Verbal descriptions of faces from memory: Are they diagnostic of identification accuracy?

There is considerable forensic import to the hypothesis that the quality of an eyewitness' description of a face is useful for predicting the accuracy of a subsequent identification. Although previous studies have failed to support this hypothesis, those studies were designed to test only whether witnesses who are good describers also are good identifiers. The current study used 88 different target faces and found a significant point-biserial correlation between description accuracy and identification accuracy. This relationship was not due to a process wherein good describers are good identifiers, however, but was due to the fact that faces that are better described are better identified; a relationship that could not be tested in the designs of previous studies. The quality of a subject's description of a given face did no better than did a second subject's description of that face in terms of predicting the former subject's identification accuracy. Because the description-identification relationship was found to be mediated by target factors (rather than subject characteristics), it was suggested that different assessments (e.g., of target-face uniqueness) could better predict identification accuracy.