Objectives This study quantifies the quality of the median voter model (MVM) and analyzes whether personal characteristics explain legislative shirking. Methods We employ a natural measure for divergence between politicians and voter preferences by matching final roll‐call votes of politicians with referendum results. The relevance of personal characteristics for legislative shirking is analyzed by means of logit regressions. Results The MVM outperforms a random decision benchmark when predicting the behavior of politicians regarding the revealed preferences of their constituencies, but the model fails to account for a substantial part of its theoretical prediction. Personal characteristics of politicians do not explain overall legislative shirking. Conclusions Policy predictions based on the MVM should be carefully considered because of failing convergence. Majority elections tend to crowd out personal characteristics of politicians as potential explanations for legislative shirking.