The Impact of the Party System on Electoral Volatility A Cross-Country Analysis of Inter-Election Switching Yves Dejaeghere 1 & Ruth Dassonneville 12 1 Center of Citizenship and Democracy, University of Leuven ² FWO – Research Foundation Flanders Paper prepared for presentation at the EPOP 2012 conference 7 th -9 th September, Oxford
38
Embed
The Impact of the Party System on Electoral Volatility · vote switching. Third, electoral systems have been found to have a considerable impact on the degree to which voters develop
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
The Impact of the Party System on
Electoral Volatility
A Cross-Country Analysis of Inter-Election Switching
Yves Dejaeghere1 & Ruth Dassonneville
12
1Center of Citizenship and Democracy, University of Leuven
² FWO – Research Foundation Flanders
Paper prepared for presentation at the EPOP 2012 conference
7th
-9th
September, Oxford
1
Abstract
In the literature on electoral volatility and party defection, structural elements have been put forward
as crucial variables. Especially the party system is suggested to be of importance for understanding
differences in levels of volatility between countries. First of all, the electoral system has been shown to
have an impact on levels of volatility. In majoritarian and highly disproportional systems, electoral
volatility proved to be more pronounced. With regard to the party system two dimensions can be
distinguished. First, the number of parties within an electoral system is expected to be related to levels
of electoral volatility. It is argued that the more options voters have, the more they will be inclined to
switch. Second, the extent to which a party system is polarized matters as well. The more polarized a
party system is, the larger ideological distances between parties are. Therefore, switching parties
implies a more pronounced ideological shift for a voter and should therefore become less probable.
Although these variables have been empirically studied separately, there has not yet been a large
comparative investigation including them in one analysis with cross-national data on the individual
level. Using the second and third module of the CSES project this paper investigated volatility for
25531 respondents in 32 elections between 2000 and 2010. Using multilevel models that include
country level variables while controlling for important individual level characteristics we have an
optimal control of the simultaneous effect of these separate variables.
Our results show that the effect of individual-level variables such as education and a persons’
satisfaction with democracy remain strong predictors of electoral volatility even in a cross-national
analysis. Of the variables on the contextual level proportionality and the number of parties seem to
have an effect on switching parties between elections. Volatility is higher in more proportional
systems. Furthermore, it seems that the sheer number of parties increases the propensity to change a
voters choice regardless of their polarization. This last finding is a refutation of a longstanding claim
in the literature that it is the distance between the parties rather than the number that influences
volatility. Because we tested three different measures for polarization separately that can be found in
the literature, this can be considered a robust finding. We furthermore find a cross-level interaction
effect that shows that satisfaction with the way democracy works in a country leads to different odds
of switching votes depending on the effective number of parties in that election.
2
1. Introduction
In his seminal 1979 paper Morgens Pedersen investigated the effect the format of a party system has
on electoral volatility. For Pedersen (1979: 14), it seemed there was a clear linkage ‘between the
number of options open to the individual voter, and the voters’ propensities to transfer votes between
parties.’. The main limitation of his analysis was the use of aggregate data whereas the effect of the
party systems worked through individual level dynamics. He stated himself (1979: 16) that his
hypotheses ‘in principle lend themselves to a test on the basis of individual level data.’, which he
didn’t have. While in the last thirty years this article has been cited in abundance there has as yet never
been a test of the claims made in it with the data they are intended for.
Pedersen assumed it is not specifically the number of parties in itself that cause more volatility, but the
fact that a voter perceives less average distance between parties when there are more of them. If
parties are ideologically close to each other, voters will be more likely to consider leaving the party
they voted for previously. This implies that Pedersen’s hypothesis also involved the polarization of the
party system. Pedersen himself, however, considered the connection between the number of parties
and polarization as a truism, as did many authors in his time. This was challenged in later empirical
research and the number of parties and polarization are treated as distinct elements of a party system in
current research and in this paper. We will assess whether the link Pedersen observed between
volatility at an aggregate level and the characteristics of the party system is indeed one driven by
individual level dynamics and caused by polarization in the party system rather than by the number of
parties as such. To do this we will use the data from the Comparative Study of Electoral Systems
(CSES). We start the paper with an overview of the literature on volatility, in which we have attention
for both the contextual and individual causes of volatility. Next we present the data used for the
analyses and elaborate on the method used. After describing the results we end with some concluding
remarks and suggestions for further research.
