Might the Rorschach Be a Projective Test After All? Social Projection of an Undesired Trait Alters Rorschach Oral Dependency Scores

Abstract

The degree to which projection plays a role in Rorschach (Rorschach, 1921/1942) responding remains controversial, in part because extant data have yielded inconclusive results. In this investigation, I examined the impact of social projection on Rorschach Oral Dependency (ROD) scores using methods adapted from social cognition research. In Study 1, I prescreened 85 college students (40 women and 45 men) with the ROD scale and a widely used self-report measure of dependency, the Interpersonal Dependency Inventory (IDI; Hirschfeld et al., 1977). Results show that informing participants who scored low on the IDI that they were in fact highly dependent led to significant increases in ROD scores; I did not obtain parallel ROD increases for participants who scored high on the IDI or for participants who received low-dependent feedback. In Study 2, I examined a separate sample of 80 prescreened college students (40 women and 40 men) and showed that providing low self-report participants an opportunity to attribute dependency to a fictional target person prior to Rorschach responding attenuated the impact of high-dependent feedback on ROD scores. These results suggest that projection played a role in at least one domain of Rorschach responding. I discuss theoretical, clinical, and empirical implications of these results.

ACKNOWLEDGMENTS

I thank Erin Hughes, Deanna Kloss, Mei Ng, and Natalie Regier for help in collecting data for these studies and Ellen Hartmann, Don Viglione, and several anonymous reviewers for helpful comments on earlier versions of this article.

Notes

¹Although Exner (1989) reported four studies, the third was a control experiment used to determine whether various types of percepts were equally easy to generate. Results of this study do not address the issue of projection in Rorschach responding.

²Because no gender differences were anticipated and preliminary analyses revealed no main effects or interactions involving gender, I did not include this variable in the main analyses in either study.

³In these reliability analyses, I computed r using summary scores across all protocols, whereas I computed kappa using scores assigned to each individual response. (N of protocols was 170 in Study 1 and 160 in Study 2; total number of responses was 4,250 in Study 1 and 4,000 in Study 2.)

⁴According to Cohen's (1988) guidelines, ds of .20, .50, and .80 are considered to reflect small, medium, and large effect sizes, respectively.

⁵In fact, if generalized semantic priming was responsible for these results, the opposite pattern would have been obtained: Because studies have shown that self-relevant traits are easily primed in individuals who score high on self-report measures of those traits (Niedenthal, 1990; Stapel & Koomen, 1999), a trait priming explanation would predict greater ROD score increases in high-IDI than low-IDI participants.

⁶Descriptions of the dependent and angry target persons are in the Appendix. Following Srull and Wyer (1979), female participants read identical paragraphs except that the name “Donna” was used in place of “Donald” throughout.

⁷In this context, it might also be useful to describe the first phase of Rorschach responding as the “attribution” phase rather than the “response” phase.

⁸Because the methods used in Study 1 differ from those used in most clinical settings, the generalizability of these findings to in vivo clinical assessments remains open to question. In this context, however, it is important to note that the ROD means and score distributions in Studies 1 and 2 were comparable to those typically obtained in samples of clinical (Bornstein et al., 2000) and nonclinical participants (Bornstein et al., 1997).