R in Rorschach research: a ghost revisited

As interest in Rorschach research has increased, an old issue-whether differences in R require that other scores be normalized in relation to R when Rorschach data are analyzed-has been resurrected. Most who advocate normalizing or partialing for R cite Cronbach (1949) or Fiske and Baughman (1953) as the basis for their argument. Although Cronbach offered several useful suggestions regarding this problem, the data base to which he referred and that used by Fiske and Baughman allowed for a much greater variation in R than occurs when Rorschach's standard instructions are applied. Intercorrelations between R and most parametric Rorschach variables are not significant among groups of nonpatients, schizophrenics, and depressives. Thus, although the issue of R can be critical when some variables are included in a data matrix, most analyses can proceed without partialing or normalization. When nonparametric variables are included in an analysis, partialing or normalizing is inappropriate because the resulting distributions will be distorted.