In this paper we analyse systematically through Monte Carlo simulations the consequences of misspecified heterogeneity on the most popular linear panel data models. We also illustrate our findings, through the estimation of a well known investment demand model.