Modern econometrics stresses the diagnostic testing of estimated models as an important part of the model-building process. In a survey article published in this journal (Pagan and Vella 1989), methods for testing the validity of the assumptions underlying the censored regression of the Tobit model were presented. This paper questions the numerical results presented there. Indeed, my results concerning the tests when applied to Fair's well-known model for infidelity (1978) are the opposite of those presented in this journal.
