The Study of Less Important Elections

Full text

The character and the importance of the electoral contest is a group of contextual
variables that have become a staple in comparative analysis and, together with
the political climate, constitute the core dimensions of multi-level electoral
research. The character of the electoral contest is defined by the perceived politi-
cal importance (or salience) of the office to be filled, which differs between the
different types of elections. Elections of members of national parliaments or of
a president with executive powers are contests of high salience (or ‘high stimulus
elections’), while all other types of elections are of low salience (or ‘low stimu-
lus elections’).
The impetus for systematic analyses of these different types of elections mostly
originates in efforts to understand the results of US midterm elections, a research
stream that was initiated in the mid 1950s by US scholars. As is well known, US
midterm elections are characterized by stable patterns of lower participation rates
compared with the preceding presidential election on one hand, and losses for
the party of the president on the other. An enormous number of studies shed light
on this classic feature of US politics. We can distinguish three main approaches:
the original ‘surge and decline’ theory of the Michigan school (Campbell 1960);
the ‘referendum’ theory (Tufte 1975); and the ‘balancing’ theory (Alesina and
Rosenthal 1989, 1995).
This perspective on electoral research, as many others, was successfully
exported beyond the confines of US elections. In the late 1970s we observe a
spillover of the model to the study of European electoral politics. A first effort
to study the ‘midterm loss’ phenomenon in a European multi-party environment
was published by Dinkel (1977, 1978). He analyzed the results of German state
4
The Study of Less Important
Elections
Hermann Schmitt and Eftichia Teperoglou
BK-SAGE-ARZHEIMER_ETAL-160488-Chp04.indd 56 2/13/2017 1:44:41 PM

The STudy of LeSS ImporTanT eLecTIonS 57
elections (Landtagswahlen) on the basis of Campbell’s surge and decline theory.
Three years later, another step towards a general theory of voting behavior in less
important elections was proposed in a study of the mechanisms behind the vari-
ous national results of the first direct elections of the members of the European
Parliament (EP) in June 1979. Writing in the aftermath of the first EP electoral
contest, Reif and Schmitt (1980) presented the first supra-national election as an
example of a series of national second-order elections. The distinction between
first-order and second-order elections is now frequently used in scholarly anal-
yses, and the theoretical framework originally developed by Reif and Schmitt
today is probably the most prominent in this research area.
The main aim of this chapter is to review the state of the art in electoral
research related to voting behavior in second-order elections and to outline the
intellectual and theoretical trajectory of the second-order election (SOE) model
from the first EP election of 1979 to the last (at the time of writing) in 2014. The
chapter is structured in four sections. The first outlines the roots of the theories of
‘low stimulus’ elections. In the second section we briefly describe the application
of the paradigm of midterm losses in European contexts. The third section offers
a detailed presentation of the SOE model. The review of the literature includes
studies of voting behavior in European Parliament elections and – to some degree
also – in other SOEs such as regional elections and by-elections in Europe and
beyond. The fourth section discusses the limits of the SOE model.
Three ConCepTions of Less imporTanT eLeCTions
The surge and decline theory
In 1960, Angus Campbell formulated the ‘surge and decline’ theory according to
which partisanship is only fully mobilized at the time of presidential elections.
This implies that mobilization is significantly less at the time of midterm elec-
tions. This proposal was seeking to explain a repeated pattern in US politics: the
lower turnout and the losses of the presidential party in midterm elections as
compared with the preceding presidential election. These fluctuations are deter-
mined by short-term political forces (A. Campbell 1960: 397). There are three
main factors in his model: level of political stimulation, political interest, and
party identification. His analysis is focused on two fundamental distinctions. The
first one is between ‘high stimulus’ or ‘surge’ elections and ‘low stimulus’ or
‘decline’ elections. The main difference between these two types of elections is
related to the importance of the contest (and its outcomes) as it is perceived by
the electorate. Presidential elections are ‘high stimulus’ contests, while midterm
elections are ‘low stimulus’ (see A. Campbell 1960 and later, among many
others, the overview by J. Campbell 1997a, Marsh 2007 and Marsh and
Mikhaylov 2010). The motivation to cast a vote in ‘high stimulus’ presidential
BK-SAGE-ARZHEIMER_ETAL-160488-Chp04.indd 57 2/13/2017 1:44:41 PM

The SAGe hAndbook of elecTorAl behAviour
58
elections is higher: along with it comes a surge of political information, and an
intensive campaign which generates higher levels of electoral participation. The
reason for all of this is that voters elect a president – the head of the executive.
Under these circumstances, many voters who may not vote in a ‘low stimulus’
environment can be motivated to turn out. With an eye on the partisan shifts
between presidential and midterm election, the second distinction in Campbell’s
study is between ‘core’ and ‘peripheral’ voters. The latter category refers to
those voters who are lacking a sustained interest in politics, who are more erratic
in terms of their political behavior (if not apolitical), and whose party attachment
is low or absent. While these voters tend to abstain in midterm elections, many
of them are likely to vote in presidential elections. This is because in the ‘low
stimulus’ elections the level of political information is lower, and the campaigns
do not provide sufficient stimulation to go to the polls. Therefore, the lower
levels of participation in midterm elections are explained by the abstention of
‘peripheral’ voters. ‘Core’ voters are habitual voters, with a stronger party iden-
tification than ‘peripheral’ voters and a comparatively high level of political
information. They tend to participate irrespectively of the salience of an election.
Core voters in a ‘high stimulus’ election may depart from their normal partisan
behavior. Even if they identify with the disadvantaged party, they might in a
presidential election vote for the candidate who is advantaged by the circum-
stances (Campbell 1960: 401), just in order to return to their habitual behavior
in the next ‘low stimulus’ midterm election. Therefore, the losses for the presi-
dential party in midterm elections are the mirror image of the surge of support in
the previous ‘high stimulus’ election. Presidential election results can thus be
seen as a departure from an equilibrium that is re-established in the next midterm
election (Marsh 2007: 76).
The revised surge and decline theory
A quarter of a century later, James Campbell (1997a, 1997b) proposed a revised
version of the original surge and decline theory. While he maintained the earlier
description of ‘high’ and ‘low stimulus’ elections, he considered in addition the
prior presidential election in order to explore the roots of midterm losses. A
major difference from the original theory is the consideration of short-term
forces on voting behavior in midterm elections. He introduced the concept of
potential cross-pressures between short-term political forces and party identifi-
cation. The original surge and decline theory takes into account how these cross-
pressures can affect party choice, but is silent about their effect on electoral
participation (Campbell 1997a: 90). He argued that the difference in the result
between the presidential and midterm elections is not caused by the turnout
effect but, on the contrary, derives from voters whom James Campbell calls ‘dis-
advantaged partisans’. Unlike the original theory, the revision hypothesizes that
the surge of information in a high stimulus election affects the turnout of
BK-SAGE-ARZHEIMER_ETAL-160488-Chp04.indd 58 2/13/2017 1:44:41 PM

The STudy of LeSS ImporTanT eLecTIonS 59
‘peripheral partisans’ as well as the vote choice of independents and weaker
partisans. Disadvantaged partisans (like advantaged partisans of the winning
presidential party) contribute to the midterm loss of the presidential party mainly
through their participation and not through their vote choice (Campbell 1997a:
103). The revised theory thus claims that short-term forces influence the vote
choice on the side of independents and the decision to participate or not on the
side of partisans (Campbell 1997a: 104).
The presidential-coattail explanation of midterm losses
Campbell’s original ‘surge and decline’ theory did not come out of the blue. It
had a predecessor in the presidential-coattail explanation of midterm losses (see
Bean 1950; Miller 1955; also Hinckley 1967 and Calvert and Ferejohn 1983).
Indeed, it has been argued that Campbell (1960) just took the withdrawn coattail
approach one step further (Erikson 1988: 1013).
The basic assumption of the coattail voting approach is that the losses suf-
fered by the president’s party in midterm elections are a result of the absence of
presidential ‘coattails’1. Congressional candidates of the president’s party when
running in the on-year elections take advantage of ‘coattail effects’. However,
this advantage is lost when running in subsequent midterm elections and there-
fore, so the argument goes, they are experiencing a ‘midterm defeat’. The coat-
tail voting approach does not consider determinants of individual vote choices
or decisions to turn out. It focuses almost exclusively on presidential candidates
(Campbell 1997a: 17). This is where ‘surge and decline’ and ‘coattail’ differ. On
the other hand, there are also some common characteristics. In both approaches
the losses in the midterm election are linked to a previous ‘surge’ in the presi-
dential election; both identify short-term political forces in on-year elections
which create favorable circumstances for the party of the winning presidential
candidate; and both argue that the variation in midterm seat losses should be
proportionate with the gains of the previous ‘high stimulus’ election (Campbell
1997a: 16–17)2.
The referendum theory
The second main approach for understanding divergent voting patterns in mid-
term elections is the referendum theory. It was proposed by Tufte (1975, 1978).
However, under this theoretical framework we can also subsume some additional
approaches (or amendments to the original theory) such as Kernell’s negative
voting theory (1977) and the strategic politician approach by Jacobson and
Kernell (1981). The referendum theory of midterm losses is probably the main
alternative approach to the original surge and decline theory. As presented above,
the latter approach identifies the roots of the midterm losses in the previous
presidential election. By contrast, the referendum theory attributes midterm
BK-SAGE-ARZHEIMER_ETAL-160488-Chp04.indd 59 2/13/2017 1:44:41 PM

