Changing crises, changing votes?
Problem priorities, party competence, and electoral behavior in Germany 2009–2017
Agatha Kratz, Maria Preißinger, Harald Schoen
This is an unedited pre-print. Please cite the original study: Agatha Kratz, Maria Preißinger,
Harald Schoen (2020): ‘Changing crises, changing votes? Problem priorities, party
competence, and electoral behavior in Germany 2009–2017’, in: 'The Changing German
Voter',Oxford University Press, edited by: Rüdiger Schmitt-Beck, Sigrid Roßteutscher,
Harald Schoen, Bernhard Weßels, and Christof Wolf.
In Germany, the occurrence of several successive crises between 2008 and 2017—the
economic and financial crisis, the Euro crisis, and the refugee crisis—went hand in hand with
high electoral volatility. This contribution examines if these phenomena were related. An
event-driven vote model suggests a causal chain connecting crises to vote switching over
changes in individual problem priorities and party competence attributions. Using data from
panel surveys, we demonstrate that this chain is actually weaker than the event-driven vote
model suggests. First, problem priorities and party competence attributions were not purely
event-driven but rather endogenous to political predispositions and policy preferences.
Second, although changes in party competence attributions promoted vote switching, other
influential factors limited their impact. In effect, the sequence of the three crises contributed
to the high level of electoral volatility only moderately through changes in problem priorities
and party competence attributions.
Keywords: economic crisis, Euro crisis, refugee crisis, inter-election switching, issue
ownership, electoral volatility
In the decade following 2008, politics in Germany hardly qualified as “politics as usual”.
According to contemporaneous accounts, the period was overshadowed by a series of crises
(e.g., Zohlnhöfer 2011; Zohlnhöfer and Saalfeld 2015, 2019). It began with the world
financial and economic crisis, which hit Germany severely. As an aftershock, beginning in
2010, the European sovereign debt crisis put a strain on the European Monetary Union
(EMU). Though not fully resolved yet, the Euro crisis was subsequently overshadowed by the
refugee crisis, which peaked in fall 2015. This sequence of crises attracted considerable media
attention (e.g., Kratz and Schoen 2017: 50) and provided an important background for the
federal elections between 2009 and 2017. In these elections, aggregate volatility of the votes
reached the highest level since the early days of the Federal Republic: almost half of the
voters switched parties from one election to the next (Blumenstiel and Wiegand 2014; Schoen
2019b). Given this striking simultaneity of the succession of several severe crises and the
rising electoral volatility, it is tempting to speculate that the latter was a consequence of the
former. At face value, it certainly seems highly appealing to interpret the three crises as
stimuli that caused voters to switch their party preferences.
A plausible way to conceive of the link between the sequence of crises and electoral volatility
starts with the idea that the three crises made issues related to the respective crisis salient for
voters and thus led to changes in the public agenda. If voters attributed the competence to
solve these changing problems to different parties, the shifts in issue salience may well have
led to changes in individual-level vote choice. Though convincing at first sight, this line of
reasoning loses some of its initial appeal when taking a closer look. To begin with, voters’
problem priorities and competence attributions may to some extent reflect not the flow of
external events but rather personal dispositions, such as partisanship or policy preferences
(e.g., Bellucci 2006; Stubager and Slothuus 2013). Some people might be concerned about
similar problems at all times and deem parties competent that correspond to their party
attachments and policy preferences. Moreover, even if they changed, competence attributions
might not necessarily make a crucial difference in voters’ decision-making. Instead, their
impact on vote choices might be trumped by other short-term factors, such as candidate
orientations, or by long-term factors, such as party identifications (e.g., Campbell et al. 1960).
In effect, event-driven changes in voters’ problem priorities and competence attributions may
not be very effective in changing vote choice. Accordingly, the three crises might ultimately
not have contributed significantly to the shifts in voting behavior that were observed between
2009 and 2017.
In this chapter, we examine the relationship between the succession of crises experienced by
German voters between 2008 and 2017 and the changes in voters’ behavior during this period.
Specifically, we explore whether the succession of crises led to changes in voting along the
lines outlined above. In the next section, we discuss the causal chain that links changing crises
to changes in voting behavior as well as possible objections that might be raised against it. We
then derive expectations from these considerations for the federal elections 2009, 2013, and
2017. After a brief description of the research design, we present the results from analyses
using cross-sectional data as well as panel data. The chapter concludes by summing up key
From 2008 to 2017, the German public was confronted with a sequence of crises that had the
potential to lead to major shifts in the electorate’s problem priorities (e.g., Zohlnhöfer 2011;
Zohlnhöfer and Saalfeld 2015, 2019). These changes, in turn, may have brought about shifts
in voters’ perceptions of which parties were most competent to address the problems
dominating their agenda. As competence attributions provide clear cues for electoral choices
(e.g., Bélanger and Meguid 2008; Green and Hobolt 2008), these changes may finally have
led to shifts in voters’ choices, and consequently also in aggregate electoral outcomes. This
would be well in line with the observation that this period was characterized by high inter-
election volatility. In effect, as a series of events, the three successive crises may have had a
major impact on shifts in popular attention, party support, and ultimately election outcomes.
This line of reasoning can be characterized as an “event-driven vote model” of crisis-related
This model builds on a set of assumptions that are not uncontested, however. For instance,
existing research suggests that problem priorities are to some extent endogenous to voter
characteristics rather than completely driven by external factors. Values and policy-related
predispositions appear to make voters chronically attentive to certain topics, and inattentive to
others (Boninger et al. 1995; Kratz and Schoen 2017; Rössler 1997). Voters might also follow
party cues when identifying political problems, particularly during campaigns (Bélanger and
Meguid 2008; Bellucci 2006; also see, e.g., Clarke et al. 2004; Damore 2004). This suggests
that voters’ problem perceptions not only might have been responsive to the three crises but
also were conditioned by policy- and party-related predispositions. Given this potential
endogeneity, some voters’ problem priorities might indeed have remained unchanged over
time, irrespective of external events. This, in turn, would limit the potential of crises to bring
about major shifts in party support.
Concerning the perceived competence of parties to tackle a specific issue (on controversies
about this concept, see, e.g, Green and Jennings 2017b), the event-driven vote model works
nicely with the issue ownership perspective. According to this view, many voters share
stereotypes about parties’ competence in tackling certain problems (see, e.g., Budge and
Farlie 1983; Damore 2004; Petrocik 1991, 1996; Petrocik et al. 2003).1 Whether a sequence of
crises with accompanying shifts of problem priorities brings about changes in party support
should therefore depend on the nature of the focal problems and related party images. If
1 To avoid conceptual confusion, we do not use the term “issue ownership” to denote competence attributions of
individual voters (e.g., Bélanger and Meguid 2008; Stubager 2018) but to describe a party’s advantage in the
distribution of competence attributions at the aggregate level.
voters consider the same party capable of addressing all issues that become salient across
several successive crises, no change in party support is to be expected despite the shifting
public agenda. However, when different parties are considered competent in dealing with the
various problems that relate to these crises, electoral support should undergo change.
