Does Productivity Impact the Stability of Rorschach Scores?

Abstract

Research suggests that productivity could impact the stability of Rorschach scores. To explore for this effect, we conducted secondary analyses of test–retest data gathered using the Rorschach Comprehensive System (Exner, 2003) and available for 75 French, nonpatient adults (Sultan, Andronikof, Réveillère, & Lemmel, 2006). We examined how response frequency (R) impacted stability using hierarchical regression models. Results on 83 variables from the lower part of the structural summary showed that stability was impacted by the mean level of productivity in 12 variables with medium to large effects (including Zf, HVI, and W location). Stability was also impacted by variations of productivity in 9 variables with medium to large effects (including Passive Movement, D Location, or Human Contents). Higher mean R and variability of R impacted stability levels negatively. Transforming scores into proportions (i.e., dividing scores by R) was beneficial for some important variables (including FM+m, Zf, DQ+). Procedures should be developed to limit productivity and control for R variations across time if one wishes to derive more reliable descriptions of individuals from the Rorschach.

Acknowledgments

A preliminary version of this article was presented at the 8th Congress of the European Rorschach Association in Padua, Italy (August 2006). We thank other contributors whose input was of immense value at an earlier stage of the study: Anne Andronikof, Christian Réveillère, Gilles Lemmel, Damien Fouques, and Thomas Saïas. This project received support from the Rorschach Research Foundation. Aspects of the analyses benefited from fruitful discussions with Cato Gr⊘nner⊘d, Donald Viglione, and three anonymous reviewers.

Editor's Note: Bill Kinder served as the final Editor for this manuscript.

Notes

¹We thank an anonymous reviewer for drawing our attention to this phenomenon.

²We also computed proportions in which the numerator was the same as in the preceding fraction, but the denominator was the sum of both elements. This did not yield any significant difference in subsequent analyses. The following results are based on the common adjustment referred to in the body of the article.

∗p < .05.

∗∗p < .01.

∗∗∗p < .001.

³This information is presented here in some detail because there was discussion during the review process around the relative importance of main effects versus interaction effects in these analyses.

⁴As detailed in Sultan et al. (2006), mean-level changes were very limited in this sample. As a consequence, for a given score, absolute stability and relative stability estimates were close to each other. In fact, within the set of 83 variables considered here, a correlation of .60 was found between the test–retest rs (indicative of relative stability) and the coefficient of absolute variation |T2 – T1|/T1 (indicative of absolute stability).

⁵The z-axis in these figures is plotted with larger positive values toward the front and increasingly negative values toward the back.

∗p < .05.

∗∗p < .01.

∗∗∗p < .001.

∗p < .05.

∗∗ p < .01.

∗∗∗p < .001.