Académique Documents
Professionnel Documents
Culture Documents
_______________________________________________________________
Table of Contents
1. Construct validity of the French version of the Dyadic Adjustment Scale....................................................
Bibliography...................................................................................................................................................... 17
20 October 2012
ii
ProQuest
Document 1 of 1
20 October 2012
Page 1 of 17
ProQuest
Table 3
Show less AbstractThe Dyadic Adjustment Scale (DAS) and its revised version (RDAS) are measures of the
quality of dyadic relationships. The goal of this study was to assess the properties of the French version of this
self-rating instrument by testing the most commonly proposed models of the English version using confirmatory
factor analysis. Our study sample included 1,131 parents of school children recruited in the general population
in the French-speaking part of Switzerland. Data analysis revealed an excellent fit of the unique-factor solution
for both the DAS and RDAS. Alternatively, our analyses also showed a good fit for the hierarchical solution of
the DAS. These results provide evidence for similar psychometric properties of the French version of the DAS
as compared to the original English version. Construct validity of the French version of the Dyadic Adjustment
ScaleA series of psychiatric conditions including mood disorders have been consistently found to be associated
with marital difficulties (Fendrich, Warner, &Weissman, 1990; Gotlib &Whiffen, 1989; Hoover &Fitzgerald, 1981;
Rutter, 1990). Marital discord may also be a salient risk factor for psychopathology in offspring (Zahn-Waxler et
al., 1988). In an effort to move beyond individually-oriented focus for the understanding of dyadic processes,
concepts such as dyadic coping and dyadic adjustment have been established. Dyadic coping is a process in
which the stress-related signals of one partner and the coping reactions of the other partner to these signals are
taken into consideration (Bodenmann, 1997). Indeed, a partners well-being and satisfaction highly depends on
the others well-being and satisfaction as well as on integration in the social environment (Bodenmann, 1997).
Dyadic adjustment is also conceptualized as a process rather than an unchanging state (Spanier, 1976), which
necessitates continual adaptation to the partner of the dyad. One instrument designed to measure marital
functioning is the Dyadic Adjustment Scale (DAS) (Spanier, 1976). According to its originators, this factoranalytically derived self-rating instrument has a four-factor structure. These factors were labeled: Dyadic
Consensus, Dyadic Satisfaction, Dyadic Cohesion and Affectionate Expression. Dyadic Consensus determines
the degree of agreement between the partners on matters within the relationship such as money, religion,
recreation, friends, household tasks and time spent together (Spanier, 1989). Dyadic Satisfaction measures the
amount of tension within the relationship and the degree to which an individual has considered ending the
relationship (Spanier, 1989). Dyadic Cohesion assesses interests and activities shared by the couple, and
Affectionate Expression quantifies the level of satisfaction regarding sex and the expression of affection (
Spanier, 1989). The four dimensions were found to have high internal reliability estimates (Cronbach alpha
coefficients between 0.73 for Affectionate Expression and 0.90 for Dyadic Consensus) and high
intercorrelations (average = 0.68; Spanier, 1976). The instrument was initially developed as a research tool but
is also applicable in clinical settings. Although it measures marital quality rather than all dimensions of dyadic
relationships (Thompson, 1988), it has been used in more than a thousand studies on heterosexual and
homosexual dyadic adjustment (Kurdek, 1992; Spanier, 1988, 1989). Numerous validation studies of the
original DAS have been performed using both exploratory and confirmatory factor analysis. The controversial
results of these studies were extensively summarized by Eddy, Heyman, and Weiss (1991) and Busby,
Christensen, Crane, and Larson (1995). Only two studies could replicate a four-factor solution which was similar
to the original solution (Gentili, Contreras, Cassaniti, &DArista, 2002; Shek, 1995). Three out of the four original
factors were further replicated in two more studies (Crane, Busby, &Larson, 1991; Spanier &Thompson, 1982).
However, Sharpley, and Cross (1982) could identify only one global factor. In fact, other investigators have also
suggested the existence of only one overall factor (Antill &Cotton, 1982; Kazak, Jarmas, &Snitzer, 1988).
