Does Democratic Consolidation Lead to a Decline in Voter Turnout? Global Evidence Since 1939


By Filip Kostelka, Institutions and Political Economy Research Group (IPERG), University of Barcelona & Centre d’études européennes, Sciences Po, Paris


Voting rates in new democracies often decline dramatically. For instance, in less than twenty years after democratization, voter turnout fell by 17.5 percentage points (pp) in Portugal (which has held democratic elections since 1975), 29 pp in El Salvador (1982), 30 pp South Korea (1988), and 47 pp in Romania (1990). Both researchers and journalists usually ascribe these declines to citizens’ disillusionment with the functioning of the new democratic regime and raising apathy in the face of reduced electoral stakes. Democratic consolidation, the process through which democracy becomes established, is believed to depress turnout.

However, this conventional interpretation is problematic. First, it does not explain the striking variation that exists among new democracies. My review of 91 democratic consolidations that took place between 1939 and 2015 shows that, in half of new democracies, turnout declined little or not at all (see Figure 1). For instance, in Spain, which, like neighbouring Portugal, experienced a military dictatorship that lasted several decades before democratizing in the mid-1970s, turnout decreased by only 2.6 pp. This is 7 times less than in Portugal. Second, a number of recent studies, including those employing a pre-/post-election panel design, have found no causal relationship between democratic dissatisfaction and voter turnout.

Figure 1: Voter Turnout Change in the First Six Democratic Elections


In a new study, I offer an alternative explanation of the deep participation declines that sometimes occur in new democracies. I argue that these declines originate in exceptionally high levels of participation at the beginning of the democratic transition, which are provoked by the democratization context. If regime change is a revolution or if citizens are used to voting from the authoritarian era, the democratization context is strongly mobilizing and conducive to electoral participation. In such cases, the voting rate in the founding democratic elections exceeds the “standard” level that would be expected in an established democracy with the same characteristics as the new democracy at hand. Later on, however, as the democratization context loses salience, voter turnout returns to the “standard” level, which is determined by the same factors as in established democracies. Only on face value, it seems that participation is depressed by the democratic consolidation context but, in reality, the dynamic is driven by what happened before the process of consolidation has even started.

To test both the conventional and my alternative accounts, I compiled an original dataset that covers most legislative elections held in new and established democracies between 1939 and 2015. I run two regression analyses. In the first, I test whether, in the first democratic elections, voter turnout reaches higher levels that those that would be expected in established democracies. I find that it is the case and that, as hypothesized, participation is particularly high after bottom-up democratizations, and when the preceding authoritarian regime held elections and forced its citizens to vote in them.

In the second analysis, I model voter turnout dynamics in new democracies. The results reveal that the initial participation surplus, which I call as the democratization bonus, translates into declines. This makes sense of the stark contrast between Portugal and Spain. In Portugal, the democratization process was mainly driven by the democratic opposition, which rendered the Portuguese founding election particularly mobilizing. In Spain, the democratization mover was the authoritarian regime, which limited the euphoria and stakes in the Spanish founding election.

Once this democratization bonus is controlled for, democratic consolidation seems to depress voter turnout only in post-communist democracies. Additionally, like in established democracies, voting rates in new democracies have tended to decrease since the 1970s regardless of the democratization and democratic consolidation contexts.

Altogether, the results untangle the complexity of voter turnout dynamics in new democracies. They show that a single occurrence of a voter turnout decline in new democracies may stem from one to three sources. In combination, the three sources can contribute to particularly sharp drops in electoral participation, such as that observed in Romania. They also explain why we have seen so many dramatic declines since the beginning of the third wave of democratization in 1974: prior to the third wave, opposition-driven democratizations from electorally mobilized dictatorships were rare, no consolidating democracy had to cope with Communist legacies, and the global environment was pushing turnout up in all democracies.

For more details, read Kostelka, Filip. 2017. “Does Democratic Consolidation Lead to a Decline in Voter Turnout? Global Evidence Since 1939.” American Political Science Review. Doi:

Posted in Blog | Comments closed

What’s Happening in Catalonia? The Crackdown May Increase Support for an Unpopular Plan

By Eric Guntermann, Université de Montréal

People all over the world this week read about Spanish police arresting Catalan government officials and confiscating ballots for an independence referendum. Rather than harming the Catalan government, however, I argue that the crackdown may help it raise support for its unpopular roadmap towards independence.

