The reliability, validity, and utility of three data modes in assessing marital relationships

Data from individual interviews, four questionnaires, and two analogue situations for 12 married couples were examined by six clinical judges and evaluated with regard to various aspects of interpersonal functioning. Orders of presentation were counterbalanced in a Latin square design. Ratings were made after each kind of data had been reviewed, after a case conference, and following discussion leading to consensus. Reliabilities were computed for the sets of variables and for each data mode. The convergent and discriminant validities of judgments based on the three data modes were examined by means of a multitrait-multimethod matrix. The utilities of the data modes and of increasing amounts of information were gauged by correlating partial estimates with average judgments of all judges based on all data following case conference. Among the findings were reasonably high levels of reliability for some aspects of interpersonal functioning but very low agreement regarding recommended treatment, quite high convergent validity across the three data modes, and modest but significantly superiority of the interview over the other procedures for assessing marital relationships.