Finally, in a recent study using confirmatory factor analysis technique, Lim and Ivey (2000) failed to replicate the
20 October 2012
Page 2 of 17
ProQuest
postulated four-factor structure of the DAS.Based on the conflicting results regarding the postulated factor
structure of the instrument, Spanier (1988) and Thompson (1988) have both stated that the DAS works best as
a summary measure and that the use of the subscales is questionable. Consequently, a shorter version is
generally considered sufficient for the measurement of a unique adjustment dimension. Sharpley and Cross
(1982) proposed the Abbreviated Dyadic Adjustment Scale (ADAS) including the overall Item 31 as well as the
6 items which best discriminated between distressed and non distressed couples. They also found that the
overall Item 31 accounted for 75% of the variance of the overall DAS score. Moreover, Goodwin (1992) found
that this item successfully classified 93,4% and 84,3% of married couples as satisfied or distressed in two
separate samples. Finally, Hunsley, Pinsent, Lefebvre, James-Tanner, and Vito (1995) tested four proposed
short forms of the DAS, including the ADAS and the use of the overall Item 31, using a Canadian sample of 196
married or cohabiting couples. The authors, who established correlations between the four forms of the DAS
and other family measures, concluded that only the ADAS and the satisfaction subscale were adequate
substitutes of the full DAS.In order to resolve discrepancies across studies with respect to the number of
underlying dimensions of the DAS, Sabourin, Lussier, Laplante, and Wright (1990) suggested a hierarchical
model including four first-level factors, which combine to form a second-level global factor. Similarly, using
males (n = 1,307) and females (n = 1,515) recruited from the community, Eddy et al. (1991) found the
hierarchical solution to best fit their data. The hierarchical model was tested again by Kurdek (1992) using
confirmatory factor analysis technique. In this study on heterosexual and homosexual couples the hierarchical
solution appeared more robust than the one-factor model. Busby et al. (1995) also considered the DAS a
hierarchical measure with first and second-order constructs. They argued that the problems with the highly
overlapping subscales occurred because some of the subscales were hybrids which contained both first and
second-order items. In other words, some of the items were homogeneous and others were related, but they
were not alternative forms of each other. Consequently, the authors suggested a revised DAS scale (RDAS)
containing seven pairs of relatively homogenous items, which formed seven first-order factors labeled Decision
Making, Values, Affection, Stability, Conflict, Activities and Discussion. Confirmatory factor analysis on data of
distressed and nondistressed individuals revealed an acceptable fit of this hierarchical model including the
seven first-order factors which formed the three second-order factors: Consensus, Satisfaction and Cohesion.
Internal consistency estimates for the second-order subscales were in the acceptable range: 0.81 for
Consensus, 0.85 for Satisfaction and 0.80 for Cohesion (Busby et al., 1995).The DAS has been translated into
French (Baillargeon, Dubois, &Marineau, 1986). Three studies on the French version using exploratory factor
analysis have also reported conflicting findings. Baillargeon et al. (1986) found three out of the four original
factors. However, in another study using distressed couples, Bouchard, Sabourin, Lussier, Wright, and Boucher
(1991) could only identify one factor, Affectionate Expression. The authors hypothesized that the original fourfactor structure of the instrument might be reduced to a single factor structure in distressed samples. Sabourin,
Bouchard, Wright, Lussier, and Boucher (1988) further tested the French version in a clinical sample of 100
Canadian couples. This analysis could not replicate the postulated four-factor structure and found a strong first
factor in males, females as well as the overall sample, which accounted for 59,4% to 64,3% of the total
variance.The goal of the present study was to determine the nature of the constructs of the French translation of
the DAS by examining the relative adequacy of the most commonly proposed models of the original English
version in a large Swiss sample recruited from the general population using confirmatory factor analysis
technique. The models tested included: a) the unidimensional model (Sharpley &Cross, 1982); b) the oblique
20 October 2012
Page 3 of 17
ProQuest
four-factor model as suggested by the originators of the instrument (Spanier, 1976); and c) the higher-order
model consisting of four first-order factors hierarchically related to a second-order factor (Sabourin et al., 1990).