However, objections can be raised against these assumptions. From the perspective of valence
politics (e.g., Clarke et al. 2004, 2009; Whiteley et al. 2013), there is no agreed-upon link
between problems and parties that are considered competent in tackling them (see, e.g.,
Anderson and Hecht 2012; Clarke et al. 2011; Lewis-Beck et al. 2013; Nezi 2012 on
economic voting). Following this line of reasoning, shifts in problem priorities do not
necessarily benefit a specific party or hurt others at the polls. This depends on, e.g., whether
governing parties succeed in tackling problems or not.
Competence attributions might also be endogenous to political predispositions such as party
attachments, values, and policy preferences (e.g., Bellucci 2006; Kuechler 1991; Schoen and
Rudnik 2016; Stubager and Slothuus 2013). Party attachments may incline voters to consider
their own party competent in tackling all problems. In addition, competence attributions might
also reflect policy preferences (Nezi and Katsanidou 2014; Stokes 1963). If different problem
perceptions bring the same policy preferences into play, competence attributions will remain
stable because voters will always prefer parties with a certain policy approach. However, this
prediction does not hold if a party changes its approach to a problem, e.g., by changing its
goals or its ideas about the right means to deal with it (Green and Jennings 2017a). In this
case, even identical policy preferences may lead to changes in competence attributions. These
considerations suggest that changes in problem priorities must not necessarily be
accompanied by changes in competence attributions and subsequent party switches at the
Finally, there is also the possibility that voters do not incorporate competence attributions in
their party choices, even if they entail clear cues to that end, because other factors weigh more
strongly in their electoral behavior (e.g., Campbell et al. 1960; Miller and Shanks 1996). For
example, vote choice may be affected by other short-term factors, such as candidate
orientations, tactical considerations, campaigns, events, or media coverage in the run-up to the
election (e.g., Dilliplane 2014; Schoen et al. 2017). Moreover, long-term political
predilections such as party identification may shape vote choice, irrespective of competence
attributions (e.g., Bartels 2002). In addition, campaigns that cover other topics than those
relating to a crisis may limit the electoral impact of competence attributions related to it. In
effect, changes in competence attributions do not always lead to changes in voting behavior.
Taken together, the event-driven vote model suggests strong repercussions of the successive
crises on voting behavior and election outcomes. Our discussion demonstrated that this model
builds on a causal chain and a set of assumptions that do not necessarily hold in every case.
We now turn to discuss the sequence of crises that may have affected the federal elections
from 2009 to 2017 in light of the theoretical discussion.
Problem priorities and competence attributions in Germany in times of crises
The financial and economic crisis that started in 2008 may have made economic problems
prevalent in the 2009 election. The Euro crisis should have given rise to the perception that
the European Monetary Union (EMU), the common currency, and bailout programs for
indebted countries were the most important problems in the 2013 election. Finally, the influx
of refugees may have made immigration the top priority on the public agenda in the 2017
election. However, since the three crises did not coincide with these elections but peaked well
before them, the shares of voters considering the respective problems as the most pressing
ones might already have decreased somewhat before they were called to the ballots. In effect,
the impact of the crises on changes in problem priorities may therefore have been limited and
further curtailed by the endogeneity of problem priorities to voter characteristics. Party
attachments may have affected problem priorities if parties emphasized or downplayed certain
topics in their campaigns. In the 2009 campaign, parties did not differ in their emphasis on
economic issues (Krewel et al. 2011). During the 2013 and 2017 campaigns, the governing
parties attempted to “dethematize” the EMU and immigration, whereas the AfD highlighted
them (Krewel 2014; Schoen 2019a). If these communication strategies were successful,
supporters of the governing parties should have been less likely and AfD adherents more
likely to consider these topics important. As concerns policy preferences, attitudes toward
state intervention into the economy may have affected people’s inclination to consider issues
related to the financial and economic crisis important in 2009. Given the multi-faceted nature
of the Euro crisis, attitudes toward public debt, European integration, open borders, and –
more broadly – national sovereignty may have made voters inclined to deem this issue
important in 2013. In 2017, opposition to immigration should have made people more likely
to consider this topic important. If voters conceived of the Euro crisis and the refugee crisis in
terms of national sovereignty, demarcation, and openness (Hellwig 2014; Hooghe and Marks
2018; e.g., Kriesi et al. 2008; Teney et al. 2014), the same policy-related predispositions may
have been invoked and voters may have considered these problems important in both the 2013
and 2017 elections. In addition, the endogeneity of problem perceptions to issue positions
may have even led voters to regard the same topics as important across all three federal
elections, irrespective of the flow of events.
Regarding competence attributions, the issue ownership model suggests that widely held party
stereotypes come into play as well. In 2009, the CDU/CSU was considered most competent in
dealing with economic issues (Forschungsgruppe Wahlen 2019). Although there is no clear
evidence on the Euro crisis, the CDU/CSU might have been considered most capable of
dealing with this issue due to its general image as being competent in matters of economics
and national sovereignty (Engler et al. 2019: 322). However, by giving up its strict anti-
bailout policy during the Euro crisis, the CDU/CSU might have undermined this stereotype.
Turning to the issue of immigration, the CDU/CSU was traditionally considered competent by
the majority of voters (Pardos-Prado et al. 2014). Under the leadership of Chancellor Angela
Merkel, however, the CDU had given up its long-held restrictive stance on immigration in
2015, which led to fierce debates within the CDU and the CSU (Mader and Schoen 2019;
Schoen and Gavras 2019). This may have eroded the traditional ownership pattern with regard
to this issue. In effect, deviations from traditional party images suggest crisis-driven changes
in competence attributions and thus a potential for changes in voting behavior.
If party attachments played a role, partisans should have displayed a general tendency to
consider their own party the most capable one, irrespective of the problem at hand. Following
this line of reasoning, the three successive crises did not have the potential to lead to any
changes in voting behavior. Given the fissures within parties during the Euro and the refugee
crises, especially within CDU and CSU, some voters may have lacked clear partisan cues.
