No Balance, No Problem: Evidence of Partisan Voting in the 2021 Georgia U.S. Senate Runoffs
| Author | Isaac Hale,Carlos Algara,Cory L. Struthers |
| DOI | 10.1177/1532673X211070819 |
| Published date | 01 July 2022 |
| Date | 01 July 2022 |
| Subject Matter | Articles |
Article
American Politics Research
2022, Vol. 50(4) 443–463
© The Author(s) 2022
Article reuse guidelines:
sagepub.com/journals-permissions
DOI: 10.1177/1532673X211070819
journals.sagepub.com/home/apr
No Balance, No Problem: Evidence of
Partisan Voting in the 2021 Georgia U.S.
Senate Runoffs
Carlos Algara
1
, Isaac Hale
2
, and Cory L. Struthers
3
Abstract
Recent work on American presidential elections suggests that voters engage in anticipatory balancing, which occurs when
voters split their ticket in order to moderate collective policy outcomes by forcing agreement among institutions controlled by
opposing parties. We use the 2021 Georgia U.S. Senate runoffs, which determined whether Democrats would have unified
control of the federal government given preceding November victories by President-elect Biden and House Democrats, to
evaluate support for anticipatory balancing. Leveraging an original survey of Georgia voters, we find no evidence of balancing
within the general electorate and among partisans across differing model specifications. We use qualitative content analysis of
voter electoral runoff intentions to support our findings and contextualize the lack of evidence for balancing withan original
analysis showing the unprecedented partisan nature of contemporary Senate elections since direct-election began in 1914.
Keywords
electoral balancing, policy balancing, voter electoral knowledge, partisanship, nationalized elections
Introduction
Split-ticket voting and its causes have been a topic of intense
debate in American politics studies. Split-ticket voting occurs
when a voter casts ballotsfor candidates of different partiesfor
multiple offices. The identified causes of split-ticket voting in
the U.S. include weak mass partisanship (Beck et al., 1992),
the strength of incumbency (Burden & Kimball, 2009),
ideologically indistinct candidates (Burden & Kimball, 2009),
ambivalence of voters towards the parties (Mulligan, 2011),
and “policy balancing”(Alesina et al., 1991;Fiorina, 1992).
However, while split-ticket voting was prevalent in post-
WWII congressional elections (Alesina et al., 1991;Fiorina,
1992), there has been a sharp drop in split-ticket voting in
recent years (Jacobson, 2017). Voters have become more
partisan (Bartels, 2000;Fiorina & Abrams, 2008;Davis &
Mason., 2016;Burden & Kimball, 2009;Abramowitz &
Webster, 2015), the incumbency advantage has decreased
(Jacobson, 2015), candidates are increasingly ideologically
distinct (e.g., Hetherington, 2001;McCarty et al., 2006), and
voters are seldom ambivalent between the parties (Smidt, 2017).
Despite decreases in split-ticket voting, recent studies have found
that “policy balancing”—which occurs when voters split their
ticket in an intentional effort to moderate future legislative policy
decisions through divided government (Fiorina, 1992;Carsey &
Layman, 2004;Lewis-beck & Nadeau, 2004;Lacy et al., 2019)—
persists (e.g., Bafumi et al., 2010;Bailey & Fullmer, 2011;
Erikson, 2016). These findings are striking given increasing levels
of straight-ticket voting. An assessment of American National
Election Study survey data since 1952 shows that the 2012, 2016,
and 2020 elections featured levels of straight-ticket voting of
90%, 85%, and 89%, respectively (compared to 70-85% from
1972–2008), which produced the greatest congruence between
House district-level presidential vote choice and the partisanship
of the legislators elected to represent voters in that district in both
the House and Senate since 1900 (Algara & Johnston, 2021).
1
In
other words, not only has straight-ticket voting increased to levels
approaching 90%, but over 90% of legislators hail from
constituencies carried by their party’s presidential nominee in the
previous election.
In this study, we use data from an original survey to
analyze whether voters in Georgia’s 2021 U.S. Senate
election runoffs cast their ballots for Republican candidates
David Perdue and Kelly Loeffler in order to balance against
1
Department of Politics & Government, Claremont Graduate University,
Claremont, CA, USA
2
Department of Politics, Occidental College, Los Angeles, CA, USA
3
Department of Public Administration and Policy, University of Georgia,
Athens, GA, USA
These authors contributed equally to this work.
Corresponding Author:
Carlos Algara, Department of Politics & Government, Claremont
Graduate University, Claremont, CA, USA.
Email: carlos.algara@cgu.edu
the incoming Democratic president and House majority, and
further evaluate whether sophisticated voters were more
likely to engage in such behavior. A central critique of an-
ticipatory policy balancing theory is that it requires voters to
engage in excessive high-level reasoning; that is, voters
seeking to balance policy through divided government in a
presidential election must forecast which party will likely
control the House and the Senate in order to determine how to
cast their vote for the White House. Our study overcomes this
challenge by using a most-likely case of policy balancing
where voters had perfect information about how their votes
would affect the probability of divided government. We find
no evidence of policy balancing, even under these ideal
conditions.