Similarly, for the RDAS, we tested: a) the unidimensional model, b) the oblique seven-factor model as
suggested by Busby et al. (1995) and c) the hierarchical model including the seven first-order and three secondorder factors as suggested by Busby et al. To the best of our knowledge, no study has yet tried to replicate the
postulated structure of the RDAS. Method ParticipantsIn 1997 a population-based study of all school-children in
the 2nd/3rd (mean age 9.0 years; SD = 0.75 years) and 6th/7th grades (mean age 13.1 years; SD = 0.78
years) of several communities was conducted in the suburbs of Lausanne (Switzerland). This study included the
completion of self-report questionnaires by children and their parents. Participation rates were 66% for the
younger and 67% for the older age group of children, 48% for mothers and 34% for fathers. Children completed
the study questionnaires at school while parents did so at questionnaire completion sessions or at home. The
parental data of both 2nd/3rd and 6th/7th grade children were used in the present analysis. Informed consent
was obtained from the parents before participation in the study. The French version of the DAS (Baillargeon et
al., 1986) was included in the self-report battery assessing psychological and familial characteristics. Analyses
were conducted on 1,131 completed DAS questionnaires. Table 1 provides a description of the sample. The
mean age of the husbands was two years higher than that of the wives. More than two thirds of the husbands
held managerial or professional specialty positions, whereas only 29% of the wives held such positions and
38% were housewives. Of the sample, 72% were Swiss and French-speaking. Almost all couples were married
and the majority had two (52%) or three children (26%).
the seven-factor model, the items were allowed to only load on one first-order factor (Decision Making-Items 12
and 15; Values- Items 3 and 7; Affection-Items 4 and 6; Stability-Items 16 and 20; Conflict-Items 21 and 22;
Activities-Items 24 and 28; Discussion-Items 25 and 27). In a second step, we also tested the hierarchical
model as suggested by Busby et al. This model has three second-order factors Consensus, Satisfaction and
Cohesion. Consensus encompasses the first-order factors Decision Making, Values and Affection, Satisfaction
Stability and Conflict, and Cohesion Activities and Discussion. As some authors found gender differences in
their analyses (Kazak et al., 1988), analyses were also performed separately. However, as we found no
differences, we present only the results of the combined sample in this paper. For the confirmatory analyses,
similar to Cole (1987), three pre-established criteria were used as overall indicators of the goodness-of-fit of the
data: a) goodness-of-fit index (GFI) >0.85; b) goodness-of-fit index adjusted for degrees of freedom (AGFI)
>0.80 and, c) root mean-square residual (RMR) <0.10. Whereas the GFI and AGFI criteria are general
indicators of the fit (varying from 0 to 1) between the proposed models and the data, the RMR criterion is an
indicator of the average residual variance and should therefore be low to indicate a good fit. Furthermore,
considering our large sample size, we did not use the criterion of a non-significant p -value for a good fit as it is
almost impossible to be met in large samples (Cole, 1987).Data were analyzed using the SAS CALIS procedure
REG from the Statistical Analysis System, and the models tested were covariance structure models. In order to
maximize the available information, missing values were replaced by the median of the item if the subject
completed at least 90% of the questions of the DAS. As the responses to all the items showed a skewed
distribution, we followed the recommendation of Sabourin et al. (1990) and unweighted least-squares (ULS)
rather than the maximum likelihood method were used to estimate the models. Scale reliability of the retained
models was assessed by calculating the alpha coefficients (Cronbach, 1951). Results 1. Unidimensional DAS
modelThe scree test of the 32 DAS items suggested that only one factor with an eigenvalue of more than 11.0
be extracted from the data (Figure 1).
20 October 2012
Page 5 of 17
ProQuest
20 October 2012
Page 6 of 17
ProQuest
20 October 2012
Page 7 of 17
ProQuest
20 October 2012
Page 8 of 17
ProQuest
20 October 2012
Page 9 of 17
ProQuest
Page 10 of 17
ProQuest
estimations of the items were larger than 0.30, thereby suggesting that no items be dropped from the model.
