Evaluating the quality of training courses: a comparison of different designs for evaluating training

This paper aims to show that some designs for evaluating training are more suitable than others in a quality approach. Effectively, several practices of evaluation are incompatible. These are presented in the first section: Kirkpatrick’s fourlevel training evaluation model (Reaction, Learning, Behavior and Results), Kraiger’s comprehensive model based on learning outcomes, as well as the more recent designs taking into account multiple stakeholder perceptions (Michalski & Cousins). On the basis of these works and of some original practices, we attempt to show the aspects which are the most adapted to a total quality approach.