Assessment center construct validity : establishing expectations based on the dimension activation theory

The purpose of this study was to examine the construct validity of assessment center dimension ratings within the confines of an extended trait activation theory. Specifically, previous findings of high within exercise rating correlations have led researchers to conclude that ratings are affected by halo. Conversely, the extended trait activation theory suggests that high correlations are a function of the different levels of activation potential for various dimensions rated in a given exercise. For dimensions having stronger activation potential, it is expected that high levels of between subject rating variance will evidence discriminant validity. However, it is expected that dimensions with lower levels of activation potential will show lower levels of rating variance. This central tendency is expected to be the source of high within exercise rating correlations. Performance based dimension ratings for four distinct exercises were gathered from 97 individuals participating in developmental and selection assessment centers with trained assessors serving as raters. Exploratory factor analyses were conducted for each exercise to determine the necessity of one factor (supporting a halo theory) versus more than one factor (supporting the extended trait activation theory). Moreover, dominance analysis revealed the importance of each dimension for predicting overall exercise performance for each exercise. For these same exercises, 11 subject matter experts familiar with the exercises and dimensions provided ratings of the relative activation potential of each dimension for IV each exercise separately. It was expected that the relative dimension activation scale would correlate with the dimension dominance as revealed in the dominance analysis of actual assessment center ratings. Furthermore, it was expected that the mean variance of activated dimensions would be significantly higher than the mean variance of nonactivated dimensions in each exercise. These same analyses were used in comparing variance results with a scale of exercise primacy provided by the original exercise creator, as welt. This three point scale was expected to correlate with dimension activation ratings and similar results were anticipated.