Theoretical Framework
Traditional Theories of “Balancing”of Partisan
Institutional Control
A long debate in American politics concerns whether some
voters engage in “policy balancing”and whether such be-
havior can influence election outcomes. At the core of the
policy balancing model is the assumption that voters abandon
sincere preferences and cast a split-ticket to strategically
“balance”extreme partisan positions through divided gov-
ernment or split delegations (e.g., Fiorina, 1988,1989,1992;
Alesina & Rosenthal, 1995;Lacy & Paolino, 1998;Kedar,
2009), offsetting an ideologue’s position “by choosing an
opposite ideologue”(Alesina et al., 1991, p. 29). Notably,
balancing voters are not attempting to achieve more moderate
policy by voting for more moderate candidates. In a given
contest, a balancing voter may actually select a candidate
further from their ideological ideal point. The policy-
moderating goal of the balancing voter is achieved by gen-
erating additional intra-chamber partisan parity or by creating
divided government—forcing ideologues to compromise or
forcing gridlock that preserves the policy status quo.
Policy balancing voters have “nonseparable”preferences,
wherein their preference for a candidate of either party de-
pends on either the extant partisan distribution of institutional
control in the expected outcome (Lewis-beck & Nadeau,
2004;Smith et al., 1999). To moderate expected policy
outputs, voters with an ideological position between the two
parties will, for instance, vote Republican in a midterm house
election if the presidency is held by the Democratic Party.
Such behavior is distinct from behavior based on “separable”
preferences, where voters cast straight tickets with little re-
gard to whether their vote leads to unified or divided gov-
ernment (Lewis-beck & Nadeau, 2004;Smith et al., 1999).
Policy balancing theory has largely been developed across
three electoral contexts: midterm elections (“retrospective
balancing”), Senate elections (a version of retrospective
balancing), and presidential elections (“anticipatory balanc-
ing”). In each context, scholars have investigated whether
policy balancing explains split-ticket voting. Midterm and
Senate election contexts can be considered “retrospective
balancing”because the party being balanced against is known
and fixed (i.e., the presidency or second Senate seat, re-
spectively, is not open for re-election), allowing voters to cast
their ballot according to the observed policy positions of the
standing politician. Presidential elections, on the other hand,
are “anticipatory”. Potential “policy-balancing”voters have
no fixed evaluation point for congressional or presidential
party control and must split their ticket according to per-
ceptions of likely outcomes across all three institutions.
A key difference between retrospective and anticipatory
balancing models is the information available to voters and,
relatedly, the sophistication necessary to engage in balancing
behavior. In retrospective contexts, policy balancing voters
cast their ballot for the “out”party to moderate the ideological
position of the politician not up for re-election (Alesina et al.,
1991;Alesina & Rosenthal, 1995;Fiorina, 1992). All voters
need to know is the party of that politician and vote for the
alternative party (Butler & Butler, 2006;Heckelman, 2000).
Yet anticipatory balancing puts additional cognitive demands
on the balancing voter (Segura & Nicholson, 1995;Erikson,
2016). Here, voters must simultaneously evaluate the prob-
ability that the Republican or Democratic Party wins the
presidency and House in order to condition their vote for the
Senate (Alesina & Rosenthal, 1995;Mebane, 2000;Scheve &
Tomz, 1989).
Despite the many classic studies (Fiorina, 1988,1996) that
find evidence of policy balancing (both retrospective and
anticipatory), many subsequent studies have found no such
support. Instead, many have argued that American voters
have neither the interest nor the sophistication necessary to
engage in this kind of strategic behavior. Voters are instead
increasingly relying on partisan cues when choosing how to
vote (Alvarez & Schousen, 1993;Born, 1994;Burden &
Kimball, 1998,2002;Sigelman et al., 1997;Segura &
Nicholson, 1995). Scholars note that partisanship is a
stronger voting heuristic in contemporary congressional
elections (Hetherington, 2001), as increased polarization
between the parties helps clarify the link between individual
candidates and their parties (Stone, 2017).
2
Even when voters
do not fully rely on partisan cues and ultimately do engage in
split-ticket voting (which has sharply decreased in recent
decades, see Figure 10), these authors have argued that other
factors such as candidate resources (Segura & Nicholson,
1995) and incumbency differentials (Jacobson, 2015;Stone,
2017) explain this behavior—not policy balancing.
3
However, some recent work—using a variety of mea-
surement approaches—has found evidence of balancing even
in the more cognitively challenging anticipatory settings.
Using spatial proximity scores, Lewis-beck & Nadeau (2004)
show that a quarter of the electorate in the 1992 and 1996
presidential elections were voters who prefer divided gov-
ernment and split their ticket. Saunders et al. (2005) assemble
a series of left-right policy scales to measure a respondent’s
444 American Politics Research 50(4)
To continue reading
Request your trial