The SAGe hAndbook of elecTorAl behAviour
60
losses to the political conditions prevailing at the time of the midterm election
(Campbell 1997a: 9).
The referendum theory was originally developed by Tufte in 1975. In this arti-
cle we can find the most straightforward definition for the midterm vote: ‘the mid-
term is neither a mystery nor an automatic swing of the pendulum; the midterm
vote is a referendum’ (Tufte 1975: 826). This insight was carried forward in the
work of 1978, in which the broader theoretical framework of the economic voting
theory is developed: ‘When you think economics, think elections, when you think
elections, think economics’ (Tufte 1978: 65). This is perhaps the most often-cited
quote in any economic voting textbook (for the economic voting theory see among
many others Lewis-Beck 1988; Lewis-Beck and Stegmaier 2000; Anderson 2000;
van der Brug etal. 2007). Tufte introduces two main explanatory variables in his
analysis of midterm election outcomes. The first one is the electorate’s evalua-
tion of the president at the time of the election. The popularity of the president
is measured by the standard question of the approval or disapproval of the presi-
dent’s job performance. The second explanatory variable is the performance of the
economy in the year prior to the midterm election. Midterm elections are charac-
terized as ‘referendum on the government’s performance, in which voters express
their approval or disapproval through voting for or against the presidential party’
(Marsh 2007: 77). This insight was generalized by showing that the referendum
mechanism is able to account for variation in the aggregate outcomes of US elec-
tions for the House, Senate, governorships, and upper and lower houses of state
legislatures (Simon 1989; Simon etal. 1991). This mechanism was also identified
in sub-national elections in Argentina (Remmer and Gélineau 2003). However,
presidential approval cannot account for everything. Other scholars suggest that
voting in gubernatorial contests reflects a combination of national and state evalu-
ations (Svoboda 1995). Moreover, the referendum element is somehow dismissed
when it is difficult for voters to assign responsibility to one party or to another
(e.g. in situations of divided government, see e.g Lowry etal. 1988).
By introducing economic conditions in his analysis of the magnitude of the
losses in midterm elections, Tufte follows, among others, the work of Kramer
(1971). According to him, the best measure of economic conditions is the pre-
election changes in real income per capita (1975: 816), when compared with
the impact of inflation and unemployment. The main hypothesis of the referen-
dum theory then is as follows: ‘The lower the approval rating of the incumbent
President and the less prosperous the economy, the greater the loss of support for
the President’s party in the midterm congressional elections’ (Tufte 1975: 817).
His theory is based on two main assumptions. The first is that the president’s
popularity declines through the term of office and it is relatively low at the time
of the midterm contest (for similar conceptions of an electoral cycle see Mueller
1973 and Stimson 1976). The second is that the state of the economy, because of
appropriate political interventions, tends to be better at the time of the presiden-
tial election.
BK-SAGE-ARZHEIMER_ETAL-160488-Chp04.indd 60 2/13/2017 1:44:41 PM

The STudy of LeSS ImporTanT eLecTIonS 61
One of the main differences from the surge and decline theory is that Tufte’s
referendum theory does not consider a turnout factor in his model of midterm
losses. Tufte measures the congressional vote as the difference between the
national congressional vote for the president’s party and the party’s normal con-
gressional vote (the latter being calculated as the average of its share of the popu-
lar vote over eight previous both on-year and off-year congressional elections).
According to him, this standardization is necessary in order to account for the
variation in the electoral performance of the Democratic and Republican parties
over time (Tufte 1975: 815).
Negative voting theory
This theory was originally proposed by Kernell (1977), who argues that voters
give more attention to failures of the incumbent than to his or her achievements
(Kernell 1977: 52). He explicitly refers to The American Voter (Campbell etal.
1960), according to which a change in the party vote is mainly driven by negative
rather than positive evaluations (see Kernell 1977: 51; Campbell etal. 1960:
554–6). Under the negativity hypothesis, midterm losses are a result of the fact
that judgments on the president’s performance are characterized by a kind of
punitive response (Marsh 2007: 78). In other words, there are more voters dis-
satisfied with the performance of the president than are satisfied, and so the
referendum character of midterm elections tends to be biased against the presi-
dential party. Contrary to Tufte’s work, in Kernell’s analysis turnout does play a
role. According to him, voters who disapprove of the performance of the incum-
bent administration, irrespective of their party identification, turn out in greater
proportions (Kernell 1977: 56). Moreover, those who are dissatisfied with the
incumbent administration are more likely to vote against the congressional can-
didates of the presidential party than satisfied voters are likely to vote for them
(Kernell 1977: 52). Finally, Kernell postulates that dissatisfaction with the presi-
dential performance is more influential in producing defections than approval of
it is. Consequently, he predicts a higher defection rate among voters who identify
with the president’s party and are dissatisfied, compared with those who identify
with the other party but approve of the record of the incumbent administration
(Kernell 1977: 58).
The strategic politician theory
Jacobson and Kernell (1981, 1982) argue that both the ‘surge and decline’ and
the ‘referendum theory’ concentrate on understanding the voters’ choices much
more than on the motivations of politicians. They argue that both the candidates
and those who recruit and finance them do consider the political climate at
midterm. Following Tufte, the political climate at midterm is indicated by the
popularity of the president and by economic performance measures. The
BK-SAGE-ARZHEIMER_ETAL-160488-Chp04.indd 61 2/13/2017 1:44:41 PM

The SAGe hAndbook of elecTorAl behAviour
62
quality of the candidates and the availability of money and other campaign
resources are strategic choices that depend upon these conditions: strong can-
didates will be recruited and the campaign will be well financed when the
economic conditions are good and the president enjoys high levels of popular-
ity. The opposite happens when the political climate is less favorable (Jacobson
and Kernell 1982: 424). Therefore, the outcome of congressional contests is
linked to these strategic choices. As J. Campbell notes, the theory by Jacobson
and Kernell boils down to an elite version of the referenda approach
(J. Campbell 1997a: 75).
The roots of balancing: The presidential penalty theory
Robert Erikson proposed a presidential penalty explanation for midterm losses.
According to this, the electorate punishes the presidential party for being the
party in power, regardless of the quality of its performance (Erikson 1988: 1013).
He identifies two reasons for this behavior. The first is related to the negative
voting approach by Kernell (1977) and assumes that voters at midterm are prone
to cast a protest vote. The second reason is based on some sort of low-cost ration-
ality on the side of the electorate: voters prefer that the same party does not
control both the presidency and Congress. Because of this, they vote against the
president’s party at midterm in order to balance the control of the two institutions
between Democrats and Republicans (Erikson 1988: 1014) in order to promote
policy moderation. He refers to similar ‘balancing’ phenomena in Canadian
provincial elections where the ruling party at the national level tends to suffer
losses. There are of course similar cases further afield such as German state
elections – we will come back to these shortly.
The Alesina and Rosenthal model
The theoretical argument for balancing was further elaborated by Alesina and
Rosenthal (1995). They identify two types of voters: extremists and moderates.
The first are those who want a unified government ruled by their preferred party.
Therefore, they tend to support the same party both in the presidential and in the
midterm contest. Moderate voters, by contrast, split their ticket by supporting
one party for the presidency and another one for the Congress – in an attempt to
moderate either a too-liberal or a too-conservative policy. An important element
in the Alesina and Rosenthal model is the ‘electoral surprise’. They argue that
midterm losses are mainly produced by the sub-group of moderates who were
surprised by the outcome of the presidential election. The probability of splitting
the ticket and counterbalancing the president’s policies is higher when the
degree of electoral surprise is high (Scheve and Tomz 1999: 510). Therefore,
midterm losses are linked to the electoral behavior of the most surprised
moderates.
BK-SAGE-ARZHEIMER_ETAL-160488-Chp04.indd 62 2/13/2017 1:44:41 PM