This may have undermined the role of party attachments on competence attributions in these
Turning to effects of policy preferences on competence attributions, attitudes toward state
intervention into the economy might have affected competence attributions in the case of the
financial crisis in 2009. However, parties widely agreed on how to handle this issue, with the
FDP being the main exception. Accordingly, these preferences should not have made a major
difference in competence attributions in 2009, though opponents of state intervention may
have considered the FDP particularly competent. In the European sovereign debt crisis, CDU,
CSU, SPD, the Greens, and the majority of the FDP backed a policy course that implied
conditional support for EMU countries under financial strain. People who preferred a more
generous policy may have considered the Left Party competent, which demanded a move
away from austerity measures. Economic conservatives, on the other hand, worried about the
possibility of joint liability between the Eurozone states. This criticism was voiced mainly by
the AfD (which was founded in early 2013). The rank-and-file of CDU, CSU, and FDP took a
rather mixed position on the Euro crisis (Zimmermann 2014). Attitudes toward public debt,
European integration, and – more broadly – demarcation and openness may thus have affected
competence attributions as well. In the refugee crisis, champions of liberal immigration laws
might have considered the Greens, the SPD, or the Left Party – all of which preferred
openness – most competent. By contrast, the AfD might have been an option for opponents of
liberal immigration laws (see Engler et al. 2019). If directional attitudes toward the
oppositional goals of openness and demarcation played a role not only in this conflict but also
in the European sovereign debt crisis, they may have attracted the same people to the pro and
con camps on these issues.
In sum, the event-driven vote model suggests that the sequential crises between 2008 and
2017 changed voters’ problem priorities and competence attributions and thereby accounted
for huge shifts in party support in Germany from 2009 to 2017. On the other hand, substantive
and theoretical objections can be raised against this view. Aside from the temporal distance
between the crises’ peaks and the ensuing federal elections, they concern the possible impact
of political predilections such as party attachments and policy preferences on problem
priorities and competence attributions as well as the limited effect of competence attributions
on vote choice. The following analysis confronts these diverging expectations with empirical
Data and methodology
To explore whether and how the three crises affected problem priorities, competence
attributions, and finally voting behavior, we primarily use panel data from the GLES Election
Campaign Panels 2009, 2013, and 2017 (Rattinger et al. 2016; Rattinger et al. 2018). As a
special feature, respondents from 2009 (2013) were re-interviewed in 2013 (2017), thereby
turning the campaign panels into inter-election panels. Since we are interested in intra-
individual change between two federal elections, we restrict our analysis to the waves fielded
in the week after the respective election.2 Because the crises and the boost in media attention
they triggered peaked quite some time before the next federal elections, the findings for the
election years should be put into perspective. We therefore track the evolution of problem
priorities in the whole period by means of a cumulated file containing 32 online tracking
surveys conducted in the GLES framework (Roßteutscher et al. 2019). With approximately
four cross-sectional surveys each year, these data permit us to cover the period from 2009 to
2017 quite extensively.
To measure problem priorities, we employed an open-ended question asking respondents to
report what they saw as the most and the second-most important problems in Germany (on
methodological issues see, e.g., Johns 2010; Min et al. 2007; Wlezien 2005). For all three
elections, an integrated but adaptable coding scheme was used. To measure problem priorities
related to the financial and economic crisis, we employed the categories “the current state of
the economy” and “unemployment”. To address the Euro crisis, we relied on a wider range of
categories, which included mentions of the common currency, financial aid programs for
indebted Euro countries, fiscal policy, and public debt in general. Finally, to capture concerns
referring to the refugee crisis, we opted for a broad indicator including all references to
immigration and right-wing extremism (since the latter includes concerns about hostility
toward foreigners; see Table A1 in the Online Appendix for details). In the analysis, we
combined all mentions of most and second-most important problems, to the effect that one
person can have more than one problem priority at the same time. Moreover, we assigned all
mentions not referring to any of the three crises to a residual category of “other problems”. In
analyses of inter-election dynamics, problem priorities are considered stable if a person
mentioned the same crisis or nothing crisis-related at all in both years.
2 In 2009 and 2013, 720 respondents participated in the respective interviews; for 2013 and 2017, we have 1,617
cases. Participation in these surveys requires online access; therefore, the results cannot be generalized to the
German electorate without caution. The same applies to the online tracking surveys mentioned below. In
addition, panel attrition and panel effects are potential issues, although some additional analyses do not hint at
massive problems (see Table A3 in the Online Appendix).
Competence attributions were elicited by means of a follow-up question to these problem
priorities, asking which party was most competent in tackling the problem the respondent had
mentioned before. In analyses of inter-election dynamics, we consider competence
attributions as stable if the same party was named or if no party was deemed competent in
both post-election interviews. If a respondent did not name any political problem in one or
both years, the variable is set to “missing”.
Vote choices in the 2009, 2013, and 2017 elections were captured by vote recall questions
asked in the week after the respective election. We consider voting behavior in two
consecutive elections stable if a voter chose the same party or did not vote in either election.
Inter-election switchers comprise all respondents who voted for different parties or
participated in only one of the two elections (for the question wording, see Table A2 in the
We begin our analysis by studying aggregate shifts as well as individual changes of problem
priorities, competence attributions, and voting behavior across the three elections. Relying on
data from the GLES online tracking surveys, Figure 1 reports the evolution of problem
priorities at the aggregate level from 2009 to 2017. The results show how the three successive
crises led to shifts in voters’ problem priorities. Obviously, the crises-induced alterations in
public opinion had peaked already before the three election days. On the other hand, at the
time of the 2009 and 2017 elections, still half of the electorate was concerned about the
economic and the refugee crisis respectively, whereas a third worried about Euro issues in
2013.3 Hence, the crises and their nature appear to have played a genuine role in shaping the
3 The evidence from the panel data supports this conclusion. However, the share of respondents naming the
economic crisis in 2009, the Euro crisis in 2013, and the refugee crisis in 2017 is somewhat higher than in the
tracking surveys (see Table A3 in the Online Appendix). We cannot rule out that these differences between the
At the same time, (changes in) voters’ problem priorities were not perfectly aligned with the
sequence of the crises. By far not all voters were concerned about the respective crises that
overshadowed the three elections; many considered other problems pressing. For example,
some people had been concerned about immigration already before 2015. In effect, changes in
problem priorities appear to have responded to the flow of events but do not perfectly reflect
it. Hence, the link between the sequential crises and changes in voting behavior appears not
- Figure 1 -
To study this chain in more depth, we turn to intra-individual change. Building on data from
the inter-election panel surveys, the left-hand column in Table 1 reports that about 60 percent
of respondents changed their problem priorities from one election to the next. Leaving aside
that this measure is agnostic about the substance of the problems mentioned as well as about
how they related to the crises, changes in problem priorities had a considerable potential to
stimulate vote switching. At the same time, we see that changing crises did not necessarily
lead to changing problem priorities on election days. Consequently, the evidence on intra-
individual change squares nicely with the aggregate-level results shown before.