2. Theory and Hypotheses
In liberal democracies, political parties are the organizational expression of ideological diversity in the
political sphere (Ware, 1996). Elections are the arena where parties, on the basis of these ideologies,
try to win over as many voters as possible. They are expected to place themselves at the most
advantageous position in the ideological space in order to maximize their potential voters (Downs,
1957). Downs compares the ideological space with a consumer market, but unlike what holds there,
political parties cannot be considered interchangeable goods. As the seminal work of Lipset & Rokkan
a decade later (1967) pointed out, several cleavages exist in western societies. These deeply-rooted
cleavages arose because of historical reasons and dominate the political sphere through the presence
3
political parties. The cleavage structure causes voters to be strongly aligned to particular parties and
parties therefore did not (and do not) go into battle over the total electorate at each election (as
companies towards consumers). They can count on a specific slice of the electorate that is bound to
them for ideological reasons. Although some change could occur at the fringes of an electoral space, a
large section of society was said to be ‘frozen’ into these cleavages (Mair, 2001). But as Blondel
(1968) pointed out a year after Lipset & Rokkan, some fluctuation is absolutely necessary to keep
democratic systems structurally balanced. If the party system is frozen and the same parties remain a
majority, this will undermine the belief in the usefulness of elections to achieve ones goal in society.
Blondel therefore sees a certain degree of electoral volatility between parties as benign to democracy
and especially so in majoritarian systems: “The distortions of two-party systems are potentially
serious, if the divine hand does not act to make the pendulum swing.” (1968: 198).
Pedersen was the first to try to investigate the issue of volatility in a large comparative manner. Since
the publication of Pedersen’s study, several scholars have drawn attention to a further trend of
Ls -0.052* (0.026) -0.046° (0.025) -0.058* (0.027)
ENEP 0.161* (0.066) 0.171* (0.067) 0.134° (0.072)
Max. left-right
distance
-0.041 (0.061)
Polarization 0.005 (0.085)
Compactness 0.245 (0.230)
Model fit statistics
N individuals 25,531 25,531 25,531 25,531 25,531
N elections 32 32 32 32 32
BIC 29,658 29,640 29,655 29,656 29,565
Elections σ² 0.278 0.278 0.168 0.171 0.165
Significance levels: ° p < 0.1, * p < 0.05, ** p < 0.01, *** p < 0.001. Maximum likelihood estimation using adaptive Gauss-Hermite approximation with 11 quadrature points
in Stata. Data: CSES Module 2 and 3. Level-2 variance in the nullmodel (σ²null): 0.290.
18
Figure 2. Estimated effect of disproportionality on electoral volatility
Estimated probability of vote switching for different levels of Ls, based on 10,000 simulated observations from
Model V. Data: CSES Module 2 and 3.
The effect of the effective number of parties in an electoral system on electoral volatility appears to be
even stronger. As evident from Figure 3, the proportion of voters changing votes is estimated to range
between about 22% when the ENEP is two to over 40% in systems where there are nine effective
electoral parties. Uncertainty about the estimated effect of the number of parties, however, is
particularly high for large party systems of which only very few are included in the dataset.
0
.1
.2
.3
.4
.5
Pro
bab
ility
of
vo
latilit
y
0 2 4 6 8 10 12 14 16
Ls
19
Figure 3. Estimated effect of effective number of parties on electoral volatility
Estimated probability of vote switching for different levels of ENEP, based on 10,000 simulated observations
from Model V. Data: CSES Module 2 and 3.
Not finding a significant effect of the degree of polarization of an electoral system on probabilities of
vote switching might suggest that there is simply no effect of polarization on volatility. The non-
finding might also be a consequence of a poor operationalisation of the concept. Using three different
measures to capture polarization, however, we limited the impact of measurement-specific issues. A
third possible explanation or the non-finding is that the effect of polarization on volatility varies
between individuals. For polarization to affect voting behavior, voters should be aware of where
parties are situated in the ideological space. Given that this is cognitively demanding, we might
therefore hypothesize this effect to be dependent on how politically sophisticated voters are.
Additionally, based on theories of strategic voting, we expected more vote switching in highly
disproportional systems. The effect of the least squares index on volatility is in opposite directions,
however. Given that strategic voting as well is quite demanding, we can expect this effect as well to be
dependent on voters’ level of political sophistication.
Investigating such varying effects, cross-level interactions are tested. Therefore, we start from Model
Vxi in Table 2 and add interactions between political knowledge and disproportionality, the effective
number of parties and compactness. Doing so, we test for the three types of contextual variables
whether their effect varies depending on voters’ level of political sophistication. As results in Table 3
make clear, the cross-level interactions are not significant, indicating that the effects of the contextual
variables do not vary according to voters’ level of political sophistication. Clearly, when finding
effects of institutional variables, these effects hold for the whole electorate and are not dependent on
0
.1
.2
.3
.4
.5
Pro
bab
ility
of
vo
latilit
y
2 3 4 5 6 7 8 9
ENEP
20
what voters know about the political system. This is apparent from the variance coefficient for the
effect of political knowledge, even before including cross-level interactions variance is limited to 0.2%
Significance levels: ° p < 0.1, * p < 0.05, ** p < 0.01, *** p < 0.001. Maximum likelihood estimation using adaptive Gauss-Hermite approximation with 11 quadrature points
in Stata. Data: CSES Module 2 and 3. Level-2 variance in the nullmodel (σ²null): 0.290. Original variance in effect of political knowledge: 0.002.