The STudy of LeSS ImporTanT eLecTIonS 63
expLaining midTerm Losses eLsewhere: from ameriCa To
european muLTi-parTy sysTems
In the late 1970s the theoretical arguments about midterm losses were imported
from the US to European multi-party systems. They were applied with one main
amendment, which originates in the fact that ‘low stimulus’ elections in Europe
are not always held at midterm of the national electoral cycle. Based on the elec-
toral cycle literature for the government popularity in the US (Goodhart and
Bhansali 1970; Stimson 1976) and the UK (Miller and Mackie 1973), it has been
argued that the outcomes of less important elections differ depending on whether
they take place at the beginning, in the middle or towards the end of the national
electoral cycle.
‘Low stimulus’ elections in the German context
Dinkel (1977) pioneered this modification of the analysis of midterm losses. He
observed that the performance of the federal government parties in German state
elections (Landtagswahlen) depends on the timing of such an election within the
federal electoral cycle. Testing a cyclical model of vote losses at state elections,
he concluded that the greater the distance to the federal election, the greater the
losses of the government parties, while their probability of winning a state elec-
tion is greatest at the beginning or at the end of the federal electoral cycle. He
concludes that ‘less important elections – above all state elections – are system-
atically influenced by the superior constellation in the federal parliament (…),
the losses of the governing parties are a function of time within the electoral
cycle at the federal level’ (Dinkel 1978: 63)3.
Dinkel’s model has been tested by a number of scholars. Anderson and Ward
(1996) studied German state elections (and also British by-elections), which they
classified as ‘barometer elections’. These are ‘elections that reflect changes in
citizens’ attitudes toward the government in response to changing political and
economic conditions, absent the direct opportunity to install a new executive or
remove the party in power’ (Anderson and Ward 1996: 448). More focused on
Dinkel’s hypothesis of an electoral cycle, Jeffery and Hough analyze the perfor-
mances of the main government party, the main opposition party, and smaller
parties in relation to the place of the state election in the federal electoral cycle.
They conclude, for the period between 1991 and 1998, that ‘there is no clear
pattern in the relation of Land to federal elections … the trend is one where
the federal–Länder electoral relationship has become less clear and predictable.
Länder elections seem to have become less second-order than they were’ (Jeffery
and Hough 2001: 93). This has been confirmed on the basis of a time series
analysis of ‘Politbarometer’ data by Schmitt and Reif (2003), and was tentatively
explained by the irritations that German re-unification introduced in the formerly
BK-SAGE-ARZHEIMER_ETAL-160488-Chp04.indd 63 2/13/2017 1:44:41 PM

The SAGe hAndbook of elecTorAl behAviour
64
highly stable and predictable German electoral system (see similarly Hough and
Jeffrey 2006: 119). This is also why the results of studies trying to identify a bal-
ancing mechanism in German state elections (Lohmann etal. 1997; Gaines and
Crombez 2004; Kern and Hainmueller 2006) are to be taken with a grain of salt.
Presidential elections in European premier-presidential
regimes
In quite a number of European electoral systems, the post of the president is
filled by way of direct election although the presidential powers are not decisive
for determining the policies of the national government. As these presidential
elections are clearly ‘less important’ than legislative elections, the usual expecta-
tions about their results should apply. However, it is not so much the electoral
behavior of citizens that has been at the center of scholarly attention in this field
(see however Tavits 2009; Magalhães 2007; van der Brug etal. 2000), but rather
the activities of directly and indirectly elected presidents (such as the frequency
of their vetoes, e.g. Köker 2014) or their constitutional powers (Roper 2002).
There is also a certain predominance of country studies and country chapter
books (e.g. Elgie 1999; Hloušek 2013), while truly comparative studies are only
recently becoming more prominent. In short, the study of less important elec-
tions of presidents in premier-presidential systems has been subjected to consti-
tutional and agency issues more than to electoral research. Part of the explanation
of that phenomenon could be that presidential elections per se are more idiosyn-
cratic – because they are tied to the personality of the presidential candidates –
than parliamentary elections are, and regularities might be harder to detect there.
Another part of the story could be that a large number of premier-presidential
systems emerged in the post-communist Eastern Europe after the fall of the Iron
Curtain – a territory in which a stable party system and well-established partisan
ties among citizens are still in the making. This again could have had a detrimen-
tal effect on generalizable findings. It is true at the same time that this research
area is waiting for a comprehensive study pursuing an electoral behavior agenda.
A new kind of ‘less important’ elections was introduced in the European
electoral system in 1979 when the European Parliament was directly elected for
the first time. The holding of European Parliament elections and the subsequent
study of them in the tradition of the US midterm elections research have con-
tributed enormously to the establishment of the study of less important elections
outside the United States.
The seCond-order eLeCTion modeL
Writing in the aftermath of the first supra-national election of the members of the
European Parliament (EP), Reif and Schmitt regarded this new ‘type’ of election
BK-SAGE-ARZHEIMER_ETAL-160488-Chp04.indd 64 2/13/2017 1:44:41 PM

The STudy of LeSS ImporTanT eLecTIonS 65
not primarily as European but as a pale reflection of national elections (Marsh
and Franklin 1996: 11). They distinguished between two types of elections:
‘first-order elections’ (FOE) and ‘second-order elections’ (SOE). The view of EP
elections as SOE captures two elements: that they are secondary compared with
national parliamentary elections and that they are ‘national’ rather than
‘European’ contests (Hix and Marsh 2007: 496). First-order national elections
are those which determine the composition of national governments. These are
national parliamentary or presidential elections, depending on the political
regime. Second-order national elections, by contrast, are less important elec-
tions. In addition to EP elections, other examples of SOE are local, municipal
and regional elections, by-elections in Britain, and midterm elections in the US
and in many presidential systems of Latin America. Actually, mid-term elections
as they are known in the Americas have always been the prototype and prime
example of SOE.
The main hypothesis and also the novelty of the second-order election model
(SOE model) compared with the previous studies of less important contests is
not the distinction of the elections as such, but their reference to different arenas
of a system of multi-level electoral politics. In order to understand the results of
EP elections, one first has to appreciate the decisive role of the political situa-
tion in the first-order political arena at the time when the second-order elections
are being held (Reif and Schmitt 1980: 8). Following Dinkel’s model, Reif and
Schmitt related the extent of the losses for the incumbent party/parties to the tim-
ing of the EP election in the ‘first-order’ electoral cycle. The SOE model does
not focus on economic performance variables. Therefore, if we want to identify a
‘referendum’ element in the SOE model, it is probably best captured by the timing
of the EP elections in the first-order electoral cycle (Marsh 2007: 74). The second
main amendment of the second-order elections model to the broader paradigm of
midterm losses is the inclusion of party properties beyond the government-oppo-
sition status. Irrelevant in the US two-party context, ‘party size’ is of particular
importance in the analysis of ‘low stimulus’ elections in multi-party systems.
The central assumption of the SOE model is that in all these less important
elections, EP elections among them, there is less at stake because the results will
not determine the partisan composition of the executive of the first-order politi-
cal arena (Reif and Schmitt 1980: 9). To be sure, the classification of different
types of elections according to the ‘less at stake’ criterion does not lead to a
simple dichotomy because not all first-order elections are equally important and
not all second-order elections are equally unimportant (van der Eijk etal. 1996).
In addition, the ‘less at stake’ dimension should not be misunderstood as a static
element of the model. This is all the more true in an unfinished electoral system
such as the European Union. It was only in the 2014 election that the selection of
the president of the European Commission was linked (so far indirectly) with the
electoral success of the lead candidates of the EP political groups (Schmitt and
Teperoglou 2015; Schmitt etal. 2015).
BK-SAGE-ARZHEIMER_ETAL-160488-Chp04.indd 65 2/13/2017 1:44:41 PM

