of email use [26];

technology adoption and usage intentions [73]; the

impact of Internet usage on the salary gap [28]; and

motivation to telecommute [11].

(b) C

areer orientations and voluntary turnover inten-

tions [37] and strategies employed by successful

female IT professionals [58].

(c) C

omparative studies that offer insights into cultural

or workplace differences of status, stress, voluntary

turnover intentions and salary in Australia and New

Zealand [70,74], the United Kingdom [61], Singa-

pore [69], Germany [56], and Finland [72].

(d) W

orkplace conditions of women in IT, including

discriminatory practices that reinforce work prac-

tices and job segregation [77]. Impediments to the

advancement of women and established discrimi-

natory practices have, of course, existed in other

fields and it seems that women in IT positions

encounter similar barriers, obstacles to advance-

ment, lack of quality mentoring, and lack of choice

assignments. As a consequence, women tend to

receive lower salaries, are seen as having less

favorable promotion chances, and are filling fewer

management positions [38].

Several authors (e.g., [13,32]) have focused on

identifying workplace conditions that constitute orga-

nizational barriers, and may explain women’s reluc-

tance to enter or remain in the IT workforce. Ahuja

proposed a set of barriers to the advancement of women

in IT including social factors (e.g., anxiety and limited

self-efficacy), structural factors (e.g., occupational

culture and lack of role models), and work–family

conflict (which has only recently been addressed).

2.2. Quality of work life and work–family conflict

Quality of work life is defined as satisfying an

employee’s needs via the resources, activities and

outcomes that arise from involvement in the workplace

[66], and has been found to influence intention to quit

[15,43]. There are several different perspectives of

quality of work life [46]. The spillover model [39]

suggests that satisfaction experienced in one domain

(e.g., work) may have a positive or negative effect on

other domain(s) (e.g., family life) and vice versa [67].

Consistent with this model, work–family conflict is

defined as an individual’s cognitions about how work

and family roles exert incompatible pressures [21]. In

other disciplines such conflicts have been identified as a

source of stress [8,24,54] and can affect organizations in

terms of the lost time, reduced productivity [41], and

eventually voluntary turnover [25,63] of employees.

Researchers have also found that an individual’s

quality of work life is influenced by his or her work

experience and future career expectations [16,34].

Work–family conflict may also negatively affect career

advancement [59,68]. In their study of female bank

managers Liff and Ward [44] found that the women’s

colleagues and bosses did not see senior management as

an appropriate objective for them. These messages were

conveyed by negative comments about women who had

been promoted, and via the assumption that all woman

would want to have children and give them priority over

work.

3. Method

3.1. Overview

In our study, we use a rich qualitative method and a

cognitive perspective in measuring the concepts and

linkages regarding work and family, in order to

D.J. Armstrong et al. / Information & Management 44 (2007) 142–153144

Table 1

Participant job titles

Job title Number of participants

Administrator 1

Associate Programmer 1

Database Analyst 1

Data Entry Clerk 1

Enterprise Reporting Group 1

IS Procurement 1

Programmer 4

Programmer/Analyst 4

Project Coordinator 1

Project Leader 3

Repairs Field Personnel 1

Sales 1

Senior Business Analyst 2

Senior Programmer/Analyst 2

Senior Project Leader 1

Supervisor 1

Systems Administrator 1

understand how the workplace issues women perceive

interact and influence voluntary turnover and perceived

promotion opportunities. Causal mapping is a collection

of techniques that allows a researcher to capture the

cognitions of an individual [36], or group [12,22], by

evoking the concepts from the participants, and the

structure of the cause and effect relationships into which

the participants placed them [7,53]. We used causal

mapping to study the cognitive representations about

women by eliciting perceptions of the issues they face

and casting them into structural representations. The

approach was consistent with the procedure detailed

elsewhere [4,55].

3.2. Data collection

Our data source was six focus groups totaling 39

women employed in the IT field at a Fortune 500

manufacturing organization. Narratives were gathered

from the respondents through open-ended focus group

interviews.

3.2.1. Sample

The focal organization employs more than 120,000

people in 22 states and countries. The IT department

serves 8500 people located at corporate headquarters.

All participants are at corporate headquarters in the

mid-south region of the United States, thus although the

organization was a multinational, this was not reflected

in the ethnicity of our respondents. The total number of

employees in the IT department was 300, and there were

73 female IT employees. The sample consisted of 39 of

these women, a 53% response rate. Our sample size

while small was consistent with the sample sizes in

other qualitative studies (e.g., [2,31]), and met accepted

key features in qualitative research sampling criteria

[20,51].

The company’s human resources director sent

information about the study to all of the organization’s

female IT employees. She then contacted the potential

participants and assigned them to the focus groups. We

admit that the grouping of the subjects might have

influenced the group dynamics if informal leaders

within each group had led a discussion.1 However, by

having two researchers present in each meeting we

sought to minimize this problem by drawing out quieter

participants and redirecting the discussion if any one

participant appeared to be taking control.

1 We would like to thank an anonymous reviewer for this insightful

comment.

The majority of participants had only worked in IT

for the focal company (49%) or for the focal company

and one previous employer (31%). The ‘‘average’’

participant had worked in IT for 8 years, with 54%

having worked in IT for 5 years or less. Various job titles

were represented; the most frequently given were

programmer, programmer/analyst, project leader,

senior business analyst, and senior programmer analyst.

