Do Voters Discount Political Scandals over Time?

• Author: Miguel M. Pereira,Nicholas W. Waterbury
• Published date: 01 September 2019
• Date: 01 September 2019
• DOI: 10.1177/1065912918795059
• Subject Matter: Articles

795059PRQXXX10.1177/1065912918795059Political Research QuarterlyPereira and Waterbury
research-article2018
Article
Political Research Quarterly
2019, Vol. 72(3) 584 –595
Do Voters Discount Political
© 2018 University of Utah
Article reuse guidelines:
sagepub.com/journals-permissions
Scandals over Time?
https://doi.org/10.1177/1065912918795059
DOI: 10.1177/1065912918795059
journals.sagepub.com/home/prq
Miguel M. Pereira1 and Nicholas W. Waterbury1
Abstract
This paper explores how political scandals are discounted over time. Previous research has shown that voters respond
disproportionately to recent economic conditions when evaluating incumbents. We argue that voters discount not
only the performance of incumbents in office but also information about their personal character, largely due to
accessibility biases. Building on a comprehensive database of congressional scandals covering the last four decades, we
show that the electoral consequences of political scandals fade fairly quickly. Only cases emerging in the election-year
systematically affect the vote share of incumbents. Moral scandals are the exception, with negative effects persisting
over the entire term. In line with the mechanism proposed, additional analyses suggest this pattern results from
disproportional levels of media attention, making moral scandals more easily retrieved from memory. The results
broaden our understanding of the nature of myopic voting and provide an explanation for the increasing reliance on
negative campaigning.
Keywords
scandals, congress, voting behavior, myopic voting
In the early 2000s, Representative John E. Sweeney was
scandal-ridden incumbents lost, on average, five points in
a rising star among New York Republicans. Apart from a
vote share. However, it remains unclear whether the tim-
successful political career in his home state, he had been
ing of scandal outbreaks plays a role in voters’ assess-
affectionately dubbed “Congressman Kick Ass” by ments of incumbents. While Sweeney’s scandal occurred
President George W. Bush.1 However, after three suc-
less than a month from the election, Kennedy’s occurred
cessful reelections, in 2006, Sweeney was challenged by
half a year before. Did this difference allow for Kennedy’s
Democrat Kirsten Gillibrand. In the midst of a conten-
transgressions to be overlooked?
tious campaign, the incumbent was suddenly charged
In this study, we examine how time moderates the
with domestic assault, and embroiled in a financial scan-
effect of scandals on voting behavior. Our argument is
dal about improperly using funds for a ski trip. These sto-
twofold. First, we expect scandal-ridden politicians to be
ries, emerging less than a month before Election Day,
punished less the longer the time between the scandal
sank Sweeney’s campaign. He was defeated, and his
outbreak and the following election. This argument builds
political career was over. In the same year, Congressman
on an extensive literature on myopic voting (e.g., Achen
Patrick Kennedy was also involved in a scandal. A simi-
and Bartels 2016; Tufte 1978). For decades, researchers
larly popular representative, Kennedy admitted to a pre-
have noticed that voters respond disproportionately to
scription drug addiction after being involved in a car
election-year economic dynamics. As Healy and Lenz
accident outside the United States Capitol. He would later
(2014, 31) summarized, “voters . . . reward incumbents
enter a treatment program. Despite initially intense media
not broadly for economic growth throughout incumbents’
coverage, six months later, Kennedy was reelected.
terms, but narrowly for conditions in the six months or
Cases like these, among several others, beg the ques-
year before Election Day.” We argue that a similar phe-
tion, “When do voters punish incumbents involved in
nomenon takes place with political scandals.
political scandals?” Previous research has shown that
scandals harm the electoral prospects of Congress mem-
1Washington University in St. Louis, MO, USA
bers (Carlson, Ganiel, and Hyde 2000; Doherty, Dowling,
and Miller 2011; Funk 1996; Groseclose and Krehbiel
Corresponding Author:
1994; Jacobson and Dimock 1994; Peters and Welch
Miguel M. Pereira, Department of Political Science, Washington
University in St. Louis, Campus Box 1063, One Brookings Drive, St.
1980; Welch and Hibbing 1997). In a recent contribution,
Louis, MO 63130, USA.
Basinger (2013) shows that in the last four decades,
Email: m.pereira@wustl.edu

