Recently models with possibly non-normally distributed disturbances have attracted more attention. For such models independence and uncorrelatedness are not equivalent. In this paper we give an example that illustrates the potential importance of distinguishing between true independence and only uncorrelatedness.