22
Additionally, besides theories linking political sophistication to electoral volatility, political
disengagement has been shown to explain vote switching as well. As apparent from results in Table 2,
external political efficacy and voters’ level of satisfaction with democracy are indeed strong predictors
of volatility. With regard to political disengagement as well, we might wonder whether the same
mechanisms hold and function likewise across the electoral contexts observed. Therefore, focusing on
satisfaction with democracy as the key variable measuring political disengagement, cross-level
interactions with the contextual variables are tested as well.
As is clear from Table 4 there is a significant interaction effect between satisfaction with democracy
and the number of political parties in an election. For disproportionality and compactness no
significant interactions can be found. The effect of satisfaction with democracy therefore varies
depending on how many parties voters can choose between. In order to understand this positive and
significant interaction effect, some quantities of interest were simulated and graphically presented in
Figure 4.xii
The figure illustrates the effect of the number of parties for voters who are not at all
satisfied with the way democracy functions and for voters who are very satisfied with the functioning
of democracy in their country. As can be seen, for the very dissatisfied voters, the number of parties
hardly affects their probability of switching parties. Dissatisfied voters have a high probability of
switching overall. For satisfied voters, on the other hand, the number of parties has a strong positive
effect on their probability of volatility. The interaction effect clearly illustrates that reward-punishment
mechanisms function mostly in contexts where clarity of responsibility is high. As can be observed, in
a context with a limited number of parties, where clarity of responsibility can be suggested to be high,
whether or not voters are satisfied with democracy matters a lot for their probability of switching
parties. In low clarity of responsibility contexts on the other hand, with more than six effective parties,
whether or not voters are satisfied does not affect their chances of switching. In such an electoral
system that is crowded with parties, all voters have a high probability of switching and these switches
are not dependent on how satisfied voters are with democracy.
23
Figure 4. Estimated effect of effective number of parties on electoral volatility by level of
satisfaction with democracy
Estimated probability of vote switching for different levels of ENEP, based on 10,000 simulated observations
Significance levels: ° p < 0.1, * p < 0.05, ** p < 0.01, *** p < 0.001. Maximum likelihood estimation using adaptive Gauss-Hermite approximation with 11 quadrature points
in Stata. Data: CSES Module 2 and 3. Level-2 variance in the nullmodel (σ²null): 0.290. Original variance in effect of satisfaction with democracy: 0.016
25
5. Conclusions
We started the paper from the conclusion that contextual theories of electoral volatility are usually
explained by mechanisms at the level of individual voters. Despite this reasoning, the link between
contextual variables and electoral volatility has so far only been tested at an aggregate level.
Addressing this gap in the literature and by means of a large scale and cross-country dataset of voter
surveys, we investigated whether disproportionality, the number of parties and polarization can be
related to volatility at the voter level as well. Additionally, we incorporate the impact of individual
characteristics and the interaction between these and the contextual variables as well in our analyses.
At the individual level, contrary to suggestions of curvilinear effects of political sophistication on
volatility, we found a significant linear effect of political knowledge. More knowledgeable voters
switch parties significantly less. Surprisingly, the effect of education is in opposite directions.
Therefore, our findings suggest that how political sophistication is operationalized can be expected to
affect results investigating the link between sophistication and electoral volatility. For theories of
political disengagement, results were in expected directions. Satisfaction with democracy and high
levels of external political efficacy suppress vote switching.
For the institutional variables, contrary to what was expected, voters in highly disproportional systems
are not significantly more volatile. Although scholars reason that more strategic voting and weaker
party attachments in disproportional systems are conducive for volatility, there is on the contrary less
volatility when levels of disproportionality are higher. The process of dealignment of the past decades
might be an explanation for not finding more vote switching in more disproportional systems. As party
attachment across the Western industrialized have weakened, differences with the highly
disproportional systems in this regard might have become negligible.
The number of parties in an electoral system is, as hypothesized, a significant predictor of volatility.
The more options voters have in the polling booth, the higher their probability of switching parties.
Contrary to the reasoning framed by Pedersen for explaining this, differences in levels of polarization
do not explain this difference. Even though we used three different measures of polarization, none of
them was found to be significantly related to levels of volatility. In this regard, the use of a one-
dimensional LR scale implies a reduction of the party-space and the options voters have in any given
election. Although many authors contend that it still captures the most salient dimension of voter
placement compared to parties, other authors state multidimensional scales would do better. Especially
Kriesi et al (2006), show using newspaper data that the cultural axis has become more salient in recent
times. By using media analysis, they draw the issue space as it is presented to voters rather than what
is found in manifestos or pre-defined survey scales. It would be interesting for future research to add
these multidimensional scales to these analysis.
