Construct Validity of the Short Forms of the Dyadic Adjustment Scale

Abstract

In the past two decades there has been a dramatic increase in research on marital quality both as a construct in its own right and as a dependent variable in models of marital functioning. An important prerequisite for any empirical investigation of marital quality is the availability of a valid indicator of the construct of interest; thus, much attention has been paid to the delineation of the components of marital quality and to the development of appropriate and sensitive measures of these components (cf. Sabatelli, 1988). Unfortunately, researchers have as yet been unable to operationalize the construct in a manner that is both unambiguous and widely accepted. Marital quality has been defined as both a unidimensional and a multidimensional construct (Glenn, 1990; Norton, 1983). The unidimensional approach focuses on the global subjective evaluation of the relationship, usually through indices of general marital satisfaction (Fincham & Bradbury, 1987; Norton, 1983). The multidimensional approach combines information from various domains of marital life, such as intimacy, communication, sexuality, and conflict resolution. This approach is perhaps best typified by Spanier's (1976) work on the construct of marital adjustment and the development of the Dyadic Adjustment Scale (DAS). Although the DAS is the most commonly used self-report measure of marital adjustment (Glenn, 1990), numerous criticisms have been raised over the years about Spanier's definition of marital adjustment and about the DAS. These include conceptual arguments about the appropriateness of defining marital quality as a multidimensional construct and measurement concerns about item scaling and subscale weighting (Fincham & Bradbury, 1987; Norton, 1983; Sabatelli, 1988). Despite these criticisms, the DAS has been used in hundreds of studies of marital functioning and marital therapy (Spanier, 1985). Although cognizant of the relevance of the criticisms we outlined above, in this article we examine a different type of recurrent criticism of the DAS, namely its length. As a result of its apparent success, many researchers have suggested ways of assessing dyadic adjustment with shortened versions of the DAS. The 32-item measure is relatively short, but it has been argued that a shorter scale assessing dyadic adjustment would be invaluable for family researchers and practitioners engaged in survey research, clinical screening, evaluation of change in marital adjustment over the course of an intervention, and any other activity in which the time available for completing questionnaires is severely constrained (Schumm et al., 1985; Schumm et al., 1986; Sharpley & Cross, 1982). At present, there are various short form versions of the DAS that have been used or suggested for use in the research literature. For example, based on their discriminant function analysis and factor analysis of DAS reports from 95 married individuals, Sharpley and Cross (1982) proposed that both a one-item (item 31) and a six-item (items 8, 10, 11, 25, 27, and 28) version could adequately replace the full DAS. Subsequently, using data from 545 people currently or recently in dyadic relationships, Sharpley and Rogers (1984) provided some evidence for the utility of a seven-item shortened DAS that combined the single. item and the six-item options. Based on data obtained from 39 couples who met through a dating agency and 162 randomly selected married individuals, Goodwin (1992) recently reported results to support the utility of the single-item approach proposed by Sharpley and Cross (1982). The single-item version (Clark, Worthington, & Danser, 1988), the six-item version (Hansen, 1987), and the seven-item version (Hansen, 1985; Martin, 1985) have all been used by researchers as substitutes for the full DAS. Finally, based on confirmatory factor analytic results of data from 538 married couples and 197 cohabiting homosexual couples, Kurdek (1992) proposed that the Satisfaction subscale (items 16 to 23 and items 31 and 32) could replace the full DAS with little loss of information. …