Assessment center ratings: Models and process

Assessment center ratings made by 22 assessors of 67 candidates for the position of Police Sergeant in a midwestern metropolitan community were analyzed. Additional data of assessor differences in cognitive complexity, assertiveness, participation in consensus discussion, use of schemata, and police background were also investigated for impact upon ratings. Five areas were explored: (1) issues of measurement, models, and reliability; (2) issues of cognitive processes involved in individual ratings; (3) issues of individual differences between assessors in cognitive and communication styles; (4) issues of group processes, influence, and communication; and (5) an integration of these four areas into a composite model of ratings. In the first of these areas, there was a convergence from results of linear structural equation analysis, factor analysis, and MTMM analysis showing the dominance of exercise (context) variance in ratings. However, the model which supplied the best fit to the observed rating data was one in which both exercise (contextual) and ability (dimension) factors were included. Reliability analysis showed differences in obtained reliabilities with those dimensions and exercises that are more commonly understood displaying higher reliabilities. In regard to individual cognitive processes, ratings data and questionnaire responses supported the early development and use by assessors of schemata for overall performance in exercises. Individual differences in cognitive complexity and assertiveness did not affect ratings as hypothesized. Group process and influence effects were investigated by means of Interaction Process Analysis of 95 taped consensus dimension discussions and content analysis of post assessment center questionnaire responses. These analyses supported a progressive stage model of discussion and a tendency for agreement. In an integration of process and individual differences data with rating data into a composite model, linear structural equation analysis (LISREL) suggested a model wherein participation, communication style, and police professional impacted upon the effects of group influence. However, only individual ratings substantially affected final performance evaluations of candidates. Additional research into assessment center processes and continued efforts to link cognitive and group process theory to this methodology are suggested