Electoral Studies 40 (2015) 158e169
Contents lists available at ScienceDirect
Electoral Studies
journal homepage: www.elsevier.com/locate/electstud
Regional variation in voter turnout in Europe: The impact of
corruption perceptions
€ m a, *, Daniel Stockemer b
Aksel Sundstro
a
b
Department of Political Science, The Quality of Government Institute, University of Gothenburg, Box 100, S-405 30 Gothenburg, Sweden
School of Political Studies, The University of Ottawa, 75 Laurier Avenue, East Ottawa, ON K1N 6N5, Canada
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 3 December 2013
Received in revised form
2 July 2015
Accepted 18 August 2015
Available online 21 August 2015
While much of the literature on voter turnout focuses on institutional and socioeconomic factors related
to the ‘input’ side of the political process, we examine the ‘output’ side, and advance this field of research
by studying the impact of corruption perceptions on turnout in the most recent national legislative
elections across 170 European regions. Using data from a novel measure of regional perceived corruption
of government services and the electoral process, together with several control variables, we find
through multilevel modeling that regional quality of government positively impacts regional turnout. In
more detail, our results indicate that citizens' perceptions of corruption make them, in the aggregate, less
likely to cast a ballot.
© 2015 Elsevier Ltd. All rights reserved.
Keywords:
Voter turnout
Corruption perceptions
Subnational variation
Europe
Political participation
1. Introduction
After 50 years of debate (see, among others, Huntington, 1968;
Lipset and Lenz, 2002; Rose-Ackerman, 2008), there is now some
strong consensus in the literature that bad governance practices in
general, and corruption in particular, are plagues that must be
combated. Bribery and corrupt governments have a corrosive
impact on a wide range of societal features, such as economic
growth (Mauro, 1995), interpersonal trust (Rothstein and Uslaner,
2005), and infant mortality rates (Gupta et al., 2000). Yet, it is
currently less clear whether citizens' perceptions of corruption in
the state apparatus, more generally, and in the electoral process,
more precisely, impact turnout rates in national legislative
elections.
The growing literature on corruption perceptions and voter
turnout has presented some conflicting theoretical expectations
and empirical findings. On the one hand, the majority view contends that corrupt governance decreases voter turnout as citizens
distrust the political system when bribery is widespread
(Kostadinova, 2009; Stockemer et al., 2013). On the other hand, the
minority view states that corruption increases electoral
* Corresponding author.
€m).
E-mail address: [email protected] (A. Sundstro
http://dx.doi.org/10.1016/j.electstud.2015.08.006
0261-3794/© 2015 Elsevier Ltd. All rights reserved.
participation because politicians can ‘buy’ votes by granting porkbarrel spending and favoring specific constituents (Karahan et al.,
2006; Escaleras et al., 2012). Moreover, some studies have found
mixed support, in which the relationship between corruption and
turnout is contingent upon how these concepts are operationalized
and measured (Stockemer, 2013; Birch, 2010).
Most studies in this literature have been conducted on the national level, thus overlooking significant variation in both corruption and turnout within countries. For example, differences in
regional perceived corruption are sometimes more pronounced
within regions of the same country than between countries (e.g.
France's regions are roughly situated between the fiftieth and
eightieth percentiles in their relative corruption level across the
approximately 170 regions studied in this article). There are hence
theoretical reasons to believe that there are important differences
within nations in this regard that hardly are captured by crosscountry indicators of corruption. Similarly, regional turnout rates
frequently vacillate considerably within the same country (e.g. on
average, there is a more than 10 percentage point gap within the
same country and this gap sometimes reaches 17 or 18 percent,
such as in Portugal and Austria, respectively). Only a more finegrained comparative analysis at the regional level can capture
these differences (Rydgren, 2007).
Using data from a novel survey on citizens' perceptions of
quality of government in 170 regions across 18 European countries
€m, D. Stockemer / Electoral Studies 40 (2015) 158e169
A. Sundstro
we study the impact of perceived corruption on voter turnout in
these regions' most recent national legislative elections. Using a
Hierarchical Linear Model (HLM) and controlling for five nationallevel factors (compulsory voting, electoral system type, presidentialism, official voting age, and years of EU membership) and
four regional-level socioeconomic and political factors (economic
development, economic growth, population density, and the
closeness of these election races), we find that differences in
perceived regional corruption levels have a significant and substantial influence on voter turnout: Corruption perceptions suppresses citizens' electoral participation.
This article proceeds as follows: We first place our study within
the scarce, albeit growing, literature that discusses the link between corruption in government on the one hand and aggregate
voter turnout on the other. After describing the design and objective of our study, we present the scope of our analysis, the variables,
and our methods. We then report our results, before we conclude
by discussing the relevance of our findings and suggesting avenues
for future research.
2. Previous literature
Alongside political and individual rights, free and fair elections
are the main characteristic of a democracy (Huntington, 1991).
Through elections, citizens can choose their representatives, control their governments, and make their preferences heard. In
addition, (high) electoral participation can be a healthy repertoire
of contention and a vent for citizens to express their discontent
with the performance of politicians (Powell, 1982; Putnam et al.,
1983). In contrast, ‘low voter turnout can act as a signal that
something is wrong e not with the voters who fail to turnout or
with the society of which they are part, but with a political system’
(Franklin, 2004, 219). Given its primary importance, it is not surprising that scholars of political science (for an overview, see for
example, Geys, 2006) have devoted considerable attention to
studying variances in voter turnout in electoral democracies across
time and space.
Previous studies have focused mainly on institutional factors
(e.g. compulsory voting or the electoral system type), which are
frequently termed the ‘input’ side of the democratic process, to
explain variation in voter turnout between and, to a lesser degree,
within countries (Blais, 2006). There are fewer studies that discuss
the ‘output’ side of the political system, such as the influence of the
quality of public services and the trustworthiness, honesty, and
professionalism of the public employees on citizens' willingness to
participate in the electoral process (Rothstein and Solevid, 2013, 2).
In fact, only a limited number of studies have focused on citizens'
perceptions of corruption when exploring between-country differences in voter turnout.
The few existing studies discussing the link between perceived
corruption in government and voter turnout germinate from the
literature that evaluates the link between perceptions of democracy
or the quality of institutions and voter turnout (Abramson and
Aldrich, 1982; Anderson and Guillory, 1997; Anderson and
€nlund and Seta
€la
€, 2007;
Tverdova, 2003; Clarke et al., 2004; Gro
Karp and Banducci, 2008). Ezrow and Xezonakis (2014), for
example, present the two countervailing views regarding the influence of citizens' satisfaction with democracy on electoral
participation. They describe the first perspective as follows: ‘citizens who are more satisfied with democracy tend to be more
politically engaged, and thus they are more likely to turnout to vote’
(2014, 1). The counter perspective proposes that the electorate
should participate more in the political process if satisfaction with
democracy is lower. In Ezrow and Xezonakis (2014, 2) words,
‘dissatisfaction generates demand for change in the electorate,
159
which in turn mobilizes citizens to engage in, amongst other forms
of participation, voting’.
