| Sumario: | Many military selection systems operate on a large scale, and this requires pooling of data from multiple raters. When individual raters vary in the quality of their ratings, the overall performance of the selection system can be adversely affected. This study uses multilevel logistic regression to investigate rater effects. Canadian military interviewers (n = 59) completed a self-report questionnaire. Their ratings of interviewees, as well as subsequent training performance of interviewees, were obtained from an archival source. In multilevel models, individual differences in interviewer leniency, but not validity, were observed. These findings have important theoretical and practical consequences.
|
|---|