See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/263055015

The Work–Family Conflict Scale (WAFCS): Development and Initial Validation

of a Self-report Measure of Work–Family Conflict for Use with Parents

Article  in  Child Psychiatry and Human Development · June 2014

DOI: 10.1007/s10578-014-0476-0 · Source: PubMed

CITATIONS READS

6 8,805

5 authors, including:

Divna Haslam Ania Filus

The University of Queensland University of Southern California
22 PUBLICATIONS   289 CITATIONS    60 PUBLICATIONS   168 CITATIONS   

SEE PROFILE SEE PROFILE

Alina Morawska Matthew R. Sanders

The University of Queensland The University of Queensland
112 PUBLICATIONS   1,627 CITATIONS    436 PUBLICATIONS   15,286 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Giving Voice to Working Mothers: A Consumer Informed Study to Program Design for Working Mothers View project

Bridging the gap between parents and parenting interventions: A model of initial parental engagement View project

All content following this page was uploaded by Matthew R. Sanders on 04 September 2014.

The user has requested enhancement of the downloaded file.

Child Psychiatry Hum Dev
DOI 10.1007/s10578-014-0476-0
ORIGINAL ARTICLE
The Work–Family Conflict Scale (WAFCS): Development
and Initial Validation of a Self-report Measure of Work–Family
Conflict for Use with Parents
Divna Haslam • Ania Filus • Alina Morawska •
Matthew R. Sanders • Renee Fletcher
Ó Springer Science+Business Media New York 2014
Abstract This paper outlines the development and vali-
dation of the Work–Family Conflict Scale (WAFCS)
designed to measure work-to-family conflict (WFC) and
family-to-work conflict (FWC) for use with parents of
young children. An expert informant and consumer feed-
back approach was utilised to develop and refine 20 items,
which were subjected to a rigorous validation process using
two separate samples of parents of 2–12 year old children
(n = 305 and n = 264). As a result of statistical analyses
several items were dropped resulting in a brief 10-item
scale comprising two subscales assessing theoretically
distinct but related constructs: FWC (five items) and WFC
(five items). Analyses revealed both subscales have good
internal consistency, construct validity as well as concur-
rent and predictive validity. The results indicate the
WAFCS is a promising brief measure for the assessment of
work–family conflict in parents. Benefits of the measure as
well as potential uses are discussed.
Keywords Work-to-family conflict
Family-to-work
conflict
Working parents
Validation
Measure
Introduction
The work–family conflict literature is a burgeoning field of
research of interest to researchers, organisations and clini-
cians. Historically work–family conflict was defined as the
extent to which functioning in one domain (e.g., work)
D. Haslam (&)
A. Filus
A. Morawska

M. R. Sanders
R. Fletcher
Parenting and Family Support Centre, School of Psychology,
The University of Queensland, Brisbane 4072, Australia
e-mail: d.haslam@uq.edu.au
negatively impacted the other [1]. Subsequent research has
established there are two directional components of work–
family conflict that have different antecedents and conse-
quences and can be conceptualised as separate but related
constructs [2–4]. These are family-to-work conflict (FWC)
(i.e. family engagement negatively impacting work) and
work-to-family conflict (WFC) (i.e. work negatively
impacting family). High levels of work–family conflict (in
either or both directions) are associated with negative out-
comes for individual employees, their families, and organi-
sations in which they work [2, 5]. These findings have led to
an increasing focus on the development of interventions, and
the identification of policy and organisational factors (e.g.,
option to telecommute) that serve to reduce employees’
experience of work–family conflict [6]. Parents are at par-
ticular risk of work–family conflict with parents of young
children reporting high levels of conflict that have the
potential to negatively impact functioning across both work
and family domains [7]. Clinical interventions for working
parents now target work–family conflict as an outcome [8, 9],
however no well-validated measures of work–family conflict
for clinical use with parents exist. This paper outlines the
development and validation of a specific measure of work–
family conflict for use with parents.
A number of measures of work–family conflict exist that
have high reliability and are commonly used in the literature
[10], however none of these have been specifically designed
from the outset for clinical use with parents to assess longi-
tudinal intervention outcomes. Intervention measures need
to be brief so they can be used in combination with longer
batteries; and benefit from having wider response ranges to
avoid anchor bias and to ensure they are sensitive enough to
detect subtle intervention changes [11]. Of the two most
utilised measures of work–family conflict one is too long for
clinical use [12] and the other [13] has a limited response
123

