Estimating Voters’ Taste for Risk:
Candidate Choice under Uncertainty
Version 1.1
Preliminary draft, Please do not cite, Comments welcome. 1
Adam J. Berinsky
Department of Politics, Princeton University
and
Jeffrey B. Lewis
Department of Political Science, UCLA
July 2, 2002
1Prepared for presentation at the 2002 Annual Summer Political Methodology Meetings, Seattle,
WA. Email: berinsky@princeton.edu and jblewis@polisci.ucla.edu.
1 Introduction
When citizens enter the voting booth on election day, they do so in an environment where the
stakes are often unclear. Over the last 40 years, political scientists have been interested in
the choice dynamics at work in such situations. Voters may be uncertain of the competence
of the candidates, the likelihood that a given candidate will win the election, and even of
their own policy positions. But a central concern to political scientists has been uncertainty
about the positions taken by candidates on the controversies of the day, and the effect of
such uncertainty on the attractiveness of candidates to voters.
Recent work in political science has taken up the question of issue voting under conditions
of uncertainty—situations in which voters have imperfect information about the policy posi-
tions of candidates. Models that recognize this principle are certainly realistic portrayals of
the campaign environment. But these studies may be limited in important respects. Specifi-
cally, to date, the study of vote choice under uncertainty has made a common assumption of
quadratic preferences implying that citizens behave in a very risk-averse manner when cast-
ing votes (see Davis, Hinich, and Ordeshook 1970; Enelow and Hinich 1981; Bartels 1986;
Alvarez 1997, Hinich and Munger 1994, though see Shepsle 1972). . But this is simply a
modelling choice. Many other utility functions consistent with “proximity” voting could be
chosen, including functions that imply risk neutral and risk acceptant behavior in the sense
developed below.1
The assumption of risk-aversion is not simply a t