Interpretation of Interaction Effects in Logit and Probit Analyses

Abstract

Scholars have argued that more restrictive registration laws most drastically deter the least educated citizens from political participation. Others, however, argue that the most educated, rather than the least educated, are most drastically impeded by restrictive registration requirements. These opposing conclusions have dramatically different implications concerning registration reform in the United States. In this analysis, we urge scholars to take the arguments made by Nagler more seriously, and we argue that past models have not fully considered the inherently nonlinear functional form of the logit and probit models. Using graphical displays, we show that citizens with moderate levels of education are actually those who are “hardest hit” by restrictive closing dates. Consequently, we moderate all prior conclusions and show evidence that it is neither the most nor the least educated who are the “hardest hit” by early closing dates.