Page 11 of 17
ProQuest
an excellent fit of the unique-factor solution for both scales with all three pre-defined criteria met. For the DAS,
the application of the four-factor solution only slightly increased the fit. However, the four-factor model revealed
high correlations between factors indicating considerable overlap between dimensions. Indeed, these factor
intercorrelations (ranging from 0.64 to 0.81) were higher than those reported by Spanier (1976; average
intercorrelation = 0.68) or by Antill and Cotton (1982; intercorrelations ranging from 0.17 to 0.67). Resolving the
problem of overlapping first-order factors, the hierarchical model introducing a unique second-order factor
revealed an almost identical fit as compared to the oblique four-factor solution.Given the similar fit between the
one-factor solution and the complex hierarchical model, both constructs appear to be acceptable solutions for
the French version of the DAS. This supports results of previous studies of the original DAS that found either
evidence for a unique-factor model (Antill &Cotton, 1982; Kazak et al., 1988; Sharpley &Cross, 1982) or a
hierarchical solution (Eddy et al. 1991; Kurdek, 1992; Sabourin et al., 1990). Furthermore, these findings are
compatible with those reported in one of the previous studies testing the French version of the DAS (Sabourin et
al., 1988).In contrast, despite the efforts of Busby et al. (1995) to create a new DAS version with more
homogenous subscales, our analysis only revealed an excellent fit of the unique-factor solution, whereas the
seven-factor model as well as the hierarchical model did not fit our data. The discrepant findings between the
study of Busby et al. and the present study, could be attributable to methodological differences with respect to
sample selection. Indeed, Busby et al. recruited both graduate students and distressed couples seeking
therapy, whereas our sample included mostly married couples recruited from the general population.Given the
good fit of both the unique-factor and the hierarchical model of the DAS, the scale can be considered either as
reflecting one unique dyadic factor or four dimensions which form together a global factor. If one wants to
measure only a unique dyadic dimension, a more parsimonious instrument than a 32-item scale should be
sufficient (Thompson, 1988). Indeed, several items are good predictors of the overall dimension. Item 18
explains 64% and Items 18 and 4 further explain up to 75% of the variance of the total score. As Figure 3
shows, when Items 31 and 10 were also added to an abbreviated scale, 84% of the total variance was
accounted for, whereas the addition of more items only slightly increased the explained variance. Therefore, a
4-item version (Items 18, 4, 31 and 10) may be sufficient to measure overall dyadic adjustment in Frenchspeaking populations. These proposed items assess more general impressions of the relationship and could be
labeled evaluative items as opposed to descriptive items (Norton, 1983). Fincham and Bradbury (1987) have
emphasized the fact that global evaluation items of marital quality generally account for most of the variance in
its measures.
20 October 2012
Page 12 of 17
ProQuest
References
1. Antill, J. K., & Cotton, S. (1982). Australian Psychologist.
20 October 2012
Page 13 of 17
ProQuest
2. Baillargeon, J., Dubois, G., & Marineau, R. (1986). Canadian Journal of Behavioral Science.
3. Bodenmann, G. (1997). Revue Europenne de Psychologie Applique.
4. Bouchard, G., Sabourin, S., Lussier, Y., Wright, J., & Boucher, C. (1991). Revue Canadienne de
Counseling.
5. Busby, D. M., Christensen, C., Crane, D. R., & Larson, J. H. (1995). Journal of Marital and Family
Therapy.