range of 5 points making it less likely to detect small inter-
vention changes. Empirical literature clearly indicates that
5-point (or less) scales do not provide sufficient sensitivity to
detect small, clinically significant differences [14–16].
Therefore, 7-point response formats allow for greater vari-
ability and clinical movement increasing scale sensitivity to
detect small changes. Additional limitations of measures in
the field include: (a) a lack of validation with parent specific
samples [12, 17, 18]; (b) a lack of rigorous psychometric
evaluation that recent advances in statistics have afforded
[19]; and (c) measures that assess work–family conflict as a
single domain [20, 21].
In this paper we sought to develop and validate a brief
measure of work–family conflict that can be available in the
public domain and that has robust psychometric properties
ensuring it can be adequately used in empirical research with
parents across a range of fields including parenting inter-
vention research. In doing so we overcome the limitations
with existing measures and add to the literature in the
development and evaluation of a parent specific work–
family conflict measures for clinical and research use. In
developing the tool we sought to address a number of specific
criteria that would allow it to be effective across research,
clinical and organisational settings. Specifically these crite-
ria included that the measure: be brief; be easy to administer,
score and interpret; cover both facets of work–family con-
flict; be change sensitive; and have adequate psychometric
properties including reliability, face validity, and construct,
concurrent and predictive validity. We also sought to
develop a measure that was general enough for use across a
variety of occupational types and across a range of family
environments, and one that did not require sophisticated
reading and comprehension skills.
This study aimed to assess the psychometric properties of
the Work and Family Conflict Scale (WAFCS, pronounced
‘waff-cuss’). Specifically we sought to: (1) apply principles
of measure development as outlined by Cohen et al. [22] to
create a brief self-report, user friendly, public domain mea-
sure of work–family conflict for parents that could be used in
research, clinical and organisational settings (detailed steps
have been outlined in the first paragraph of the Procedure
section); (2) determine the construct, concurrent and pre-
dictive validity of the WAFCS and; (3) determine the
internal consistency of the scale.
Methods
Participants
Two separate samples of parents were used and the detailed
demographic characteristics of both samples are presented
in Table 1. Both samples (Study 1 and 2) consisted of
123
Child Psychiatry Hum Dev
parents recruited Australia-wide from adverts in schools
and day care centres, online forums, and parenting news-
letters. The only eligibility criteria were that parents have a
child between ages 2–12 years, be able to read English and
be computer literate enough to complete a questionnaire
online. The samples had similar demographic characteris-
tics. The majority of parents were mothers (with an average
age of 39.35 years (SD = 5.76) in Study 1 and 35.47 years
(SD = 6.51) in Study 2 and of Caucasian/Australian
background. Children’s ages ranged from 2 to 12 years old
and both samples had good gender ratios of target children.
Further, for both samples a majority of parents had a uni-
versity degree, and were married. All parents surveyed
were working parents with the majority employed full or
part-time. For both groups, a range of socio-economic
backgrounds were sampled. In Study 1 and 2, 33.40 and
54.90 %, respectively, reported that they earn enough to
comfortably purchase most of the things they really want,
46.90 and 35.20 % declared that their earnings allow them
to purchase only some things that they want, while 10 and
9.50 % reported they don’t have enough money to pur-
chase much of anything they really want.
Procedure
We used the following procedure to design the measure:
(1) definition of constructs; (2) review of existing mea-
sures; (3) generation of initial item pool; (4) input and
feedback from key experts; (5) input and feedback from
parents, and; (6) initial piloting to assess psychometric
properties. We theorised that WFC and FWC are separate
but related constructs [5]. Items were included to assess
potential strain-based, time-based and behaviour based
aspects of conflict however our goal was to develop a
coherent assessment of WFC and FWC regardless of the
cause of the conflict. We reviewed the work–family con-
flict literature and existing validated measures including
those by key experts in the field. An initial item pool of 20
items was generated on the basis of this review in the
context of our focus on parenting. The initial scale was
disseminated to a number of international experts for
feedback and to ensure wording and content were cultur-
ally relevant. Four in-depth semi-structured interviews
were conducted with parents to gain feedback about ease of
understanding, the completion process and face validity.
Parents completed the questionnaire without additional
instruction and were then requested to note anything that
was unclear or ambiguous. They were then asked a series
of questions designed to elicit feedback (e.g., Is there
anything that would make the survey easier to complete? Is
there anything missing from the questionnaire that is
important to you?). Feedback from the interviews indicated
the measure was easy to understand and had high face
Child Psychiatry Hum Dev

Table 1 Participants’ demographic information

Study 1 Study 2
Mothers Fathers Total Mothers Fathers Total
N = 286 N = 14 N = 305 N = 251 N = 12 N = 264
(93.8 %) (4.90 %) (100 %) (95.1 %) (4.5 %) (100 %)

Age
Mean (SD) 39.35 (5.76) 40.80 (7.41) 39.47 (5.93) 35.47 (6.51) 36.67 (3.54) 35.51 (6.40)
Range 24–54 24–49 24–58 21–56 32–41 21–56
Child’s age
Mean (SD) 7.29 (2.80) 8.50 (2.80) 7.35 (2.78) 4.40 (2.73) 5.04 (2.73) 5.37 (3.09)
Range 2–12 3–12 2–12 2–12 2–10 2–12
Child’s gender
Girls (%) 58.40 46.70 7.70 48.60 33.30 48.10
Number of children 2.25 (.92) 2.07 (1.07) 2.05 (.97) 1.94 (.92) 1.94 (.68) 1.96 (.91)
Ethnicity
Caucasian/Australia (%) 83.60 40.00 80.30 79.70 83.30 79.90
Aboriginal & Torres Strait Islander (%) 1.70 6.70 1.60 – – –
Asian (%) 1.40 – 2.00 3.20 – 3.0
Other (%) 2.10 13.30 2.60 16.70 16.70 16.70
Education
University degree (%) 55.30 40.00 54.10 64.20 91.70 65.50
High school (%) 23.70 33.40 24.00 12.00 8.30 11.70
Trade/technical college (%) 21.00 26.70 21.00 23.90 – 22.70
Relationship status
Married (%) 75.20 53.30 73.10 70.50 75.00 70.80
Cohabitating (%) 10.50 20.00 10.80 15.50 16.70 15.50
Divorced (%) 9.80 20.00 10.20 8.00 8.30 8.00
Widow/er (%) .30 – .30 .40 – .40
Single 2.80 – 3.00 5.20 – 4.90
Work status
Working full-time (%) 33.20 46.70 33.40 40.60 91.7 42.8
Working part-time (%) 36.70 26.70 35.70 53.8 18.3 51.9
Casual work (%) 30.10 21.4 29.18 – – –
– 5.6 5.3
– – –
Finance
Enough to purchase most of the things (%) 34.60 20.00 33.40 54.20 75.00 54.90
Enough to purchase only some things (%) 46.90 53.30 46.90 35.50 25.00 35.20
Not enough to purchase most of the things (%) 18.50 26.70 18.70 10.00 – 9.50
The numbers may not add up to 100 % due to the missing data

validity however some parents recommended rewording
items slightly. Several items were modified in response to
expert and parent feedback and the order of items was
changed. This resulted in the 20-item Work–Family Con-
flict Questionnaire (WAFCS) [23] that was subjected to
further analysis.
To ensure the revised measure could be understood by
parents with low education it was assessed for readability
using the Flesch–Kincaid grade tests. This test assesses
comprehension difficulty and provides an estimate of
education grade level (grade 1–12) required for under-
standing. A Flesch–Kincaid score of 9.8 (possible range
1–12) was obtained indicating the measure could be easily
understood by a student aged 13–15 years or someone with
a grade 9 level education.
Following the measure development and consultation
process initial piloting was conducted to assess the psy-
chometric properties of the measure. As it was planned to
rigorously evaluate the factor structure of WAFCS with
both exploratory and confirmatory factor analysis (CFA),