Table 1 also demonstrates that, according to the panel-based evidence, party competences
were less flexible than problem priorities. Between 2009 and 2013, 38 percent changed
competence attributions, whereas this rate rose to 48 percent in the following period. The rate
of changes in voting behavior resembles the results for party competences more than those for
problem priorities. The evidence shows that 42 percent cast votes for different parties or
switched between abstention and a party vote in 2009 and 2013; from 2013 to 2017 the rate
online tracking surveys and the panel data are partly due to inconsistencies in the coding process. Although using
identical coding schemes, the problem priorities in the two datasets were coded by different institutes.
was even 47 percent. In comparison to prior German federal elections, these rates are quite
high (e.g., Rattinger and Schoen 2009; Schoen 2003). However, they fall short of approaching
the rates of change in problem priorities. The results suggest that not all changes in problem
priorities went hand in hand with changes in competence attributions, nor did they ultimately
result in vote switching.
- Table 1 -
In order to further explore the relationship between (changes in) problem priorities,
competence attributions, and vote choice, we cross-tabulated the three change variables at the
individual level and identified eight trajectories. Table 2 reports the marginal distributions of
these trajectories during the two periods under study. Between 2009 and 2013, 12 percent of
the respondents experienced changes in all three respects, whereas this share amounted to 18
percent during the following electoral cycle (2013–2017). Thus, the link from changing
problem priorities to changing competence attributions to vote switching appears to hold for
some people. For many it does not, however. More than half of the people who changed their
problem priorities from one election to the next did not change competence attributions. Some
ten percent of the respondents in both periods changed problem priorities and competence
attributions, but not vote choice. About 20 percent of the respondents changed problem
priorities without changing competence attributions or vote choice. Looked at from a different
angle, 42 (2009–2013) and 47 (2013–2017) percent of the respondents changed voting
behavior from one election to the next, but only 12 and 18 percent also changed problem
priorities and competence attributions, respectively. Accordingly, changes in the latter two
appear to be neither a necessary nor a sufficient condition for changes in voting behavior.
Thus, the relationship between problem priorities, competence attributions, and vote choice is
far from perfect.
- Table 2 -
Clearly, the sequence of crises did not translate into changes of voting behavior in the
straightforward way suggested by the event-driven vote model. To better understand the –
obviously at best imperfect – connections between crises-induced problem priorities and
related competence perceptions, we now turn to an exploration of how the mechanisms
discussed above contribute to weakening the linkage between the flow of events and electoral
choices. We start with examining the evolution of specific problem priorities between two
elections at the individual level by cross-tabulating the mentions in successive elections.
The evidence reported in Table 3 indicates, again, that some voters did not alter their problem
priorities between elections, although the situational context changed significantly from one
crisis to the next. This particularly applies to economic and immigration issues, whereas Euro-
related mentions appear to be rather time-bound. If the Euro crisis and the European refugee
crisis related to the same overarching demarcation vs. openness cleavage (e.g., Hooghe and
Marks 2018), endogeneity might have developed in a more overarching way. From this
perspective, endogeneity of problem priorities appears more widespread between 2013 and
2017 because transitions from Euro-related to immigration issues do no longer count as
changes. Empirically, some three in four respondents who had mentioned Euro-related issues
in 2013 switched to immigration as problem priority in 2017. The fact that this rate is higher
than the proportion of respondents mentioning immigration in 2017 in the sample as a whole
might be read as supporting the notion that there is some affinity between these topics.4
- Table 3 -
4 Looking at all mentions in the same interview, we found very small proportions of respondents worrying about
both crises at the same time (see Table A3 in the Online Appendix). In particular, hardly anyone cared about the
Euro crisis in 2017 anymore.
How about party attachments and policy preferences – did they also affect problem priorities?
Using binary logistic regression models, we predicted crisis-related issue salience by
predispositions and policy preferences. For each crisis, we took the same set of
predispositions and policy preferences in order to compare their effects. Party identification
was measured by means of the German standard instrument; for the purpose of our analysis,
this information was turned into a set of dummy variables for the CDU/CSU, SPD, FDP,
Greens, The Left, and – starting from its formation in 2013 – AfD. To measure policy
preferences, we relied on attitudes toward state intervention into the economy indicated on a
bipolar scale in 2009 and a 5-point Likert scale in 2013. We also included attitudes toward
public debt (only available in 2009). To measure attitudes toward financial aid for indebted
Euro countries, we relied on an item that asked respondents how they saw Germany’s
involvement in such a program. In addition, we included attitudes toward European
integration and immigration.5 For these variables, we used – as far as possible6 – data from
the panel wave conducted at the preceding federal election, i.e., before the respective crises
unfolded and four years before problem priorities were measured. As these attitudes may have
undergone some change in the course of four years (see Gärtner et al. in this volume),7 we are
likely to get rather conservative estimates of the effects of predispositions. As controls, we
included age, education, region, and unemployment in our models. All variables were
rescaled to run from zero to one (for the question wording, see Table A2 in the Online
Appendix). Figure 2 reports the respective average marginal effects (for the regression
coefficients of the complete models see Table A4 in the Online Appendix).
5 Some attitudes were measured using a split-half design with 7- and 11-point response scales in the 2009 data.
Because the results do not differ substantially between response scales, we treat them as interchangeable.
6 Lacking data from a 2005–2009 panel survey, in the analyses of the 2009 election we relied on predictor
variables measured in 2009. As the debate about bailout programs was not anticipated in 2009, we also had to
employ a measure of attitudes toward financial help programs for indebted Euro countries from 2013.
7 Empirically, some attitudes prove stable over the period under observation (immigration, r=0.7), while others
do so less (state intervention into economy, r=0.4).
The effects of policy preferences and party identification on problem priorities in many cases
do not pass conventional levels of statistical significance – even if we employ the 90-percent
level in 2009 and 2013 in order to account for the rather low numbers of observations. To
some degree, this result reflects the conservative setup of our analysis that led us to use
predictor variables (where applicable) that were measured four years before the respective
election. Still, the evidence suggests that party identifiers of the CDU/CSU were particularly
likely to consider Euro-related topics in 2013 and immigration in 2017. Given the CDU and
CSU’s attempts at downplaying rather than highlighting these topics, this finding may appear
somewhat surprising. However, supporters of the leading governing party may tend to care
especially for problems the government attempts to address. Moreover, support for relaxing
immigration policies made voters less likely to be concerned about immigration in 2017 and
about the economy in 2009. Although we find a slight tendency to but not a robust effect on
mentioning Euro-related topics in 2013, support of more open immigration policies appears to
have made voters less likely to mention problems relating to any of the three crises as
important. Looked at from a different perspective, opposition to immigration increased voters’
likelihood to mention these problems and thus appears to be a common source of concerns
voiced in these elections. In a similar vein, opposition to giving financial aid to EU members
in need appears to have made people more likely to regard immigration as a problem in 2017.