So what happens if the government or elections are perceived as
corrupt? Does the resulting dissatisfaction with the government's
performance trigger increased mobilization to oust corrupt leaders
and re-implement transparency, or does it make citizens discouraged and apathetic about the political system? The literature
measuring the link between perceived corruption and electoral
turnout is not unanimous on that point. The minority view sees
perceived corruption as a mobilizing agent for increased citizen's
involvement and proposes three rather different causal mechanisms to support this view. The first mechanism advances the
simple argument that perceived corruption in the public is likely to
enrage citizens, who should then express their dissatisfaction on
Election Day by casting a ballot for a non-corrupt politician
(Kostadinova, 2009). Bauhr and Grimes (2014) describe this as a
process of indignation. In contrast, Escaleras et al. (2012) make the
argument that in corrupt and clientelistic societies, voters can be
bought off. According to this train of thought, ‘corruption increases
the value of office-holding, it also increases the demand for votes
on the part of incumbents wanting to retain their profitable positions of political power … this produces more electoral effort and
higher voter turnout in corrupt than in non-corrupt countries’
(Karahan et al., 2006, 88).
The third argument is rather indirect and suggests that once
corruption is contained in the eyes of the voters, voter turnout
should be lower. For example, explaining voter turnout in the
second half of the nineteenth century in the US, Heckelman (1995)
argues that bribery was a common method of harvesting votes
prior to the reform of secret voting, which most US states adopted
in the 1880s. The author finds that after this reform, the direct
importance of bribery diminished. More precisely, Heckelman
(1995, 107) concludes that political parties stopped offering payments in return for votes after they were ‘no longer able to verify
the voters’ choices'. Hence, a reform that reduced corrupt practices
indirectly resulted in a decreased turnout. Schaffer (2002, 2007)
further explores the finding that a ‘clean up’ of the election process ought to keep people away from the polls. Analyzing the
processes through which reduced corruption e and hence the
reduced possibilities for vote buying e affects turnout, he concludes that ‘lowered turnout in more transparent and better performing systems may thus be either an unintended consequence or
an intended effect of clean election reforms’ (2002, 80).
By far, the majority of studies adhere to the view that corruption
is a plague that directly diminishes citizens' participation in elections. As the pioneer in this discussion, Putnam (1993) states that
high electoral turnout requires elected decision-makers that are
trustworthy and act in accordance to the will of their constituency.
Drawing on Putnam, a number of studies spell out the reasoning
behind why corruption perceptions are likely to have a negative
influence on turnout. For example, Inman and Andrews (2010)
argue that ‘citizens who interpret corruption as an indication that
they cannot trust their leaders may be less likely to participate
because corruption signals to them that the government will not
respond to their concerns’ (7). Stockemer et al. (2013) outline a
similar argument: ‘if politicians at various levels of government
cannot be trusted, then citizens have no interest to interact with
them; a feature which should further decrease individuals’ motivation to vote for any of the existing corrupt parties or politicians'
(4). Similarly, Chong et al. (2013) argue that ‘if people vote, not for
personal benefit and not because of the possibility of being the
marginal voter, but rather out of utility garnered through fulfillment of civic duty, then information about a corrupt political process may lower the value of that civic duty for the individual, and
thus lower voter turnout’ (2e3).
160
€m, D. Stockemer / Electoral Studies 40 (2015) 158e169
A. Sundstro
In addition, corruption and bribery are seldom only present in
the realm of elected officials e they also infest government administrations and the work of public officials (e.g. Rothstein and
Teorell, 2008). In this sense, citizens in a corrupt context may
realize that while their vote could possibly remove elected leaders
perceived as corrupt, they might not change the behavior of corrupt
bureaucrats by attending the electoral booth. According to
Kostadinova (2009) this feeling that corruption is ‘sticky’ and
cannot be easily removed from the democratic process undermines
the faith of voters in the democratic process, and consequently,
weakens their desire to participate in politics' (693). In Bauhr and
Grimes (2014) terminology, increased perceived political corruption in the political and social realm could even lead to a process of
resignation in the population. Finally, perceived corruption in government might also have a contagious impact on individual
behavior. If citizens perceive administrations to be corrupt, they
might lose the scruple of bribery or fraudulent activities and engage
in these activities themselves. In other words, if individuals feel
that it is acceptable for public officials to play unfairly, they might
also be more inclined to bend the rules themselves (Rothstein and
Eek, 2009). Hence, perceptions of corruption in the bureaucracy
might spread to behavior in the general population e risking the
reduction of interpersonal trust, which, in turn, could lead to
decreased participation in elections (Rothstein and Uslaner, 2005,
41).
Several of the existing empirical studies (Hellman et al., 2000;
Tucker, 2007) find that widespread citizenry perceptions of
governmental corruption lead to fewer citizens turning out on
Election Day. For example, McCann and Dominguez (1998, 499)
report that ‘perceptions of electoral fraud affects the likelihood of
voting: the greater the expectation of fraud, the lower the likelihood of voting’. Similarly, Simpser (2005) finds perceived corruption to be negatively related to turnout in a large-N study
comprising of both autocracies and democracies. Likewise,
Stockemer et al. (2013) find in their study of a large number of
national legislative elections in electoral democracies that as corruption increases in the eyes of the citizenry, the percentage of
voters who go to the polls decreases.
Other studies find some more nuanced results. For example,
Birch (2010), using data from the Comparative Study of Electoral
Systems for a large number of new and established democracies,
finds that perceptions of governments as unresponsive, ineffective,
and corrupt negatively affect the likelihood that voters go to the
polls. However, Birch (2010, 1607) adds to this general finding that
‘actual electoral misconduct will in some cases increase rates of
electoral participation, but perceptions of malpractice will have the
countervailing effect of decreasing it’. Equally detailed, Stockemer
(2013) e using three corruption measures (i.e. the International
Country Risk Guide (ICRG) measure and the measures from the
World Bank and Transparency International) across a large sample
of presidential regimes e finds that only the ICRG corruption
measure has the expected negative influence on voter turnout. The
other two measures are unrelated to macro-level turnout.
Kostadinova (2009) also reports mixed results based on a sample of
post-communist countries. In her study, she finds that corruption's
mobilizing effect is slightly weaker than its deterring effect on voter
turnout (see also Kostadinova, 2003).1
In addition to these studies with mixed results, there are three
regional- and district-level studies that find empirical support for
the minority view, that is, that higher corruption leads to higher
turnout. For example, Escaleras et al. (2012) find that governmental
1
Using a similar sample of post-communist countries, Pacek et al. (2009) find no
correlation between higher corruption and decreased citizen engagement.
corruption raises electoral participation in gubernatorial elections
in the US between 1979 and 2005 e a finding that was advanced in
the early 1980s by Johnston (1983). In addition, Stockemer and
Calca (2013a) develop the argument that while nationally high
corruption leads to low average turnout, the subnational relationship is reversed; that is, the authors find in their case study on
Portugal that districts with higher corruption actually have higher
turnout. While limited in number, the few studies with a subnational focus entertain the possibility that relationships at the national and subnational level are not identical. To highlight, it is
theoretically possible that overall corruption levels decrease electoral participation, but in the most corrupt parts within countries
citizens might actually rush to the polls. For example, perceived
corruption levels might have varied impacts on citizens in different
regions, because investment projects might get delayed or there
could be inefficiencies that damage government functioning.
In sum, the empirical national-level evidence points toward a
negative impact from corruption on turnout, whereas at the
regional level this relationship is questioned. However, regionallevel studies have been few in number and restricted to single
countries. There is also a difference between the few regional-level
studies and the national-level analyses, in that all cross-nationallevel studies use data on perceived corruption, whereas the few
studies with a regional focus (i.e. Escaleras et al., 2012; Stockemer
and Calca, 2013a) use measures that capture the number of individual officials indicted because of corruption charges. These two
types of measures are not necessarily comparable, as perceptions of
corruption might actually differ from publicly known instances of
corruption. We find it plausible that voters' perceptions of corruption are more important than the public knowing the number of
people facing corruption charges. Citizens make up their mind on
whether to participate or stay at home on Election Day, based on
their view of how honest, trustworthy, and transparent the electoral process and the government are. This applies even more so, if
we consider that figures on officials with corruption charges might
be quite difficult to interpret. For example, it is unclear whether a
high number of indicted public officials is a sign that the system
works well in fighting corruption, or whether it indicates that the
system is truly corrupted.