123

two indecent samples of parents were needed [24].
Therefore, two online surveys were created simultaneously
and two large parent samples were recruited. For both
samples parents were directed to a website where they read
a brief information sheet and provided informed consent
prior to completing the questionnaire anonymously. For
child related questions parents were asked to identify a
target child based on the child who had the next birthday.
This ensured all child related questions related to the same
child. Both samples were advised they were participating in
a study designed to assess a new measure of work and
family conflict and to explore the relationship between
work–family conflict and other family related variables.
The first sample was used to conduct exploratory factor
analysis (EFA) and the second to conduct the CFA.
Measures
Demographics were assessed using the Family Background
Questionnaire [25].
The Work–Family Conflict Scale (WAFCS) [23] is a
short 10-item measure assessing WFC (five items) and
FWC (five items). Respondents are asked to rate their level
of agreement with each item on a 7-point scale form 1
(very strongly disagree) to 7 (very strongly agree). Sample
items include: ‘‘My work prevents me spending sufficient
quality time with my family’’ (Work-to-family subscale)
and ‘‘My family has a negative impact on my day to day
work duties’’ (Family-to-work subscale). For each subscale
items are summed to provide the total WFC score (range
7–35) and FWC score (range 7–35) with higher scores
indicating higher levels of conflict. The final measure is
outlined in ‘‘Appendix’’.
Frone’s WAFCS [13] is a 12-item self-report measure
that assesses the extent to which an individual experiences
WFC and FWC. Items consist of common scenarios or
statements about work and home life (e.g., ‘‘After work I am
too tired to do some of the things I’d like to at home’’).
Respondents rate how often each statement applies to them
on a scale from 1 (never) to 5 (very often). Item scores are
averaged to provide a score for levels of FWC and WFC.
Research shows that the scale has good internal consistency
(a [.80) and factor validity [4]. For the current study
internal consistency was adequate for both the FWC
(a = .84) and WFC (a = .80) subscales.
The Parenting Scale [26] is a 30-item scale that assesses
three dysfunctional discipline styles (over-reactivity, lax-
ness and verbosity) and provides a total score of dysfunc-
tional parenting. Parents are asked to envisage how they
would act in response to a certain situation. Each item uses
a 7-point scale with two anchors where one represents an
effective parenting practice and the other a less effective
approach. For example, on the item ‘‘when my child
123
Child Psychiatry Hum Dev
misbehaves’’ ratings range from 1 (I do something right
away) to 7 (I do something about it later). Research has
shown this scale has good internal consistency (with alphas
[.70), is change sensitive and can differentiate between
clinical and non-clinical populations [27]. High internal
consistency was obtained for the current study (a = .72).
The Eyberg Child Behaviour Inventory [28] is a parent
report measure consisting of 36 items that assess the fre-
quency and severity of behaviour problems in children ages
2–16 (Intensity subscale) as well as the extent to which
parent finds the behaviours troublesome (Problem sub-
scale). Only the Intensity scale was used for the current
validation as this was considered a more sensitive measure
of child behaviour problems than the dichotomous problem
scale. Parents rate the frequency of different problem
behaviours on a 7-point scale, from 1 (never) to 7 (always)
yielding an Intensity score. The subscale has high internal
consistency (a = .92) and good test–retest reliability
(r = .86) and is change sensitive [29]. The internal con-
sistency for the current study was high for both the Prob-
lem (a = .92) and Intensity (a = .94) subscales.
The Parenting Tasks Checklist [30] is a 28-item ques-
tionnaire that assesses parental levels of self-efficacy by
asking parents to rate their confidence in dealing with dif-
ferent types of behaviour problems in children (Behaviour
subscale) and across different community settings (Setting
subscale) on a 100-point scale from O (very Certain I could
not do it) to 100 (very certain I could do it). Scores are
summed to provide a Total score where higher scores indi-
cate higher levels of parental self-efficacy. The scale has high
internal consistency (a [ .90). For the purpose of this study
the Total score was used. High internal consistency was
found for the current study (a = .85).
Analytical Procedure
Construct Validity
The factor structure of WAFCS was tested with EFA fol-
lowed by CFA in Mplus version 7.1 [31]. First we con-
ducted EFA using the first data set (first study) to identify
the factor structure for the scale and reduce the number of
items. Although we hypothesized the measure should
reveal a two-factor structure, we decided to run the EFA
first due to a lack of previous exploration of the measure.
Furthermore, our goal was to reduce the number of items
into a shorter format that can be easily used in larger
research studies. The GEOMIN (oblique) factor rotation
was used to increase interpretably of the factors while also
allowing the factors to correlate. We determined the
number of factors using the scree-plot, the Kaiser’s criteria
of eigenvalues [1.0, comparison of fit indices as well as
the conceptual meaning behind the factors. Items with a
Child Psychiatry Hum Dev
factor loading [.45 were considered as a part of a factor
[32]. In the next step we performed CFA using the second
data set (second study) to confirm the factor structure
revealed through EFA.
For all models tested in this paper, we employed robust
maximum likelihood estimator MLR, which produces
standard errors and fit indices that are robust in relation to
non-normality of observations [33]. Model fit was mea-
sured by using the scaled Chi square (v2) statistic, the
comparative fit index (CFI), the root mean square error of
approximation (RMSEA) with 90 % confidence interval,
and the standardized root mean square residual (SRMR).1
For the model to be considered to have acceptable fit:
RMSEA and SRMR should be \.08 with CFI [.90 [34,
35]. Models were respecified based on Modification Indi-
ces (MIs), inspection of standardized residuals and theo-
retical considerations [36]. To assess the extent to which
newly specified model exhibits an improvement over its
predecessor, we used the v2 difference test (Dv2) for nested
models, and Akaike information criterion (AIC) and
Bayesian information criterion (BIC) values to compare
non-nested models. The Chi square difference test was
calculated using the scaled Chi square and formulas
developed by Satorra and Bentler [33]. Smaller values of
AIC and BIC indicate better fit of the model [37, 38].
The assessment of construct validity included investi-
gation of the convergent and discriminant validity [39].
Three approaches were applied to assess convergent
validity: (1) we evaluated the statistical significance and
magnitude of factor loadings for each latent construct [40];
(2) checked that the estimate of the AVE estimates (aver-
age variance extracted) representing the average amount of
variation that a latent construct is able to explain in the
observed variables are above .50 [41]; (3) tested that esti-
mates of composite reliability (CR) representing the overall
reliability of a collection of heterogeneous yet similar items
were above .70 [41].
We also employed three common techniques to assess
discriminant validity: (1) we examined that the correlation
between the two latent constructs is not close or equal to
the value of 1.00; (2) applied the v2 difference test [42];2
(3) assessed if the average variance extracted estimates
(AVE) for each construct are higher than to the shared
variance between the constructs (the squared interconstruct
correlation estimate SIC; Fornell and Larcker [41]).
1
The CFI, RMSEA and SRMR fit indices were all affected by the
Satorra–Bentler scaling correction for the Chi square statistic.
2
For the chi square difference test a constrained model, in which the
correlation between the factors is fixed at 1.00 is compared to the
original model’s v2 where the correlation between the constructs is
estimated freely. Significantly lower chi square value of the
unconstrained model implies good discriminant validity.
Work-to- Family-to-
family conflict work conflict
Parental Coercive
Confidence Parenting
Child Behaviour
Fig. 1 Hypothesized model of associations between family-to-work,
work-to-family conflict, parental confidence, coercive parenting and
child behaviour problems
Concurrent Validity
To assess the concurrent validity of WAFCS we evaluated
the relationships between WAFCS and Frone’s subscales
[13]. Frone’s is a well-established measure of work–family
conflict. Pearson product-moment correlations were cal-
culated in SPSS v.20 to assess associations between
WAFCS and Frone’s subscales.
Predictive Validity
We also assessed the predictive validity of WAFCS by
examining associations between WFC and FWC subscales
of WAFCS and parental confidence, coercive parenting and
children’s behaviour problems constructs as measured by
PTC, PS, and ECBI Intensity, respectively. Structural
equation modelling in Mplus version 7.1 was performed to
evaluate the theorised associations between the constructs
(schematic model of associations is depicted in Fig. 1).
Based on the literature we predicted that higher levels of
conflict in either direction would be associated with poorer
parental confidence and more coercive parenting style (two
consecutive mediators), which, in turn, will negatively
impact child behaviour problems [2, 9, 43, 44]. To test the
hypothesized mediation effects thebootstrap method with
5,000 bootstrap samples was used [45].
Reliability
Due to the limitations associated with Cronbach’s alpha
coefficient when the assumptions of tau-equivalence and
uncorrelated errors are violated [46, 47] we assessed the
internal consistency of the measures by calculating the H
coefficient [48]. The range and interpretations of H
123