This finding may be read as indicating some relationship between the Euro crisis and the
refugee crisis in voters’ minds. Thus, we find some, albeit not very strong evidence for the
idea that people with specific political views are particularly (un)likely to pick up problems
related to the three crises.8
8 Additional analyses (not presented) suggest that opposition to relaxing regulations on immigration and to
financial aid for EU member countries in need made voters more likely to deem immigration an important
problem in 2013. This result squares nicely with the idea that predispositions made voters also somewhat less
likely to pick up problems in line with the sequence of changing crises.
- Figure 2 -
Party attachments and policy preferences may also exert influence on evaluations which party
is the most competent one. They may therefore explain the weak link between individual
changes in competence attributions and vote choice. In exploring this question empirically,
we have to consider the specifics of data collection. As respondents were asked about
competence attributions as a follow-up to problem priorities, we have information on party
competence attributions for a crisis only if a respondent regarded issues associated with this
crisis as (second-) most important problems. Accordingly, the number of cases available for
the following models is reduced by half (economy 2009, immigration 2017) or even by two-
thirds (Euro 2013). Given the small number of respondents, we treated CDU/CSU and “no
party” as separate outcome categories in all analyses, while we had to put all other parties in a
residual category. The only exception is the AfD, which was mentioned rather often as being
most competent to deal with the immigration issue in 2017. We use a multinomial logistic
regression, employ the same set of predictor variables as in the analysis of problem priorities,
and again present average marginal effects (regression coefficients of the complete models in
Table A5 in the Online Appendix). Since they are not of substantive interest, we do not
display the effects for the residual category “other party” in the presentation of the results.
According to the results in Figure 3, party attachments made a difference in voters’
perceptions of which party was most competent to tackle important problems. In 2009 and
2017, adherents of the CDU/CSU (as measured four years earlier) were considerably more
likely to deem this party most capable of tackling problems concerning the economy (2009)
and immigration (2017). Given the CDU’s policy shift on immigration in 2015 and the
ensuing debates, the latter finding is particularly noteworthy. The 2013 analysis yields a
similar pattern, but the effect does not pass conventional levels of statistical significance.
Furthermore, party identifiers are less likely to deem no party competent. This applies not
only to CDU/CSU attachments, but also to attachments to parties such as SPD, FDP, and The
Left. Although the data do not allow us to study competence attributions for each party
separately, we conclude that party attachments made a difference in competence attributions
and may have served as an impediment to vote switching.
- Figure 3 -
Competence attributions were also affected by some policy preferences.9 Socio-economic
policy preferences do not make a difference. However, opposition to financial aid for indebted
EU countries made voters more likely to judge no party as competent in 2009 and 2013 and to
consider the AfD most competent in dealing with immigration in 2017. A similar pattern
applies to opposition to the deepening of European integration. Finally, opposition to relaxing
immigration regulations made voters more likely to deem the AfD most competent in
addressing the immigration issue in 2017. Although the effects are not overly strong, the
results indicate that competence attributions were affected by (positional) policy preferences.
Moreover, the pattern of effects across topics and elections suggests that a combination of
opposition to European integration, financial aid, and immigration made people particularly
eager to deem the AfD competent and that this party appears to have filled a representational
gap (Rohrschneider and Whitefield 2017).
Our analysis thus lends credence to the idea that party attachments and policy preferences
affect competence attributions. In the former case, the effects mean that – despite changing
problem agendas – voters stick to a party when it comes to tackling pressing problems. If
anything, the result will be lesser incentives for changes in voting behavior. The implications
of the role of policy preferences are less clear: they rather depend on whether policy
9 Lacking a sufficient number of observations, we were unable to include distances between voter positions and
(perceived) party positions.
preferences fit best with the platform of the same party in every election or not. The electoral
implications of endogenous competence attributions thus vary across sources of endogeneity.
Even if competence attributions change from one election to the next, they do not necessarily
lead to changes in voting behavior, however. As voting behavior can be affected by a plethora
of factors (Campbell et al. 1960; Miller and Shanks 1996), other factors may prove influential
in shaping vote choice or may even make changes in competence attributions completely
ineffective at the polls. To explore this possibility, we ran a simple model of vote switching
that includes change in competence attributions, change in evaluations of chancellor
candidates, and party identification as measured at the first election of the respective pair of
elections. As concerns candidate attitudes, we focused on evaluations of the Chancellor
candidates of the CDU/CSU and SPD, which were registered on 11-point feeling
thermometers. We calculated absolute differences for the CDU/CSU and SPD candidates for
the period of 2009 to 2013 and 2013 to 2017. Of course, this is not a comprehensive model of
vote choice. For the present purposes, however, it may suffice to analyze whether party
identification as a long-term political predisposition and candidate orientations as a factor that
is susceptible to short-term influences affect vote switching and how their impact compares to
the influence of changes in competence attributions. Table 4 reports average marginal effects
from logistic regressions of inter-election vote switching in 2009–2013 and 2013–2017.
The results demonstrate that the three types of factors made a difference in vote switching in
both pairs of elections. Partisan independents had an above-average likelihood of vote
switching, while SPD supporters were less likely to switch. Changing candidate evaluations
increased the likelihood of vote switching in both pairs of elections considerably, though the
effect is confined to attitudes toward the CDU/CSU candidate between 2009 and 2013.
Consequently, other factors than changes in competence attributions affected vote choice,
thereby decreasing the impact of the latter. Despite that, changes in competence attributions
are clearly related positively to changes in vote choices in both elections.10 We thus conclude
that changes in competence attributions made voters more likely to change their voting
behavior from one election to the next but that they are by far not the only factor.
- Table 4 -
In sum, the analysis demonstrated that the causal chain from a sequence of crises via changing
problem priorities and changing competence attributions to changing voting behavior proved
quite tenuous in the cases under study. Voters’ problem priorities were not fully aligned with
the sequence of crises. Moreover, changes in problem priorities did not always translate into
changes in competence attributions or changing voting behavior. These imperfections reflect,
at least partially, endogeneities stemming from political predispositions.