The SAGe hAndbook of elecTorAl behAviour
66
Initially, Reif and Schmitt’s effort to understand EP elections results focused
on the following dimensions of electoral contests: the specific-arena dimension,
the institutional-procedural dimension, the campaign, the main-arena political
change dimension and the social and cultural change dimension. Of these the one
that has been most thoroughly studied is probably that of the electoral campaign
(e.g. de Vreese etal. 2007). Most of the studies that followed focused on aggre-
gate predictions of the SOE model based on the less-at-stake dimension. Until the
last EP election of 2014 the model consistently received empirical support (e.g.
Freire 2004; Hix and Marsh 2011; Norris 1997; Reif 1985, 1997; Schmitt and
Teperoglou 2015; Schmitt 2005, 2009; Teperoglou 2010; van der Brug and van
der Eijk 2007; van der Eijk and Franklin 1996).
Aggregate hypotheses of the second-order election model
1 turnout is lower than in first-order elections,
2 national governmental parties lose, following the first-order electoral cycle,
3 large parties do worse,
4 small parties (and in particular new flash and ideologically extreme parties) fare better.
Turnout
Much as in US midterm elections, turnout is expected to be lower in second-
order elections compared with the preceding national first-order contest (Reif
1985: 15). Low turnout does not necessarily indicate a lack of legitimacy of EU
politics, a point that is usually raised by media commentators shortly after each
EP election. One of the main reasons for the lower participation levels is linked
to Campbell’s seminal work (1960): the lack of politicization and electoral mobi-
lization in these second-order elections. For Schmitt and Mannheimer (1991),
the ‘habitual voting’ mechanism is in line with the SOE model. Based on the
post-electoral surveys of the European Elections Study 1989, participation in
that EP election was primarily found to be the result of ‘habitual voting’; citizens
went to the polls just because they were used to doing so on election day, regard-
less of their attitudes towards the EU and their more general political involve-
ment (see also Franklin 2004; Franklin and Hobolt 2010).
Some studies have insisted on an alternative set of factors explaining low
levels of turnout in EP elections. ‘Voluntary abstainers’ as they have been called
by Blondel cum suis (Blondel etal. 1997, 1998) are those citizens who decide
to abstain due to matters related to European politics. For them this is a con-
scious decision related to the lack of viable choice options that can represent
their Euroscepticism without undesired undertones of an e.g. extreme-left or
extreme-right political orientation. So far, the observed effects of attitudes
toward the EU on turnout are typically small – but, given the growing policy
reach of the Union and the increasing opposition this is evoking in the citizenry,
this is not necessarily the case. In the past, the weakness of the ‘European’
BK-SAGE-ARZHEIMER_ETAL-160488-Chp04.indd 66 2/13/2017 1:44:41 PM

The STudy of LeSS ImporTanT eLecTIonS 67
explanation of low turnout has been frequently addressed (e.g. Schmitt and van
der Eijk 2007). Nevertheless, other studies (e.g. Schmitt 2005) that include the
new post-communist member states show that abstentions in EP elections in
Eastern Europe are partly driven by EU-scepticism, while the same does not
hold in older member states. Flickinger and Studlar (2007) find that in the old
EU member states higher aggregate levels of support for EU membership cor-
relate with higher turnout rates, while in the new member states national politics
factors (e.g. trust in government) play a more prominent role. Alternative expla-
nations focus on structural-level causes of abstention in EP elections as the EU
expands (e.g. Franklin 2007a and 2007b).
Franklin, van der Eijk, and Oppenhuis (1996) have argued that the frequency
of the elections may affect electoral participation and some recent studies inves-
tigate whether the second-order nature of EP elections may also depress turnout
in national elections (Franklin and Hobolt 2010). Their main conclusion is that,
due to the failure to mobilize new voters in these elections, it is more likely for
these second-order non-voters not to vote in subsequent national elections either.
Another study (Schmitt etal. 2008, 2009) relates turnout in SOE to dissatisfac-
tion with one’s last first-order vote choice.
Many studies have focused on the individual-level determinants of turnout
(e.g. Franklin etal. 1996; Schmitt 2005). Macro-level systemic and contextual
factors, such as compulsory voting, the size of the party system, the nature of
the electoral system, Sunday voting, last first-order participation rate, concurrent
national elections, and post-communist environment, can be studied in tandem
with individual-level characteristics, such as interest in politics, campaign mobi-
lization, attitudes towards the EU, party attachment etc., as well as with socio-
demographic characteristics.
Time and party size as moderators of parties’
electoral fortunes
Another main hypothesis of the SOE model is that we observe in the EP elec-
tions different electoral performance of the incumbent party (or parties) as a
function of the time distance of the SOE from the previous FOE. The losses for
governing parties will be greater the closer a SOE is located around the midterm
between two FOE. Reif and Schmitt introduced in the analysis of ‘low stimulus’
elections the performance not only of the governmental parties, but also that of
the opposition parties. They hypothesize that these parties will suffer losses too.
While governing parties in EP elections tend to lose votes regardless of their size
(and their party family; see Hix and Marsh 2007: 503), only big opposition par-
ties usually see their electoral percentages shrink (Marsh 2005). Another new
variable in the theoretical paradigm of ‘less important’ elections is the perfor-
mance of the small parties. According to the SOE model, smaller parties are
expected to perform better in the EP elections compared with a first-order
BK-SAGE-ARZHEIMER_ETAL-160488-Chp04.indd 67 2/13/2017 1:44:41 PM

The SAGe hAndbook of elecTorAl behAviour
68
election. Schmitt (2005: 662) suggested using the index of the effective number
of electoral parties (ENEP, introduced by Laakso and Taagepera 1979) to test this
hypothesis. If small parties perform better in the EP election, the ENEP will be
higher compared with that for the preceding FOE. The main rationale behind this
hypothesis is that there is a desire to switch from a strategic FOE to a sincere
SOE vote choice when there is less at stake because voters here can express their
‘true first preference’ even if the chosen party will not gain representation
(Schmitt etal. 2008, 2009).
Do all the small parties, irrespective of their party family, perform better in an
EP election? The findings are contradictory. Reif (1997) claims that small par-
ties’ gains are greater for extreme parties and new parties, while Marsh (2005)
does not find any correlation of electoral success with party extremity. Hix and
Marsh (2007) find that anti-EU parties on average do much better in European
than in national elections. Curtice (1989: 227) suggests that Green parties consis-
tently perform better in EP elections, while Marsh (1998) observes only modest
support.
Main amendments/revisions of the original second-order
election model
From the first study of the 1979 election onwards, we can identify over time
two main amendments of the original model. The first one is that subsequent
studies have offered typologies of voting in second-order elections. Van der
Eijk and Franklin (1996) discovered that one of the major types of voting in
EP elections is a swing into a protest vote against the incumbent government,
the political class, the programs, and/or the candidates of the parties voters
would normally vote for, or to indicate support for a particular policy. Using a
term from the lexicon of football hooliganism, this has been characterized as
‘voting with the boot’ (e.g. Oppenhuis et al. 1996: 301–4; Franklin 2005).
Thus, less salient EP elections offer voters the opportunity to express a (tem-
porary) dissatisfaction with the party that they usually vote for by deciding
either to defect (i.e. vote for another party) or to abstain. Another form of vote
switching is to vote for the party they prefer most, given the fact that the EP
election outcomes thus far cannot be expected to have severe political conse-
quences. Voters do not tend to vote strategically (in terms of not wasting one’s
vote on a party that is unlikely to be a viable contender for government office,
or to vote for a party that might have a destabilizing effect in Parliament) and
vote sincerely ‘with the heart’ (by supporting the first electoral preference).
The reverse holds for ‘voting with the head’ (Oppenhuis etal. 1996: 301–4;
Marsh and Franklin 1996: 16–21; Franklin 2005). Here, voters do behave stra-
tegically by supporting an opposition party to the previously preferred govern-
ment party – thus signaling their discontent with government policy making at
the time of the second-order election.
BK-SAGE-ARZHEIMER_ETAL-160488-Chp04.indd 68 2/13/2017 1:44:41 PM

The STudy of LeSS ImporTanT eLecTIonS 69
For an electoral ‘honeymoon’ period some scholars (Reif and Schmitt 1980)
suggest that the victorious government parties will receive greater or near identi-
cal support in an EP election shortly after the first-order election (post-electoral
euphoria). Others (van der Eijk and Franklin 1996) argue that more voters will
tend to ‘vote with their heart’ during this period. For the later term, van der Eijk
and Franklin (1996) again argue that the closer the EP election comes to the next
FOE, the more likely it is that citizens ‘vote with the boot’ (or ‘strategic defec-
tion’; see Schmitt etal. 2008, 2009), while Reif and Schmitt (1980) claim that
these elections are characterized by some recovery in national government popu-
larity, so that parties will tend to lose fewer votes than during mid-term elections.
For the intermediate period, there is a consensus that losses will be greater than
either early or late (e.g. Reif 1985; Van der Eijk and Franklin 1996). Here we
mainly observe instrumental voting against the government (‘cyclical signaling’;
see Schmitt etal. 2008, 2009).
The aggregate predictions of the second-order election model have ‘overshad-
owed’ the somewhat hidden ‘micro-foundations’ of the SOE model. The second
main revision of the model since its inception in the early 1980s is this research
orientation. These are the hypotheses about the motivations and intentions of indi-
vidual voters that drive their behavior in a second-order election (e.g. Carrubba and
Timpone 2005; Hobolt and Wittrock 2011; Hobolt and Spoon 2012; Schmitt etal.
2008, 2009, and Weber 2011). We can identify three different mechanisms that are
likely to affect inter-election vote patterns according to the SOE model: mobiliza-
tion, sincere voting, and strategic voting (Schmitt etal. 2008, 2009). Mobilization
only has an impact on SOE abstentions, while sincere as well as strategic voting
choices are expected to affect both SOE abstainers and SOE defectors. A series
of micro-level hypotheses can be built around these three mechanisms. We can
distinguish between two main groups of predictions: the first is about turnout, and
the second is about vote switching (Schmitt etal. 2008, 2009).
Micro-level hypotheses of the second-order
election model
Turnout:
1 Mobilization: FOE voters abstain in SOE simply because of the lack of politicization and
mobilization,
2 Sincere abstention: FOE voters abstain in SOE because of their Euroscepticism and the lack of
viable choice options (e.g. Blondel etal. 1997),
3 Strategic abstention: FOE voters abstain in order to signal discontent with the performance of
their previous first-order party choice at the time of the SOE,
4 Cyclical signaling: strategic abstentions are expected to increase with decreasing distance to
first-order midterm. This happens because of the tendency of government popularity to evolve
in a cyclical manner, with a depression shortly after midterm (e.g. Stimson 1976)
BK-SAGE-ARZHEIMER_ETAL-160488-Chp04.indd 69 2/13/2017 1:44:41 PM

