A strategy for comparing two sets of results (one based upon early childhood recollections (ECR) and another upon video taped (VT) group behavior) from the Perceptual Characteristics Rating Scale was developed. The null distribution of the mean deviation was estimated by randomly matching an ECR response vector with a VT response vector. To evaluate the degree of relationship of ECR's and VT's for each subject, a correct pairing was made. The observed values of the mean deviations were compared to the null distributions and the associated empirical probabilities were converted to chi squares to allow summary tests of "no relationship." (Author)