In Germany, a series of crises coincided with a high level of electoral volatility between 2008
and 2017. This observation squares nicely with the idea that political events in general and
political crises in particular have the potential to bring about large shifts in voting behavior
(Campbell et al. 1960: 151). One way of linking a sequence of crises to shifts in voting
behavior is to regard crises as leading to changes in voters’ problem priorities and competence
attributions, which in turn result in changing voting behavior. The analysis demonstrated that
the causal chain underlying this event-driven vote model appears to work for some voters, but
by far not for all. Quite often, problem priorities were not congruent with the respective crises
and might rather reflect personal circumstances and preferences. Further complexities arise
from competence attributions, which appeared to be somewhat endogenous to political
10 We acknowledge that there are many issues in disentangling specific effects of, e.g., attitudes toward
candidates and issues because they are interrelated. Including the policy preferences (integrated in the above
models) does not alter the results substantially.
predispositions and did not always affect vote choice strongly. As a result, less than half of
the voters who changed problem priorities simultaneously changed voting behavior; neither
did changes in competence attributions necessarily go hand in hand with changes in voting
behavior. Putting the findings together, the causal assumptions underlying the event-driven
vote model proved fragile and appear to have less explanatory power than one might expect at
From a substantive perspective, the findings suggest that voting behavior is less responsive to
the flow of external events than the event-driven vote model suggests. This imperfection
implies that changes in society, even crises, may not necessarily translate into large shifts in
electoral support. Therefore, the party system appears to be to some extent insulated against
these external influences. Regardless of whether one considers this a lack of responsiveness or
a valuable source of stability, the drivers of this disconnect include voter characteristics such
as values, party attachments, and policy preferences, which give rise to motivated reasoning
(Taber and Lodge 2006; Skitka et al. 2002). Accordingly, the explanatory power of the event-
driven vote model may vary across cases, depending on the circumstances. An electorate
comprising dyed-in-the-wool partisans with well-aligned policy convictions is unlikely to
exhibit patterns derived from the event-driven vote model. However, crises with severe
consequences for people’s everyday lives and clear attribution of responsibility to partisan
actors may cause strong responses even by such an electorate. Studying these processes in
comparative perspective is thus a valuable way forward to better understand the way crises
produce repercussions in the electoral arena and on voting behavior in particular.
Anderson, Christopher J., and Jason D. Hecht. 2012. “Voting when the Economy Goes Bad, Everyone
Is in Charge, and No One Is to Blame: The Case of the 2009 German Election.” Electoral
Studies 31 (1): 5–19.
Bartels, Larry M. 2002. “Beyond the Running Tally: Partisan Bias in Political Perceptions.” Political
Behavior 24 (2): 117–150.
Bélanger, Éric, and Bonnie M. Meguid. 2008. “Issue Salience, Issue Ownership, and Issue-Based Vote
Choice.” Electoral Studies 27 (3): 477–491.
Bellucci, Paolo. 2006. “Tracing the Cognitive and Affective Roots of 'Party Competence': Italy and
Britain, 2001.” Electoral Studies 25 (3): 548–569.
Blumenstiel, Jan Eric, and Elena Wiegand. 2014. “Wechselwähler.” In Rüdiger Schmitt-Beck, Hans
Rattinger, Sigrid Roßteutscher, Bernhard Weßels and Christof et al. Wolf (eds.), Zwischen
Fragmentierung und Konzentration: Die Bundestagswahl 2013. Baden-Baden: Nomos, pp.
Boninger, David S., Jon A. Krosnick, and Matthew K. Berent. 1995. “Origins of Attitude Importance:
Self-Interest, Social Identification, and Value Relevance.” Journal of Personality and Social
Psychology 68 (1): 61–80.
Budge, Ian, and Dennis J. Farlie. 1983. Explaining and Predicting Elections: Issue Effects and Party
Strategies in Twenty-Three Democracies. London: Allen & Unwin.
Campbell, Angus, Philipp E. Converse, Warren E. Miller, and Donald Stokes, E. 1960. The American
Voter. Chicago: University of Chicago Press.
Clarke, Harold D., David Sanders, Marianne C. Stewart, and Paul Whiteley. 2004. Political Choice in
Britain. Oxford: Oxford University Press.
Clarke, Harold D., David Sanders, Marianne C. Stewart, and Paul Whiteley. 2009. Performance Politics
and the British Voter. Cambridge: Cambridge University Press.
Clarke, Harold D., Thomas J. Scotto, and Allan Kornberg. 2011. “Valence Politics and Economic Crisis:
Electoral Choice in Canada 2008.” Electoral Studies 30 (3): 438–449.
Damore, David F. 2004. “The Dynamics of Issue Ownership in Presidential Campaigns.” Political
Research Quarterly 57 (3): 391–397.
Dilliplane, Susanna. 2014. “Activation, Conversion, or Reinforcement? The Impact of Partisan News
Exposure on Vote Choice.” American Journal of Political Science 58 (1): 79–94.
Engler, Fabian, Svenja Bauer-Blaschkowski, and Reimut Zohlnhöfer. 2019. “Disregarding the Voters?
Electoral Competition and the Merkel Government’s Public Policies, 2013–17.” German
Politics 28 (3): 312–331.
Forschungsgruppe Wahlen, Mannheim. 2019. Politbarometer. Langzeitentwicklung: Wirtschaft und
_Themen_im_Ueberblick/Wirtschaft_und_Soziales/#WirtschEnt. July 1, 2019.
Gärtner, Lea, Harald Schoen, and Alexander Wuttke. 2020. “Ties That No Longer Bind?: Effect and
Responsiveness of Party Attachement in a Period of Crises.” In Rüdiger Schmitt-Beck, Sigrid
Roßteutscher, Harald Schoen, Bernhard Weßels and Christof Wolf (eds.), The Changing
German Voter. Oxford: Oxford University Press, pp. XX-XX.
Green, Jane, and Sarah Hobolt. 2008. “Owning the Issue Agenda: Party Strategies and Vote Choices in
British Elections.” Electoral Studies 27 (3): 460–476.
Green, Jane, and Will Jennings. 2017a. “Party Reputations and Policy Priorities: How Issue Ownership
Shapes Executive and Legislative Agendas.” British Journal of Political Science 49 (2): 443-
Green, Jane, and Willl Jennings. 2017b. The Politics of Competence: Parties, Public Opinion and
Voters. Cambridge: Cambridge University Press.
Hellwig, Timothy. 2014. “The Structure of Issue Voting in Postindustrial Democracies.” The
Sociological Quarterly 55 (4): 596–624.
Hooghe, Liesbet, and Gary Marks. 2018. “Cleavage Theory Meets Europe’s Crises: Lipset, Rokkan, and
the Transnational Cleavage.” Journal of European Public Policy 25 (1): 109–135.
Johns, Robert. 2010. “Measuring Issue Salience in British Elections: Competing Interpretations of "Most
Important Issue".” Political Research Quarterly 63 (1): 143–158.