See Table 1 for a complete listing of all participants’

titles. Only 15% were self-identified managers, and they

supervised between 3 and 11 employees. Some of the

women were single, some had grown children, some

had babies or were pregnant, and some were in the

middle of parenting. Thus, the data evoked in the

interviews were the salient and intertwined concepts

generated by a heterogeneous group of women working

in the IT field.

3.2.2. Focus groups

Narratives were gathered during six 1 h long focus

group meetings held in workplace conference rooms

over a 2-week period with between four and eight

women in each group. This arrangement allowed us to

interact face-to-face with the women and elicit their

views on the workplace issues they faced. The focus

group format is designed to allow respondents to build

on one another’s comments [76].

Our procedure consisted of open-ended questions

with probes [62]. The 39 women discussed several

questions of relevance to our study; for example, one

Technical Writer 1

Undeclared 11

Total 39

D.J. Armstrong et al. / Information & Management 44 (2007) 142–153 145

question was: ‘‘What challenges do you think women in

IT face that their male peers do not?’’ All questions put

to the groups focused on work aspects; no questions

asked about family or leisure activities (see Appendix A

for the interview guide). The themes related to work–

family conflict were generated by the respondents.

Based on the answers to a question, follow up probes

were asked to elicit details of the respondents’

experiences. Audiotapes made during the meetings

were transcribed, resulting in between 14 and 22 pages

per session.

3.3. Deriving revealed causal maps

The first task in deriving the maps was to identify the

causal (cause–effect) statements from the transcripts.

These statements are identified through the use of key

words (such as because or so). Due to differences in the

authors’ backgrounds, all four of us independently

coded each interview text. Once each author coded the

texts, comparisons were made for agreement and

disagreement. Where disagreement occurred, the dis-

crepancies were resolved through discussion; thus the

authors’ biases were neutralized. The causal statements

were then separated into cause and effect phrases which

were used to construct raw causal maps.

In a separate process, the relevant concepts were

identified (using the participants’ statements). One

researcher read all the transcripts and inductively

generated a list of preliminary coding categories, then

the other authors read the material to verify their face

validity and assess the parsimony and coverage of the

categories. Scott’s pi [65] was calculated using 3 of the

6 transcripts and 11 of the coding categories. It

estimates the reliability of the coding process; i.e.,

whether the reviewers agreed on the categories more

often than they would if they had been randomly

selecting concepts. A heuristic for content analysis is to

require a reliability coefficient of 0.90 or more when

using Holsti’s formula, and approximately 0.75 or more

when using pi or alpha [35]. For our study, Scott’s pi

ranged from 0.72 to 0.77 between the coders, indicating

an acceptable level of reliability.

Congruent with studies in the management and IT

literature [29,30], and to reinforce conceptual validity,

the groupings were examined for theoretical and logical

relevance, and placed into categories to create a coding

scheme. Once the coding scheme was completed, the

causal statements for each respondent were placed into

the appropriate categories. Using the categories in place

of the cause and effect phrases in the raw maps, the

causal maps were developed. A causal map was

developed for each of the six focus groups. The six

maps were then aggregated by adding together the

concepts and linkages of each causal map [52]. The

union of all concepts and linkages from the individual

maps were placed on the final aggregate map.

As the concepts emerged from the participants, the

point of redundancy represented the point at which

further data collection would not provide additional

concepts [27]. The point of redundancy was computed

by aggregating the concepts mentioned by each

participant. No new concepts were elicited from the

sixth focus group, so redundancy was reached by the

fifth focus group interview eliciting a total of 63

concepts. This suggested that the sample of six focus

groups (39 women) was sufficient to capture all of the

relevant concepts in the sample.

After the maps had been developed, the next task was

to analyze the maps based on past research using this

method [23]. There are two aspects of analyzing a

causal map: its content and structure. The content

analysis consists of identifying and defining the

concepts contained in the domain under study. The

structural analysis consists of analyzing the linkages

between the concepts. The measures used in the

structural analysis are from social network analysis

and include the adjacency and reachability matrices,

and centrality measures. See [60] for details of the

measures used.

Reachability is an indicator of the total strength of

the connection between concepts. The higher the

reachability, the greater the strength of the connection

(direct effect) and/or the greater the number of paths

that can be used to connect the concepts (indirect

effects) [64]. Reachability is reported on the linkage

between the nodes on the revealed causal map. The first

step in calculating the reachability matrix is to develop

an adjacency matrix which captures the direct relation-

ships between two concepts. We were interested in the

occurrence and strength of the relationship so the

adjacency matrix we used contained values between 0

and 22 (in each cell). Centrality is a measure of the

relative importance of a concept or how involved it is in

the cognitive structure. It is a ratio of the sum of the

linkages involving the concept divided by the total

linkages in the matrix [40].