In the last regional election, held in 2015, secessionist parties argued that a majority of votes would give them a mandate to make Catalonia an independent state. However, they did less well than expected and needed the radical secessionist Popular Unity Candidacy (CUP) to govern. Together they received a majority of seats (53.3%) but not of votes (47.9%).

Nevertheless, they saw the result as a mandate to move unilaterally towards independence. They eventually announced a referendum on independence for October 1, 2017, even though the Spanish government insisted doing so was unconstitutional.

However, the Catalan population has been reluctant to support the regional government’s plan. In a survey Making Electoral Democracy Work (MEDW) ran prior to the Spanish national election held in 2016, only 43.3 % of respondents supported the government’s unilateral independence plan. In fact, more Catalans strongly opposed it (28.0%) than strongly supported it (24.2%). Catalan government surveys also showed that support for independence fell below 40% in 2015.


Parties and Ambivalent Citizens

In a paper that was recently accepted for publication at Party politics, I show that conflict between opposing party positions leads people who prefer one side to the other to adopt that party’s positions as their own. Thus, conflict between regional nationalist parties and the national government led by the People’s Party (PP), which is extremely unpopular in Catalonia, should increase support for the nationalists’ plan.  

In Catalonia, many citizens are ambivalent. In separate questions, the MEDW survey asked Catalans how attached they are to Catalonia and Spain on a scale from 0 to 10. I created a score representing how much more (or less) a respondent identifies with Catalonia than with Spain by subtracting identification with Spain from identification with Catalonia.

I plot the proportions with each value in Figure 1. We can see that overall Catalans identify more with Catalonia than with Spain. The mean value is 1.3. However, many Catalans have ambivalent identities and may be influenced by the PP’s opposition to the Catalan government’s actions.

Figure 1: Relative Identification with Catalonia and Spain



Survey Experiment: Does Opposition from the PP Increase Support for the Catalan Government’s Plans?

An experiment was added to the MEDW survey to see how Catalans respond to PP opposition to the Catalan government’s actions. Half of the 543 respondents were randomly assigned to a control group that read a statement that the “Catalan Government has decided to take steps towards the independence of Catalonia without an agreement with the Spanish Government”. The other half were assigned to a treatment group that read the same statement followed by “the Partido Popular (PP) opposes this decision”.

Respondents in both groups were asked whether they support or oppose the Catalan government’s actions. I focus on the half of Catalans with the most ambivalent identities, who, as argued above, are the most susceptible to being influenced by the PP’s opposition.

Figure 2 shows the responses among these participants. In the control group, 60.0% oppose the plan, while 27.09% support it. In the treatment group, opposition drops to 53.8% and support rises to 33.8%.

Figure 2: Effect of PP Opposition on Support for Unilateral Steps Towards Independence (Ambivalent Identifiers)


To test for significance, I created a numeric version of the support variable (coded from 1 to 4, where higher numbers indicate greater support for unilateral independence). The difference between the two groups is 0.32 (p<0.05). Thus, ambivalent respondents who read about the PP’s opposition to unilateral independence were significantly more supportive of the Catalan government’s plan than those who did not read about that opposition.

Given that a simple statement that the PP opposes the secessionists’ plan influences support for it, it is likely that the powerful assertion of the PP’s opposition we saw this week will have a much stronger effect.


For the article referenced above, see Guntermann, Eric (2017). “Party influence where predispositions are strong and party identification is weak: Assessing citizens’ reactions to party cues on regional nationalism in Spain” Party Politics,

Posted in Blog | Comments closed

Was my decision to vote (or abstain) the right one?

By André Blais, Fernando Feitosa and Semra Sevi, Université de Montréal 

How do individuals, after the election, evaluate their decision to vote or abstain? This is the question our study examines in our recent publication in Party Politics.  