26
Investigating cross-level interactions between the contextual variables and voter specific covariates,
we did not find voters’ level of political knowledge to affect reactions to the institutional context.
Therefore, how voters respond to the electoral system is a general mechanism and does not seem to
depend on voters’ understanding and knowledge of politics. When looking at disengagement with
politics we do find a significant cross-level interaction with the number of parties in an electoral
system. Clearly, in small party systems, satisfaction with democracy determines to a large extent
whether or not voters switch parties from one election to another. In large party systems, on the other
hand, no significant differences in levels of volatility can be found between very satisfied and
dissatisfied voters. This finding suggests that when voters have a lot of parties to choose between,
switching is no longer driven by dissatisfaction. This obviously raises the question what reasons they
do switch for in such a context.
27
References
Abedi, A. (2002). Challenges to Established Parties: The Effects of Party System Features on the
Electoral Fortunes of Anti-political-establishment Parties, European Journal of Political Research, 41
(4), 551-583
Alvarez, R. & Nagler, J. (2004). Party System Compactness: Measurement and Consequences.
Political Analysis, 12 (1), 46-62.
Bardi, L., & Mair, P. (2008). The Parameters of Party Systems. Party Politics, 14 (2), 147-166.
Bartolini, S. & Mair, P. (1990). Identity, Competition and Electoral Availability. The Stability of
European Electorates 1885-1985. Cambridge University Press, Cambridge.
Berelson, B.R., Lazarsfeld, P.F. & McPhee, W.N. (1963). Voting: A Study of Opinion Formation in a
Presidential Campaign. University of Chicago Press, Chicago.
Bischoff, C.S. (2012). Electorally Unstable by Supply or Demand? An Examination of the Causes of
Electoral Volatility in Advanced Industrial Democracies. Public Choice, forthcoming.
Blais, A., & Gschwend, T. (2010). Strategic Defection Across Elections, Parties and Voters. In: R. J.
Dalton & C. J. Anderson (Eds.), Citizens, Context and Choice. How Context Shapes Citizens’
Electoral Choices. (pp. 176-193). Oxford: Oxford University Press.
Blondel, J. (1968). Party Systems and Patterns of Government in Western Democracies. Canadian
Journal of Political Science, 1 (2), 180-203.
Bowler, S., Lanoue, D.J. & Savoie, P. (1994). Electoral Systems, Party Competition and Strength of
Partisan Attachment: Evidence from Three Countries. The Journal of Politics, 56 (4), 991-1007.
Budge, I., & McDonald, M.D. (2006). Choices Parties Define. Policy Alternatives in Representative
Elections, 17 Countries 1945-1998. Party Politics, 12 (4), 451-466.
Campbell, A., Converse, P., Miller, W. & Stokes, D. (1960). The American Voter. The University of
Chicago Press, Chicago.
Clarke, H.D., Sanders, D., Stewart, M.C., Whiteley, P. (2004). Political Choice in Britain. Oxford:
Oxford University Press.
The Comparative Study of Electoral Systems. (2007). CSES Module 2 Full Release [dataset].
The Comparative Study of Electoral Systems. (2012). CSES Module 3 third Advance Release
[dataset].
Curini, L., & Hino, A. (2012). Missing Links in Party-System Polarization: How Institutions and
Voters Matter. The Journal of Politics, 74 (2), 460-473.
Crepaz, M.M. (1990). The Impact of Party Polarization and Postmaterialism on Voter Turnout. A
Comparative Study of 16 Industrial Democracies. European Journal of Political Research, 18 (2),
183-205.
28
Crewe, I. & Denver, D. (1985). Electoral Change in Western Democracies: Patterns and Sources of
Electoral Volatility. Croom Helm, London.
Dalton, R.J. (1984). Cognitive Mobilization and Partisan Dealignment in Advanced Industrial
Democracies. Journal of Politics, 46 (2), 264-284.
Dalton, R.J. (2007). Partisan Mobilization, Cognitive Mobilization and the Changing American
Electorate. Electoral Studies, 26 (2), 274-286.
Dalton, R.J. (2008). The Quantity and the Quality of Party Systems: Party System Polarization, Its
Measurement, and Its Consequences. Comparative Political Studies, 41 (7), 899-920.
Dalton, R.J., McAllister, I. & Wattenberg, M. (2000). The Consequences of Partisan Dealignment. In:
R.J. Dalton & M. Wattenberg (Eds.) Parties without Partisans. Oxford University Press, Oxford.
Dalton, R.J. & Weldon, S.A. (2005). Public Images of Political Parties: A Necessary Evil? West
European Politics, 28 (5), 931-951.
Dassonneville, R. (2012). Electoral Volatility, Political Sophistication, Trust and Efficacy. A Study on
Changes in Voter Preferences during the Belgian Regional Elections of 2009. Acta Politica, 47 (1),
18-41.
