Accountability, Framing Effects, and Risk-Seeking by Elected Representatives: An Experimental Study with American Local Politicians

• Author: Peter John Loewen,Lior Sheffer
• Published date: 01 March 2019
• DOI: 10.1177/1065912918775252
• Date: 01 March 2019
• Subject Matter: Articles

775252PRQXXX10.1177/1065912918775252Political Research QuarterlySheffer and Loewen
research-article2018
Article
Political Research Quarterly
2019, Vol. 72(1) 49 –62
Accountability, Framing Effects, and
© 2018 University of Utah
Article reuse guidelines:
sagepub.com/journals-permissions
Risk-Seeking by Elected Representatives: https://doi.org/10.1177/1065912918775252
DOI: 10.1177/1065912918775252
journals.sagepub.com/home/prq
An Experimental Study with American
Local Politicians
Lior Sheffer1 and Peter John Loewen2
Abstract
Risk management underlies almost every aspect of elite politics. However, due to the difficulty of administering
assessment tasks to elites, direct evidence on the risk preferences of elected politicians scarcely exists. As a result,
we do not know how consistent are politicians’ risk preferences, and under what conditions they can be changed.
In this paper, we conduct a survey experiment with 440 incumbent local politicians from across the United States.
Using a modified version of the Asian Disease framing experiment, we show that gain/loss frames alter the stated risk
preferences of elected officials. We further show that priming democratic accountability increases the tendency to
engage in risky behavior, but that this shift in preference only occurs in those politicians who are interested in seeking
reelection. These results inform several political science theories that assume stable risk preferences by political
elites, or that make no risk assumptions whatsoever. They also provide insights into the role of political ambition and
accountability in structuring the behavior of political elites.
Keywords
experiments, accountability, risk, framing, elites, progressive ambition
Introduction
2013; McDermott, Fowler, and Smirnov 2008). Moreover,
we do not know well how elements inherent to political
Virtually all political decisions involve some degree of risk
life—such as operating under public scrutiny and having
and uncertainty. As such, the risk preferences of elite deci-
to stand for reelection—impact risk-seeking by politi-
sion-makers are important theoretical determinants in
cians. This is of particular importance not only because it
models that explain political outcomes, such as govern-
means we are lacking insight into how political account-
ment spending, adoption of public policies, violent conflict
ability impacts officeholders, but also because evidence
escalation and resolution, international negotiations, and
from other domains suggests that accountability results in
elections and campaigning (Druckman and McDermott
marked preference change by amplifying certain choice
2008; Kam 2012; Levy 1997; McDermott 2001; see
biases and heuristics (Lerner and Tetlock 1999, 2003).
McDermott, Fowler, and Smirnov 2008 for an extensive
Absent direct evidence, most theories of elite choice do
review).1 Furthermore, the unique features of the environ-
not thoroughly consider the mutability of politicians’ risk
ment in which politicians make decisions, and the system-
preferences. Instead, representatives are frequently
atically different personality profiles of those who select
into elected positions, suggest that the way politicians
assess risky choices may be substantially different from
1Department of Political Science, University of Toronto, Ontario,
what we know about the behavior of nonelites.
Canada
2
Despite the consequentiality and centrality of risk in
School of Public Policy and Governance and Department of Political
Science, University of Toronto, Ontario, Canada
political decision-making, little evidence exists on the
risk preferences of politicians,2 and, in particular, there is
Corresponding Author:
no direct evidence on the degree to which political incum-
Lior Sheffer, Department of Political Science, University of Toronto,
Sidney Smith Hall, 100 St. George Street, Toronto, Ontario, Canada
bents are able to resist classic framing effects when mak-
M5S 3G3.
ing risky choices (Hafner-Burton, Hughes, and Victor
Email: lior.sheffer@utoronto.ca