3. Scope of the analysis
In the ensuing analysis, we combine the small but growing
scholarly focus on corruption perceptions as an explanation for
electoral participation with the literature focusing on regional differences in turnout. Examining the relationship between perceived
corruption and electoral turnout on the regional level in a
comparative perspective has several advantages. First it allows us to
test two countervailing arguments: (i) It might be the case that
while overall high levels of perceived corruption might decrease
turnout, citizens in more corrupt areas might actually rush to the
polls, because the perceived high corruption in their area might
trigger a comparative disadvantage for their region (e.g. investments might go elsewhere). (ii) It may also be the case that
individuals become particularly disillusioned when they perceive
their local administration and state apparatus to be nontransparent
and corrupted and therefore abstain from voting.
Second and more theoretically, it makes sense to study the
relationship between perceived corruption and turnout at the
subnational level. We believe that citizens' perceptions of the level
of corruption in government are in many ways a product of
everyday life experiences that for most people take place right
where they live, in their town or region. As Lipsky (1980) argues,
citizens most often come into contact with street-level bureaucrats
in their dealings with agents of the state. Consequently, citizens in
€m, D. Stockemer / Electoral Studies 40 (2015) 158e169
A. Sundstro
regions where corruption in government is perceived to be widespread will be more likely to be confronted with officials asking for
bribes than citizens in regions with lower levels of perceived corruption. This reasoning directly implies that citizens living in highly
corrupt regions are more prone to lose their faith in the public
administration and the democratic process as compared to citizens
in less corrupt regions of the same country. To illustrate, this would
imply that citizens in southern Italy, for instance, would have a
different set of experiences in regards to corruption among local
civil servants than compared to citizens in the northern part of this
country e differences that may translate to varying willingness to
participate in elections.
Third, analyzing the influence of perceived corruption on
turnout at the regional level allows us to not only measure corruption where it matters the most e in the environment closest to
individuals e it also allows for a more fine-grained turnout model,
more generally. For instance, turnout rates within European
countries in this study vary significantly (often by more than 15
percentage points). The same applies to differences in perceived
corruption in government, which change by more than 30 percent
for regions within one country (see also Tabellini, 2005). In addition, some of the components of electoral turnout, including population density and economic growth, differ more strongly within
than between countries. For example, for the former indicator,
population density, differences between regions are often 10 times
larger than differences between countries. More theoretically, the
densely populated areas in Europe should generate cosmopolitan,
egalitarian, and multicultural values e three characteristics that
could influence turnout. However, these differences in values and
beliefs in the citizenry that might possibly be triggered by urbanization can only be captured subnationally.
Fourth, switching the level of analysis from the national to the
subnational level also results in an increase in degrees of freedom.
Instead of comparing 18 countries, we can compare more than 170
regions: an endeavor that should trigger more viable statistical
results.
4. Control variables
To assess the relative importance of regional perceived corruption we also include a number of national- and regional-level factors that might influence regional electoral participation. In more
detail, we add four regional-level variables, or Level 1 indicators, to
our multilevel model: regional GDP per capita, economic growth,
population density, and the closeness of the election. We also add
five national-level factors, or Level 2 indicators, to our model:
compulsory voting, electoral system type, presidentialism, national
voting age, and years of EU membership.
4.1. Regional-level covariates
Pertaining to the first indicator, economic development, we
assume that richer regions have higher turnout. The rationale for
this hypothesis is that affluent parts of a country should have a
more educated workforce, more civil society organizations, and
more citizens that adhere to post-materialist values (Inglehart,
1997) e all factors that should be beneficial for increased citizen
participation in politics including elections (Wyman et al., 1998,
67). Second, we also expect that significant changes in economic
development (e.g. fast economic decline) could influence voters'
likelihood of rushing to the polls. As such, by including data on
economic growth, we control whether the specific economic situation occurring when an election took place had a mobilizing or
depressing effect on regional-level turnout. Third, we include a
control variable for population density and assume turnout to be
161
higher in more densely populated areas: highly populated areas
have higher concentrations of electors. Accordingly, it makes more
sense for parties to campaign in densely populated areas, because
they reach more individuals (Lipset, 1981). In contrast, more rural
areas are not political hotspots; political rallies and events normally
do not take place there, and citizens are not as politically socialized
as they are in cities (Oliver, 2001). Summing up the potential
importance of population density for political participation studies,
Geys (2006) writes that a variable measuring the composition ‘of
the population [is] indispensable to any future analysis of turnout
(at whatever level of analysis) to avoid misspecification’ (653). Our
last regional control variable is the closeness of the election. We
believe that a close election should increase individuals' likelihood
of casting their ballot (Blais, 2000, 60). For one, a rational choice
argument would predict that the more narrow an election, the
more a vote counts (Powell, 1986). Moreover, a close election ought
to trigger higher campaign activities by parties, which in turn
should push more citizens to actually cast their ballots (Geys, 2006).
4.2. National-level covariates
For the first national indicator, compulsory voting, we expect to
see a positive relationship (Jackman, 1987; Franklin, 1999, 2004). If
voting is mandatory, citizens are forced by law to cast their ballot. If
they fail to do so, they face sanctions such as fines. Hence, voting no
longer becomes a choice but a duty prescribed by law. For the
second national-level variable, a country's electoral system type,
we assume that proportional representation (PR) systems have
higher turnout than majoritarian systems. Powell (1986) summarizes the logic behind this hypothesis, stating that in countries with
‘proportional representation from the nation as a whole or from
large districts, parties have an incentive to mobilize everywhere. In
contrast, in single-member districts some areas may be written off
as hopeless [or safe]’ (21). Hence parties in these districts have little
to no incentive to mobilize (Blais and Carty, 1990). Although later
studies (e.g. Bowler et al., 2001; Blais and Aarts, 2006) have found
less support for the finding that PR systems bolster turnout, we still
assume a positive relationship between PR and higher turnout. In
the analysis we differentiate between three types of electoral system types: PR systems, mixed systems, and majoritarian systems.
The third national-level control variable that we include in the
analysis is a dummy variable for semi-presidentialism. We suggest
that turnout should be lower in legislative elections in the two
semi-presidential systems, Romania and France. While legislative
elections in a parliamentary system fill both the executive and the
legislature, they fill only the legislature in France and Romania
(Marsh, 2008). Consequently, legislative elections become less
important or second order in these semi-presidential systems (Reif
and Schmitt, 1980). Hence legislative turnout should be lower,
because the stakes of the election are not as high as for the firstorder (semi-) presidential election (Stockemer and Calca, 2013b).
Fourth, we include the variable national voting age. Because
there is an age gap in voting, with older generations turning out in
higher numbers than younger generations, we assume that
lowering the voting age to 16 will decrease turnout rates (Wagner
et al., 2012). Fifth, we control for the number of years a country has
been an EU member. Longstanding EU members such as Germany,
France, or the Netherlands are consolidated democracies with an
established party system. First, voting in these longstanding democracies has become a habit for generations (Franklin, 2004).
Second, as a result of the predictable political environment, individuals have forged ties to parties (Wessels and Franklin, 2009).
However, the same might not be true for recent EU members. As
democracies for only one generation, countries such as Bulgaria or
Romania have had a shorter democratic experience and a less
162
€m, D. Stockemer / Electoral Studies 40 (2015) 158e169
A. Sundstro
developed party system than the founding members of the European Economic Commission (Kostadinova and Power, 2007).
