Examining the validity of self-report assessments of family functioning: a question of the level of analysis.

Family assessment instruments attempt to measure family functioning at a particular level of the family system: individual, dyad, or family as a whole. This article introduces the concept of level validity, that is, whether an assessment measures family functioning at the level that it was intended to measure. The authors argue that whenever higher-order factors (e.g., dyadic subsystems) are the target of a measure, these factors should explain variance that is independent of their lower-order constituents (e.g., individual-level characteristics). Previously published data targeting dyadic subsystems within the family were reanalyzed using a model that controls for lower-order effects. Dyad-level factors rarely emerged independent of individual-level factors and, when they did, they did not replicate across samples. The results suggest that level validity should be tested and reported along with other aspects of construct validity before accepting such measures as valid assessments of family functioning.