This thesis is an investigation of whether assessors' ratings of dimensions are affected by situation, using trait activation theory as a conceptual framework. An experimental 2 x 2 study was designed with a candidate in a vignette who participants assessed on behaviour and performance ratings on two dimensions. Vignette conditions differed by the level of performance the manager demonstrated for the two different dimensions, and by the dimension-relevance. As hypothesized, it was found that differences in dimension-relevance affected dimension ratings of performance but not dimension ratings of behaviour. These results suggest that task-based assessment centres (ACs) may provide more valid assessments of performance than dimension-based ACs. However, given the current popularity and support of the latter, best practices for dimension-based ACs are recommended that should appropriately account for the effect from differences in situation across exercises on assessors' performance ratings of ACs.