This paper studies the effects of unions on the structure of wages, us
ing an estimation technique that explicitly accounts for misclassifica
tion errors in reported union status, and potential correlations betwe
en union status and unobserved productivity. The econometric model is
estimated separately for five skill groups using a large panel data se
t formed from the U.S. Current Population Survey. The results suggest
that unions raise wages more for workers with lower levels of observed
skills. In addition, the patterns of selection bias differ by skill g
roup. Among workers with lower levels of observed skill, unionized wor
kers are positively selected, whereas union workers are negatively sel
ected from among those with higher levels of observed skill.