Deegan-Krause, K., & Zsolt, E. (2010). Agency and the Structure of Party Competition: Alignment,
Stability and the Role of Political Elites. West European Politics, 33 (3), 686-710.
Downs, A. (1957). An Economic Theory of Democracy. Harper and Row, New York.
Ezrow, L. (2007). The Variance Matters: How Party Systems Represent the Preferences of Voters. The
Journal of Politics, 69 (1), 182-192.
Ezrow, L. (2008). Parties’ Political Programmes and the Dog that Didn’t Bark: No Evidence that
Proportional Systems Promote Extreme Party Positioning. British Journal of Political Science, 38 (3),
479-497.
Farrell, D.M. (2001). Electoral Systems. A Comparative Introduction. Palgrave Macmillan,
Basingstoke.
Franzmann, S., & Kaiser, A. (2006). Locating Political Parties in Policy Space. A Reanalysis of Party
Manifesto Data. Party Politics, 12 (2), 163-188.
Freire, A. (2008). Party Polarization and Citizens’ Left-Right Orientations. Party Politics, 14 (2), 189-
209.
Gallagher, M. (1991). Proportionality, Disproportionality and Electoral Systems. Electoral Studies, 10
(1), 33-51.
Granberg, D. & Holmberg, S. (1991). Election Campaign Volatility in Sweden and the United States.
Electoral Studies, 10 (3), 208-230.
Grönlund, K. & Milner, H. (2006). The Determinants of Political Knowledge in Comparative
Perspective. Scandinavian Political Studies, 29 (4), 386-406.
29
Hazan, R. (1997). Centre Parties: Polarization and Competition in European Parliamentary
Democracies. London: Pinter.
Hooghe, L., Bakker, R., Brigevich, A., De Vries, C., Edwards, E., Marks, G., Rovny, J., Steenbergen,
M., & Vachudova, M. (2010). Reliability and Validity of the 2002 and 2006 Chapel Hill Expert
Surveys on Party Positioning, European Journal of Political Research,49 (5), 687-703.
Karp, J.A. & Banducci, S.A. (2008). Political Efficacy and Participation in Twenty-Seven
Democracies: How Electoral Systems Shape Political Behaviour. British Journal of Political Studies,
38 (2), 311-334.
Key, V.O. (1966). The Responsible Electorate: Rationality in Presidential Voting 1936-1960. Harvard
University Press, Cambridge.
Kriesi, H., Grande, E., Lachat, R., Dolezal, M., Bornschier, S., & Frey, T. (2006). Globalization and
the Transformation of the National Political Space: Six European Countries Compared. European
Journal of Political Research, 45 (6), 921-956.
Kuhn, U. (2009). Stability and Change in Party Preference. Swiss Political Science Review, 15 (3),
463-494.
Laakso, M., & Taagepera, R. (1979). “Effective” Number of Parties – A Measure with Application to
West Europe. Comparative Political Studies, 12 (3), 3-27.
Lachat, R. (2007). A Heterogeneous Electorate. Political Sophistication, Predisposition Strength and
the Voting Decision Process. Nomos, Baden-Baden.
Lewis-Beck, M.S. & Stegmaier, M. (2000). Economic Determinants of Electoral Outcomes. Annual
Review of Political Science, 3 (1), 138-219.
Lijphart, A. (1999). Patterns of Democracy. Government Forms and Performance in Thirty-Six
Countries. New Haven: Yale University Press
Lipset, S. & Rokkan, S. (1967). Party Systems and Voter Alignments: Cross-National Perspectives.
Free Press, New York.
Luskin, R.C. (1990). Explaining Political Sophistication. Political Behavior, 12 (4), 331-361.
Mair, P. (1995). Political Parties, Popular Legitimacy and Public Privilege, West European Politics, 18
(3), 40-57.
Mair, P. (2001). The Freezing Hypothesis: An Evaluation. In: L. Karvonen & S. Kuhnle (Eds). Party
Systems and Voter Alignments Revisited. London: Routledge.
Marks, G. (2007). Triangulation and the Square-root Law. Electoral Studies,26 (1), 1-10.
Marshall, J., & Fisher, S. (2010). Economic Globalization and the Decline of Electoral Turnout:
Compensation, Constraint and Ownership. Paper presented at the 2010 annual meeting of the
American Political Science Association, Washington (DC).
Marthaler, S. (2008). The Paradox of the Politically Sophisticated Partisan: The French Case. West
European Politics, 31 (5), 937-959.
30
Pedersen, M. (1979). The Dynamics of European Party Systems: Changing Patterns of Electoral
Volatility. European Journal of Political Research, 7 (1), 1-26.
Powell, G.B. & Whitten, G.D. (1993). A Cross-National Analysis of Economic Voting: Taking
Account of the Political Context. American Journal of Political Science, 37 (2), 391-414.