coefficient values is exactly the same as for the popular
Cronbach’s alpha coefficient. The H coefficient ranges
from 0 to 1 with the values above .70 are considered good
indicators of internal consistency [48].
Results
Data Screening
As far as the first data set is concerned (Study 1), three
hundred and seventy parents responded to the survey, but
of these 59 did not reply to any of the WAFCS items and
were excluded from further analysis. This gave a total
sample of N = 311 with 3.35 % of missing data. Little’s
MCAR test indicated that the data were missing completely
at random [v2(290) = 187.91, p = 1.00]. For the second
data set (Study 2), two hundred eighty-five people
responded to the survey, of these 17 provided only some
demographic data and did not respond to any of the
questionnaires; thus they were excluded from further
analyses. This gave a total sample of N = 268 with 7.39 %
missing values that were missing completely at random
[Little’s MCAR test: v2(55) = 64.73, p = .17]. Full
Information Maximum Likelihood (FIML) procedure was
used to handle missing data, which has been shown to
outperform traditional approaches for handling missing
data [49].
Raw data was examined for departures from both uni-
variate and multivariate normality, and for the presence of
potential outliers. For both data sets, most WAFCS items
showed significant univariate and multivariate skew and
kurtosis.3 The univariate outliers were detected and as a
result 24 (0.39 %) extreme data points in the first data set
and 4 (0.07 %) extreme data points in the second data set
were transformed by changing the value to the next high-
est/lowest (non-outlier) number. In addition, using squared
Mahalanobis distances (D2) six multivariate outliers were
detected in the first data set and four in the second. They
were removed from further analyses giving a sample size
of N = 305 for the first study and of N = 264 for the
second study.
Exploratory Factor Analysis
Examination of eigenvalues suggested a possible three-
factor solution (eigenvalues [1), whereas examination of
the scree-plot suggested a two-factor solution. To
3
For more information please contact the corresponding author.
123
Child Psychiatry Hum Dev
determine the optimal number of factors we examined a
series of alternative models up to three factors and selected
the most parsimonious model with acceptable fit that pro-
vided a conceptually and theoretically coherent and
meaningful factor solution. None of the models showed
excellent (see Table 2). The three-factor solution showed
better fit to the data than the one and two-factor solutions,
however, the third factor consisted of only four items, out
of which two did not load significantly on this or any other
factor. In addition, our evaluation of the item content
suggested that only two out of three factors were mean-
ingfully interpretable. Therefore we decided to drop the 2
items that did not load significantly on any of the factors
and proceed with the two-factor solution as the best rep-
resentation of the data. The revised two-factor model
showed good fit to the data (see Table 2). Each factor had
an eigenvalue greater than 1 (7.60 and 2.02, respectively)
and the correlation between the two factors was moderate
(r = .56, p \ .001). The factor loadings of the WAFCS
scale after oblique (GEOMIN) rotation are presented in
Table 3. All the items had significant and meaningful
loadings on the designated factors. To reduce the length of
the scale we made an a priori decision to retain only the
strongest items, with the factor loadings C.70 (50 % or
more variance of the item explained by the factor). This
gave the final 10-item scale (see Table 3).
Confirmatory Factor Analysis
Analysis started with testing the hypothesized two-factor
10-item model of WAFCS (see Table 4, Model A). The
overall fit of this model was not acceptable. The exami-
nation of standardised residuals indicated that the model
adequately accounted for the associations between the
variables. The MIs indicated that the model could be
improved by allowing the correlation between error terms
of items 17 and 18, and 2 and 5. Items 17 ‘‘If I could relax
more at home I wouldn’t be as stressed and irritable at
work’’ and 18 ‘‘It is difficult to concentrate at work because
I am so exhausted by family responsibilities’’ both referred
to parent feeling stressed and exhausted at work due to
family responsibilities; items 2 ‘‘There is no time left at the
end of the day to do the things I would like at home’’ and 5
‘‘My family misses out because of my work commitments’’
both referred to not having enough quality time with one’s
family due to work obligations, which made the corrections
theoretically sensible. The revised model showed good fit
to the data. In addition the Chi square difference test
between Model B and Model C was 27.62 (df = 1) indi-
cating a significant improvement (p \ .001) of model fit.
Therefore Model C was chosen as an adequate description
of the data. Its graphic illustration is presented in Fig. 2.
Child Psychiatry Hum Dev

