A Meta-Analysis of an Implicit Measure of Personality Functioning: The Mutuality of Autonomy Scale

Abstract

The Mutuality of Autonomy scale (MA) is a Rorschach variable designed to capture the degree to which individuals mentally represent self and other as mutually autonomous versus pathologically destructive (Urist, 1977). Discussions of the MA's validity found in articles and chapters usually claim good support, which we evaluated by a systematic review and meta-analysis of its construct validity. Overall, in a random effects analysis across 24 samples (N = 1,801) and 91 effect sizes, the MA scale was found to maintain a relationship of r =.20, 95% CI [.16,.25], with relevant validity criteria. We hypothesized that MA summary scores that aggregate more MA response-level data would maintain the strongest relationship with relevant validity criteria. Results supported this hypothesis (aggregated scoring method: r =.24, k = 57, S = 24; nonaggregated scoring methods: r =.15, k = 34, S = 10; p =.039, 2-tailed). Across 7 exploratory moderator analyses, only 1 (criterion method) produced significant results. Criteria derived from the Thematic Apperception Test produced smaller effects than clinician ratings, diagnostic differentiation, and self-attributed characteristics; criteria derived from observer reports produced smaller effects than clinician ratings and self-attributed characteristics. Implications of the study's findings are discussed in terms of both research and clinical work.

Notes

One example of a study that was not considered to have used the original scale is Spear and Sugarman's (1984) study in which they extended the MA continuum from 7 to 10 scale points. Another is Gluckman's (1992) dissertation in which a structured procedure was used to inquire about any movement in a response to see if it could develop into an MA codeable response.

In one instance we used a dissertation rather than the published article associated with that study. Zodan (personal communication, January 2, 2014) clarified that her dissertation research (Zodan, 2010) expanded the sample size and extended the scope of the published Zodan, Charnas, and Hilsenroth (2009) article.

Some samples used Ns that varied from one publication to another and sample sizes were averaged. Because of this, in one of the moderator analyses the sample size exceeds what we report for the N of the entire meta-analytic sample (N = 1,822 vs. N = 1,801).

The number of samples in this analysis is 25 because there was both a publication (Leifer et al., 1991) and a dissertation (DeSousa, 1993) using the same participants.