Respondents were asked, right after the election, how satisfied they are with their decision to vote (or not to vote)” We hypothesize that satisfaction (or lack of it) is driven primarily by voters’ motivation to vote. The study includes interest in politics, a sense of civic duty, and party attachment as motivating factors for voting, and it also examines how the cost of voting and socio-demographic factors affect voters’ satisfaction with their decision to vote or to abstain. Finally, the study examines whether priming voters about the result of the election affects their satisfaction with their decision.

We use an original dataset that is part of the Making Electoral Democracy Work Project, led by André Blais. We analyze a total of 22 surveys conducted in 5 different countries (Canada, France, Germany, Spain and Switzerland) in national, supra-national and sub-national elections held between 2011 and 2015. The surveys ask respondents, after the election, to indicate whether they think that their decision to vote (or abstain) was a very good decision, a fairly good decision, a fairly bad decision, or a very bad decision. Half of the respondents are primed to consider the results of the election (“Given the outcome…”) before indicating their level of satisfaction.

Our study tests the four following hypotheses:

  • The stronger the motivation to vote, the more satisfied (dissatisfied) one is with her decision to vote (abstain).
  • The higher the cost of voting, the less (more) satisfied one is with her decision to vote (abstain).
  • Men, older and better educated citizens are more satisfied with their decision to vote or abstain.
  • Priming ‘given the outcome’ decreases (increases) satisfaction among voters (abstainers).

We find that an overwhelming majority of those who voted in an election feel, ex post, that they made the right decision, while non-voters are less certain about the correctness of their choice to abstain. Table 1 shows the total distribution of responses to our satisfaction questions. Of those who voted, 71 per cent indicated that they made a very good decision and another 26 per cent said that it was a fairly good decision. An overwhelming majority of voters thus felt very positive about their decision to vote. Things were different among abstainers. Only 26 per cent of abstainers qualified their choice as a very good one, while as many as 40 per cent acknowledged that it was a very or fairly bad decision not to vote. Abstainers are thus much more critical of their choice than voters.


Table 1: Distribution of responses (in %) to satisfaction with decision to vote/abstain


to vote



to abstain


Very bad decision 1 15
Fairly bad decision 2 25
Fairly good decision 26 34
Very good decision 71 26
N 17,561 1,891


We also show that each motivating factor (interest in politics, sense of duty, party attachment) has an impact on voter (non-voter) satisfaction with their decision to vote (abstain), confirming H1. The findings present only partial support for H2, and suggest that motivational factors trump cost considerations in voters’ judgments. The data agree fully with the predictions of H3 for age, but only partially for gender and education. Finally, the prime has a substantial marginal impact, decreasing the likelihood of saying that the decision to vote was a very good decision by 10 percentage points. Interestingly, the prime does not have any impact among those who had abstained. 

Our study shows that citizens’ views about the correctness of their decision to vote or abstain depend first and foremost on the strength of their motivation to vote. Those who are highly interested in politics, who believe that they have a moral obligation to vote, and who feel close to a party have no doubt that their decision to vote was the right one while the verdict is much less positive when they failed to vote.


For more details, see Blais, A., Feitosa, F., and Sevi, S. (2017). “Was my decision to vote (or abstain) the right one?” Party Politics,


Posted in Blog | Comments closed

MEDW-APSA Mini-Conference on Political Behavior This Friday (September 1st) in San Francisco!

By André Blais (University of Montreal) and Filip Kostelka (University of Montreal; Sciences Po, Paris)


Political scientists from all over the world will present their research this Friday at a mini-conference on political behavior organized by the Making Electoral Democracy Work project in San Francisco. The mini-conference is incorporated in the American Political Science Association (APSA) meeting and takes place at the Westin St. Francis Hotel (California East room) from 8 AM to 5:30 PM.  All participants in the APSA meeting are warmly encouraged to attend the mini-conference. The programme is as follows:

8:00am-9:30am Panel 1: Participation and Partisanship,

Panel Chair: André Blais (University of Montreal)

Panel Discussant: Indridi H. Indridason (UC Riverside)


  1. Elections Activate Partisanship Across Countries

Shane P. Singh (University of Georgia), Judd Thornton (Georgia State University)

  1. Does Too Much Democracy Kills Participation? Election Frequency and Voter Turnout in Canada and Germany

Filip Kostelka (University of Montreal and Sciences Po, Paris), Alexander Wüttke (University of Manheim)