Kratz, Agatha, and Harald Schoen. 2017. “Just Like Leaves in the Wind? Exploring the Interplay of
Media Coverage and Personal Characteristics in Affecting Issue Salience.” In Harald Schoen,
Sigrid Roßteutscher, Rüdiger Schmitt-Beck, Bernhard Weßels and Christof Wolf (eds.), Voters
and Voting in Context: Multiple Contexts and the Heterogeneous German Electorate. Oxford;
New York: Oxford University Press, pp. 43–70.
Krewel, Mona. 2014. “Die Wahlkampagnen der Parteien und ihr Kontext.” In Rüdiger Schmitt-Beck,
Hans Rattinger, Sigrid Roßteutscher, Bernhard Weßels and Christof et al. Wolf (eds.), Zwischen
Fragmentierung und Konzentration: Die Bundestagswahl 2013. Baden-Baden: Nomos, pp. 35–
Krewel, Mona, Rüdiger Schmitt-Beck, and Ansgar Wolsing. 2011. “The Campaign and its Dynamics at
the 2009 German General Election.” German Politics 20 (1): 28–50.
Kriesi, Hanspeter, Edgar Grande, Romain Lachat, Martin Dolezal, Simon Bornschier, and Timotheos
Frey. 2008. West European Politics in the Age of Globalization. Cambridge: Cambridge
Kuechler, Manfred. 1991. “Issues and Voting in the European Elections 1989.” European Journal of
Political Research 19 (1): 81–103.
Lewis-Beck, Michael S., Richard Nadeau, and Martial Foucault. 2013. “The Compleat Economic Voter:
New Theory and British Evidence.” British Journal of Political Science 43 (2): 241–261.
Mader, Matthias, and Harald Schoen. 2019. “The European Refugee Crisis, Party Competition, and
Voters’ Responses in Germany.” West European Politics 42 (1): 67–90.
Miller, Warren E., and J. Merrill Shanks. 1996. The New American Voter. Cambridge, Mass.: Harvard
Min, Young, Salma I. Ghanem, and Dixie Evatt. 2007. “Using A Split-Ballot Survey to Explore the
Robustness of the 'MIP' Question in Agenda-Setting Research: A Methodological Study.”
International Journal of Public Opinion Research 19 (2): 221–236.
Nezi, Roula. 2012. “Economic Voting under the Economic Crisis: Evidence from Greece.” Electoral
Studies 31 (3): 498–505.
Nezi, Roula, and Alexia Katsanidou. 2014. “From Valence to Position: Economic Voting in
Extraordinary Conditions.” Acta Politica 49 (4): 413–430.
Pardos-Prado, Sergi, Bram Lancee, and Iñaki Sagarzazu. 2014. “Immigration and Electoral Change in
Mainstream Political Space.” Political Behavior 36 (4): 847–875.
Petrocik, John R. 1991. “Divided Government: Is It All in the Campaigns?” In Gary W. Cox and Samuel
Kernell (eds.), The Politics of Divided Government. Boulder, CO: Westview Press, pp. 13–38.
Petrocik, John R. 1996. “Issue Ownership in Presidential Elections, with a 1980 Case Study.” American
Journal of Political Science 40 (3): 825–850.
Petrocik, John R., William L. Benoit, and Glenn J. Hansen. 2003. “Issue Ownership and Presidential
Campaigning, 1952-2000.” Political Science Quarterly 118 (4): 599–626.
Rattinger, Hans, Sigrid Roßteutscher, Rüdiger Schmitt-Beck, Harald Schoen, Bernhard Weßels,
Christof Wolf, Jan Eric Blumenstiel, Lea Gärtner, Agatha Kratz, Thomas Plischke, Maria
Preißinger, Elena Wiegand, and Alexander Wuttke. 2018. Repeatedly Questioned Respondents
of the Short-Term Campaign Panel 2013 and 2017 (GLES). GESIS Data Archive, Cologne.
ZA6827 Data file Version 1.0.0, doi: 10.4232/1.13129.
Rattinger, Hans, Sigrid Roßteutscher, Rüdiger Schmitt-Beck, Bernhard Weßels, Christof Wolf, Markus
Steinbrecher, Thomas Plischke, and Elena Wiegand. 2016. Repeatedly Questioned Respondents
of the Short-Term Campaign Panel 2009 and 2013 (GLES). GESIS Data Archive, Cologne.
ZA5757 Data file Version 1.1.0, doi: 10.4232/1.12536.
Rattinger, Hans, and Harald Schoen. 2009. “Ein Schritt vorwärts, zwei zurück: Stabiles und wechselndes
Wahlverhalten bei den Bundestagswahlen 1994 bis 2005.” In Oscar W. Gabriel, Bernhard
Weßels and Jürgen W. Falter (eds.), Wahlen und Wähler. Analysen aus Anlass der
Bundestagswahl 2005. Wiesbaden: VS Verlag für Sozialwissenschaften, pp. 78–102.
Rohrschneider, Robert, and Stephen Whitefield. 2017. “Party Positions about European Integration in
Germany: An Electoral Quandary?” German Politics 26 (1): 83–103.
Rössler, Patrick. 1997. Agenda-Setting: Theoretische Annahmen und empirische Evidenzen einer
Medienwirkungsypothese. Westdeutscher Verlag: Opladen.
Roßteutscher, Sigrid, Rüdiger Schmitt-Beck, Harald Schoen, Bernhard Weßels, Christof Wolf, Simon
Henckel, Ina Bieber, Melanie Dietz, Philipp Scherer, and Lars-Christopher Stövsand. 2019.
Long-Term Online-Tracking, Cumulation 2009-2017, Core Questions (GLES). GESIS Data
Archive, Cologne. ZAXXXX Data file Version X.X.X, doi: XX.XXXX/X.XXXXX.
Schoen, Harald. 2003. Wählerwandel und Wechselwahl: Eine vergleichende Untersuchung. Wiesbaden:
Schoen, Harald. 2019a. “Not a Powerful Electoral Issue Yet: On the Role of European Integration in the
2017 German Federal Election.” Journal of European Public Policy 26 (5): 717–733.
Schoen, Harald. 2019b. “Wechselwähler.” In Sigrid Roßteutscher, Rüdiger Schmitt-Beck, Harald
Schoen, Bernhard Weßels, Christof Wolf and et al. (eds.), Zwischen Polarisierung und
Beharrung: Die Bundestagswahl 2017. Baden-Baden: Nomos, pp. 145–154.
Schoen, Harald, and Konstantin Gavras. 2019. “Eher anhaltende Polarisierung als vorübergehende
Verstimmung: Die Flüchtlingskrise und die Bürgerurteile über die Große Koalition zwischen
2013 und 2017.” In Reimut Zohlnhöfer and Thomas Saalfeld (eds.), Zwischen Stillstand,
Politikwandel und Krisenmanagement: Eine Bilanz der Regierung Merkel 2013-2017.