The SAGe hAndbook of elecTorAl behAviour
70
Vote switching:
1 Strategic defection: people defect from their FOE vote because of their intention to signal
discontent with the performance of their previous vote choice (see also ‘voting with the boot’),
2 Cyclical signaling: strategic defection should increase with decreasing distance to first-order
midterm,
3 Systemic signaling: strategic defection should increase with increasing clarity of policy respon-
sibilities,
4 Sincere defection: voters support a different party in a SOE because they are closer to it on policy
grounds. We can identify two possible reasons for this. The first one is that in a SOE there is
no danger of wasting one’s vote. The second reason is that arena-specific issues might call for
arena-specific vote choices (e.g. Clark and Rohrschneider 2009)
By-elections and regional elections as SOE
The outcomes of the British by-elections have been analyzed under the ‘midterm
losses’ paradigm mainly by Mughan (1986)4. He built three models of ‘govern-
ment vote losses’: the turnout model (referring among others to Campbell 1960),
the referendum model and the personal vote model. For the latter, Mughan refers
to the work by Cain, Ferejohn, and Fiorina (1984) in which we can also find the
roots of the ‘balancing’ theory (for more details on testing the models, see
Mughan 1986: 767–72). He concluded that by-elections could be seen as
‘national referendums on government performance’ (Mughan 1986: 772–3).
A number of different studies have tested the applicability of the SOE model
to regional elections in Europe, North and Latin America (among others, see
Erikson and Filippov 2001; Jeffery and Hough 2003; Pallarés and Keating 2003;
Schakel and Jeffery 2013; Thorlakson 2015; Remmer and Gélineau 2003).
Overall, two of the SOE hypotheses have been corroborated almost universally:
the first is that turnout is lower. However, provincial elections in Canada seem to
be an exception to this rule; here the turnout difference does not always favor fed-
eral elections, probably due to the felt importance of provincial politics and the
less-than-perfect nationalization (i.e. structural congruence) of provincial party
systems (Thorlakson 2015: 172). The second is that anti-government swings are
observed, and opposition parties fare better – in relative terms. This expectation
is supported by a comparison of Canadian federal and sub-national election out-
comes (Erikson and Filippov 2001). While government parties regularly tend to
lose, the notion of an electoral cycle does not always help in understanding the
SOE electoral results (Schakel 2014: 4; Johnston 1999 for the Canadian case;
Schmitt and Reif 2003 for Germany).
Some other studies have compared different types of SOE in an attempt to
depict a ‘hierarchy’ among them: for the British local and EP elections, Heath,
McLean, Taylor, and Curtice concluded that ‘if elections to the European
Parliament are regarded as second-order, then we might think of elections to local
council as “one and three-quarters order”’ (1999: 391). Another study comparing
BK-SAGE-ARZHEIMER_ETAL-160488-Chp04.indd 70 2/13/2017 1:44:41 PM

The STudy of LeSS ImporTanT eLecTIonS 71
regional and EP elections in Greece, Spain and Portugal indicates that there is a
high dissimilarity between these contests and that voters treat each SOE differ-
ently. Some voters in local elections choose one of the larger parties, instead of
switching to smaller ones, as they do in the European elections. Overall, munici-
pal elections appear to be two-faced by nature. They maintain characteristics of a
first- and second-order election (Skrinis and Teperoglou 2008).
The LimiTs of The seCond-order eLeCTion modeL: when ep
eLeCTions progressiveLy Lose Their seCond-orderness
We identify two main sources that can diminish the adequacy of the SOE model.
The first one is that the utility of the theoretical paradigm by Reif and Schmitt is
getting weaker when applied to analysis of the EP results in the post-communist
countries. The second one has to do with the skeleton of analysis of the model
itself: the ‘less at stake’ dimension. In the 2004 EP election we observed an
inconsistency in testing the model between the old Western EU member coun-
tries (plus Cyprus and Malta), and the eight member countries of Eastern Europe
(see Schmitt 2005; Marsh 2005; Koepke and Ringe 2006). The losses of govern-
mental parties did not follow the FOE cycle and, in some of these eight countries,
smaller parties did not perform better compared with the previous FOE. The
explanation for these ‘failures’ is linked to the fact that the SOE model has been
built upon consolidated electoral and party systems, while in these eight and now
eleven (including Bulgaria, Croatia, and Romania) additional EU countries a
stable and consolidated party system has yet to develop (Schmitt 2005: 666).
Aiming to justify these exceptions, some scholars have tried to refine the SOE
model by introducing in the analysis some new variables, most of them from the
broader ‘referendum theory’ for US midterm losses5. They have related the per-
formance of government not to the timing of the EP election, but rather to the
state of the economy (using indicators such as unemployment, economic growth,
and inflation; see Kousser 2004; Marsh 2005; Schakel 2014 for regional elec-
tions in the post-communist countries). However, the results are inconclusive.
Marsh (2005) states that the timing of the election, rather than economic perfor-
mance, is a better proxy for the governmental losses, while the other two studies
highlight the role of the economy in determining the vote in ‘low stimulus’
elections.
Moving to the second main possible limitation for the SOE model, its main
assumption that there is less at stake in the EP elections compared with the FOEs
is arguably ignoring the growing policy authority of the EU level of European
governance, as well as efforts to strengthen the authority of the European
Parliament in the EU-level policy-making process. We might possibly identify a
‘threat’ for the SOE model coming from the politicization of European integra-
tion in general (Hooghe and Marks 2009): political conflicts increasingly cross
BK-SAGE-ARZHEIMER_ETAL-160488-Chp04.indd 71 2/13/2017 1:44:41 PM

The SAGe hAndbook of elecTorAl behAviour
72
national boundaries on the ‘higher level’ of the multi-level system of European
governance.
A central question for the 2014 EP election was whether changing socio-polit-
ical circumstances and periods of economic turmoil, such as those experienced
across Southern Europe, have the potential to change the second-order character
of the contest. Have these EP elections become ‘critical elections’ in the sense
of V. O. Key (1955)6 or have the SOE characteristics prevailed? A major conclu-
sion for the study of these elections in the countries of Southern Europe is that
the contest has maintained its characteristics of a classic SOE. Nevertheless, we
can also observe indications of a more critical contest in which the EU divide
becomes more polarized and important in determining voting choices. This how-
ever is concentrated primarily on FOE under conditions of high electoral mobi-
lization. European Parliament elections then primarily serve as ‘confirmatory’
elections in which realignments that occurred in the first-order electoral arena
are ‘rubber-stamped’ (see Teperoglou etal. 2015; Schmitt and Teperoglou 2015).
ConCLusion
In this chapter we have made an effort to review the findings of more than half
a century of research into electoral behavior in less important elections –
midterm elections in the US, and various types of second-order elections in
Europe (sub-national elections of the members of state parliaments, or simulta-
neous supra-national elections of the members of the European Parliament). The
particular property of this field of electoral research is that electoral decisions –
to participate or not in a contest and, if so, which party to support – is never
analyzed in isolation, but usually with reference to another such decision in an
election of greater importance. Therefore, we can also think about this stream of
analyses as focusing on inter-election voting patterns.
All started in the 1950s in the US when the ‘surge and decline’ mechanism of
electoral support between on-year elections (presidential elections) and off-year
elections (midterm congressional elections) was identified. Pretty much a child
of its time, this understanding of inter-election variation of voting behavior con-
centrated on the informational and mobilization aspects of the game. In this view,
an equilibrium of partisan sorting was reached in ‘low stimulus’ midterm elec-
tions while, two years later, at ‘high stimulus’ presidential elections, the swing of
the pendulum was affected by short-term factors of various kinds. More particu-
larly, issue and leadership effects were observed to gain greater importance com-
pared with the normal vote at midterm, which was mostly driven by partisanship.
Some twenty years later, in the 1970s, a rational-choice-inspired perspective on
inter-election change gained in prominence – the ‘referendum’ approach, which
focused more on the economic achievements and shortcomings of the incumbent
administration at midterm than on partisanship and mobilization.
BK-SAGE-ARZHEIMER_ETAL-160488-Chp04.indd 72 2/13/2017 1:44:41 PM