Table 2 Exploratory factor analysis fit statistics

Model v2 df CFI SRMR RMSEA RMSEA 90 % CI AIC BIC

One-factor solution 984.65*** 170 .693 .10 .13 .118–.133 20,601.38 20,824.41
Two-factor solution 482.22*** 151 .875 .05 .08 .077–.094 20,036.61 20,330.25
Three-factor solution 340.83*** 133 .922 .04 .07 .062–.081 19,892.99 20,253.54
Two-factor solution (16-item model) 268.94*** 89 .915 .04 .08 .070–.093 15,790.98 16,025.15
All models based on N = 305
v2 Satorra–Bentler scaled Chi square, df degrees of freedom, CFI comparative fit index, SRMR standardized root mean square residual, RMSEA
root mean square error of approximation, CI confidence interval, AIC Akaike’s information criterion, BIC Bayesian information criterion
*** p \ .001

Table 3 Results of the exploratory factor analysis with MLR Esti- subscales were above .70 (.88 and .86, respectively). In
mator and Geomin (Oblique) rotation addition, all items had significant loadings [.45 on the
Factor designated factors [32].
The correlation between the two factors was moderate
WtF FtW
(r = .49, p \ .001). The Chi square difference test was
1. My work prevents me spending sufficient quality .83 -.16 significant [scaled Dv2(1) = 359.40 - 90.30 = 198.03,
time with my family p \ .001], indicating that the better model is the one in
2. There is no time left at the end of the day to do the .75 -.05 which the two constructs are viewed as distinct, yet cor-
things I’d like at home
related factors. The SIC between the two factors reached
3. I am often distracted by thoughts about work while .70 .04 the value of .24, which was much lower than the AVE
spending time with my family
estimates for both subscales.
4. I often complete work tasks outside of work hours .66 .00
5. My family misses out because of my work .92 -.04
commitments Concurrent Validity
6. My work performance suffers because of my .83 .06
personal and family commitments As expected, the WFC subscale correlated highly with the
7. When work and family commitments clash I usually .65 .04 Frone’s subscale measuring WFC (r = .82, p \ .001) and
fulfil work commitments first lower with the Frone’s subscale measuring FWC (r = . 62,
8. At the end of the day I am too tired to enjoy .59 .16 p \ .001). Further, the FWC subscale correlated highly
spending time with my family
with the Frone’s subscale measuring FWC (r = .73,
9. If I didn’t work I’d be a better partner or parent .50 .14
p \ .001) and much lower with the Frone’s subscale
10. Working often makes me irritable or short .63 .13
measuring WFC (r = .35, p \ .001).
tempered at home
13. I often arrive at work late or in bad mood because .14 .56
of something that has happened at home Predictive Validity
14. My family commitments stop me from spending as -.08 .70
much time at work as I would like The model testing the relationships between FWC and
15. If I did not have a family I’d be a better employee -.09 .86 WFC, parental confidence, coercive parenting practices
16. My family has a negative impact on my day to day .00 .84 and child behaviour problems is presented in Fig. 3. The
work duties PTC, PS and ECBI scales all consisted of more than 20
17. If I could relax more at home I wouldn’t be as .12 .70 items, therefore for these scales we formed item parcels as
stressed and irritable at work the indicators of latent constructs in order to control for
18. It is difficult to concentrate at work because I am .06 .72 inflated measurement errors and improve the psychometric
so exhausted by family responsibilities
properties of the variables [50, 51]. Each parcel represented
Coefficients C .45 are in bold. Coefficients C .70 are underlined the average of items included. Examination of VIF statis-
WtF Work-to-family conflict, FtW Family-to-work conflict tics indicated no multicollinearities between the variables;
values below the cut-off threshold of 10 [32, 51].
Convergent and Discriminant Validity The hypothesized model showed good fit to the data [v2
(287) = 553.33, p \ .001; CFI = .922; SRMR = .059;
The AVE estimates for both subscales exceeded the cut-off RMSEA = .059 (90 % CI .052–.067)]. The analyses
value of .50 (.60 and .56 for WFC and FWC subscales, revealed that FWC was a significant and negative predictor
respectively). Furthermore, the CR estimates for both of parental confidence implying that the more frequently

123
 Child Psychiatry Hum Dev

Table 4 Confirmatory factor analysis of the factor structure of WAFCS

Model v2 df Dv2 Ddf CFI SRMR RMSEA RMSEA AIC BIC
90 % CI

A WAFCS
Initial 2-factor model 150.81*** 34 .890 .08 .12 .100–.138 7,677.97 7,786.38
B WAFCS
2-factor model with correlated errors 112.52*** 33 36.77*** 1 .925 .08 .09 .080–.120 7,635.40 77.47.31
between items 17 & 18
C WAFCS
2-factor model with correlated errors 90.38*** 32 27.62*** 1 .945 .07 .08 .066–.108 7,612.49 7,727.99
between items 17 & 18, 2 & 5
All models based on N = 264
v2 Satorra–Bentler scaled Chi square, df degrees of freedom, CFI comparative fit index, SRMR standardized root mean square residual, RMSEA
root mean square error of approximation, CI confidence interval, AIC Akaike’s information criterion, BIC Bayesian information criterion
*** p \ .001

My work prevents me spending sufficient quality time with my family

.83***

There is no time left at the end of the day to do the things I’d like at home
.56***

Work to family
-.52*** I am often distracted by thoughts about work while spending time with my family
.74*** Conflict