Hence, turnout should be lower in these new member states.
5. Operationalizations
5.1. Dependent variable
We measure voter turnout, our dependent variable, in its standard form, by the percentage of registered citizens that cast their
ballot in their country's national legislative elections aggregated by
region. The data on voter turnout came from the European Election
Database, where we selected official turnout data for the most
recent available elections.2 Since the electoral cycles in the examined 18 countries are not synchronized, the most recently available
figures on turnout range from the year 2009 in two countries
(Germany and Bulgaria) to the year 2013 (for Italy and Austria). The
dependent variable is illustrated graphically in Fig. 1. Fig. 2 illustrates regional differences of turnout in relation to the national
means. For a full list of how turnout rates are dispersed in the
different regions, see Appendix 2.
5.2. Independent variable
Gauging perceptions of corruption, the main independent variable of this study, requires a bit more conceptual discussion. Corruption e most often defined as ‘the misuse of public power for
private gain’ (Treisman, 2007)3 e is a multifaceted concept where
researchers often make the general distinction between where it
occurs and in what scale. Whereas small-scale corruption often
involves public officials, large-scale corruption involves politically
elected decision-makers during the shaping of policy. Small-scale
corruption may take place in both administrative and political
spheres. In public administrations, an example is asking citizens for
bribes to gain access to services that they are legally entitled to. In
the political sphere, an example of small-scale corruption is vote
buying in the electoral process (e.g. Hellman et al., 2000; Wilson
and Damania, 2005).4
Because large-scale corruption is relatively absent in Europe, we
measure different types of small-scale corruption. In more detail,
we use the European Quality of Government Index (EQI), which
was compiled in 2009 by researchers at the Quality of Government
Institute, at the University of Gothenburg (see Charron, 2010; 2013;
Charron et al., 2013; Charron and Lapuente, 2013).5 To construct the
EQI, the researchers at the Quality of Government Institute implemented a survey where around 34,000 citizens in 18 countries
participated in telephone-based interviews carried out by a survey
company based in France. The respondents e all over 18 years of
2
This information is available online (see http://www.nsd.uib.no/european_
election_database/). For the French case, where no data were available on the
regional level, we calculated this measure by aggregating data from provinces,
available from the Ministry of Interior (see http://www.interieur.gouv.fr/Elections/
Les-resultats). For Austria, Italy, Romania, and Spain the regional data on turnout
were taken from the project Election Resources on the Internet (see http://www.
electionresources.org/).
3
For a discussion on definitions of corruption, see Johnston (1996) and Warren
(2013).
4
While vote buying may lead associations to less developed countries in Africa
and Asia, European countries such as Bulgaria recently witnessed those practices
during elections (Transparency International, 2013).
5
Aside from its regional focus, the EQI has one other advantage over the existing
indicators (e.g. Transparency International or the World Governance indicators),
who mainly use elite surveys to measure a country's corruption level. The EQI
directly asks citizens, who by definition are also possible voters, their perceptions of
corruption and good governance.
age e where selected randomly and interviews where carried out in
the local language.6
The survey focused on citizens' perceptions of and experiences
with three types of services (public education, public health care,
and law enforcement). All participants were asked several questions related to how they would numerically evaluate these services according to quality, impartiality, and corruption. In total, the
questionnaire consisted of 16 independent survey questions related
to these three pillars. In this article, we only focus on the five items
directly related to the concept of corruption (see Rothstein and
Teorell, 2008; Agnafors, 2013 for a discussion).
The five items we use from the EQI survey fall conceptually into
two different categories: (i) perceptions of small-scale corruption
in public services and (ii) perceptions of small-scale corruption in
the electoral process. Regarding the first category, four questions
focused on different sectors (public education, public health care,
and law enforcement).7 Specifically, three items asked respondents
to answer if ‘Corruption is prevalent in the police force in my area/
in my area's local public school system/in the public health-care
system in my area’ (disagree/agree, 0e10), respectively. A fourth
item asked the respondents, ‘In your opinion, how often do you
think other citizens in your area use bribery to obtain public services?’ (never/very often, 0e10). Regarding the second category, the
item focusing on the election process asks participants to ‘respond
to the following: Elections in my area are honest and clean from
corruption’ (never/very often, 0e10).
While these five items are theoretically distinct from each other,
they are highly correlated. Measures of their reliability coefficient
indicate that they are well suited for an aggregate index (Cronbach's Alpha amounts to .90). For our analysis we aggregated these
items into a composite regional indicator of corruption perceptions.
The indicator is standardized so that the mean is 0 and that the
standard deviation is 1. Lower levels equal more corruption (see
also Charron and Lapuente, 2013). In our analysis we include all
regions from the 18 countries that have been covered by the EQI
data collection, where turnout figures were available. (For a more
detailed description on the survey see Charron et al., 2013.)
5.3. Control variables
We measure our first control variable, regional-level economic
development, in its standard form by the GDP per capita per region.
As it is the standard procedure in turnout studies, we log transform
the variable.8 Second, economic growth is calculated based on the
same data as economic development. The formula for growth is the
following: GDPt GDPt1/(GDPt1)* 100. Third, to account for the
possible impact from population density, we gauge the average
number of individuals per square kilometer per region.9 Fourth, we
6
The survey was implemented randomly by the ‘next-birthday’ method. This
procedure is an alternative to other approaches e such as quota sampling e and is
generally seen as an efficient way to generate representative samples through
surveys (see e.g. Salmon and Nichols, 1983; Battaglia et al., 2008). This method
entails that the interviewer ask to speak with the person in the household who has
the next birthday (and make an appointment if this person is not available). The
researchers chose this method, although a quota method may render a more evenly
spread demographic sample, because the next birthday method was deemed to
generate a better range of opinions.
7
Performing a principal component analysis, Charron et al. (2011, 132) find that
these survey questions load on a latent variable, labeled corruption.
8
The variable is an average of the GDP per capita in these regions between the
years 2007 and 2009. The data were generously made available from Charron et al.
(2013) and originated from the official figures reported to the Eurostat database.
9
The data on population density were generously made available from Charron
et al. (2013) and originated from the official figures reported to the Eurostat
database for the year 2010.
€m, D. Stockemer / Electoral Studies 40 (2015) 158e169
A. Sundstro
163
Fig. 1. Voter turnout in 170 European regions, percentages. Comments: The data refer to the most recent national legislative elections in the 18 countries. The regions in the upper
right corner are French and Spanish territories. Source: The European Election Database and the Ministry of Interior in France and Italy.
measure the closeness of the election by subtracting the vote share
of the runner-up party from that of the winning party (see e.g.
Stockemer and Scruggs, 2012).10
Pertaining to national-level control variables, we first create an
ordinal variable in our dataset that captures variances in compulsory voting legislations. We code Belgium (the only country with
enforced compulsory voting) as 1; Greece and Italy are coded .5
(Greece has compulsory voting on paper but does not enforce it,
and Italy has abolished it); all other countries are coded 0.11 To
gauge different election system types, we create two dummy variables, one for PR systems and one for mixed systems. The
remaining category, majoritarian systems, serves as the reference.12
The third national-level control variable we include is a dummy
variable measuring semi-presidentialism. Fourth, to account for the
possible effect from voting age on turnout, we create a dummy
variable where we code all regions in Austria, the only country
where the voting age is 16, as 1. All other regions in all other
countries are coded 0. Lastly, we operationalize the variable ‘old’
and ‘new’ EU member state by the number of years a country has
been a member of the EU. (See Appendix 3 for some univariate
statistics of the independent variables.)
