Abstract: As high-fidelity simulations, assessment centers (ACs) are designed to measure a number of dimensions in various job-related exercises. Over the past 30 years, a great deal of research has indicated that ACs exhibit high content and criterion-related validity, but low construct-related validity. AC ratings substantially reflect the effects of the exercises and not the dimensions they were designed to reflect. This so-called AC “construct validity puzzle” drew a lot of attention from researchers. Three main research perspectives have gradually emerged: dimension-based approach, exercise-based approach, and interactionist approach, which focus on controlling factors causing measurement errors to improve the quality of dimension measurement, removing the dimensions from AC framework and focusing solely on the exercises or tasks, and taking the interaction between dimensions and exercises into account, respectively. Future research should pay more attentions to the exercise-based approach beyond the traditional dimension-based approach, and place great emphasis on the interactionist approach.

Key words: assessment centers, construct-related validity, dimension-based approach, exercise-based approach, interactionist approach