Abstraction,Associative Focusing,and Similarities Tests: Towards More Clinically Sensitive and Better Understood Tests

Abstract

A proposed associative focusing hypothesis that describes the response process in similarities tests was investigated in a series of 10 experiments. The experiments ranged from general experimental studies of the hypothesis to investigations of the clinical utility of a new unconventional similarities test. Exp. I confirmed specific predictions derived from the associative focusing hypothesis. Exp. II permitted accurate prediction of similarities item difficulty from association data. Exp. III provided evidence that an extraneous confounding variable was built into the very structure of conventional similarities tests. All three experiments were replicated, with almost identical results. Two clinical experiments suggested that a new similarities test, from which the confounding variable was excluded, was more sensitive than conventional similarities tests to diminished abstraction ability in schizophrenics. The third clinical experiment extended the investigation of these issues with a sharply contrasting sample of patients, differing markedly in age, chronicity of illness, and use of medication. The experiment also studied the assessment of impaired abstraction ability in patients with presumed brain dysfunction. The fourth clinical experiment provided further evidence that the new test is free of contamination by word association.