The STudy of LeSS ImporTanT eLecTIonS 73
At the same time, this kind of electoral research crossed the Atlantic and
spilled over to Europe in order to inspire the study of electoral behavior in less
important elections there. Applying US-grown models of political behavior in
less important elections, there were important institutional and procedural differ-
ences to overcome. One was that ‘midterm’ elections in Europe are not always
held at midterm, but at various times in the national electoral cycle. Another was
that party politics in Europe was (and still is) somewhat more variegated com-
pared with its counter-image in the US: coalition governments are the rule rather
than the exception in Europe, and there are relevant parties of very different sizes
and ideological extremity which are competing for electoral representation.
These were some of the challenges that inspired a European answer to the
US-based midterm elections literature. There are, however, two more. One addi-
tional challenge was the introduction of a series of less important elections in the
calendar of national electoral politics of the member countries of the European
Union (which called itself European Community at the time): direct elections
of the members of the European Parliament. After decades of deliberation and
political controversy, this project – which aimed at democratizing European level
politics – finally took off in June 1979. The scholarly answer here was a pro-
posal to consider these new electoral contests as additional national second-order
elections held simultaneously in the whole Union but at different places in the
national electoral cycles.
An additional challenge for this field of studies was presenting itself roughly
a quarter of a century later. After the collapse of the Soviet empire, many post-
communist Eastern European countries became members of the European Union,
and were included in the electoral process of selecting their representatives for the
European Parliament. The challenge here was that these newly democratizing poli-
ties had not (and in part still have not) developed the kind of consolidated party sys-
tems which the second-order election theory assumed as a stable basis of electoral
competition. As a consequence, many predictions of the SOE model – such as that
government parties lose, or that small parties have brighter prospects in these con-
tests – could not be corroborated in the elections of these new member countries.
Where does that lead us to and what are the likely challenges ahead? The study
of electoral behavior in less important elections, we believe, has three important
tasks to address. One is to increase the outreach of our theories, to increase the
level of generalization. This asks for more mutual recognition of the work that is
done in the US and elsewhere, but in addition for a new class of SOE research at
a higher level that would combine e.g. the analysis of electoral behavior in US
midterm elections, EP elections, and state elections in federal systems.
A second and related task is to include contextual parameters in the multi-level
analysis of electoral behavior in different kinds of elections. The properties of
the political regime, the electoral system applied, the format of party competi-
tion, the degree of ideological polarization, these and more contextual parameters
are expected to have an impact on the electoral choices. The broader the scope
BK-SAGE-ARZHEIMER_ETAL-160488-Chp04.indd 73 2/13/2017 1:44:41 PM

The SAGe hAndbook of elecTorAl behAviour
74
of analysis, the more promising are these multi-level approaches to the study of
electoral behavior.
And having said this, an important third task of inter-election voting studies
is to concentrate more than in the past on the micro-foundations of electoral
behavior in less important elections. How important are balancing motivations
beyond the narrow confines of a two-party system in a presidential regime
such as the US? What is the role and relevance of arena-specific issue voting in
European Parliament elections, at different times and in different member coun-
tries? Moreover, are less important elections, such as the SOE to the European
Parliament, an arena used primarily for strategic purposes – in order to signal
discontent e.g. with the policy output of the national political process, or are
they – because there is less at stake – rather an opportunity to vote sincerely? It
is not that we do not know enough to answer these questions. But we certainly
do not know enough to construct a comprehensive model of inter-election voting
decisions.
Notes
1 The explanation in terms of withdrawn coattails is an example of statistical ‘regression to the mean’
(Erikson 1988: 1012).
2 The coattail voting approach has also been tested in some Latin American countries. Evidence from
Brazil suggests that we have to take into account the extent of political decentralization, which is
likely to produce ‘reverse coattail effects’ (Ames 1994; Samuels 2000).
3 G. Fabritius (although employing a qualitative analysis) arrives at similar conclusions; see G. Fabritius
‘Sind Landtagswahlen Bundesteilwahlen?’, Aus Politik und Zeitgeschichte, B21/79: 23–38.
4 We refer to by-elections in those parliamentary systems where there is no possibility to fill automati-
cally a casual vacancy (caused either by the death or resignation of a member of the Parliament) by
the next person on the list (Mughan 1986: 762).
5 Refinements of the original SOE model could also be formulated by studying the SOE micro-foundations;
see among others Hobolt etal. 2008; Weber 2007.
6 According to Key (1955), elections are critical when the traditional coalitions between social groups
and their political agents are subject to profound and lasting realignments.
referenCes
Alesina, A. and Rosenthal, H. (1989) ‘Partisan cycles in congressional elections and the macroeconomy’,
American Political Science Review, 83(2): 373–98.
Alesina, A. and Rosenthal, H. (1995) Partisan Politics, Divided Government, and the Economy. New York:
Cambridge University Press.
Ames, B. (1994) ‘The reverse coattails effect: local party organization in the 1989 Brazilian presidential
election’, American Political Science Review, 88(1): 95–111.
Anderson, C. J. and Ward, D. S. (1996) ‘Barometer elections in comparative perspective’, Electoral
Studies, 15(4): 447–60.
Anderson, C. J. (2000) ‘Economic voting and political context: a comparative perspective’, Electoral
Studies, 19(2–3): 151–70.
BK-SAGE-ARZHEIMER_ETAL-160488-Chp04.indd 74 2/13/2017 1:44:41 PM

The STudy of LeSS ImporTanT eLecTIonS 75
Bean, L. (1950) The Midterm Battle. Washington: Business Press.
Blondel, J., Sinnott, R. and Svensson, P. (1997) ‘Representation and voter participation’, European
Journal of Political Research, 32(2): 243–72.
Blondel, J., Sinnott, R. and Svensson, P. (1998) People and Parliament in the European Union:
Participation, Democracy, and Legitimacy. Oxford: Oxford University Press.
Cain, B., Ferejohn, J. and Fiorina, M. (1984) ‘The constituency service basis of the personal vote for US
representatives and British members of parliament’, American Political Science Review, 78(1):
110–25.
Calvert, R. L. and Ferejohn, J. A. (1983) ‘Coattail voting in recent presidential elections’, American
Political Science Review, 77(2): 407–19.
Campbell, A. (1960) ‘Surge and decline: a study of electoral change’, Public Opinion Quarterly, 24(3):
397–418.
Campbell, A., Converse, P., Miller, W. and Stokes, D. (1960) The American Voter. New York: Wiley.
Campbell, J. (1997a) The Presidential Pulse of Congressional Elections. 2nd edn, Kentucky: The University
Press of Kentucky.
Campbell, J. (1997b) ‘The revised theory of surge and decline’, American Journal of Political Science,
31(4): 965–79.
Carrubba, C. J. and Timpone, R. J. (2005) ‘Explaining vote switching across first and second order elec-
tions: evidence from Europe’, Comparative Political Studies, 38(3): 260–81.
Clark, N. and Rohrschneider, R. (2009) ‘Second-order elections versus first-order thinking: how voters
perceive the representation process in a multi-layered system of governance’, Journal of European
Integration, 31(5): 645–64.
Curtice, J. (1989) ‘The 1989 European election: protest or green tide’, Electoral Studies, 8(3): 217–30.
de Vreese, C. H., Lauf, E. and Peter, J. (2007) ‘The media and European Parliament elections: second-rate
coverage of a second-order event?’. In W. van der Brug and C. van der Eijk (eds), European Elections
and Domestic Politics: Lessons from the Past and Scenarios for the Future. Southbend: University of
Notre Dame Press. pp. 129–47.
Dinkel, R. (1977) ‘Der Zusammenhang zwischen Bundes- und Landtagswahlergebnissen’. In M. Kaase
(ed.), Wahlsoziologie heute. Analysen aus Anlass der Bundestagswahl 1976. Opladen: Westdeutscher
Verlag. pp.348–59.
Dinkel, R. (1978) ‘The relationship between federal and state elections in West Germany’. In M. Kaase
and K. Von Beyme (eds), Elections and Parties. London: Sage. pp. 53–65.
Elgie, R. (ed.) (1999) Semi-Presidentialism in Europe. Oxford: Oxford University Press.
Erikson, R. S. (1988) ‘The puzzle of mid-term loss’, Journal of Politics, 50(4): 1011–29.
Erikson, R. S. and Filippov, M. G. (2001) ‘Electoral balancing in federal and sub-national elections: the
case of Canada’, Constitutional Political Economy, 12: 313–31.
Fabritius, G. (1979) ‘Sind Landtagswahlen Bundesteilwahlen?’, Aus Politik und Zeitgeschichte, B21/79:
23–38.
Flickinger, S. and Studlar, D. T. (2007) ‘One Europe, many electorates? Models of turnout in European
Parliament elections after 2004’, Comparative Political Studies, 40(4): 383–404.
Franklin, M. and Hobolt, S. (2010) ‘The legacy of lethargy: how elections to the European Parliament
depress turnout’, Electoral Studies, 30(1): 67–76.
Franklin, M. (2004) Voter Turnout and the Dynamics of Electoral Competition in Established Democracies
since 1945. Cambridge: Cambridge University Press.
Franklin, M. (2005) ‘The fading power of national politics to structure voting behaviour in elections to
the European Parliament’, paper presented at the Conference on the 2004 European Elections,
Central European University, Budapest.
Franklin, M. (2007a) ‘Effects of space and time on turnout in European parliament elections’. In W. van
der Brug and C. van der Eijk (eds), European Elections and Domestic Politics: Lessons from the Past
and Scenarios for the Future. Southbend: University of Notre Dame Press. pp. 13–32.
BK-SAGE-ARZHEIMER_ETAL-160488-Chp04.indd 75 2/13/2017 1:44:42 PM