4. Results

4.1. Concepts and linkages

There were 12 major concepts that constituted the

cognitive structure of the issues in the workplace. Each

D.J. Armstrong et al. / Information & Management 44 (2007) 142–153146

Table 2

Concept level classification scheme

Concept Definition Examples

Actions the company can take Actions the company can take to solve

problems, increase performance

Stagnant management holders are moved out;

So they look for you to have physical ties,

emotional ties to the area

Barriers: promotion Problems being promoted or lack of

opportunities for promotion

I don’t have a chance at promotion; Then you

just can’t go any further

Company comparisons References to other companies,

comparison to other companies

At X [different company] all, not all, most of the

majority of the management are women

Managing family

responsibilities

Care for family members (children,

home, spouse)

I do have to go get my kids; If the family

is more important

Work schedule flexibility Flexibility; give and take; come and

go when needed; reduced work schedule

If you need to leave and do some personal business;

I wanted to work part-time and still get some pay

during that time

Job qualities Qualities, skills needed for job If you are willing to put hours in; If you go to night

classes and learn it

Lack of control This is how things are, You can’t change it You can’t change it; It was out of my control

My situation is different Cannot relate to what ‘‘everyone

else’’ is experiencing

She was in a different circumstance; X and I have

special work schedules

Reasons left: job

characteristics

Left job because bored, stressed,

project ambiguity

They like the challenge; They like better opportunities;

I was bored silly; I was on call 24 h a day 7 days a week

and got calls in the middle of the night

Result of own decision People are responsible for their own

decisions; women make decisions

about family

I personally have made that decision; She just wanted

a more low-keyed approach to life

Work stress Challenges of stress There was just so much stuff going on; Things were

more competitive; Working way too many hours

Turnover Turnover, leaving organization Most of the women had left way before 15–20 years;

I had folks that worked with me leave; I left, I just quit

is included in Table 2 and the overall causal map is

presented in Fig. 1. The arrows indicate the causal

relationships (linkages) between two concepts with the

arrow initiating from the cause concept and pointing

toward the effect concept.

For our study, the reachability ranged from 0 to 0.10

and a 0.05 reachability cutoff was used to improve map

readability. The criterion was used to focus on the most

central concepts of the work–family conflict issue. The

concepts with the highest average reachability were

Barriers: Promotion, Work Schedule Flexibility, Result

of Own Decision, and Turnover. These are the most

reachable through direct and/or indirect paths. See

Fig. 1 for reachability values. For ease of understanding,

we provided the same information in Table 3. Here, the

cause concept is listed in the row and the effect concept

in the column.

Looking at Fig. 1, the Reasons Left: Job Character-

istics, Company Comparisons, and My Situation is

Different concepts were the cause concepts because all

arrows involving them emanated from them. The

Actions the Company Can Take, Barriers: Promotion,

Result of Own Decision, Lack of Control, and Turnover

concepts were effect concepts because all arrows

involving them pointed to them. Also advancement

barriers were not a strong reason for leaving as shown

by the lack of a connection between Barriers:

Promotion and Turnover.

The concepts with the highest centrality were

Barriers: Promotion at 1.015, Turnover at 0.940,

Managing Family Responsibilities and Job Qualities

both at 0.789, and Work Stress and Work Schedule

Flexibility both at 0.752. The higher centrality numbers

suggest that in this context these concepts played a

more prominent role in workplace issues than, for

example, My Situation is Different (0.301) (see

Table 4).

4.2. Loops

After the maps have been constructed patterns of

linkages emerge. In the systems dynamics literature, a

loop is defined as a closed path of linkages between

concepts that can begin and end with a concept [45,49].

The concepts within a loop directly influence each other

and the direction of the relationship is indicated. A

D.J. Armstrong et al. / Information & Management 44 (2007) 142–153 147

Fig. 1. Concept Level Causal Map. Note: Reachability values appear on the linkages as close to the arrowhead as possible.

relationship between two concepts is said to be positive

if an increase in the cause concept is accompanied by an

increase in the effect concept. Conversely, a relationship

is said to be negative if an increase in the cause is

accompanied by a decrease in the effect. In our study,

Table 3

Reachability matrix for aggregated concept level causal map

Construct A B C D

A. Actions the company can take – – – –

B. Barriers: promotion – – – –

C. Company comparisons – – – –

D. Job qualities 0.05 0.07 – –

E. Lack of control – – – –

F. Managing family responsibilities 0.06 0.09 – 0.05

G. My situation is different 0.06 0.09 – 0.05

H. Reasons left: job characteristics – 0.05 – –

I. Result of own decision – – – –

J. Turnover – – – –

K. Work schedule flexibility – 0.06 – –

L. Work stress 0.05 0.08 – –

Note: 0.05 was used as the reachability cutoff.

the causal statements were directionally coded (posi-

tive, negative, or neutral) independently by all four

authors. Comparisons were made for agreement and

disagreement, and discrepancies were resolved through

discussion.

E F G H I J K L

– – – – – – – –

– – – – – – – –

– – – – 0.05 0.05 0.05 –

– – – – 0.07 0.07 0.07 0.05

– – – – – – – –

0.05 – – – – 0.10 0.10 0.07

– 0.05 – – 0.10 0.10 0.10 0.06

– – – – 0.06 0.05 0.05 –

– – – – – – – –

– – – – – – – –

– – – – – 0.07 – 0.05

– 0.05 – – 0.08 0.08 0.08 –

D.J. Armstrong et al. / Information & Management 44 (2007) 142–153148

Table 4

Centrality

Concept Centrality

Barriers: promotion 1.015

Turnover 0.940

Managing family responsibilities 0.789

Work stress 0.789

Work schedule flexibility 0.752

Job qualities 0.752

Lack of control 0.677

Company comparisons 0.564

Actions the company can take 0.526

Result of own decision 0.526

Reasons left: job characteristics 0.451

My situation is different 0.301

Loops with an even number of negative relationships

or no negative relationships are known as positive,

deviation amplifying, reinforcing, or ‘vicious cycles’

[75]; an increase [decrease] in the value of one variable

will lead to increases [decreases] in the other variables.