  1. The Correlates of Duty: Universal or Context Specific?

Laura Stephenson (University of Western Ontario)

  1. Looking in a Carnival Mirror: Ideology and Protest Participation in Old and New Democracies

Filip Kostelka (University of Montreal and Sciences Po, Paris), Jan Rovny (Sciences Po, Paris and University of Gothenburg)


09:30am-09:45am Break


09:45am-11:15am Panel 2: French Elections

Panel Chair: Martial Foucault (Sciences Po, Paris)

Panel Discussant: Sylvain Brouard (Sciences Po, Paris)


  1. Aging, habit and turnout. New evidence from 12 voting rounds in France

Jean-Yves Dormagen (Université de Montpellier)

  1. From votes to seats. The 2017 French legislative elections

Annie Laurent (Université de Lille), Bernard Dolez (Université Paris 1)

  1. Performance voting and the selection of alternatives.

Nicolas Sauger (Sciences Po, Paris)

  1. The Wealth Effect on the 2017 French Presidential Outcomes

Martial Foucault (Sciences Po, Paris)


11:15am-11:30am Break


11:30am-1:00pm Panel 3: Public Opinion and Satisfaction with Democracy

Panel Chair: Laura Stephenson (University of Western Ontario)

Panel Discussant: Damien Bol (King’s College London)


  1. Does Ideological Congruence Matter? Assessing Its Impact on Satisfaction with Democracy

Eric Guntermann (University of Gothenburg)

  1. Revisiting the notion of electoral winner

Jean-François Daoust (University of Montreal)

  1. Conventional wisdom or paradigm shift? Getting at the root causes of temporal variability in public opinion towards immigration

Steven Vanhauwaert (Centro de Investigación y Docencia Económicas), Patrick English (University of Manchester)


1:00pm-2:15pm Lunch Break


14:15pm-3:45pm Panel 4: How do voters decide?

Panel Chair: Filip Kostelka (University of Montreal and Sciences Po, Paris)

Panel Discussant: Ruth Dassonneville (University of Montreal)


  1. How government alternation shapes voter incentives to engage in compensational voting

Carolina Plescia (University of Vienna) and Francesco Zucchini (University of Milan)

  1. What Are the Causes of Voters’ Indecision? A Study of Late Deciders in Western Democraties

Simon Willocq (Université Libre de Bruxelles)

  1. Electoral Accountability in France

Martial Foucault (Sciences Po, Paris), Romain Lachat (Sciences Po, Paris), Guy Whitten (Texas A&M University)


3:45pm-4:00pm Break


4:00pm-5:30pm Panel 5: Study of Elections, Parties and Electoral Rules

Panel Chair: Jan Rovny (Sciences Po, Paris and University of Gothenburg)

Panel Discussant: Shane P. Singh (University of Georgia)


  1. Voting and satisfaction with democracy under proportional representation: Does ballot structure matter?

Damien Bol (King’s College London), André Blais (Université of Motnreal), Lidia Nunez (Université libre de Bruxelles), Jean-Benoit Pilet (Université libre de Bruxelles)

  1. Negative Campaigning in Multi-Party Contests

Charles Crabtree (Pennsylvania State University), Matt Golder (Pennsylvania State University), Thomas Gschwend (University of Manheim) & Indridi H. Indridason (UC Riverside).

    3. The Effects of Survey Mode and Sampling in Belgian Election Studies: A Comparison of a National Probability Face-to-Face Survey and a Non-Probability Internet Survey

Ruth Dassonneville (Université de Montréal), Kris Deschouwer (Vrije Universiteit Brussel) and Marc Hooghe (University of Leuven).


If you have questions about the mini-conference, do not hesitate to contact Filip Kostelka (filip.kostelka@umontreal).


Posted in Blog | Comments closed

MEDW Survey Data Now Available in Open Access!