The SAGe hAndbook of elecTorAl behAviour
76
Franklin, M. (2007b) ‘Turning out or turning off? How the European parliament elections of 2004 shed
light on turnout dynamics’. In M. Marsh, S. Mikhaylov and H. Schmitt (eds), European Elections after
Eastern Enlargement: Preliminary Results from the European Election Study 2004. Mannheim:
CONNEX Report Series. pp. 53–70.
Franklin, M., van der Eijk, C. and Oppenhuis, E. (1996) ‘The institutional context: turnout’. In C. van der
Eijk and M. Franklin (eds), Choosing Europe? The European Electorate and National Politics in the
Face of Union. Ann Arbor: University of Michigan Press. pp. 306–32.
Freire, A. (2004) ‘Second-order elections and electoral cycles in democratic Portugal’, South European
Society and Politics, 9(3): 54–79.
Gaines, B. J. and Crombez, C. (2004) ‘Another look at connections across German Elections’, Journal of
Theoretical Politics, 16(3): 289–319.
Goodhart, C. A. E. and Bhansali, R. J. (1970) ‘Political economy’, Political Studies, 18: 43–106.
Heath, A., McLean, I., Taylor, B. and Curtice, J. (1999) ‘Between first and second-order: a comparison of
voting behaviour in European and local elections in Britain’, European Journal of Political Research,
35(3): 389–414.
Hinckley, B. (1967) ‘Interpreting House midterm elections: toward a measurement of the in-party’s
“expected” loss of seats’, American Political Science Review, 61(3): 694–700.
Hix, S. and Marsh, M. (2007) ‘Punishment or protest? Understanding European Parliament elections’,
Journal of Politics, 69(2): 495–510.
Hix, S. and Marsh, M. (2011) ‘Second-order effects plus pan-European political swings: an analysis of
European Parliament elections across time’, Electoral Studies, 30(1): 4–15.
Hloušek, V. (ed.) (2013) Presidents above Parties? Presidents in Central and Eastern Europe, their Formal
Competencies and Informal Power. Brno: Masaryk University Press.
Hobolt, S. and Spoon, J-J. (2012) ‘Motivating the European voter: parties, issues and campaigns in
European Parliament elections’, European Journal of Political Research, 51(6): 701–27.
Hobolt, S. and Wittrock, J. (2011) ‘The second order election model revisited: an experimental test of
vote choices in European parliament elections’, Electoral Studies, 30(1): 29–40.
Hobolt, S., Spoon, J-J. and Tilley, J. (2008) ‘A vote against Europe? Explaining defection at the 1999 and
2004 European Parliament elections’, British Journal of Political Science, 39(1): 93–115.
Hooghe, L. and Marks, G. (2009) ‘A post-functionalist theory of European integration: from permissive
consensus to constraining dissensus’, British Journal of Political Science, 39(1): 1–23.
Hough, D. and Jeffrey, Ch. (2006) ‘Germany: An erosion of federal-Länder linkages. In Devolution and
Electoral Politics, eds. D. Hough and Ch. Jeffrey. Manchester: Manchester University Press.
Jacobson, G. C. and Kernell, S. (1981) Strategy and Choice in Congressional Elections. New Haven: Yale
University Press.
Jacobson, G. C. and Kernell, S. (1982) ‘Strategy and choice in the 1982 congressional elections’, Political
Science and Politics, 15(3): 423–30.
Jeffery, C. and Hough, D. (2001) ‘The electoral cycle and multi-level voting in Germany’, German Politics,
10: 73–98.
Jeffery, C. and Hough, D. (2003) ‘Regional elections in multi-level systems’, European Urban and
Regional Studies, 10(3): 199–212.
Johnston, R. (1999) ‘Business cycles, political cycles and the popularity of Canadian governments,
1974–1998’, Canadian Journal of Political Science, 32(3): 499–520.
Kern, H. L. and Hainmueller, J. (2006) ‘Electoral balancing, divided government and “midterm” loss in
German elections’, The Journal of Legislative Studies, 12(2): 127–49.
Kernell, S. (1977) ‘Presidential popularity and negative voting: an alternative explanation of the midterm
congressional decline of the president’s party’, American Political Science Review, 71(1): 44–66.
Key, V. (1955) ‘A theory of critical elections’, The Journal of Politics, 17(1): 145–55.
Köker, P. (2014) Veto et Peto: Patterns of Presidential Activism in Central and Eastern Europe. London:
University College London (PhD diss.).
BK-SAGE-ARZHEIMER_ETAL-160488-Chp04.indd 76 2/13/2017 1:44:42 PM

The STudy of LeSS ImporTanT eLecTIonS 77
Koepke, J. R. and Ringe, N. (2006) ‘The second-order election model in an enlarged Europe’, European
Union Politics, 7(3): 321–46.
Kousser, T. (2004) ‘Retrospective voting and strategic behavior in European Parliament elections’,
Electoral Studies, 23(1): 1–21.
Kramer, G. H. (1971) ‘Short-term fluctuations in US voting behavior. 1986–1964’, American Political
Science Review, 65(1): 131–43.
Laakso, M. and Taagepera, R. (1979) ‘Effective number of parties: a measure with application to West
Europe’, Comparative Political Studies, 2(1): 3–27.
Lewis-Beck, S. (1988) Economics and Elections: The Major Western Democracies. Ann Arbor: University
of Michigan Press.
Lewis-Beck, M. and Stegmaier, M. (2000) ‘Economic determinants of electoral outcomes’, Annual
Review of Political Science, 3(1): 183–219.
Lohmann, S., Brady, D. W. and Rivers, D. (1997) ‘Party identification, retrospective voting, and moderating
elections in a federal system – West Germany 1961–1989’, Comparative Political Studies, 30(4):
420–49.
Lowry, R. C., Alt, J. E. and Ferree, K. E. (1998) ‘Fiscal policy outcomes and electoral accountability in
American states’, American Political Science Review, 92(4): 759–74.
Magalhães, P. C. (2007) ‘What are (semi-) presidential elections about? A case study of the Portuguese
2006 elections’, Journal of Elections, Public Opinion and Parties, 17(3): 263–91.
Marsh, M. and Mikhaylov, S. (2010) ‘European Parliament elections and EU governance’, Living Reviews
in European Governance, 5(4): 5–30.
Marsh, M. (1998) ‘Testing the second-order election model after four European elections’, British Journal
of Political Science, 28(4): 591–607.
Marsh, M. (2005) ‘The results of the 2004 European parliament elections and the second-order model’.
In O. Niedermayer and H. Schmitt (eds), Europawahl 2004. Wiesbaden: VS Verlag fur
Sozialwissenschaften. pp. 142–58.
Marsh, M. (2007) ‘Vote switching in the European parliament elections: evidence from June 2004’. In
M. Marsh, S. Mikhaylov and H. Schmitt (eds), European Elections after Eastern Enlargement:
Preliminary Results from the European Election Study 2004. Mannheim: CONNEX Report Series. pp.
71–95.
Marsh, M. and Franklin, M. (1996) ‘The foundations: unanswered questions from the study of European
elections, 1979–1994’. In C. van der Eijk and M. Franklin (eds), Choosing Europe? The European
Electorate and National Politics in the Face of Union. Ann Arbor: University of Michigan Press. pp. 11–33.
Miller, W. E. (1955) ‘Presidential coattails: a study in political myth and methodology’, Public Opinion
Quarterly, 19(4): 353–68.
Miller, W. L. and Mackie, M. (1973) ‘The electoral cycle and the asymmetry of government and opposition
popularity’, Political Studies, 21(3): 263–79.
Mueller, J.E. (1973) War, Presidents, and Public Opinion, New York: Wiley.
Mughan, A. (1986) ‘Toward a political explanation of government vote losses in midterm by-elections’,
American Political Science Review, 80(3): 761–75.
Norris, P. (1997) ‘Nomination: second order elections revisited’, European Journal of Political Research,
31(1–2): 109–114.
Oppenhuis, E., van der Eijk, C. and Franklin, M. (1996) ‘The party context: outcomes’. In C. van der Eijk
and M. Franklin (eds), Choosing Europe? The European Electorate and National Politics in the Face
of Union. Ann Arbor: University of Michigan Press, pp. 287–306.
Pallarés, F. and Keating, M. (2003) ‘Multi-level electoral competition: regional elections and party
systems in Spain’, European Urban and Regional Studies, 10(3): 239–55.
Reif, K. (1985) ‘Ten second-order national elections’. In K. Reif (ed.), Ten European Elections. Campaigns
and Results of the 1979/81 First Direct Elections to the European Parliament. Aldershot: Gower. pp.
1–36.
BK-SAGE-ARZHEIMER_ETAL-160488-Chp04.indd 77 2/13/2017 1:44:42 PM