A system that contains a vicious cycle will continue

until it is destroyed or some dramatic change occurs to

break the cycle. In contrast, loops with an odd number

Fig. 2. Work–family conflic

of negative relationships are known as negative,

deviation dampening, deviation counteracting, or

balancing cycles [48].

When analyzing the maps it was discovered that three

loops resulted from the participants’ responses, all with

positive components, resulting in deviation amplifying

loops, or vicious cycles. These three loops together

created a system. Each element in the loops was strongly

linked to voluntary turnover (and promotional barriers).

The loop system is represented in Fig. 2.

The first loop occurred between Managing Family

Responsibilities, Work Schedule Flexibility, and Work

Stress (see Fig. 3A). The Managing Family Responsi-

bilities concept contained statements addressing chil-

dren, family, being a mother and sharing responsibilities

with a spouse (although this was not necessarily equal).

The Managing Family Responsibilities concept was so

labeled because the participants focused on how they

managed their family responsibilities (not on the

pressure of balancing the roles). A work–family conflict

concept did not appear on the map because the number

of statements that addressed work interfering with

family and family interfering with work did not reach

the threshold needed to appear on the map.

t system causal map.

D.J. Armstrong et al. / Information & Management 44 (2007) 142–153 149

Fig. 3. (A) Loop 1. (B) Loop 2. (C) Loop 3.

The Work Schedule Flexibility concept contained

two themes: reduced work hours (i.e., part time work)

and flexibility in time at work. Within this concept

many of the respondents focused on flextime. The

women did not want to work less hours, but to work

when needed and also deal with family concerns as

needed. The Work Stress concept contained statements

about multiple responsibilities at work, competition,

working long hours, deadlines, and the amount of

pressure felt on the job. The second loop occurred

between Managing Family Responsibilities, Job

Qualities, and Work Stress (see Fig. 3B). The Job

Qualities concept contained statements that addressed

learning, using one’s intelligence, challenge, putting

forth effort, and self-reliance. The third loop occurred

between Managing Family Responsibilities, Job

Qualities, Work Schedule Flexibility, and Work Stress

(see Fig. 3C).

5. Discussion

Women in IT perceive the interaction of work and

family as an interconnected system of loops constituted

by the concepts of family (Managing Family Respon-

sibilities), work (Job Qualities), stress (Work Stress),

and time (Work Schedule Flexibility). The women

appear to perceive the interaction between work and

family as directly and indirectly impacting both

advancement opportunities and voluntary turnover.

In the first loop we saw that strong direct and indirect

linkages emerged between Managing Family Respon-

sibilities, Work Stress, and Work Schedule Flexibility.

Managing Family Responsibilities deals with issues in

the women’s home life (e.g., sick children, adequate

daycare, sharing responsibilities with a spouse) while

Work Stress primarily dealt with issues in the workplace

(e.g., meeting deadlines, working overtime). Work

Schedule Flexibility was the bridge between the work

and family roles and involved issues such as being able

to arrange work schedules and taking time off work for

personal business. Flexible work schedules have been

associated with higher organizational commitment and

job satisfaction for women [64], and consistent with our

findings lower stress [3], and lower work–family

conflict [33].

Although a flexible schedule provides many benefits,

there could also be drawbacks in terms of promotion

opportunities. Researchers have found that participants

perceived female employees with flexible or reduced

work schedules as having less job-career dedication,

less advancement motivation, and higher likelihood of

turnover [18]. MacDermid et al. [47] found that the

women managers interviewed felt they had lost some

promotion opportunities because of using a reduced

work schedule. This negative repercussion of using a

flexible work schedule was consistent with the findings

of our study.

One avenue that may ease the stress of work–family

conflict for women (flextime) may not really be an

option if the woman is looking to advance within the

organization. As one woman stated,

D.J. Armstrong et al. / Information & Management 44 (2007) 142–153150

In IT everybody tends to work long hours at times,

weekends, work late, but there is no policy that

allows us to take time off to make up for any of that

. . . so you almost feel should I ask for it?

These work practices are a part of the IToccupational

culture that may be particularly demanding and present

work related stress for women. The tension created by

the need for a flexible work schedule can adversely

impact a woman’s promotional opportunities and lead

to voluntary turnover.