6. Methodology
10
While regional figures of polls before each of the elections could have been
preferable, this is not available in a comparative perspective across the 170 regions.
11
Italy abolished compulsory voting in 1993 and Greece lifted all sanctions for
nonvoting in 2000. We code these two countries as .5 because we think that
compulsory voting created a tradition to vote in these countries. We conjecture that
turning out has become habit, especially for the older generation that were politically socialized under mandatory voting (see Franklin, 2004). We also run the
model coding Italy and Greece zero. In this additional specification the influence of
compulsory voting becomes even stronger.
12
Data are taken from the Electoral System Design database at http://www.idea.
int/esd/.
We examine the influence of regional perceptions of corruption
in government on turnout by analyzing electoral participation in
170 NUTS (the Nomenclature of Territorial Units for Statistics) regions in 18 European countries.13 While these regions are at the
NUTS 1 level for 7 countries in the analysis and the NUTS 2 level for
11 others, these regions are relatively comparable at the same
territorial level; they feature the official geographic unit below the
13
Data availability of regional corruption limits us to using these countries.
164
€m, D. Stockemer / Electoral Studies 40 (2015) 158e169
A. Sundstro
Fig. 2. Regional deviations from national means of voter turnout, percentage units. Comments: The data refer to the most recent national legislative elections in the 18 countries. The
regions in the upper right corner are French and Spanish territories. Source: The European Election Database and the Ministry of Interior in France and the project Election Resources on the Internet.
national level.14 Although these regions admittedly differ in size,
population composition, and their administrative importance (see
Casellas and Galley, 1999), they are often used as units for
comparing different features across regional contexts (e.g. Marks
et al., 1996; Bottazzi and Giovanni, 2003; Brulhart and Traeger,
2005).
To measure the influence of corruption perceptions on turnout
at the regional level we engage in a two-step process. First, we
present the relationship between the two concepts in a bivariate
realm. Second, we calculate the interclass correlation coefficient
(ICC) of our two variables of interest, regional turnout and
perceived corruption via a so-called null model. The ICC, which is
calculated by dividing the variance component associated with the
intercept by the total variance, allows us to determine how much of
the variance in turnout and perceived corruption is due to national
and regional variation (see Rabe-Hesketh and Skrondal, 2012).
Third, we build a multilevel or hierarchical model. We have four
regional-level covariates (i.e. GDP per capita, economic growth,
population density, and the closeness of the race) and five static
14
NUTS 1 regions are from Belgium, Germany, Greece, Hungary, Netherlands,
Sweden, and the UK. NUTS 2 regions are from Austria, Bulgaria, Czech Republic,
Denmark, France, Italy, Poland, Portugal, Romania, Slovakia, and Spain.
national-level covariates or institutional factors (i.e. compulsory
voting, the type of electoral system, presidentialism, voting age,
and years of EU membership). Having data spanning two different
geographical levels renders Hierarchical Linear Modeling a sound
choice. In the analysis that ensues we cluster regions within
countries. Evaluating corruption's and the other regional-level
covariates' influence within their national context also avoids an
overestimation of the statistical significance of these Level 1 indicators (O'Connell and McCoach, 2008). Our multilevel model
covers 170 regions in 18 European countries. We also include a
dummy variable for Eastern Europe in the model as a Level 2 indicator because turnout there is approximately 20 percentage
points lower than in Western Europe (Stockemer, 2012).
As with many other studies in the social sciences there is the
latent danger of reversed causality in our model. It is theoretically
and empirically possible that low levels of turnout generate ‘malfunctioning institutions’ that over time spur increased corruption
perceptions among citizens. As often is the case with cross-sectional
studies, our research design limits our empirical testing of this
alternative reasoning: We therefore rest on previous research as well
as theoretical reasoning to suggest that it is corruption perceptions in
the citizenry that affect the propensity to be active in elections (and
not the other way around). First, we follow a plethora of studies (e.g.
Simpser, 2005; Tucker, 2007; Stockemer et al., 2013) that all
€m, D. Stockemer / Electoral Studies 40 (2015) 158e169
A. Sundstro
demonstrate that individuals' perceptions of government are an
important trigger for citizens to vote or stay at home on Election Day.
In all these studies the causal chain is as follows: Perceived corruption generates a direct decrease in trust of the political leaders and as
a result indirect mistrust between citizens, which then leads to a
decreased value in the civic duty of voting and lower turnout. Second,
there is experimental evidence supporting that the causal chain goes
from higher perceived corruption to lower turnout. For example,
Chong et al. (2013), who gave Mexican voters controlled information
about politicians' actual corrupt records, clearly find that when
voters are given evidence on misconduct of incumbent politicians
their actual participation in the following elections tends to decrease.
Their finding makes us more confident that the causal chain goes
from corruption to turnout and not vice-versa. To further minimize
the risk of reversed causation we made sure that the data for our
regional corruption preceded the turnout data.
7. Results
First, our bivariate analysis indicates a strong negative correlation
between higher levels of corruption and lower turnout. For example,
Fig. 3 illustrates that transparent and clean regions have over 30
percentage points more citizens turning out in elections than poorly
governed or corrupt regions. Unsurprisingly, this substantively
strong relationship is statistically significant at the 99 percent level
(see Fig. 3). In addition, we find that regions that have an aboveaverage score in corruption perceptions have approximately 14
percentage points more citizens turning out on Election Day than
regions with below-average corruption scores (i.e. the average
turnout for regions that range above the mean in corruption control
is 69.8 percent, whereas for regions that range below the mean in
corruption control, the average turnout is 55.4 percent). Based on
these two types of bivariate analyses, it seems that corruption influences turnout in European regions. Still, to gain more robust evidence of this finding, we have to test this stipulation in a multivariate
model, which we will do next. However, before doing so, we partition
the variance into regional and national level variance to determine
how much of the variance is explained by regional turnout/perceived
corruption and by national turnout/perceived corruption.
We find for both regional and national level turnout, as well as
regional and national level perceived corruption, that the variance
is larger than the corresponding standard error. This implies that
there is both statistically significant national and regional level
Fig. 3. The relationship between regional levels of perceived corruption and turnout.
Comments: The regional corruption indicator is based on five survey items measuring
perceptions of corruption in government. The data on turnout refer to the most recent
national legislative elections in the 18 countries.
Source: Charron et al., 2013; The European Election Database, the Ministry of Interior
in France and the project Election Resources on the Internet.
165
variance. The ICC then measures the variance attributable to national factors. For our two variables of interest the ICC is roughly 86.
This implies that around 14 percent of the variance in turnout, as
well as perceived corruption is explained by regional level factors.
Models 1 to 3 illustrate that corruption takes up most of this
regional variance. Moreover, the hierarchical model specifies this
impact. Our indicator of perceived corruption has the expected sign
and is statistically significant (see Model 1 in Table 1). Substantively, the perceived corruption indicator is the strongest Level 1
factor in the model. The equation predicts that for every point a
region becomes less corrupt, turnout increases by more than 5
percentage points.15 This implies that corruption perceptions alone
explain more than 15 percentage points of the variance in turnout
between regions across Europe. This also implies that within
countries, in which the score of perceived corruption sometimes
differs by more than 1 point between regions (e.g. Italy and
Bulgaria), the indicator can explain variation in turnout by up to 5
percentage points. Several country examples support this finding.
In Belgium, for example, there is a 5 percentage point gap in
turnout between the ‘cleanest’ region (Vlaams Gewest) and the
‘most corrupt’ region (Brussels). Similar relationships exist in many
other countries including Romania, France, and Greece.
