This paper addresses the question to which extent the complementarity between education and training can be attributed to differences in observable characteristics, i.e. to individual, job and firm specific characteristics. The novelty of this paper is to analyze previously unconsidered characteristics, in particular, personality traits and tasks performed at work which are taken into account in addition to the standard individual specific determinants. Results show that tasks performed at work are strong predictors of training participation while personality traits are not. Once working tasks and other job related characteristics are controlled for, the skill gap in training participation drops considerably for off-the-job training and vanishes for on-the-job training.