ABSTRACT
Do incumbents dominate non-partisan elections because of an especially large personal vote? This question has important implications for understanding the causes of incumbent success and the benefits or drawbacks of non-partisan elections. This paper uses a natural experiment, combined with three original datasets, to estimate the size, persistence, and consequences of the personal vote in a large non-partisan city election. We first use individual-level survey data to show that individuals assigned quasi-randomly to a new incumbent are substantially less likely to support the incumbent. We use a second survey, one year later, to demonstrate the persistence of this effect. Finally, we use historical election results to simulate the electoral consequences of the personal vote; we find that the personal vote is sufficiently large to affect one in four incumbent races. We conclude that the personal vote, while large and important, is not sufficient to explain incumbent dominance in non-partisan contests.
