Sample selection bias in analyses of the political democracy and income inequality relationship

Abstract

Results of studies into the relationship between political democracy and income inequality vary, with some researchers finding that democracy decreases inequality, others that democracy has no effect, and several that democracy actually increases inequality. This article explores sample-selection bias as a possible explanation for these discrepancies. We apply Heckman's two-step correction procedure to three studies: Bollen and Jackman (1985), Muller (1988), and Simpson (1990). Results indicate that even when statistically significant bias exists, coefficient estimates for political democracy are unaffected; sample-selection bias is apparently not responsible for the discrepancies.