The SAGe hAndbook of elecTorAl behAviour
78
Reif, K. (1997) ‘European elections as member state second-order elections revisited’, European Journal
of Political Research, 31(1–2): 115–24.
Reif, K. and Schmitt, H. (1980) ‘Nine second-order national elections; a conceptual framework for the
analyses of European election results’, European Journal of Political Research, 8(3): 3–44.
Remmer, K. L. and Gélineau, F. (2003) ‘Subnational electoral choice. Economic and referendum voting in
Argentina, 1983–1999’, Comparative Political Studies, 36(7): 801–21.
Roper, S. D. (2002). ‘Are all semipresidential regimes the same? A comparison of premier-presidential
regimes’, Comparative Politics, 34(3): 253–72.
Samuels, D. J. (2000) ‘The Gubernatorial coattails effect: federalism and congressional elections in
Brazil’, The Journal of Politics, 62(1): 240–53.
Schakel, A. H. and Jeffery, C. (2013) ‘Are regional elections really second-order?’, Regional Studies, 48(3):
323–41.
Schakel, A. H. (2014) ‘How to analyze second-order election effects? A refined second-order election
model’, Comparative European Politics, first published online July 7, 2014, doi: 10.1057/
cep.2014.16.
Scheve, K. and Tomz, M. (1999) ‘Electoral surprise and the midterm loss in US congressional elections’,
British Journal of Political Science, 29(3): 507–21.
Schmitt, H. and Teperoglou, E. (2015) ‘The 2014 European Parliament elections in Southern Europe:
second-order or critical elections?’, South European Society and Politics, 20(3): 287–309, doi:
10.1080/13608746.2015.1078271.
Schmitt, H., Hobolt, S. and Popa, S. (2015) ‘Does personalization increase turnout? “Spitzenkandidaten”
in the 2014 European Parliament elections’, European Union Politics, 16(3): 347–68.
Schmitt, H. and van der Eijk, C. (2007) ‘Non-voting in European Parliament elections and support for
European integration’. In W. van der Brug and C. van der Eijk (eds), European Elections and Domestic
Politics: Lessons from the Past and Scenarios for the Future. Southbend: University of Notre Dame
Press. pp. 145–67.
Schmitt, H., Sanz, A. and Braun, D. (2008) ‘The micro-foundations of second-order election theory: theo-
retical reconstructions and empirical tests based upon the European election study 2004’, paper
presented at the CSES Workshop, University of Manchester.
Schmitt, H., Sanz, A. and Braun, D. (2009) ‘Motive individuellen Wahlverhaltens in Nebenwahlen: eine
theoretische Rekonstruktion und empirische Überprüfung’. In J. W. Falter, O. Gabriel and B. Wessels
(eds), Wahlen und Wähler. Analysen aus Anlass der Bundestagswahl 2005. Wiesbaden: Verlag für
Sozialwissenschaften. pp. 585–605.
Schmitt, H. and Mannheimer, R. (1991) ‘About voting and non-voting in the European elections in June
1989’, European Journal of Political Research, 19(1): 31–54.
Schmitt, H. and Reif, K. (2003) ‘Der Hauptwahlzyklus und die Ergebnisse von Nebenwahlen’. In A. Wüst
(ed.), Politbarometer. Opladen: Leske und Budrich.pp. 239–54.
Schmitt, H. (2005) ‘The European Parliament elections of June 2004: still second order?’, West European
Politics, 28(3): 650–79.
Schmitt, H. (2009) ‘Introduction’, special issue on ‘European Parliament Elections after the Eastern
Enlargement’, Journal of European Integration, 31: 525–36.
Simon, D. M., Ostrom, C. W. and Marra, R. F. (1991) ‘The president, referendum voting, and subnational
elections in the United States’, American Political Science Review, 85(4): 1177–92.
Simon, D. M. (1989) ‘Presidents, governors, and electoral accountability’, Journal of Politics, 51(2):
286–304.
Skrinis, S. and Teperoglou, E. (2008) ‘Studying and comparing second-order elections. Examples from
Greece, Portugal and Spain’. In C. van der Eijk and H. Schmitt (eds.), The Multilevel Electoral System
of the EU. Mannheim: CONNEX Report Series No 4. pp. 163–89.
Stimson, J. (1976) ‘Public support for American presidents: a cyclical model’, Public Opinion Quarterly,
40: 1–21.
BK-SAGE-ARZHEIMER_ETAL-160488-Chp04.indd 78 2/13/2017 1:44:42 PM

The STudy of LeSS ImporTanT eLecTIonS 79
Svoboda, C. J. (1995) ‘Retrospective voting in gubernatorial elections: 1982–1986’, Political Research
Quarterly, 48(1): 135–50.
Tavits, M. (2009) Presidents with Prime Ministers: Do Direct Elections Matter?. Oxford: Oxford University
Press.
Teperoglou, E. (2010) ‘A chance to blame the government? The 2009 European election in Southern
Europe’, South European Society and Politics, 15(2): 247–72.
Teperoglou, E., Tsatsanis, E. and Nicolacopoulos, El. (2015) ‘Habituating to the new normal in a
postearthquake party system: The 2014 European Election in Greece’, South European Society and
Politics, 20 (3): 333–55.
Thorlakson, L. (2015) ‘Explaining the federal-provincial turnout gap in the Canadian provinces’,
Canadian Political Science Review, 9(1): 164–76.
Tufte, E. (1975) ‘Determinants and outcomes of midterm congressional elections’, American Political
Science Review, 69(3): 812–26.
Tufte, E. (1978) Political Control of the Economy. Princeton: Princeton University Press.
van der Brug, W. and van der Eijk, C. (eds) (2007) European Elections and Domestic Politics: Lessons from
the Past and Scenarios for the Future. Southbend: University of Notre Dame Press.
van der Brug, W., van der Eijk, C. and Franklin, M. (2007) The Economy and the Vote. Economic
Conditions and Elections in Fifteen Countries. New York: Cambridge University Press.
van der Brug, W., van der Eijk, C. and Marsh, M. (2000) ‘Exploring uncharted territory: the Irish
presidential election, 1997’, British Journal of Political Science, 30: 631–50.
van der Eijk, C. and Franklin, M. (1996) Choosing Europe? The European Electorate and National Politics
in the Face of the Union. Ann Arbor: University of Michigan Press.
van der Eijk, C., Franklin, M. and Marsh, M. (1996) ‘What voters teach us about Europe-wide elections:
what Europe-wide elections teach us about voters’, Electoral Studies, 15(2): 149–66.
Weber, T. (2007) ‘Campaign effects and second-order cycles: a top-down approach to European
Parliament elections’, European Union Politics, 8(4): 509–36.
Weber, T., (2011) ‘Exit, voice, and cyclicality: a micro-logic of midterm effects in European parliament
elections’, American Journal of Political Science, 55(4): 907–22.
BK-SAGE-ARZHEIMER_ETAL-160488-Chp04.indd 79 2/13/2017 1:44:42 PM