---

Pereira and Waterbury
585
However, we expect that biases in media coverage
varying levels of media attention, the analyses reveal that
shape how voters discount different types of scandals.
accessibility biases (Iyengar 1990) explain—at least par-
News outlets are the main channel through which voters
tially—why voters discount events that took place earlier
acquire political information (Arceneaux and Johnson
in the term. This result reveals the advantages of moving
2013; Prior 2007), and scandal coverage is not uniform
beyond economic assessments when studying myopic
(Nyhan 2015, 2017). A growing interest in covering pri-
voting.
vate misbehavior over the last fifty years has led media
The patterns uncovered in this study have relevant
outlets to devote disproportionate levels of attention to
normative implications. On one hand, our study provides
moral scandals, such as extramarital affairs (Sabato 1991;
a pessimistic view on the prospects of democratic
Streitmatter 2004). This trend toward the personalization
accountability. The myopic nature of voters’ judgments is
of politics results both from the commercial consider-
not specific to policy evaluations or economic perfor-
ations of media groups and the strategic behavior of pub-
mance. Instead, voters also discount information that is
lic officials who use their private life as an electoral asset
largely time invariant: elicited personal traits of represen-
(Holtz-Bacha 2004; McAllister 2007). Larger volumes of
tatives and their ethical conduct in office. In that sense,
media coverage, in turn, make information more easily
scandal discounting represents an acute version of myo-
retrieved from memory and ultimately more salient in
pic voting. On the other hand, the differential effects of
voters’ judgments (Iyengar 1990). As a consequence, we
moral violations suggest a path to overcome scandal dis-
expect moral scandals to have more persistent electoral
counting. If the persistent electoral penalty paid by repre-
consequences than other types of scandals.
sentatives involved in sex scandals is due to higher levels
To test these arguments, we build on Basinger’s (2013)
of media attention, as the analyses suggest, a more bal-
compilation of congressional scandals. The original data-
anced treatment of other public misconduct cases may
base includes 246 scandals involving members of the
enhance electoral accountability.
House of Representatives, between 1972 and 2010. We
In the following section, we review existing research
expanded this dataset to account for the role of time and
on the electoral consequences of political scandals. Next,
explore how the effects of timing vary by the type of
we account for the role of time and introduce our main
scandal. Importantly, the analyses focus on the timing of
arguments, building on the myopic voting literature. We
scandal outbreaks. We are interested in learning about
then test our expectations with newly collected data.
voters’ response to a revelation that occurs at different
Next, we present an analysis of timing decomposed by
stages of a congressional term, regardless of when the
type of scandal. A final empirical section offers support
event actually took place.2
for the mechanism of differential discounting. The paper
The analyses show that the effects of scandals quickly
closes with some considerations regarding the normative
fade over time. For members of the House who remained
consequences of scandal discounting.
in office, only scandals emerging within twelve months
of an upcoming election significantly decrease their elec-
The Electoral Consequences of
toral prospects. This pattern is not due to strategic retire-
Political Scandals
ment. Moreover, we show that the effects of timing are
not equal across types of scandals. While financial and
Existing research has established that politicians involved
corruption scandals have short-lived effects, cases involv-
in scandals are punished at the polls (Abramowitz 1988,
ing sexual misconduct resonate for longer time periods.3
1991; Carlson, Ganiel, and Hyde 2000; Doherty, Dowling,
Additional analyses based on Google Trends data sug-
and Miller 2011; Funk 1996; Groseclose and Krehbiel
gest that variation in scandal discounting rates is due to
1994; Jacobson and Dimock 1994; Peters and Welch
unequal levels of media attention. By comparing same-
1980; Welch and Hibbing 1997). Scandal-ridden politi-
state legislators involved in different types of scandals in
cians have a harder time raising campaign funds
the same term, we show that sex scandals systematically
(Rottinghaus 2014), face higher quality challengers
attract more online media coverage than any other form
(Lazarus 2008), and tend to win a smaller vote share in
of public misconduct. This result provides a plausible
their next election (Basinger 2013; Brown 2006).
explanation for why moral scandals...