.89***
My family misses out because of my work commitments

.83***
My work performance suffers because of my personal and family commitments
0.49***

My family commitments stop me from spending as much time at work as I would like
.65***

If I did not have a family I’d be a better employee

.85***
Family to work
My family has a negative impact on my day to day work duties conflict
.92***

.65***
If I could relax more at home I wouldn’t be as stressed and irritable at work

.49*** .65***
It is difficult to concentrate at work because I am so exhausted by family responsibilities

Fig. 2 Two-factor confirmatory factor analysis of the 10-item WAFCS with two error covariances and standardized estimates

family obligations distract parents from accomplishing
work responsibilities; the lower is parental confidence in
handling children’s behaviour problems. Further, parental
confidence turned out to be a negative predictor of coercive
parenting practices, indicating that the less confident par-
ents are in handling their child’s behaviour problems, the
more inadequate parenting practices they apply. Finally,
parental confidence was a significant negative predictor,
whereas coercive parenting was a significant positive
predictor of child behaviour problems. The proportion of
variance in child behaviour problems explained by the
collective set of predictors was 33 %.
As Fig. 2 presents, WFC was associated indirectly with
child behaviour problems through parental confidence and
coercive parenting. Thus, in the hypothesized model, the
total indirect effect of FWC on child behaviour problems
was the sum of two simple indirect effects: (1) one simple
indirect effect with one mediator (FWC ? coercive

123
Child Psychiatry Hum Dev

-.54***
.44***

Item Item Item Item Item Item Item Item Item Item
1 2 3 5 6 14 15 16 17 18

.83*** .74*** .56*** .89*** .83*** .76*** .85*** .91*** .65*** .65***

.49***
Work-to- Family-to-
family conflict work conflict

.09 -.38***

.09 .14
Parcel 1 Parcel 1

.71***
.68***

Parcel 2 Parcel 2
.69***
.81***
Parental -.37*** Coercive
Parcel 3 .93*** Confidence Parenting .72*** Parcel 3

.91*** .76***

Parcel 4 Parcel 4
.67***
.92***
-.35*** .30***
Parcel 5 Parcel 5

.17
-.05
Child Behaviour

.77*** .90*** .88*** .71*** .75*** .58***

Parcel 1 Parcel 2 Parcel 3 Parcel 4 Parcel 5 Parcel 6

Fig. 3 Structural equation model of the direct and indirect relations between family-to-work, work-to-family conflict, parental confidence,
coercive parenting and child behaviour problems

parenting ? child behaviour problems), (2) one three-path
simple indirect effect that involves two consecutive
mediators (FWC ? parental confidence ? coercive par-
enting ? child behaviour problems). To assess the sig-
nificance of indirect and total effects we performed
bootstrap procedure with 5,000 bootstrap samples from the
original dataset to compute bias-corrected 95 % confidence
intervals (CI). If the 95 % CI excludes 0, then the effect is
significant at p \ .05 [45]. The estimates and their
bootstrap confidence intervals for all indirect and total
effects are presented in Table 5. The analysis revealed that
only a three-path indirect effect (with two consecutive
mediators) was significant.
Reliability
The coefficients H for both subscales of WAFCS were well
above the cut-off value of .70 (.91 and .91 for the WFC and

123

Table 5 Total effect, total indirect effect and simple indirect effects for the relationships between family-to-work conflict, parental confidence,
coercive parenting and child behaviour problems, unstandardized estimates
TE
Family-to-work conflict ? child behaviour problems
Simple IE
Family-to-work conflict ? coercive parenting ? child behaviour problems
Three-path simple IE
Family-to-work conflict ? parental confidence ? coercive parenting ? child behaviour problems
Total IE
Family-to-work conflict ? child behaviour problems
CI Confidence interval, TE total effect, IE indirect effect
FWC subscales, respectively), indicating very good inter-
nal consistency of the measure.
Discussion
This study used an expert informant approach with con-
sumer feedback to construct and validate a brief, easy to
use measure of work–family conflict: The WAFCS. The
resultant measure comprises two subscales assessing the
related but distinct constructs of WFC and FWC. We
sought to determine the psychometric properties of the
WAFCS including the validation of the factor structure,
construct, concurrent and predictive validity as well as
internal consistency. Analysis revealed the measure has
strong psychometric properties for use with parents.
In developing the measure we harnessed the views of
experienced international parenting researchers and clini-
cians to identify an item pool of 20 items assessing dif-
ferent aspects of WFC and FWC that could be subjected to
psychometric evaluation. We then gathered feedback from
parents regarding ease of use and face validity. Items were
subject to analyses and only the strongest items were
retained. This approach has the advantage of allowing
consumer feedback (to identify face validity) and of
determining the most effective items contributing to the
robust psychometric properties observed.
The CFA supported the 10-item 2-factor model of the
WAFCS identified through EFA. The 10-item WAFCS
showed excellent psychometric properties. The factor
structure revealed excellent convergent, and discriminant
validity. The WAFCS also showed excellent concurrent
validity in terms of high correlations with an established
Frone’s scale of work–family conflict. The internal con-
sistency for both subscales was very high ([.90) indicating
good reliability of the measure.
Furthermore, WAFCS also showed excellent predictive
validity. The tested model of relationships between WFC
123
Child Psychiatry Hum Dev
Estimate 95 % Bootstrap CI
.126 -.052 to .338
.003 -.019 to .104
.031 .005–.087
.165 .069–.309
and FWC and parent and child outcomes showed excellent
fit to the data. The results showed that when both types of
conflict were tested together in one model only the FWC
had a significant indirect effect on child behaviour prob-
lems through two consecutive mediators: parental confi-
dence and coercive parenting. The indicates high FWC
decreases parental confidence in handling child behaviour
problems, which in turn increases the use of coercive
parenting practices. The more frequent use of coercive
parenting the more behavioural problems the child dis-
plays. This relationship may be mediated by other variables
such as stress or low work efficacy. For example, parents
may use limited coping resources to address FWC in an
effort reduce adverse occupational outcomes. This would
leave few resources available to address parenting chal-
lenges resulting family stress, low parenting efficacy and
increased use of coercive parenting practices. Future
research could examine the role of using advanced struc-
tural equation modelling techniques. The finding that FWC
plays a more important role in predicting child behaviour
than WFC highlights a clinical need to focus on FWC
given its potential to simultaneously benefit the workplace
and the family environment.
This paper has a number of strengths including the use
of well-validated measures, and the use of multiple samples
and rigorous statistical techniques. The findings, however,
should be considered in light of limitations. In particular,
the sample comprised mostly of educated mothers of young
children, which may impact the generality of findings. It
may be that parents with greater education experience
different levels of work and family conflict or may respond
to the scale differently. Future research with a focus on
recruiting fathers and less educated parents as well as
parents of adolescents is needed to confirm validity within
these samples. Additionally, the use of cross-sectional
samples meant no conclusions could be made about the
WAFCS’s test–retest reliability or its potential change
sensitivity. Nevertheless, this paper has made a significant
Child Psychiatry Hum Dev

