| This study examined the interrater reliability of Rorschach Comprehensive System protocol-level variables. Archival data were obtained from the following nine countries: Argentina, Belgium, Israel, Italy, Japan, Peru, Portugal, Spain, and the United States. All Rorschach protocols were administered and coded according to the Comprehensive System (Exner & Weiner, 1995). Participants from the various countries were randomly selected nonpatients. Each randomly selected Rorschach protocol was coded by at least two qualified researchers to obtain the interrater reliability statistics. Interrater reliability was calculated using Rorschach Research Utilities Version 1.0.0, 2003, developed by Harold Janson. The statistical coefficient iota, which corrects for chance, was used to calculate interrater agreement.;There were 192 protocol-level variables from the Comprehensive System that occurred in the 487 protocols used in this study. Four of the variables (2.1%) had fair agreement. Eleven variables (5.7%) had moderate agreement. Fifty-one variables (26.6%) had substantial agreement. One hundred and twenty-six variables (65.6%) had nearly perfect agreement.;The results of this study support the majority of the research on the interrater reliability of the Rorschach using the Comprehensive System. Most Rorschach variables are coded reliably, and the test as a whole is a reliable assessment instrument when using Exner's Comprehensive System (2001). However, further clarification of coding instructions is needed to increase reliability in the areas where coding instructions are unclear and reliability falls short. |