While most of our Level 1 and Level 2 control variables display
the expected sign, only a few variables are statistically significant.
Pertaining to the regional-level variables, we find that both rural
areas (i.e. regions with a lower population density) and richer areas
(i.e. regions with a higher GDP per capita) have increased turnout.
Substantively, the model predicts that a hypothetical region with a
population density of 100 has about 1.2 percentage points more
citizens turning out on Election Day than a region with a population
density of 1000, if we hold all other variables constant. This finding
does not support the hypothesized idea that the political life is
more pronounced in cities and that parties are more active there.
Rather, we find that country dwellers are in the aggregate more
likely to vote. However, this influence is quite moderate. The second
statistically significant variable, economic development, has a
relatively strong influence on turnout. For example, the model
predicts that a region with a GDP per capita of 30,000 Euro has
nearly 4 percentage points more individuals turning out than a
region with an average income of 20,000 Euro a year. The other
regional control variables are not statistically significant.
With regard to national-level covariates we find that regionallevel turnout is strongly influenced by compulsory voting. Rather
unsurprisingly, our results indicate that compulsory voting boosts
turnout by approximately 20 percentage points, a finding that is in
line with previous research (e.g. Franklin, 2004). Second, while we
find that none of other national-level variables is statistically significant, our results nevertheless indicate that several other Level 2
factors have quite a large coefficient. For example, the coefficient for
semi-presidentialism indicates that separate presidential elections
decrease turnout in legislative elections by over 8 percentage points.
We also explored the robustness of our model by disaggregating
our main independent variable. Since our perceived corruption indicator is a composite measure of two types of perceptions e corruption in public services and corruption in the electoral process e
we evaluate if these two components have different effects on
15
We also ran individual models with the five components of our aggregate
corruption index. The size of the coefficients of these five individual corruption
indicators matches that of our composite index. As an additional robustness check,
we run Model 1 as an OLS regression with the four regional-level covariates and 17
country dummy variables, with Austria serving as the reference category. The results of this additional specification confirm the results from our main model. The
regional corruption indicator has the right sign, is statistically significant, and its
substantive impact is very similar to that of our main specification.
€m, D. Stockemer / Electoral Studies 40 (2015) 158e169
A. Sundstro
166
Table 1
Results of the hierarchical linear model.
Fixed effects
Country level
Intercept
Compulsory voting
PR
Mixed electoral system
Presidentialism
Voting age at 16
Years of EU membership
East
Regional level
Corruption perceptions
Component 1: Perceived
corruption in public
services
Component 2: Perceived
corruption in electoral
process
Closeness of the election
Log GDP per capita
Economic growth
Population density
Random effect
Intercept
Residual
Model fit statistics
Log likelihood
BIC
AIC
Model 1
Model 2
Model 3
13.12 (19.50)
18.88*** (7.70)
4.71 (6.65)
9.71 (6.83)
9.14 (6.28)
1.04 (4.26)
.09 (.14)
1.94 (6.23)
19.53 (19.41)
19.82*** (7.55)
5.19 (6.43)
10.48 (6.60)
8.64 (6.10)
1.64 (4.26)
.11 (.14)
2.68 (6.05)
14.17 (20.84)
12.27*** (9.84)
4.72 (8.71)
8.12 (9.03)
10.63 (8.17)
.58 (4.54)
.01 (.18)
1.55 (8.03)
5.13*** (.95)
4.75*** (.97)
3.08*** (.95)
.01 (.05)
7.86*** (1.78)
.13 (.17)
.001* (.001)
Estimate
41.80 (19.58)
22.28 (2.60)
.01 (.05)
8.59*** (1.77)
.16 (.17)
.001* (.001)
.00 (.05)
7.73*** (1.80)
.16 (.17)
.001* (.001)
38.80 (18.34)
23.07 (2.69)
75.00 (34.39)
22.05 (2.58)
515.15
1107.35
1060.31
517.37
1111.77
1064.74
517.73
1112.49
1065.46
Comments: * ¼ p < 0.05, ** ¼ p < 0.01, *** ¼ p < 0.001.
turnout. When doing so we do not find any large differences between these measures (see Models 2 and 3 in Table 1). If anything,
the size of the coefficients indicates that perceptions of corruption in
public services have a slightly stronger suppressing effect on turnout
than perceived corruption in the electoral process. However, the
differences between these measures are not substantively large.
Theoretically, our result of a strong negative relationship between regional-level perceptions of corruption and regional-level
turnout provides robust support for the majoritarian perspective;
that is, corruption perceptions render citizens less likely to go to the
polls. While recent research (e.g. Stockemer et al., 2013) has found
that this is the case across a large sample of democracies, our
research indicates that corruption suppresses turnout at the
regional level, where people have the most contact with state officials. Hence our study finds support for the claim that perceptions
of corruption in government deter people from engaging in the
political process. In fact, the outlook of corrupt politicians in leadership and citizens' incapability to vote away bureaucrats that demand bribes for public services result in a decreased value of
voting, which, in turn, tends to make citizens less likely to cast their
ballot on Election Day. This finding becomes increasingly relevant
the more we consider that regional-level perceptions of corruption
impact variation in turnout not only between regions across Europe
but also within regions in one country.
8. Conclusions
As the first study that evaluates the link between citizens' perceptions of corruption and voter turnout at the regional level in a
comparative perspective, this study adds to our understanding of
the relationship between perceived corruption and turnout in two
ways. First, by taking into consideration that there is strong withincountry variation in perceived corruption and turnout, the design
of our study considerably reduces the risk of ecological fallacies.
While national-level studies could find that average (high) levels of
corruption could decrease turnout, it is still theoretically possible
that turnout is actually highest in the most corrupt areas of a
country. Our analysis indicates that this is not the case. In fact,
supporting the majoritarian view, we find that perceptions of corruption decrease electoral participation. In fact corruption reduces
the value of the civic duty of voting: it can spur distrust of elected
leaders and generate feelings of resignation from having corrupt
leaders no matter the outcome of the election. In this sense, our
study provides strong backing for the theory that perceptions of
bribery within the bureaucracy politically disenfranchise citizens.
Second, our analysis somewhat contradicts the few existing
regional-level studies (e.g. Escaleras et al., 2012; Stockemer and
Calca, 2013a), which measure corruption as the number of indicted
public servants, and which actually find a positive relationship between more indicted individuals and higher turnout. However, we
argue that gauging corruption by the number of indicted individuals
is difficult to interpret: It is unclear whether more individuals
persecuted because of corruption reflects effective anticorruption
efforts or a society full of corruption. Importantly, we believe that
voters base their decision to vote on perceptions. Possibly unaware of
the actual degree of corruption, citizens are likely to decide whether
to rush to the polls or to stay at home on Election Day based on their
perception of whether they think the political system and public
authorities are corrupted. We find that if they believe that this is the
case, they have a higher likelihood of staying at home.
While our study adds greatly to the burgeoning literature on
corruption and political participation, the research on corruption and
turnout is still in its early stages. For one, good regional-level data are
missing for all other regions of the globe (except for North America).
To fill this gap, it would be worthwhile to compile regional corruption/governance data for Asia, Latin America, and Africa. While this is
certainly a costly endeavor, it would benefit not only the political
participation research, but also development studies more generally.
In addition, future research should use the existing data and combine
various levels of analyses e an endeavor we could not engage in this
study, because the survey conceived by the research team only surveyed individuals on corruption and quality of governance items.
Questions on elections and voting were absent from the survey.