As the family responsibilities women face increase,

so might the importance of the IT job qualities, as can

be seen in Loops 2 and 3. Job Qualities contained

statements that addressed learning, using intelligence,

putting forth effort and time, and self-reliance. The IT

field is project-oriented with rapid technology changes

that make skills quickly obsolete, and require IT

workers to frequently re-skill. A willingness to acquire

diverse skills is an important success factor for a

woman working in IT; e.g., the linkage we found

between Job Qualities and Barriers: Promotion. Many

of the Job Qualities comments dealt with actions to

improve skill sets to further a career. Women face the

fear of their skills becoming obsolete, causing work

related stress, which interacts with the challenges of

managing family responsibilities. The importance or

attractiveness of IT job qualities also places an

increased importance on the need for flexibility in a

woman’s schedule. From the maps found in our

study, there is a vicious cycle in which work stress

plays a major role in the work–family conflict for

women in IT.

5.1. Limitations

The comments of the women in our study were made

about work in one organizational context and may not

apply to others or to all women. Our goal was to gather

contextualized knowledge from women in the IT

department and construct a domain specific mental

model. Another limitation is the female-only respon-

dent pool, although we believed that having only

women participants was not a weakness. However, we

do not know that the issues are unique to women. One

final limitation concerns the causal mapping method

employed here. The coder may impute his or her own

assumptions into the coding. As with any such project a

large number of coding choices were made, and these

could alter the analysis and results [14]. To reduce this

potential bias, we used multiple raters (across dis-

ciplines) at every stage of the process. We believe that

the method was applicable for this study and appro-

priately applied.

6. Implications and conclusion

Our study demonstrated the value of causal

mapping in exploring cognition and provided some

new insights. One insight was provided by identifying

the vicious cycles women face when dealing with the

tensions between family responsibilities and the

unique job qualities required in an IT position. One

of the unique job qualities is the need for continual

learning (IT Job Qualities), which while often desired

by IT professionals, can place additional stress on the

individual. As seen in this study, this stress may

heighten work–family conflicts experienced by

women in IT. Organizations may want to look at

their current training and educational programs within

the context of re-tooling IT workers. Are women at a

disadvantage due to work schedules or extended

leave? An emphasis on flexible training options (e.g.,

self-paced training, online training, and paired

learning) may meet the women’s re-tooling needs,

improve advancement opportunities, and ultimately

decrease voluntary turnover intentions.

Contrary to management literature which has found

that work-role characteristics exerted a stronger influence

on work–family conflict than pressures originating from

the family-role (e.g. [6,9,17]), in our study the women’s

causal statements indicated that family responsibilities

did not cause stress; only work responsibilities did. One

explanation for our finding is that for the women in this

study if family responsibilities interfere with work roles,

it is somewhat expected and manageable; whereas if

work roles interfere with family responsibilities it is more

problematic and stressful. A flexible work schedule was

one option available to the women in this study to

attenuate this stress. An interesting aspect mentioned by

the women was the lack of consistency across the focal

organization. Senior executives may want to standardize

policies across the organization on various forms of

flexible work schedules and publicize these policies to

combat potential equity concerns. Even with an official

policy regarding flexible work schedules, women in IT

recognize that they must make difficult choices. Work

practices, such as long hours, late nights, and on-call duty

are one part of the IT occupational culture [50] that may

present work related stress for women; especially if a

woman has family responsibilities. Organizations might

mitigate voluntary turnover intentions and increase

diversity by addressing female IT professionals’ con-

cerns regarding work–family conflict issues.

D.J. Armstrong et al. / Information & Management 44 (2007) 142–153 151

Acknowledgement

The authors would like to thank Manju Ahuja, Kay

M. Nelson, and the participants of the 2004 SIGIS

Cognitive Research Workshop for their helpful com-

ments on earlier versions of this paper. The authors

would also like to acknowledge the Information

Technology Research Institute at the University of

Arkansas for assistance in the funding of this project.

Appendix A

Focus group questions:

1. T

hink back to why you left other IT positions in the

past and what your colleagues told you when they

decided to leave an organization. What was going on

in the organization you were working in that may

have caused either you or your friends to decide to

leave the organization?

2. D

o you think women in the IT workplace face

different or more barriers than men? Please explain.

3. W

hat challenges do you think women in IT face that

their male peers do not? Please explain.

4. I

f you were in charge, what kind of changes would

you made in your organization to better retain and

promote women?

5. W

hat experiences have you had or what have you

heard from others that organizations are doing that

you think are especially effective in retaining female

IT employees?

References

[1] M.K. Ahuja, Women in the information technology profession: a

literature review, synthesis and research agenda, European Jour-

nal of Information Systems 11 (1), 2002, pp. 20–34.

[2] M.T. Allison, D.K. Hibbler, Organizational barriers to inclusion:

perspectives from the recreation professional, Leisure Sciences

26, 2004, pp. 261–280.

[3] E.D. Almer, S.E.Kaplan,Theeffectsofflexibleworkarrangements

on stressors, burnout, and behavioral outcomes in public account-

ing, Behavioral Research in Accounting 14, 2002, pp. 1–34.

[4] D.J. Armstrong, Causal mapping: a discussion and demonstra-

tion, in: V.K. Narayanan, D.J. Armstrong (Eds.), Causal Map-

ping for Research in Information Technology, Idea Group

Publishing, Hershey, PA, 2005, pp. 20–45.

[5] D. Arnold, F. Niederman, The global IT work force, Commu-

nications of the ACM 44 (7), 2001, pp. 30–34.

[6] S. Aryee, V. Luk, A. Leung, S. Lo, Role stressors, inter-role

conflict, and well being: the moderating influence of spousal

support and coping behaviors among employed parents in Hong

Kong, Journal of Vocational Behavior 54, 1999, pp. 259–278.