Rose, R., & Urwin (1970). Persistence and Change in Western Party Systems Since 1945. Political
Studies, 18 (3), 287-319.
Sartori, G. (1976). Parties and Party Systems: A Framework for Analysis. Cambridge: Cambridge
University Press.
Sigelman, L., & Nam Yough, S. (1978). Left-Right Polarization In National Party Systems: A Cross-
national Analysis. Comparative Political Studies, 11 (3), 355-379.
Söderlund, P. (2008). Retrospective Voting and Electoral Volatility. A Nordic Perspective.
Scandinavian Political Studies, 31 (2), 217-240.
Steenbergen, M.R. & Jones, B.S. (2002). Modeling Multilevel Data Structures. American Journal of
Political Science, 46 (1), 218-237.
Steiner, N.D., & Martin, C.W. (2012). Economic Integration, Party Polarisation and Electoral Turnout.
West-European Politics, 35 (2), 238-265.
Ware, A. (1996). Political Parties and Party Systems. Oxford: Oxford University Press.
Zelle, C. (1995). Social Dealignment versus Political Frustration: Contrasting Explanations of the
Floating Vote in Germany. European Journal of Political Research, 27 (3), 319-345.
31
APPENDIX A
module Election N Voted_now Vote_last Voted_both N_analysis
2 Australia 2004
1769 1625 1628 1525 1525
3 Croatia 2007 1004 811 732 663 571
2 Czech Republic 2002 948 695 537 489 476
3 Denmark 2007 1442 1390 1367 1332 1259
3 Estonia 2011 1000 790 638 575 434
2 Finland 2003 1196 957 903 821 707
3 Finland 2007 1283 1066 1048 970 839
2 Germany 2002 2000 1878 1784 1710 1569
3 Germany 2009 2095 1645 1571 1418 1258
3 Greece 2009 1022 902 897 823 653
2 Hungary 2002 1200 992 964 866 783
3 Iceland 2007 1595 1373 1289 1190 1023
3 Iceland 2009 1385 1266 1200 1167 940
2 Ireland 2002 2367 2019 1808 1652 1297
2 Israel 2003 1212 1081 981 906 758
3 Israel 2006 1200 965 945 836 645
2 Italy 2006 1439 1102 NA NA 499
2 New Zealand 2002 1741 1402 1456 1243 1243
3 New Zealand 2008 1149 1064 920 893 886
2 Norway 2001 2052 1698 1662 1458 1449
3 Norway 2005 2012 1795 1654 1555 1453
2 Poland 2001 1794 1035 1211 855 678
2 Portugal 2002 1303 971 973 869 704
2 Romania 2004 1913 1504 1307 1155 818
3 Slovak Republic 2010 1203 939 987 836 747
2 Slovenia 2004 1002 758 742 648 410
2 Spain 2004 1212 1077 938 897 779
2 Sweden 2002 1060 937 861 810 745
3 Sweden 2006 1547 1398 1306 1234 1094
2 Switzerland 2003 1418 1048 1003 908 355
3 The Netherlands 2006 2359 2196 1966 1922 1850
2 United Kingdom 2005 860 620 623 549 540
32
APPENDIX B
The reference brackets refers to the source(s) where the election report can be found. ES stands for the
journal Electoral Studies, EJPR for European Journal of Political Research
Australia 2004 (ES: Vol. 24, 3, pp. 545–551)
The coalition between Liberal and National might make some voters switch their first choice on the
ballot purely out of strategic reasons if the chance of winning has drastically changed in a
constituency, but we doubt this will happen more than on a very rare occasion.
New Zealand 2002 (ES: Vol. 23, 1, pp. 149–155)
Between 1999 and the current election United Future, formed in 2000 through a merger between the
United Party and the Christian-based Future NewZealand Party. So we consider previous voters from
those parties that now vote United Future as non-switchers
Czech Republic 2002 (ES: Vol. 22, 4, pp. 772–778)
Before the 1998 local and Senate elections, discontent with the opposition agreement led to the
formation of the so-called Quad Coalition of four smaller parties: the Christian Democrats (KDU-
ČSL), the Freedom Union (US), the Civic Democratic Alliance (ODA), and Democratic Union
(DEU). The Quads split again at the beginning of 2002. The ODA left the Quads. Just before the
ODA departed, the US and DEU merged, so the Quad Coalition then had only two parties and was
renamed simply the Coalition. Voters coming from the original 1998 ‘Quad’ that vote for one of the
resulting parties that split from the Quad or the Coalition that is the remainder of the Quad are non-
switchers.
Denmark 2007 (EJPR: Vol. 47, pp. 952-961)
A new party (New Alliance) was founded by a number of MP’s from existing parties, but since the
original parties still run separately we consider this not to be a party connected to the previous ones.
People voting for New Alliance are coded as switchers.
