A Short Form of the Test of Facial Recognition for Clinical Use

Summary

To explore the possibility of developing a valid abbreviated form of the Test of Facial Recognition of Benton and Van Allen for clinical use, the test performances of 185 adult patients with established diagnoses of brain disease and 151 adult patients without history or evidence of brain disease were studied. Stepwise multiple discriminant analysis on subsamples of the brain-diseased and control patients identified discriminating items which were then subjected to cross-validation on independent subsamples. The resulting short form of the test consisted of 16 items requiring 27 responses, representing a 50 percent reduction in length. Part-whole correlation coefficients between the short and long forms of the test ranged from .884 to .940 in five different samples of patients. With use of the smoothed equipercentile method of Flanagan, a table for predicting long form scores from obtained short form scores was constructed. Guidelines for the employment of the short form in clinical evaluation were presented.