The effect of reducing the variability of length of written recalls on the rank order scale values of the recalls

Abstract

Two sets of fifteen recalls of two stories were selected so that they exhibited small amounts of variability in length. These two sets of recalls were scaled for accuracy of recall by the method of rank order. Factor analytic treatment of the intercorrelations of seven accuracy of recall measures resulted in a two factor solution for both sets of recalls. Regression analysis indicated, as might be predicted, that the scaled scores were more influenced by nonlength than by length factors although even under these length restrictive conditions length of recalls still influenced judgments of accuracy of recall.