33
Norway 2005 (ES: Vol. 26, 1, pp. 219–223)
Although parties ran as possible future coalition-blocks both on left and right, all parties kept running
separately in the election.
Hungary 2002 (ES: Vol. 22, 4, pp. 799–807)
Fidesz-MPP and MDF joint forces in 2002, so previous voters from both parties voting for this block
are non-switchers. Although some MP’s of the small FKGP ran for Fidesz-MPP/MDF in 2002, we
follow the EJPR (volume 43, pp. 968-969) and consider it as a separate party.
Slovak Republic 2010 (ES: Vol. 31, 1, pp. 222–225)
Most-Híd, ran as a new party, but was led by Béla Bugár, the former long-time head of the Party of
the Hungarian Coalition (SMK). We therefore consider voters for Most-Hid coming from SMK are
non-switchers
Croatia 2007 (ES: Vol. 27, 4, pp. 752–755)
The HSS and HSLS parties formed the so-called ‘yellow-green’ coalition. Voters for HSS and HSLS
in previous election are non-switchers when voting on the coalition of both parties in the 2007 election
Iceland 2007 (ES : Vol. 27, 2, pp. 373–377)
A new party entered the electoral arena (Iceland Movement), but was not a split from an existing
party. Any new voters for this party are considered switchers.
Israel 2003 (ES: Vol. 23, 2, pp. 353–360 & EJPR: Vol. 43, pp. 1033-1040)
A number of smaller parties had splits or coalitions such as One Nation (Am Ehad), but it had no
voters in the CSES, so there is no recoding. The same goes for the Green Leaf (Ale Yarok) party and
for Herut (The National Party). Similarly Labour-Meimad 2003 was Yisrael Ahat in 1999 and
although CSES has a code for Labour-Meimad, there were no voters. Yisrael Ahat was not coded
34
separately by CSES. The parties Meimad en Gesher that also formed Yisrael Ahat have no voters in
CSES. National Unity-Israel our Home was two separate parties in 1999. Herut that was part of this
group ran separately in 2003.
Iceland 2009 (ES: Vol. 29, 3, pp. 523–526)
The Icelandic movement became a part of the Social Democratic Alliance for the 2009 election, so
voters for Icelandic movement going to SDA are non-switchers.
Poland 2001 (ES : Vol. 22, 2, pp. 367–374 & EJPR : Vol. 41, pp. 1057-1067)
SLD and UP voters 1997 are stable if they voted for the SLD-UP coalition in 2001. AWS voters in
1997 are considered stable if they voted for the AWSP (name change of what was left after several
splinter groups left the party) in 2001. Since a lot of splinter parties emerged and these new parties
were not homogeneous (MP’s from different parties joined the new formations PO and PiS), we do not
consider voters switching to all of these parties as stable. Additionaly, voters for Polish Accord
(Porozumienie Polskie) in 1997 now voting for LPR could also be considered stable, but the only
voter for that party in 1997 in the CSES did not vote in 2001.
Ireland 2002 (ES: Vol. 23, 1, pp. 155–160)
Democratic Left and Irish Labour Party merged ‘Irish Labour Party’ in 2002 and voters from these
parties going to Labour are considered stable.
Israel 2006 (ES: Vol. 26, 3, pp. 707–711 & EJPR: Vol. 46, pp. 984-992)
On 20 November 2005, Israels prime minister Sharon resigned from Likud and announced his
intention to lead a new political party: Kadima (Forward). We consider it to be a new party (as does
Abraham in the EJPR yearbook) . We consider 2003 voters of Labour/Meimad and One Nation are
stable if they voted Labour/Meimad in 2006 and same for voters from National Union/Yisrael Beitenu
if they voted Yisrael Beitenu in 2006. The 2003 voters of Ra’am are also stable if they voted
Ra’am/Ta’al in 2006.
35
Estonia 2011
There was no entry yet in ES or EJPR for this election, but the OECD election reportxiii
does not
mention any major changes in the party system
Romania 2004 (ES : Vol. 25, 2, pp. 409–415)
At a joint congress in September 2004, the Social Democrats agreed to establish an electoral alliance
with the Humanist Party of Romania (Partidul Umanist din România, PUR) called the National Union
(Uniunea Naţională). Equally the PNL (National Liberal Party) and the Democratic Party (PD) went
into the election togheter. Respondents that voted for parties that went into the election as a coaltion
are stable if they come from one of the parties forming that coalition.
Slovenia 2004 (Vol. 25, 4, pp. 825–831)
The SKD (Slovenian Christian Democrats) were still named on the ticket in 2000 but they were
already then de facto in the SLS. The New Slovenia party is a breakaway from this union by the
former SKD leader, but since both already took part as separate parties in 2000 they are to be
considered distinct parties in both elections.