Wiesbaden: Springer VS, pp. 17–37.
Schoen, Harald, Hans Rattinger, Maria Preißinger, Konstantin Gavras, and Markus Steinbrecher. 2017.
Election Campaigns and Voter Decision-Making in a Multi-Party System: Nomos.
Schoen, Harald, and Agatha Rudnik. 2016. “Wirkungen von Einstellungen zur europäischen Schulden-
und Währungskrise auf das Wahlverhalten bei der Bundestagswahl 2013.” In Harald Schoen
and Bernhard Weßels (eds.), Wahlen und Wähler. Analysen aus Anlass der Bundestagswahl
2013. Wiesbaden: Springer VS, pp. 135–160.
Skitka, Linda J, Elizabeth Mullen, Thomas Griffin, Susan Hutchinson, and Brian Chamberlin. 2002.
“Dispositions, scripts, or motivated correction? Understanding ideological differences in
explanations for social problems.” Journal of personality and social psychology 83 (2): 470.
Stokes, Donald E. 1963. “Spatial Models of Party Competition.” American Political Science Review 57
Stubager, Rune. 2018. “What is Issue Ownership and How Should We Measure It?” Political Behavior
40 (2): 345–370.
Stubager, Rune, and Rune Slothuus. 2013. “What are the Sources of Political Parties’ Issue Ownership?
Testing Four Explanations at the Individual Level.” Political Behavior 35 (3): 567–588.
Taber, Charles S, and Milton Lodge. 2006. “Motivated skepticism in the evaluation of political beliefs.”
American Journal of Political Science 50 (3): 755-769.
Teney, Céline, Onawa Promise Lacewell, and Pieter De Wilde. 2014. “Winners and Losers of
Globalization in Europe: Attitudes and Ideologies.” European Political Science Review 6 (4):
Whiteley, Paul, Harold D. Clarke, David Sanders, and Marianne C. Stewart. 2013. Affluence, Austerity
and Electoral Change in Britain. Cambridge: Cambridge University Press.
Wlezien, Christopher. 2005. “On the Salience of Political Issues: The Problem with 'Most Important
Problem'.” Electoral Studies 24 (4): 555–579.
Zimmermann, Hubert. 2014. “A Grand Coalition for the Euro: The Second Merkel Cabinet, the Euro
Crisis and the Elections of 2013.” German Politics 23 (4): 322–336.
Zohlnhöfer, Reimut. 2011. “The 2009 Federal Election and the Economic Crisis.” German Politics 20
Zohlnhöfer, Reimut, and Thomas Saalfeld (eds). 2015. Politik im Schatten der Krise: Eine Bilanz der
Regierung Merkel 2009-2013. Wiesbaden: Springer VS.
Zohlnhöfer, Reimut, and Thomas Saalfeld (eds). 2019. Zwischen Stillstand, Politikwandel und
Krisenmanagement: Eine Bilanz der Regierung Merkel 2013-2017. Wiesbaden: Springer VS.
Figure 1: Problem priorities during the crisis period 2008–2017
Data: GLES Online Tracking; weighted by a post-stratification weight. Multiple answers were possible. All
mentions for the first and second problem considered.
Table 1: Share of respondents with an individual change in problem priorities, party
competence attributions, and voting 2009–2013 and 2013–2017
2009–2013 57% 38% 42%
(325) (215) (229)
2013–2017 63% 48% 47%
(1009) (773) (730)
N (2009-2013) 572 570 545
N (2013-2017) 1608 1606 1567
Data: GLES campaign panels; weighted by post-stratification weights. Entries are relative frequencies;
corresponding absolute frequencies are in parentheses. N differs between columns because of
nonresponse to single items.
Table 2: Patterns of individual change in problem priorities, party competence, and vote
choice 2009–2013 and 2013–2017
2009–2013 2013–2017 Problem priorities Competence Votin
Yes Yes Yes 12%
Yes Yes No 9%
Yes No Yes 12%
Yes No No 24%
No Yes Yes 8%
No No Yes 10%
No Yes No 9%
No No No 16%
N 543 1565
Data: GLES campaign panels¸ weighted by post-stratification weights. Entries are relative frequencies;
corresponding absolute frequencies are in parentheses. Putting various problems into the residual category “other
problem” decreases the overall turnover rates.
Table 3: Individual trajectories in problem priorities 2009–2013 and 2013–2017
Data: GLES campaign panels; weighted by post-stratification weights. Shares do not add up to 100 percent because
multiple answers were possible. All mentions for the first and second problem considered. Putting various
problems into the residual category “other problem” decreases the overall turnover rates.
Problem priority 2013
Problem priority 2009 Euro Economy Immigration
Economy 27% 32% 16% 35% 386
Immigration 15% 31% 50% 26% 41
Other problem 23% 6% 7% 64% 160
All respondents 26% 24% 14% 43% 572
Problem priority 2017
Problem priority 2013 Euro Economy Immigration
Euro 3% 3% 73% 25% 430
Economy 2% 18% 69% 22% 343
Immigration 1% 8% 87% 12% 263
Other problem 0% 2% 56% 42% 734
All respondents 1% 5% 65% 32% 1608
Figure 2: The influence of predispositions and policy preferences on problem priorities 2009,
2013, and 2017
Data: GLES campaign panels, weighted by post-stratification weights. Entries are average marginal effects from
logistic regressions with 90 percent (Economy & Euro) and 95 percent confidence intervals (Immigration). For
party identification entries are discrete probability changes from omitted reference category (Independents). N:
Economy (399), Euro (399), Immigration (1,597). Effects of control variables (age, education, region,
unemployment) are not displayed.
Figure 3: The influence of predispositions and policy preferences on party competence
Data: GLES campaign panels. Entries are average marginal effects with 90 percent (Economy 2009 and Euro
2013) and 95 percent confidence intervals from multinomial logistic regressions. For party identification entries
are discrete probability changes from omitted reference category (Independents). N: 258 (Economy 2009), 113
(Euro 2013), 1,103 (Immigration 2017). Effects of control variables (age, education, region, unemployment) are
Table 4: The effect of change in party competence evaluations on change in vote choice
Δ Competence .15**
Δ CDU/CSU candidate .48**
Δ SPD candidate .09
ID (ref.: CDU/CSU):
- SPD -.14*
- FDP .02
- Greens .05
- The Left -.05
- Other -.03
- Independent .26**
Data: GLES campaign panels. Average marginal effects from logistic regressions of inter-election vote change.
Standard errors in parentheses. * p < 0.05, ** p < 0.01, *** p < 0.001.