Interrater reliability among three judges on 130 Defense Mechanism Test variables: A multivariate approach

The aim of this study was to examine the interrater reliability in the codi ng of the Defense Mechanism Test for 130 variables on both a variable and a subject level. Parallel with a traditional variable analysis, a multivaria te strategy for measurement of interrater reliability, built upon principal component analysis, is presented. The results showed high agreement betwee n judges with an overall interrater reliability coefficient of .98 but indi cates the need for elucidation in the coding instructions for nine single v ariables. It also showed that multivariate modeling methods for handling pe rsonality test data are highly resistant to occasional differences on singl e variables but sensitive to major differences.