Assessment centers: is it worthwhile to use a competencies-based framework?

Over the past several decades, assessment centers have enjoyed increased popularity as both development and selection tools. Indeed, this method which evaluates candidates on several job-related dimensions through various simulation exercises and by multiple trained assessors has been shown to have a substantial criterion-related validity, i.e. to be a valid predictor of performance in a wide range of jobs. In contrast, evidence for the construct-related validity of assessment centers dimensions has been more mixed. The recent spreading of competencies-based approach in HRM has led to replace specific and behavioral dimensions by broader competencies, probably again more complex and difficult to operationnalize. Even if they benefit from good face validity among human resources managers, these competencies-based assessment centers still have to demonstrate their construct-related and criterion-related validity. The purpose of the present study is to examine the construct-related and criterion-related validity of the assessment centre used to select entry-level candidates to the Belgian federal police force. Following a preseletion on the basis of a cognitive ability test, this assessment center taps seven competecies by using five techniques: a personnality inventory, two group simulations, an individual inteview and an interview by a jury. Data were obtained from 369 police officers which were selected in 2006 and which completed their traineeship one year later. As criteria of work performance, we used six training results: 4 scores of success in various achievement tests and two trainee’s behavioural evaluation by a trainer and by a professionnal tutor. Construct-related validity was tested by using a principal component analysis and a multitraits-multimethods matrix. Results confirm a well-known effect of common variance due to the method/technique. As regards criterion-related validity, we compared the respective predictive validity of two sets of predictors (competencies scores measured through different techniques versus global scores from each technique), controlling for cognitive ability score. Results show a low incremental predictive validity of assessment center scores over cognitive score but are in line with predictive values reported in similar studies. Several methodological and conceptual explanations were proposed to discuss these relatively disappointing results and to suggest future avenues for testing and improving validity of competencies-based assessment centers