Italy 2006 (ES: Vol. 27, 1, pp. 185–190)
It is important to note that the 2006 election was done using a PR system, changing from the mixed
PR+Majoritarian system used since 1993. In the centre-left camp, Romano Prodi re-launched the
Olive Tree Federation, based on the Democratic Left (DS), Daisy (DL), and the Italian Social
Democrats (SDI). The Rose in the Fist is a ‘New Labour’ kind of party founded in 2005 from minor
parties including the Radicale (code 22). The UDC is a merger between the CCD and CDU, but since
those two already participated ‘merged’ in the 2001 election there is no recoding of respondents.
36
Elections with no significant changes in the party-system
*Norway 2001 (ES: Vol. 22, 1, pp. 179–185)
*United Kindom 2005 (ES: Vol. 25, 4, pp. 814–820)
*The Netherlands 2006 (ES: Vol. 26, 4, pp. 832–837)
*New Zealand 2008 (ES: Vol. 28, 3, pp. 507–510)
*Sweden 2002 (ES: Vol. 22, 4, pp. 778–784)
*Germany 2002 (ES: Vol. 23, 1, pp. 143–149)
*Germany 2009 (ES: Vol. 29, 2, pp. 289–292)
*Sweden 2006 (ES: Vol. 26, 4, pp. 820–823)
*Spain 2004 (ES: Vol. 24, 1, pp. 149–156)
*Finland 2007 (ES: Vol. 26, 4, pp. 797–803)
*Finland 2003 (ES: Vol. 23, 3, pp. 557–565)
*Portugal 2002 (EJPR: Vol. 43, pp. 1058-1066)
*Greece 2009 (ES: Vol. 29, 2, pp. 293–296)
i Correlation between ENEP and the maximum left-right distance is -0.081 (ns), correlation between ENEP and
the Dalton measure of polarization is -0.196 (ns) and correlation between ENEP and compactness is 0.313
(marginally significant at 0.1-level). ii Although recall questions tend to underestimate the true amount of volatility (Schoen & Falter, 2001), cross-
national panel data on elections is not available. While acknowledging possible underestimation of volatility,
relying on recall questions for investigating the individual determinants of electoral volatility between elections
is regularly done within the scholarly literature (Lachat, 2009). iii
Looking at the percentage of volatile voters in the elections for which we could code vote switching, the
percentage of switchers in Canada (87.64%) was implausible. Therefore the Canadian 2004 election was not
included in the analysis. iv One could refer to the 2002 election in The Netherlands where the populist LPF party gained 17percent of the
votes out of nothing, but that also saw a 6 percentage pointincrease in turnout. v We also removed Albania (2005) from the sample as the Freedom House index for that year only classifies it as
a ‘partly free’ democracy and the OSCE election report states that this election “complied, only in part, with
OSCE commitments and other international standards for democratic elections”. The elections of Bulgaria
(2001), Czech Republic (2006) and Poland (2007) were also not added for this paper as a control of the data with
the actual election result showed some differences when aggregating the responses for vote choice for the
‘previous’ election. These will be added to the analysis for this paper when these minor issues are resolved. The
Canadian 2004 survey had only 258 respondents out of 1674 that had information on their electoral choice in
both elections, which is a too large distortion of the original sample and the actual turnout to retain it (see
endnote iii on this). vi Note that Ezrow inverses this index to become a measure of dispersion rather than compactness as this is more
straightforwardly understood. We will use the original Alvarez & Nagler formula. vii
The pearson correlation between the Dalton and Alvarez & Nagler measure is -0,854, this high correlation
makes us predict that the difference in coefficient will probably be negligible. viii
Wrong answers, don’t knows and refusals to answer the questions were all treated equally as wrong.
37
ix
Elections included are the Australia 2004, Switzerland 2003, Czech Republic 2002, West Germany 2002,
Germany 2009, Denmark 2007, Spain 2004, Estonia 2011, Finland 2003 and 2007, United Kingdom 2005,
Greece 2009, Croatia 2007, Hungary 2002, Ireland 2002, Iceland 2007 and 2009, Israel 2003 and 2006, Italy
2006, the Netherlands 2006, Norway 2001 and 2005, New Zealand 2002 and 2008, Poland 2001, Portugal 2002,
Romania 2004, Slovak Republic 2010, Slovenia 2004 and Sweden 2002 and 2006. x For disproportionality, estimated probabilities were calculated for Ls 0.5 to 17, with delta 0.5. For the effective
number of parties, estimated probabilities were calculated for ENEP from 2 to 9, with delta 0.5. xi
The cross-level interactions were tested for the other measures of polarization (with Model III and Model IV as
the base model) as well. Neither for the maximum ideological distance nor for the Dalton measure of
polarization, there was a significant interaction effect with political knowledge. xii
Estimated probabilities were calculated for ENEP from 2 to 9, with delta 0.5. This was done both for a level of
satisfaction of 1 as well as for a level of satisfaction of 5. xiii