contribution to the literature in being one of the first rig-
orous validations of a measure of work–family conflict to
be used with parents. The WAFCS appears to be a psy-
chometrically robust measure of WFC and FWC that could
be applied in a variety of research settings to examine
work–family conflict in parents of young children. Future
research is warranted to validate the measure longitudinally
and ensure it is clinically sensitive to change and to
examine whether the psychometric strengths of the WAF-
CS are retained in other samples such as less educated
samples, fathers and potentially with non-parents.
interventions and its stability over time before it can rec-
ommended for routine clinical use.
Appendix: Items Included in the Final Work
and Family Conflict Scale
For the following scale please rate how much you agree
with the following statements by circling the appropriate
number.

Very Very
strongly strongly
disagree agree
Summary
1 My work me spending sufficient 1 2 3 4 5 6 7
Parents are at particular risk of work–family conflict that quality time with my family
has the potential to negatively impact their functioning 2 There is no time left at the end of 1 2 3 4 5 6 7
the day to do the things I’d like at
across both work and family domains and ultimately bring home (e.g., chores and leisure
negative outcomes for child development and adjustment. activities)
Clinical interventions for working parents now target 3 My family misses out because of 1 2 3 4 5 6 7
work–family conflict as an outcome variable. Therefore, my work commitments
the development of an easy to administer measure of work– 4 My work has a negative impact on 1 2 3 4 5 6 7
family conflict that is valid, reliable and change sensitive my family life
and can facilitate the tracking of intervention outcomes is 5 Working often makes me irritable 1 2 3 4 5 6 7
or short tempered at home
essential. A number of measures of work–family conflict
exist however none of these have been specifically 6 My work performance suffers 1 2 3 4 5 6 7
because of my personal and
designed from the outset for clinical use with parents to family commitments
assess longitudinal intervention outcomes. This paper 7 Family related concerns or 1 2 3 4 5 6 7
outlines the development and validation of WAFCS responsibilities often distract me
designed to measure work–family conflict in parents. The at work
measure was developed using an expert informer and 8 If I did not have a family I’d be a 1 2 3 4 5 6 7
consumer feedback approach and then subjected to rigor- better employee
ous psychometric evaluation, which included exploratory 9 My family has a negative impact 1 2 3 4 5 6 7
on my day to day work duties
and CFA as well as assessment of construct, concurrent and
10 It is difficult to concentrate at work 1 2 3 4 5 6 7
predictive validity and reliability. EFA revealed 2-factor 10 because I am so exhausted by
item structure of the measure assessing theoretically dis- family responsibilities
tinct but related constructs: FWC (5 items) and WFC (5
items). The CFA supported the 10-item 2-factor model of
the WAFCS identified through EFA. The analysis implied
that the factor structure of the scale has excellent conver- References
gent and discriminant validity. Furthermore, the WAFCS
1. Greenhaus JH, Beutell NJ (1985) Sources of conflict between
showed excellent concurrent validity in terms of its asso- work and family roles. Acad Manag Rev 10:76–88
ciations with Frone’s scale (an established measure of 2. Byron K (2005) A meta-analytic review of work–family conflict
work–family conflict). WAFCS has also very good pre- and its antecedents. J Vocat Behav 67:169–198
dictive validity. Both, WFC and FWC constructs were 3. Casper WJ et al (2007) A review of research methods in IO/OB
work–family research. J Appl Psychol 92:28–43
significantly associated with the use of coercive parenting 4. Frone MR, Russell M, Cooper ML (1992) Antecedents and out-
practices as well as the severity of child behaviour prob- comes of work–family conflict: testing a model of the work–
lems. The WAFCS appears to be a psychometrically robust family interface. J Appl Psychol 77:65–78
measure of WFC and FWC that could be applied in a 5. Eby LT et al (2005) Work and family research in IO/OB: content
analysis and review of the literature (1980–2002). J Vocat Behav
variety of research settings to examine work–family con- 66:124–197
flict in parents of young children. Future research is needed 6. Ruppanner L (2013) Conflict between work and family: an
to test the change sensitivity of this measure in clinical investigation of four policy measures. Soc Indic Res 110:327–347

