The Impact of R-Optimized Administration Modeling Procedures on Brazilian Normative Reference Values for Rorschach Scores

ABSTRACT

To generate normative reference data for the Rorschach Performance Assessment System (R–PAS), modeling procedures were developed to convert the distribution of responses (R) in protocols obtained using Comprehensive System (CS; Exner 2003) administration guidelines to match the distribution of R in protocols obtained using R-Optimized Administration (Meyer, Viglione, Mihura, Erard, & Erdberg, 2011). This study replicates the R–PAS study, examining the impact of modeling R-Optimized Administration on Brazilian normative reference values by comparing a sample of 746 CS administered protocols to its counterpart sample of 343 records modeled to match R-Optimized Administration. The results were strongly consistent with the R–PAS findings, showing the modeled records had a slightly higher mean R and, secondarily, slightly higher means for Complexity and V-Comp, as well as smaller standard deviations for R, Complexity, and R8910%. We also observed 5 other small differences not observed in the R–PAS study. However, when comparing effect sizes for the differences in means and standard deviations observed in this study to the differences found in the R–PAS study, the results were virtually identical. These findings suggest that using R-Optimized Administration in Brazil might produce normative results that are similar to traditional CS norms for Brazil and similar to the international norms used in R–PAS.

Disclosure

The second author receives royalties on the sale of the Rorschach Performance Assessment System manual and associated products.

Notes

¹ SPSS syntax is available from the second author to assist with replications.

² In this data set Step 3 was close to the target sample values for the first and second substeps, although a bit farther off for the third substep. The target values were 1%, 5%, and 7%; the actual values were 0.6% (2/343 = .0058), 4.4% (15/343 = .044), and 4.7% (16/343 = .047). A viable alternative procedure would be to randomly select a specific number of protocols meeting the substep criteria rather than randomly selecting a proportion of protocols on each substep. To do this one would recognize that the sample size after Step 2 is 87% of the final sample size, with 13% still to come from protocols that have at least two responses to four, five, or six of the last nine cards (i.e., 1% + 5% + 7% = 13%). Thus, for this sample, one could say that the final sample size should be 356 (i.e., 310/.87 = 356.32) so that one should then randomly select 3, 18, and 25 protocols at each of the Step 3 substeps to reach the targets of 1% (3/356 = .0084), 5% (18/356 = .051), and 7% (25/356 = .070). We did not follow this alternative procedure because we were replicating the steps followed in the original R–PAS study

³ Although not previously reported, in the R–PAS norms this step eliminated 4.1% of the initial responses.

⁴ The flip side of this is that the protocols that are not selected to be part of the modeled records tend to be brief. In this data set, the distribution of R in the 403 protocols that were not used for modeling had M = 16.89 and SD = 2.87 (minimum = 14, maximum = 30), whereas the initial distribution of R in the 343 that were used in the modeling had M = 28.38 and SD = 10.09 (minimum = 17, maximum = 85).

⁵ Although not fully reported before, in the original data set for the R–PAS norms, the 1,396 protocols had a total of 30,499 responses. The 640 modeled protocols had 15,463 responses, such that 50.7% of the original responses were retained in the modeled data.

⁶ Technically, the boundaries of 0.85 and 1.15, which are 15% above or below a value of 1.0, are asymmetrical. To be balanced, a lower boundary of 0.85 would correspond to an upper boundary of 1.18 or an upper boundary of 1.15 would correspond to a lower boundary of 0.87. We retained the boundaries of 0.85 and 1.15 because those were the benchmarks used in the R–PAS study. In the data to be reported this asymmetry had no practical importance.