Ideally, future surveys should ask individuals questions on their
perceptions of governance and corruption, as well as questions
related to voting and vote choice. For example, such surveys would
allow us to investigate which types of individuals are most likely to
abandon the polls because of corruption. While there is still much
room for additional studies, we maintain that our research advances
the understanding of the impact of perceived corruption.
Appendix 1. The year of the national legislative elections
included in this study
Austria
Belgium
Bulgaria
Czech Republic
Denmark
France
Germany
Greece
Hungary
Italy
Netherlands
Poland
Portugal
Romania
Slovakia
Spain
Sweden
United Kingdom
2013
2012
2009
2010
2011
2012
2009
2012
2010
2013
2010
2011
2011
2012
2010
2011
2010
2010
€m, D. Stockemer / Electoral Studies 40 (2015) 158e169
A. Sundstro
Appendix 2. The turnout rates in the different regions of the
study (NUTS-codes, percentages)
167
168
€m, D. Stockemer / Electoral Studies 40 (2015) 158e169
A. Sundstro
Appendix 3. Descriptive statistics of the regional level
indicators
Turnout
Corruption perceptions indicator
Log GDP per capita
Economic growth
Population density
Electoral closeness
Mean
Standard deviation
Minimum
Maximum
64.21
.01
9.85
.41
14.48
11.24
12.46
1.00
.65
3.98
.89
8.18
30.84
3.12
7.95
11.3
11.74
.04
90.86
2.22
11.01
19
16.705
38.25
References
Abramson, P.R., Aldrich, J.H., 1982. The decline of electoral participation in america.
Am. Polit. Sci. Rev. 76, 502e521.
Agnafors, M., 2013. Quality of government: towards a more complex definition. Am.
Polit. Sci. Rev. 107 (3), 433e445.
Anderson, C.J., Guillory, C.A., 1997. Political institutions and satisfaction with democracy: a cross-national analysis of consensus and majoritarian systems. Am.
Polit. Sci. Rev. 91 (1), 66e81.
Anderson, C.J., Tverdova, Y.V., 2003. Corruption, political allegiances, and attitudes
toward government in contemporary democracies. Am. J. Polit. Sci. 47 (1),
91e109.
Battaglia, M.P., Link, M.W., Frankel, M.R., Osborn, L., Mokdad, A.H., 2008. An evaluation of respondent selection methods for household mail surveys. Public
Opin. Q. 72 (3), 459e469.
Bauhr, M., Grimes, M., 2014. The effect on societal accountability from transparency.
Governance 27 (2), 291e320.
Birch, S., 2010. Perceptions of electoral fairness and voter turnout. Comp. Polit. Stud.
43 (12), 1601e1622.
Blais, A., 2006. What affects voter turnout? Annu. Rev. Polit. Sci. 9, 111e125.
Blais, A., 2000. To Vote or Not to Vote? The Merits and Limits of Rational Choice.
Univ. Pittsburgh Press, Pittsburgh.
Blais, A., Aarts, K., 2006. Electoral systems and turnout. Acta Polit. 41, 180e196.
Blais, A., Carty, K., 1990. Does proportional representation foster voter turnout? Eur.
J. Polit. Res. 18, 167e181.
Bottazzi, L., Giovanni, P., 2003. Innovation and spillovers in regions: evidence from
European patent data. Eur. Econ. Rev. 47, 687e710.
Bowler, S., Brockington, D., Donovan, T., 2001. Election systems and voter turnout:
experiments in the United States. J. Polit. 63, 902e915.
Brulhart, M., Traeger, R., 2005. An account of geographic concentration patterns in
europe. Reg. Sci. Urban Econ. 37 (6), 597e624.
Casellas, A., Galley, C.C., 1999. Regional definitions in the European Union: a
question of disparities? Reg. Stud. 33, 551e558.
Clarke, H., Sanders, D., Stewart, M., Whiteley, P., 2004. Political Choice in Britain.
Oxford University Press, Oxford.
Charron, N., 2010. Assessing the Quality of the Quality of Government Data: a
Sensitivity Test of the World Bank Government Indicators. QoG Working Paper.
http://www.qog.pol.gu.se/digitalAssets/1357/1357980_paper-on-sensititivtytests-of-world-bank-data.pdf.
Charron, N., 2013. From Åland to Ankara: European Quality of Government Data.
2013 Data, Sensitivity Analysis and Final Results. In: QoG Working Paper Series
2013, p. 1.
Charron, N., Lapuente, V., 2013. Why do some regions in europe have higher quality
of government? J. Polit. 75 (3), 567e582.
Charron, N., Dijkstra, L., Lapuente, V., 2013. Regional governance matters: quality of
government within European union member state. Reg. Stud. 48 (1), 68e90.
Charron, N., Lapuente, V., Rothstein, B., 2011. Measuring Quality of Government and
Sub-national Variation. Report for the EU Commission of Regional Development. European Commission Directorate-General Regional Policy Directorate
Policy Development.
Chong, A., De La O, A.L., Karlan, D., Wantchekon, L., 2013. Looking Beyond the
Incumbent: Exposing Corruption and the Effect on Electoral Outcomes. NBER
Working Paper No. 17679.
Escaleras, M., Calcagno, P., Shughart II, W., 2012. Corruption and voter participation:
evidence from the US States. Public Finance Rev. 40, 789e815.
Ezrow, L., Xezonakis, G., 2014. Satisfaction with Democracy and Voter Turnout: a
Temporal Perspective (Forthcoming in Party Politics).
Franklin, M., 2004. Voter Turnout and the Dynamics of Electoral Competition in
Established Democracies Since 1945. Cambridge University Press, Cambridge,
UK.
Franklin, M., 1999. Electoral engineering and cross-national turnout differences:
what role for compulsory voting? Br. J. Polit. Sci. 29 (1), 205e224.
Geys, B., 2006. Explaining voter turnout: a review of aggregate-level research. Elect.
Stud. 25, 637e663.
€nlund, K., Set€
Gro
al€
a, M., 2007. Political trust, satisfaction and voter turnout. Comp.
Eur. Polit. 5, 400e422.
Gupta, S., Davoodi, H., Tiongson, E., 2000. Corruption and the Provision of Health
Care and Education Services. Working paper. International Monetary Fund,
Washington DC.
Heckelman, J., 1995. The effect of the secret ballot on voter turnout rates. Public
Choice 82 (1e2), 107e124.
Hellman, J.S., Jones, G., Kaufmann, D., 2000. Seize the State, Seize the Day. State
Capture, Corruption and Influence in Transition. World Bank Policy Research
Paper 2444.
Huntington, S., 1968. Political Order in Changing Societies. Yale University Press,
New Haven, CT.
Huntington, S., 1991. The Third Wave: Democratization in the Late Twentieth
Century. University of Oklahoma Press, Norman, OK.
Inglehart, R., 1997. Modernization and Postmodernization: Cultural, Economic and
Political Change in 43 Societies. Princeton University Press, Princeton.
Inman, K., Andrews, J., 2010. Political participation in Africa: evidence from survey
and experimental research. In: Paper presented at the annual meeting of the
Midwest Political Science Association. Chicago, IL, April 22e25, 2010.
Jackman, R.W., 1987. Political institutions and voter turnout in industrial democracies. Am. Polit. Sci. Rev. 81, 405e424.
Johnston, M., 1983. Corruption and political culture in America. Publius 13 (1),
19e39.
Johnston, M., 1996. The definitions debate: old conflicts in new guises. In: Jain, A.K.
(Ed.), The Political Economy of Corruption. Routledge, London, pp. 11e31.
