This research work adresses the prediction of success after a training scheme (success in university examination, integration of job seekers after a training session) on the basis of 1/ subjective indicators at the end of training: satisfaction or subjective learning and 2/ individuals' variables (self-efficacy, locus of control, self-esteem, optimism and core self-evaluation). Our results show that, compared to training evaluation models, subjective indicators at the end of training do not appear to be good predictors of success. In addition, if individuals' variables are sometimes predictable of success, this seems to be related to the mere fact that they represent high values in students with past good results (the effect of prediction disappears when the effect of the student's past academic level is neutralized). In agreement with studies carried out on Internality Norm, such variables appear to be no more than indicators of previous social success situations, but not direct determinants of social success (Dubois, 1987, 2003). In the second part of our thesis, we propose – via the theory of commitment – to impact on participants' commitment, thanks to a statement of freedom (of choice) as to the attendance to classes. Favorable effects derived from such freedom of choice – as opposed to compulsory attendance - are observed on several results collected at the end of the training (satisfaction, subjective learning). At theoretical level, It seems that trainers' evaluation assessments could affect declared trainees' individual variables, thus making them predictive of success. Evaluation interactions are thus thought to reify the link usually considered between individuals' variables and situations of success. In addition, the context of freedom thus induced might contribute to enhanced individuals' variables as well as promote the mechanisms of learning.