6. Cole, D. A. (1987). Journal of Consulting and Clinical Psychology.
7. Crane, D. R., Allgood, S. M., Larson, J. H., & Griffin, W. (1990). Journal of Marriage and the Family.
8. Crane, D. R., Busby, D. M., & Larson, J. H. (1991). American Journal of Family Therapy.
9. Cronbach, L. J. (1951). Psychometrika.
10. Eddy, J. M., Heyman, R. E., & Weiss, R. L. (1991). Behavioral Assessment.
11. Fendrich, M., Warner, V., & Weissman, M. M. (1990). Developmental Psychology.
12. Fincham, F. D., & Bradbury, T. N. (1987). Journal of Marriage and the Family.
13. Gentili, P., Contreras, L., Cassaniti, M., & DArista, F. (2002). Minerva Psichiatrica,.
14. Goodwin, R. (1992). Family Therapy.
15. Gotlib, I. H., & Whiffen, V. E. (1989). Journal of Abnormal Psychology.
16. Hoover, C. F., & Fitzgerald, R. G. (1981). Archives of General Psychiatry.
17. Hunsley, J., Pinsent, C., Lefebvre, M., James-Tanner, S., & Vito, D. (1995). Family Relations.
18. Kazak, A. E., Jarmas, A., & Snitzer, L. (1988). Journal of Family Psychology.
19. Kurdek, L. A. (1992). Journal of Family Psychology.
20. Lim, B. K., & Ivey, D. (2000). Contemporary Family Therapy.
21. Norton, R. (1983). Journal of Marriage and the Family.
22. Rutter, M. (1990). Developmental Psychology.
23. Sabourin, S., Bouchard, G., Wright, J., Lussier, Y., & Boucher, C. (1988). Science et Comportement.
24. Sabourin, S., Lussier, Y., Laplante, B., & Wright, J. (1990). Psychological Assessment.
25. Sharpley, C. F., & Cross, D. G. (1982). Journal of Marriage and the Family.
26. Sharpley, C. F., & Rogers, J. (1984). Educational and Psychological Measurement.
27. Shek, D. T. L. (1995). Journal of Clinical Psychology.
28. Spanier, G. B. (1976). Journal of Marriage and the Family.
29. Spanier, G. B. (1988). Journal of Family Psychology.
30. Spanier, G. B. (1989). Manual for the Dyadic Adjustment Scale. North Tonawanda, NY: Multi-Health
Systems, Inc.
31. Spanier, G. B., & Thompson, L. (1982). Journal of Marriage and the Family.
32. Thompson, L. (1988). Journal of Family Psychology.
33. Zahn-Waxler, C., Mayfield, A., Radke-Yarrow, M. R., McKnew, D. H., Cytryn, L., & Davenport, Y.
B. (1988). American Journal of Psychiatry.
Show less
Address for Correspondence:Caroline L. Vandeleur, DUPA, Site de Cery, Unit de Recherche en Epidmiologie
et Psychopathologie (UREP), CH-1008 Prilly, Lausanne, Switzerland
Email:E-mail: Caroline.Vandeleur@hospvd.ch
20 October 2012
Page 14 of 17
ProQuest
Subject: Factor Structure (major); Foreign Language Translation (major); Marital Relations (major); Rating
Scales (major); Statistical Validity (major); Adjustment; Dyads; Psychometrics; Test Forms; Test Validity
Classification: 2220: Tests&Testing, 2950: Marriage&Family
Age: Adulthood (18 yrs&older)
Population: Human, Male, Female
Location: Switzerland
Identifier (keyword): Dyadic Adjustment Scale, French version, martial functioning, psychometric properties,
Revised Dyadic Adjustment Scale, construct validity, factor structure
Test and measure: Dyadic Adjustment Scale
Methodology: Empirical Study, Quantitative Study
Author e-mail address: Caroline.Vandeleur@hospvd.ch
Contact individual: Vandeleur, Caroline L., Unite de Recherche en Epidemiologie et;
Psychopathologie
20 October 2012
Page 15 of 17
ProQuest
20 October 2012
Page 16 of 17
ProQuest
Bibliography
Citation style: APA 6th - American Psychological Association, 6th Edition
Vandeleur, C. L., Fenton, B. T., Ferrero, F., & Preisig, M. (2003). Construct validity of the french version of the
dyadic adjustment scale. Swiss Journal of Psychology/Schweizerische Zeitschrift Fr Psychologie/Revue
Suisse De Psychologie, 62(3), 167-175. doi: http://dx.doi.org/10.1024//1421-0185.62.3.167
_______________________________________________________________
Contact ProQuest
Copyright 2012 ProQuest LLC. All rights reserved. - Terms and Conditions
20 October 2012
Page 17 of 17
ProQuest