[7] R. Axelrod, Structure of Decision: The Cognitive Maps of

Political Elites, Princeton University Press, Princeton, NJ, 1976.

[8] S.B. Bacharach, P. Bamberger, S. Conley, Work–home conflict

among nurses and engineers: mediating the impact of role stress

and burnout and satisfaction at work, Journal of Organizational

Behavior 12 (1), 1991, pp. 39–53.

[9] A. Bakker, S.A.E. Geurts, Toward a dual-process model of

work–home interference, Work and Occupations 31 (3), 2004,

pp. 345–366.

[10] J.J. Baroudi, M. Igbaria, An examination of gender effects on

career success of information systems employees, Journal of

Management Information Systems 11 (3), 1994–1995, pp. 181–

201.

[11] R.E. Beasley, E. Lomo-David, V.R. Seubert, Telework and

gender: implications for the management of information tech-

nology professionals, Industrial Management & Data Systems

101 (8/9), 2001, pp. 477–482.

[12] M. Bougon, K. Weick, D. Binkhorst, Cognition in organizations:

an analysis of the Utrecht jazz orchestra, Administrative Science

Quarterly 22, 1977, pp. 606–639.

[13] P. Carayon, M.C. Haims, T. Kraemer, Turnover and retention of the

information technology workforce: the diversity issue, in: M.J.

Smith, G. Salvendy (Eds.), Systems, Social and Internationaliza-

tion Design Aspects of Human–Computer Interaction, Lawrence

Erlbaum Associates, Inc., Mahwah, NJ, 2001, pp. 67–70.

[14] K. Carley, Extracting team mental models through textual analy-

sis, Journal of Organizational Behavior 18, 1997, pp. 533–559.

[15] C.G. Carter, D.G. Pounder, F.G. Lawrence, P.J. Wozniak, Factors

related toorganizational turnover intentionsofLouisianaextension

service agents, in:H.L.Meadow, M.J. Sirgy (Eds.), Quality-of-Life

Studies in Marketing and Management, International Society for

Quality-of-Life Studies, Blacksburg, VA, 1990, pp. 170–181.

[16] J.A. Chatman, Improving interactional organizational research: a

model of person–organization fit, Academy of Management

Review 14 (1), 1989, pp. 333–350.

[17] R.G. Cinamon, Y. Rich, Profiles of attribution of importance to

life roles and their implications for the work–family conflict,

Journal of Counseling Psychology 49 (2), 2002, pp. 212–220.

[18] J.R. Cohen,L.E. Single,An examinationof theperceived impact of

flexible work arrangements on professional opportunities in public

accounting, Journal of Business Ethics 32, 2001, pp. 317–328.

[19] J.D. Couger, R.A. Zawacki, What motivates DP professionals?

Datamation 24 (9), 1978, pp. 116–123.

[20] S. Curtis, W. Gesler, G. Smith, S. Washburn, Approaches to

sampling and case selection in qualitative research: examples in

the geography of health, Social Science & Medicine 50, 2000,

pp. 1001–1014.

[21] L.E. Duxbury, C.A. Higgins, Gender differences in work–family

conflict, Journal of Applied Psychology 76 (1), 1991, pp. 60–74.

[22] C. Eden, S. Jones, D. Sims, T. Smithin, The intersubjectivity of

issues and issues of intersubjectivity, Journal of Management

Studies 18 (1), 1981, pp. 37–47.

[23] J.D. Ford, W.H. Hegarty, Decision maker’s beliefs about the

causes and effects of structure: an exploratory study, Academy of

Management Journal 27 (2), 1984, pp. 271–291.

[24] M.R. Frone, M. Russell, M.L. Cooper, Antecedents and out-

comes of work–family conflict: testing a model of the work–

family interface, Journal of Applied Psychology 77 (1), 1992, pp.

65–78.

[25] D.C. Ganster, J. Schaubroeck, Work stress and employee health,

Journal of Management 17 (2), 1991, pp. 235–271.

[26] D. Gefen, D.W. Straub, Gender differences in the perception and

use of e-mail: an extension to the technology acceptance model,

MIS Quarterly 21 (4), 1997, pp. 389–401.

D.J. Armstrong et al. / Information & Management 44 (2007) 142–153152

[27] B.G. Glaser, A.L. Strauss, The Discovery of Grounded Theory:

Strategies for Qualitative Research, Aldine Publishing Com-

pany, Chicago, IL, 1967.

[28] E. Goss, U. Gupta, Women and the internet: is there an economic

payoff? Communications of the ACM 46 (9), 2003, pp. 160–166.

[29] R.W. Griffeth, P.W. Hom, Retaining Valued Employees, Sage

Publications, Thousand Oaks, CA, 2001.

[30] R.W. Griffeth, P.W. Hom, S. Gaertner, A meta-analysis of

antecedents and correlates of employee turnover: update, mod-

erator tests, and research, implications for the next millennium,

Journal of Management 26 (3), 2000, pp. 463–488.

[31] E. Halliwell, H. Dittmar, A qualitative investigation of women’s

and men’s body image concerns and their attitudes toward aging,

Sex Roles 49, 2003, pp. 675–684.