123

7. Nomaguchi KM (2009) Change in work–family conflict among
employed parents between 1977 and 1997. J Marriage Fam
71:15–32
8. Haslam DM, Sanders MR, Sofronoff K (2013) Reducing work–
family conflict in teachers: a randomised controlled trial of
workplace triple P. Sch Ment Health 5:70–82
9. Hartung D, Hahlweg K (2010) Strengthening parent well-being at
the work–family interface: a German trial on workplace triple P.
J Community Appl Soc Psychol 20:404–418
10. Mesmer-Magnus JR, Viswesvaran C (2005) Convergence
between measures of work-to-family and family-to-work conflict:
a meta-analytic examination. J Vocat Behav 67:215–232
11. Nunnally JC, Bernstein IH (1994) Psychometric theory. McGraw-
Hill, New York
12. Carlson D, Kacmar KM, Williams LJ (2000) Construction and
initial validation of a multidimensional measure of work–family
conflict. J Vocat Behav 56:249–276
13. Frone MR, Yardley JK, Markel KS (1997) Developing and
testing an integrative model of the work–family interface. J Vocat
Behav 50:145–167
14. Jaeschke R, Guyatt GH (1990) How to develop and validate a
new quality of life instrument. In: Spilker B (ed) Quality of life
assessment in clinical trials. Raven Press, New York, pp 47–57
15. Diefenbach MA, Weinstein ND, O’Reilly J (1993) Scales for
assessing perceptions of health hazard susceptibility. Health Educ
Res 8:181–192
16. Cummins RA, Gullone E (2000) Why we should not use 5-point
Likert scales: the case for subjective quality of life measurement.
In: Second international conference on quality of life. National
University of Singapore
17. Matthews RA, Kath LM, Barnes-Farrell JL (2010) A short, valid,
predictive measure of work–family conflict: item selection and
scale validation. J Occup Health Psychol 15:75–90
18. Boyar SL et al (2006) Assessment of the validity of Netemeyer
et al’.s (1996) WFC and FWC scales. Int J Confl Manag 17:34–44
19. Netemeyer RG, Boles JS, McMurrian R (1996) Development and
validation of work–family conflict and family–work conflict
scales. J Appl Psychol 81:400–410
20. Kopelman RE, Greenhaus JH, Connolly TF (1983) A model of
work family and interrole conflict: a construct validation study.
Organ Behav Hum Perform 32:198–215
21. Thomas LT, Ganstr DC (1995) Impact of family supportive work
variables on work–family conflict and strain : a control per-
spective. J Appl Psychol 80:6–15
22. Cohen RJ, Swerdlik M, Sturman E (2012) Psychological testing
and assessment: an introduction to tests and measurement, 8th
edn. McGraw-Hill Higher Education, Boston
23. Haslam DM, Morawska A, Sanders MR (2010). The Work–
Family Conflict Scale (WAFCS). Parenting and Family Support
Centre, The University of Queensland, Australia
24. Kline P (1994) An easy guide to factor analysis. Taylor & Francis
Group, London
25. Sanders MR, Morawska A (2010) Family Background Ques-
tionnaire. Parenting and Family Support Centre, Brisbane
26. Arnold DS et al (1993) The Parenting Scale: a measure of dysfunc-
tional parenting in discipline situations. Psychol Assess 5:137–144
27. Locke LM, Prinz RJ (2002) Measurement of parental discipline
and nurturance. Clin Psychol Rev 22:895–929
28. Eyberg SM, Pincus D (1999) Eyberg child behavior inventory
and Sutter–Eyberg student behavior inventory—revised: profes-
sional manual. Psychological Assessment Resources, Odessa, FL
29. Webster-Stratton C, Hammond M (1997) Treating children with
early-onset conduct problems: a comparison of child and parent
training interventions. J Consult Clin Psychol 65:93–109
123
View publication stats
Child Psychiatry Hum Dev
30. Sanders MR, Woolley ML (2005) The relationship between
maternal self-efficacy and parenting practices: implications for
parent training. Child Care Health Dev 31(1):65–73
31. Muthèn LK, Muthèn BO (1998–20112) Mplus user’s guide.
Author, Los Angeles, CA
32. Stevens J (2002) Applied multivariate statistics for the social
sciences, 4th edn. Lawrence Erlbaum, Hillsdale, NJ
33. Satorra A, Bentler PM (1994) Corrections to test statistics and
standard errors in covariance structure analysis. In: Eye AV,
Clogg CC (eds) Latent variables analysis: applications for
developmental research. Sage Publications Inc, Thousand Oaks,
CA, pp 399–419
34. Browne MW, Cudeck R (1989) Single sample cross-validation
indexes for covariance-structures. Multivar Behav Res 24:445–455
35. Hu LT, Bentler PM (1999) Cutoff criteria for fit indexes in
covariance structure analysis: conventional criteria versus new
alternatives. Struct Equ Model Multidiscip J 6:1–55
36. Kline RB (2011) Principles and practice of structural equation
modeling. In: Methodology in the social sciences, 3rd edn.
Guilford Press, New York, xvi, 427 p
37. Brown TA (2006) Confirmatory factor analysis for applied
research. Guilford Press, New York
38. Schreiber JB et al (2006) Reporting structural equation modeling
and confirmatory factor analysis results: a review. J Educ Res
99:323–337
39. Churchill GA (1979) A paradigm for developing better measures
of marketing constructs. J Mark Res 16(1):64–73
40. Gerbing DW, Anderson JC (1988) An updated paradigm for scale
development incorporating unidimensionality and its assessment.
J Mark Res 25(2):186–192
41. Fornell C, Larcker DF (1981) Evaluating structural equation
models with unobservable variables and measurement error.
J Mark Res 18:39–50
42. Bollen KA (1989) Structural equations with latent variables.
Wiley series in probability and mathematical statistics Applied
probability and statistics. Wiley, New York, xiv, 514 p
43. Morawska A, Sanders MR (2007) Concurrent predictors of dys-
functional parenting and maternal confidence: implications for
parenting interventions. Child Care Health Dev 33:757–767
44. Cinamon RG, Weisel A, Tzuk K (2007) Work–family conflict
within the family. Crossover effects, perceived parent-child
interaction quality, parental self-efficacy, and life role attribu-
tions. J Career Dev 34:79–100
45. Shrout PE, Bolger N (2002) Mediation in experimental and
nonexperimental studies: new procedures and recommendations.
Psychol Methods 7:422–445
46. Cheng Y, Yuan K-H, Liu C (2012) Comparison of reliability
measures under factor analysis and item response theory. Educ
Psychol Meas 72:52–67
47. Sijtsma K (2009) On the use, the misuse, and the very limited
usefulness of Cronbach’s alpha. Psychometrika 74:107–120
48. Hancock GR, Mueller RO (2001) Rethinking construct reliability
within latent variable systems. In: Cudeck R, Toit SD, Sörbom D
(eds) Factor analysis and structural equation modeling: a Fest-
schrift honoring Karl G. Jöreskog. Scientific Software Interna-
tional, Lincolnwood, IL
49. Enders CK (2010) Applied missing data analysis. Guilford Press,
New York
50. Little TD et al (2002) To parcel or not to parcel: exploring the
question, weighing the merits. Struct Equ Model Multidiscip J
9:151–173
51. Kaplan D (1994) Estimator conditioning diagnostics for covari-
ance structure models. Sociol Methods Res 23:200–229