This paper econometrically compares the perceived marginal tax rates and the actually computed marginal tax rates and tries to find out if consumers could accurately perceive the marginal tax rates. Econometrically, the paper highlights that sample selectivity operates through unobservable elements and their correlation with unobservables influencing the variable of primary interest. Sample selection bias will not arise purely because of difference in observable characteristics. Although our paper is illustrative, it highlights the generality of the issue and its relevance to many economic examples