Karahan, G.R., Coats, M., Shugart II, W.F., 2006. Corrupt political jurisdictions and
voter participation. Public Choice 126 (1), 87e196.
Karp, J.A., Banducci, S., 2008. Getting out the vote: party mobilization in a
comparative perspective. Br. J. Polit. Sci. 38 (1), 91e112.
Kostadinova, T., 2003. Voter turnout dynamics in post-communist Europe. Eur. J.
Polit. Res. 42 (6), 741e759.
Kostadinova, T., 2009. Abstain or rebel: corruption perceptions and voting in East
European elections. Polit. Policy 37 (4), 691e714.
Kostadinova, T., Power, T., 2007. Does democratization depress participation? Voter
turnout in the Latin American and Eastern European transitional democracies.
Polit. Res. Q. 60 (3), 363e377.
Lipset, S., 1981. Political Man: the Social Bases of Politics. John Hopkins University
Press, Baltimore.
Lipset, S., Lenz, G.S., 2002. Corruption, culture and markets. In: Harrison, L.E.,
Huntington, S. (Eds.), Culture Matters: How Values Shape Human Progress.
Basic Books, New York.
Lipsky, M., 1980. Street Level Bureaucracy: Dilemmas of the Individual in Public
Service. Russell Sage Foundation, New York.
Marks, G., Nielsen, F., Salk, J., Ray, L., 1996. Competencies, cracks, and conflicts:
regional mobilization in the European Union. Comp. Polit. Stud. 29 (2), 164e193.
Marsh, M., 2008. Understanding European parliament elections. In: Kohler-Koch, B.,
Larat, F. (Eds.), Efficient and Democratic Governance in the European Union.
Connex, Mannheim.
Mauro, P., 1995. Corruption and growth. Q. J. Econ. 110, 681e712.
McCann, J.A., Dominguez, J.I., 1998. Mexicans react to electoral fraud and political
corruption. Elect. Stud. 17 (4), 483e503.
O'Connell, A.A., McCoach, D.B. (Eds.), 2008. Multilevel Modeling of Educational
Data. Volume Sponsored by AERA SIG: Educational Statisticians. Information
Age Publishing, Charlotte, NC.
Oliver, J.E., 2001. Democracy in Suburbia. Princeton University Press, Princeton.
Pacek, A.C., Pop-Eleches, G., Tucker, J.A., 2009. Disenchanted Or discerning: voter
turnout in post-communist countries. J. Polit. 71 (2), 473e491.
Powell, G.B., 1982. Comparative Democracies: Participation, Stability and Violence.
Harvard University Press, Cambridge, MA.
Powell, G.B., 1986. American voter turnout in comparative perspective. Am. Polit.
Sci. Rev. 80 (1), 17e43.
Putnam, R.D., Leonardi, R., Nanetti, R.Y., Pavoncello, F., 1983. Explaining institutional
success: the case of Italian regional government. Am. Polit. Sci. Rev. 77, 55e74.
Putnam, R., 1993. Making Democracy Work: Civic Traditions in Modern Italy.
Princeton University Press, Princeton.
Rabe-Hesketh, S., Skrondal, A., 2012. Multilevel and Longitudinal Modeling Using
Stata, third ed. Stata Press, College Station.
Reif, K., Schmitt, H., 1980. Nine second-order national elections: a conceptual
framework for the analyses of european election results. Eur. J. Polit. Res. 8 (1),
3e44.
Rose-Ackerman, S., 2008. Corruption. In: Rowley, C.K., Schneider, F.G. (Eds.),
€m, D. Stockemer / Electoral Studies 40 (2015) 158e169
A. Sundstro
Readings in Public Choice and Constitutional Political Economy. Springer, New
York.
Rothstein, B., Eek, D., 2009. Political corruption and social trust: an experimental
approach. Ration. Soc. 21 (1), 81e112.
Rothstein, B., Solevid, M., 2013. Corruption and Political Participation. Work in
progress. The Quality of Government Institute, University of Gothenburg.
Rothstein, B., Teorell, J., 2008. What is quality of government? A theory of impartial
political institutions. Gov. Int. J. Policy Adm. 21 (1), 165e190.
Rothstein, B., Uslaner, E.M., 2005. All for all: equality, corruption, and social trust.
World Polit. 58 (1), 41e72.
Rydgren, J., 2007. The sociology of the radical right. Annu. Rev. Sociol. 33, 241e262.
Salmon, C.T., Nichols, S.N., 1983. The next-birthday method of respondent selection.
Public Opin. Q. 47 (2), 270e276.
Schaffer, F., 2002. Might cleaning up elections keep people away from the polls?
Historical and comparative perspectives. Int. Polit. Sci. Rev. 23 (1), 69e84.
Schaffer, F., 2007. Lessons learned. In: Schaffer, F. (Ed.), Elections for Sale: the Causes
and Consequence of Vote Buying. Lynne Rienner, Boulder and London.
Simpser, A., 2005. Making Votes Not Count: Strategic Incentives for Electoral Corruption (PhD dissertation). Stanford University, Stanford, CA.
Stockemer, D., 2012. Citizens' support for the European Union and participation in
European parliamentary elections. Eur. Union Polit. 13 (1), 26e46.
Stockemer, D., 2013. Corruption and turnout in presidential elections e a macrolevel quantitative analysis. Polit. Policy 41 (2), 191e214.
Stockemer, D., Calca, P., 2013a. Corruption and turnout in Portugal: a municipal
level study. Crime, Law Soc. Change 60 (5), 535e548.
Stockemer, D., Calca, P., 2013b. Presidentialism and Voter Turnout in Legislative
Elections (Forthcoming in Parliamentary Affairs).
169
Stockemer, D., LaMontagne, B., Scruggs, L., 2013. Bribes and ballots: the impact of
corruption on voter turnout in democracies. Int. Polit. Sci. Rev. 34 (1), 70e92.
Stockemer, D., Scruggs, L., 2012. Income inequality and electoral turnout: new evidence on a burgeoning debate. Elect. Stud. 31 (4), 764e773.
Tabellini, G., 2005. Culture and Institutions: Economic Development in the Regions
of Europe. Bocconi University, Mimeo.
Treisman, D., 2007. What have we learned about the causes of corruption from ten
years of cross-national empirical research? Annu. Rev. Polit. Sci. 10, 211e244.
Transparency International, 2013. Bulgaria: Rock Bottom Election Turnout Exposes
Accountability Failure of Political Class. Published 14 May 2013. Available online.
http://www.transparency.org/news/feature/bulgaria_rock_bottom_
election_turnout_exposes_accountability_failure_of_pol.
Tucker, J.A., 2007. Enough! Electoral fraud, collective action problems, and postcommunist coloured revolutions. Perspect. Polit. 5 (3), 537e553.
Wagner, M., Johann, D., Kritzinger, S., 2012. Voting at 16: turnout and the quality of
vote choice. Elect. Stud. 31 (2), 372e383.
Warren, M.E., 2013. The meaning of corruption in democracies. In: Heywood, P.
(Ed.), The Routledge International Handbook on Political Corruption. Routledge,
Oxford.
Wessels, B., Franklin, M., 2009. Turning out or turning off: do mobilization and
attitudes account for turnout differences between new and established member states at the 2004 EP elections? J. Eur. Integr. 31 (5), 609e626.
Wilson, J., Damania, R., 2005. Corruption, political competition and environmental
policy. J. Environ. Econ. Manag. 49 (3), 516e535.
Wyman, M., White, S., Oates, S. (Eds.), 1998. Elections and Voters in PostCommunist Russia. Edward Elgar, London.