Developing an Alternative Rorschach Administration Method to Optimize the Number of Responses and Enhance Clinical Inferences.

Unlabelled: Variability in the number of Rorschach responses (R) has stimulated controversy among clinicians and researchers for many years, and recent research reveals that R is much more variable than previously thought. Because R is correlated with other scores, its excessive variability may reduce the reliability, validity and clinical utility of these other scores. We present two experimental studies and additional results from other clinical datasets with the aim of developing a new administration procedure to diminish variability in R by reducing the number of very short and long records. In the first experiment, protocols were obtained using standard Comprehensive System administration or an alternative where we encouraged a second response if only one was given to a card and allowed only four responses on each card. This alternative method reduced the proportion of short records but produced an undesirable number of long records. To minimize the proportion of long records, in a second experiment, we added an instruction to give two or maybe three responses per card when introducing the test. Comparisons to CS administrations revealed that this procedure reduced variability in R by limiting the proportion of both short and long records. This reduced range was largely retained in an outpatient sample of older respondents with schizophrenia and a mixed clinical sample. Thus, we recommend this method of optimizing the range of R, which has since been included with very minor changes in the Rorschach Performance Assessment System.

Key Practitioner Message: Alternative administration method successfully eliminated overly short and excessively long records. Utility is potentially increased by greatly reducing both short records that often lack reliability and validity, as well as long records that consume an excessive amount of examiner administration and scoring time. Psychometric properties and the ability to apply parametric statistics are likely increased across variables given that the distribution of R is more normal. Re-administration due to inadequate R is almost never needed. Results are consistent with the conclusion that this alternative procedure reduces examiner variability by offering simple, but explicit instructions for encouraging sufficient productivity. Overall variability of R produced using the refined alternative procedure was significantly less than that produced using the traditional CS method, although more in line with Exner's (2003) normative expectations. Suggests that when using the alternative method, R becomes less of a confound for all other scores that are moderately to highly correlated with R. Also demonstrated that the reduced variability of R and the reduced number of less useful short and long records are generalized to clinical samples. Additional research (Reese, Viglione, & Giromini, 2014) provides support for these conclusions with child clinical samples.