We compare employment and wage polarization in the U.S. using different sample periods, occupations classification and the inclusion or not of agricultural occupations. We report two main findings. First, we show that employment polarization can emerge together or without wage polarization, depending on the sample period considered. Second, we show that removing agricultural occupations changes the timing of employment polarization, making it emerge earlier, and substantially increases the degree of both employment and wage polarization with respect to the case in which they are included in the sample.