[32] K. Hemenway, Human nature and the glass ceiling in industry,

Communications of the ACM 38 (1), 1995, pp. 55–62.

[33] E.J. Hill, A.J. Hawkins, M. Ferris, M. Weitzman, Finding an

extra day a week: the positive influence of perceived job flex-

ibility on work and family life balance, Family Relations 50 (1),

2001, pp. 49–58.

[34] R. Hodson, Working in ‘‘high-tech’’: research issues and oppor-

tunities for the industrial sociologist, Sociological Quarterly 26

(3), 1985, pp. 351–364.

[35] O.R. Holsti, Content Analysis for the Social Sciences and

Humanities, Addison-Wesley, Reading, MA, 1969.

[36] Mapping the process of problem reformulation: implications for

understanding strategic thought,A.S. Huff (Ed.), Mapping Stra-

tegic Thought, John Wiley, New York, 1990.

[37] M. Igbaria, Career orientations of MIS employees: an empirical

analysis, MIS Quarterly 15 (2), 1991, pp. 151–169.

[38] M. Igbaria, J.J. Baroudi, The impact of job performance evalua-

tions on career advancement prospects: an examination of

gender differences in the IS workplace, MIS Quarterly 19 (1),

1995, pp. 107–123.

[39] J.A. Jacobs, K. Gerson, Overworked individuals or overworked

families? Work and Occupations 28 (1), 2001, pp. 40–63.

[40] D. Knoke, J.H. Kuklinski, Network Analysis, Sage Publications,

Newbury Park, CA, 1982.

[41] E.E. Kossek, C. Ozeki, Bridging the work–family policy and

productivity gap: a literature review, Community, Work &

Family 2 (1), 1999, pp. 7–32.

[42] M.A. Lemons, M.J. Parzinger, Designing women: a qualitative

study of the glass ceiling for women in technology, S.A.M.

Advanced Management Journal 66 (2), 2001, pp. 1–10.

[43] P.A. Lewellyn, E.A. Wibker, Significance of quality of life on

turnover intentions of certified public accountants, in: H.L.

Meadow, M.J. Sirgy (Eds.), Quality of Life Studies in Marketing

and Management, International Society for Quality of Life

Studies, Blacksburg, VA, 1990, pp. 182–193.

[44] S. Liff, K. Ward, Distorted views through the glass ceiling: the

construction of women’s understandings of promotion and senior

management positions, Gender, Work & Organization 8 (1),

2001, pp. 19–37.

[45] D.H. Lindsley, D.J. Brass, J.B. Thomas, Efficacy-performance

spirals: a multilevel perspective, Academy of Management

Review 20 (3), 1995, pp. 645–678.

[46] K.A. Loscocco, A.R. Roschelle, Influences on the quality of

work and nonwork life: to decades in review, Journal of Voca-

tional Behavior 39, 1991, pp. 182–225.

[47] S.M. MacDermid, M.D. Lee, M. Buck, M.L. Williams, Alter-

native work arrangements among professionals and managers,

Journal of Management Development 20, 2001, pp. 305–317.

[48] M. Maruyama, The second cybernetics: deviation amplifying

mutual causal processes, American Scientist 51, 1963, pp. 164–

179.

[49] M. Masuch, Vicious circles in organizations, Administrative

Science Quarterly 30 (1), 1985, pp. 14–34.

[50] D.E. Meyerson, J.K. Fletcher, A modest manifesto for shattering

the glass ceiling, Harvard Business Review 78 (1), 2000, pp.

126–136.

[51] M. Miles, M. Huberman, Qualitative Data Analysis: An Expanded

Sourcebook, Sage Publications, Newbury Park, CA, 1994.

[52] S. Nadkarni, F.F.H. Nah, Aggregated causal maps: an approach

to elicit and aggregate the knowledge of multiple experts,

Communications of the AIS 12, 2003, pp. 406–436.

[53] V.K. Narayanan, L. Fahey, Evolution of revealed causal maps

during decline: a case study of admiral, in: A. Huff (Ed.),

Mapping Strategic Thought, John Wiley and Sons, London,

1990, pp. 109–133.

[54] D.L. Nelson, J.C. Quick, M.A. Hitt, D. Moesel, Politics, lack of

career progress, and work/home conflict: stress and strain for

working women, Sex Roles 23 (3/4), 1990, pp. 169–185.

[55] K.M. Nelson, S. Nadkarni, V.K. Narayanan, M. Ghods, Under-

standing software operations support expertise: a causal map-

ping approach, MIS Quarterly 24 (3), 2000, pp. 475–507.

[56] V. Oechtering, R. Behnke, Situations and advancement measures

in Germany, Communications of the ACM 38 (1), 1995, pp. 75–83.

[57] W. Orlikowski, J.J. Baroudi, The information systems profes-

sion: myth or reality? Information Technology and People 4 (1),

1989, pp. 13–31.

[58] R. Pringle, S. Nielsen, L. von Hellens, A. Greenhill, L. Parfitt,

Net gains, in: Proceedings of the 7th International Conference on

Women, Work and Computerization: Charting A Course to the

Future, Vancouver, BC, June 8–11, Kluwer Academic Publish-

ers, Boston, MA, 2000.

