Presidential Vetoes: An Event Count Model

Abstract

We argue that past models of presidential veto behavior have not adequately conceptualized the fundamental nature of the dependent variable - a count of the total number of vetoes per unit of analysis. Consequently, ordinary least squares techniques have been employed when more appropriate statistical estimation techniques are warranted. Further, past conceptualization (and measurement) of theoretically relevant variables have masked important relationships such as the importance of a bicameral legislature. We show that rigorous consideration of research methodologies provides theoretical insights obscured by relying on more traditional approaches. Finally, this investigation updates understanding of the veto process through the first year of the Clinton administration.