Toward a psychometric analysis of violations of the independence assumption in process dissociation

The authors outline a psychometric analysis of effects of violating the independence assumption underlying the process-dissociation procedure. That analysis distinguishes between process dependence and aggregation bias. Process dependence results when subjects rely on a strategy that makes recollection dependent on automatic influences of memory and is reflected by a correlation that can only be imagined, not observed. Aggregation bias results when parameters from a subject-item specific psychometric model are estimated by aggregating across observed subject and item data. Quantifying the magnitude of aggregation bias also requires speculation about a correlation that is not directly observed. Easily observed correlations calculated from aggregated estimates of automatic and recollective processes over subjects or items cannot be used to diagnose process dependence and are of limited utility for diagnosing aggregation bias. A postscript responds to T. Curran and D. L. Hintzman's (1997) reply.