[59] B.R. Ragins, Power and gender congruency effects in evalua-

tions of male and female managers, Journal of Management 15

(1), 1989, pp. 65–77.

[60] C.K. Riemenschneider, D.J. Armstrong, M.W. Allen, M. Reid,

Barriers facing women in the IT work force, The DATA BASE

for Advances in Information Systems 37 (4), 2006, pp. 58–78.

[61] M. Robertson, S. Newell, J. Swan, L. Mathiassen, G. Bjerknes,

The issue of gender within computing: reflections from the UK

and Scandinavia, Information Systems Journal 11, 2001, pp.

111–126.

[62] P.H. Rossi, J.D. Wright, A.B. Anderson, Handbook of Research

Methods, Academic Press Inc., Orlando, FL, 1983.

[63] D.M. Rotondo, D.S. Carlson, J.F. Kincaid, Coping with multiple

dimensions of work–family conflict, Personnel Review 32 (3),

2003, pp. 275–296.

[64] T. Scandura, M. Lankau, Relationships of gender, family respon-

sibility, and flexible work hours to organizational commitment

and job satisfaction, Journal of Organizational Behavior 18,

1997, pp. 377–391.

[65] W.A. Scott, Reliability of content analysis: the case of nominal

scale coding, Public Opinion Quarterly 19, 1955, pp. 321–325.

[66] M.J. Sirgy, D. Efraty, P. Siegel, D.J. Lee, A new measure of quality

of work life (qwl) based on need satisfaction and spillover theories,

Social Indicators Research 55 (3), 2001, pp. 241–259.

[67] G.L. Staines, Spillover versus compensation: a review of the

literature on the relationship between work and non-work,

Human Relations 33, 1980, pp. 111–129.

[68] S.S. Stautberg, Status report: the corporation and trends in family

issues, Human Resource Management 26 (2), 1987, pp. 277–290.

D.J. Armstrong et al. / Information & Management 44 (2007) 142–153 153

[69] M. Tan, M. Igbaria, Turnover and remuneration of information

technology professionals in Singapore, Information & Manage-

ment 26 (4), 1994, pp. 219–229.

[70] E.M. Trauth, Odd girl out: an individual differences perspective

on women in the IT profession, Information Technology and

People 15 (2), 2002, pp. 98–118.

[71] G.E. Truman, J.J. Baroudi, Gender differences in the information

systems managerial ranks: an assessment of potential discrimi-

natory practices, MIS Quarterly 18 (2), 1994, pp. 129–141.

[72] M. Vehvilainen, Gender and computing in retrospect: the case of

Finland, IEEE Annals of the History of Computing 21 (2), 1999,

pp. 44–51.

[73] V. Venkatesh, M.G. Morris, P.L. Ackerman, A longitudinal field

investigation of gender differences in individual technology

adoption decision-making processes, Organizational Behavior

and Human Decision Processes 83 (1), 2000, pp. 33–60.

[74] L. von Hellens, S. Nielsen, Australian women in IT, Commu-

nications of the ACM 44 (7), 2001, pp. 46–52.

[75] K.E. Weick, Causal loops and control, The Social Psychology of

Organizing, 2nd ed., Addison-Wesley, New York, 1979.

[76] B.E. Whitney, Principles of Research in Behavioral Science,

Mayfield, Mountain View, 1996.

[77] R. Woodfield, Women and information systems development:

not just a pretty (inter) face, Information Technology & People

15 (2), 2002, pp. 118–138.

[78] J.L. Wynekoop, D.B. Walz, Revisiting the perennial question:

are IS people different? The DATA BASE for Advances in

Information Systems 29 (2), 1998, pp. 62–72.

Deborah J. Armstrong is an assistant

professor of Information Systems in the

Sam M. Walton College of Business at

the University of Arkansas. Her research

includes issues at the intersection of IS

personnel and mental models involving

the human aspects of technology, change,

learning and cognition. Her articles have

appeared in journals such as the Journal of

Management Information Systems, Com-

munications of the ACM, and Sex Roles among others, and she has

co-edited a book Causal Mapping for Research in Information

Technology.

Cynthia K. Riemenschneider is Associate

Professor of Information Systems in the Sam

M. Walton College of Business at the Uni-

versity of Arkansas. Her publications have

appeared in Information Systems Research,

IEEE Transactions on Software Engineer-

ing, Journal of Management Information

Systems, Sex Roles, and others. She currently

conducts research on IT personnel issues, IT

adoption, and women in IT.

Myria Allen (Ph.D. Communication, Uni-

versity of Kentucky, 1988) is an Associate

Professor in the Department of Communi-

cation at the University of Arkansas, Fayet-

teville. Her research interests include

organizational communication and issues

related to gender and technology. Her

research has been published in the Journal

of Applied Communication Research, Man-

agement Communication Quarterly, Com-

munication Monographs, Sex Roles, and Communication Yearbook

16.

Margaret Reid is Professor of Political

Science at the University of Arkansas. Her

current research focuses on gendered work-

places and complexities involving public-

nonprofit alliances. She recently published

a book with Kerr and Miller, Glass Walls and

Glass Ceilings. Her work has been published

in Public Administration Review, Women &

Politics, Urban Affairs Review, Sex Roles

and numerous other outlets.