Designing self-report instruments for family assessment.

Abstract

As family medicine research expands, more and better self-report measures for quantifying family systems variables will be required. These measures should be developed from an accepted paradigm for instrument design that includes the following. They should be: 1) grounded in a contemporary conceptual framework; 2) constructed following a rational theoretical design that includes the sequential stages of construct identification, construct definition, item generation, item editing, item formatting, and item scoring; 3) analyzed empirically at the item level for item distribution characteristics, item-scale correlation, and item-response style correlations; 4) analyzed at the scale level for scale distribution characteristics, scale-scale correlations, and scale-response style correlation; 5) revised based on editing and empirical analysis; and 6) psychometrically evaluated for reliability (eg, internal consistency, test-retest) and validity (construct related, criterion related, and content related). Normative data from various reference populations must also be obtained to provide a basis for interpreting scores. Finally, test manuals must be written to summarize test development data and provide information on administration procedures, scoring, and interpretation.