The presence of omitted variables calls into question the use of a simple linear regression model y=_0+_1 x+u unless one uses homogenous subsamples as a means of controlling for other variables. So, in a wage equation where the focus is on the impact of education, the sample could be restricted say to females with the same occupation and level of experience.

Is this true, false or not certain? Why?