Construct Validity of Spouses' Relative Influence Measures: An Application of the Direct Product Model

This article investigates the construct validity of three measures of spouses' relative influence. A joint decision-making exercise was developed and completed by 65 couples in which relative influence was measured by two self-report measures and one outcome measure. The self-report measures were a balanced 5-point "who won" continuum and a 100-point constant sum scale. The out-come measure was derived from part-worth utility weights yielded from dummy-coded ordinary least squares regressions. The direct product model (DPM) was used to analyze the resulting multitrait-multimethod data. The model assumes that correlations among measures are influenced by method effects. Results provide evidence of convergent validity among methods and trait discriminant validity. However, the communality index suggests that the outcome measure captured a different dimension of relative influence than did the self-report measures. Implications of these findings for research in joint decision making are discussed.