---

50
Political Research Quarterly 72(1)
conceptualized as having stable decision-making styles,
determinants of political elites’ decision-making has been
with predictable—and often different—risk preferences:
relatively scarce.4 In recent years, however, scholars have
depending on the study, they are either risk-favoring
called for a more rigorous investigation of politicians’
(Downs and Rocke 1994; Taliaferro 2004), risk-averse
behavioral traits, emphasizing the need to establish if and
(Hood 2002; Weaver 1986), or a mix of these two types
how known findings on biases and heuristics from psy-
(Cox and McCubbins 1986; see also Mercer 2005, 4–5 for
chology apply to elite decision-makers—and how they
additional examples). Neither of these, however, is con-
impact policy outcomes (Druckman and Lupia 2012;
gruent with evidence from psychology on the malleable
Hafner-Burton, Hughes, and Victor 2013).
nature of individuals’ risk calculi (McDermott, Fowler,
A growing number of studies directly address this
and Smirnov 2008). As far as we can tell, two main chal-
empirical gap by recruiting political elites and having
lenges account for the absence of evidence on the risk
them take part in controlled experiments or behavioral
preferences of elected politicians: first, the difficulty in
tests.5 For example, recent efforts have been made to
administering established experimental protocols to suffi-
obtain valid measures of politicians’ personality profiles
ciently large numbers of incumbent politicians (Druckman
using the Big Five personality traits in the United States
and Lupia 2012), and, second, the attendant difficulty of
(Dietrich et al. 2012), Germany (Best 2011), and Italy
inferring risk preferences from observational data.
(Caprara et al. 2010). Other studies have attempted to
Addressing this empirical gap, in this paper, we pro-
assess to what degree political elites have different levels
vide direct evidence on the stability of elected officials’
of power motivation (Renshon 2015), stress (Sherman et
preferences when making risky policy choices under dif-
al. 2012), overconfidence (Johnson et al. 2006), self-
ferent conditions. We use a public choice task derived
interest and capacity for strategic inference (LeVeck et al.
from the classic Mysterious Disease experiment first
2014), and cooperative behavior and reciprocity in public
introduced by Tversky and Kahneman (1981) (originally
goods provision (Enemark et al. 2016). Assessing risky
called “Asian Disease”), which we embedded in a large-
alternatives and choosing between them is inherent to
scale survey of incumbent municipal politicians in the
almost every aspect of political life, including voting,
United States.3 We find that when reasoning through con-
collective action, and political mobilization (Druckman
sequential public problems, these elected officials’ risk
and McDermott 2008; Eckles et al. 2014; Kam 2012). In
preferences are strongly conditioned by whether choices
theories of elite decision-making in politics, elected poli-
are framed in terms of gains or losses. We also demon-
ticians’ preferences for risk-taking frequently appear as a
strate that priming the public nature of the task at hand
determinant cause or mediating factor in explanations of
and its potential electoral consequences—that is, increas-
outcomes, such as initiation of wars and foreign military
ing implied accountability for the decision—results in
interventions (Downs and Rocke 1994; Taliaferro 2004),
significantly more risk-taking. Furthermore, this effect
escalating commitment to failed financial courses of
only obtains for incumbents who intend to seek reelec-
action (Dur 2001), and the use of blame avoidance tactics
tion. Our findings, thus, provide important insights into
in face of scandals (Dewan and Myatt 2007; Hood 2002).
the formation of policy preferences among political elites.
In some of these cases, politicians are depicted as having
a constant, often unspecified, base level of risk affinity or
Theoretical Background
aversion, which increases or decreases the likelihood of
adopting certain choices (Fearon 1995). In other accounts,
The personality characteristics and cognitive styles of
politicians’ preference for risk is modified by situational
political actors are central to the study of political choice.
determinants such as elections, economic conditions,
Attention deficits, trust dispositions, and susceptibility to
military success/failure, and media attention (see Levy
various behavioral anomalies, such as hyperbolic dis-
2003; Mercer 2005 for reviews).
counting and framing effects, are used to explain out-
In addition to risk-taking preferences being central to
comes as varied as public opinion formation, vote choice,
politics, there is also evidence that politicians manage
and government spending patterns (Druckman and Lupia
risk differently than nonpoliticians. As Hafner-Burton,
2000; Jones and Baumgartner 2005; Zaller 1992). Such
Hughes, and Victor (2013, 370) suggest, in politics,
insights are frequently ascribed to regular citizens in their
“experienced elites might manage . . . risks differently
capacity as voters, political opinion holders, and eco-
from decision makers selected randomly.” We identify
nomic decision-makers.
four broad categories of factors that may influence...