By André Blais (University of Montreal), Laura B. Stephenson (University of Western Ontario), Damien Bol (King’s College London), and Filip Kostelka (University of Montreal) on behalf of the MEDW Team  

The MEDW team is proud to announce that all survey data from the Making Electoral Democracy Work project are now available in open access in the Harvard Dataverse. In total, this represents 32 studies from national, subnational and supranational elections, which were conducted in seven established democracies between 2010 and 2016 (see the list below). The core data module is an aggregated dataset that combines information from 27 surveys fielded in 11 regions from five countries (Canada, France, Germany, Spain, and Switzerland) and that spans over 40 000 respondents. The project dataverse also includes 5 “special datasets” (additional studies conducted in Belgium, France, Germany, Spain and Sweden) that are not incorporated in the aggregated dataset but contain a large number of questions asked in the regular MEDW surveys.

The data are characterized by several unique features that may be of interest to political scientists and students of political behavior, mass political attitudes, and multi-level governance. They notably provide the same kind of information for different levels of government, which allows for cross-level comparison and the study of cross-level interactions. The topics covered by the surveys include satisfaction with democratic and government performance, political efficacy, election importance, perceptions of the voting act, attachment to different levels of government, perceived corruption, value orientations on economic and social issues, closeness to political parties, exposure to campaign information and mobilization efforts, electoral and non-electoral political participation, pre-election voting intentions, and reported vote choice (see the aggregated dataset codebook).  Almost all of the studies are two-wave panels, in which a pre-election questionnaire was administered approximately one week before the election and a post-election questionnaire was launched in the period immediately after the election. While most attitudes were measured in the first wave, electoral participation, vote choice and non-electoral political participation were measured in the second wave. This allows for greater confidence in the direction of causality (i.e., from attitudes to behavior). Last but not least, the questionnaires feature several experimental questions where half of the respondents were asked a different wording of the same question or the same questions in a different order.

The MEDW project dataverse contains the Aggregated Dataset folder and 6 (sub) dataverses with election-specific datasets: “MEDW Canada”, “MEDW France”, “MEDW Germany”, “MEDW Spain”, “MEDW Switzerland” and “MEDW Special Datasets”. All datasets are in Stata (13) format and are supplemented with codebooks, questionnaires (in the original language and / or in English), and technical reports.

For more information, access the project dataverse and do not hesitate to contact André Blais (

 We thank all members of the MEDW team who contributed to the project and to the production of the data. They include Christopher H. Achen, John Aldrich, Christopher Anderson, Vincent Arel-Bundock, David Austen-Smith, Laurie Beaudonnet, Stefanie Beyens, Marian Bohl, Maxime Coulombe, Charles Crabtree, William Cross, Fred Cutler, Jean-François Daoust, Ruth Dassonneville, Kris Deschouwer, Delia Dumitrescu, Jim Engle-Warnick, David Farrell, Fernando Feitosa, Benjamin Ferland, Martial Foucault, Carol Galais, François Gélineau, Elisabeth  Gidengil, Matt Golder, Sona Golder, Thomas Gschwend, Marc Guinjoan, Eric Guntermann, Phillip Harfst, Maxime Héroux-Legault, Rafael Hortala-Vallve, Indridi Indridason, Ekrem Karakoc, Anja Kilibarda, Hanspeter Kriesi, Simon Labbé St-Vincent Simon, Romain Lachat, Ignacio Lago Peñas, Jean-François Laslier, Jean-Michel Lavoie, Andrea Lawlor, Louis Massicotte, Mike Medeiros, Claude Montmarquette, Alexander Morin-Chassé, Ben Nyblade, Scott Pruysers, Paul Quirk, Victoria Savalei, Semra Sevi, Shane Singh, Karine van der Straeten, Katherine Sullivan, Charles Tessier,  Tom Verthé, and Steffen Zittlau.

The 32 MEDW studies were transformed into 23 datasets that are archived in the project dataverse. This is their exhaustive list:

“Making Electoral Democracy Work” [aggregated dataset]

In the (sub)dataverse  “MEDW Canada”:

“MEDW 2011 Ontario Provincial Election Study”

“MEDW 2012 Quebec Provincial Election Study”

“MEDW 2015 Canadian Federal Election Study”


In the (sub)dataverse “MEDW France”:

 “MEDW 2012 French Legislative Election Study”

“MEDW 2014 French Municipal Election Study”

“MEDW 2014 French European Election Study”


In the (sub)dataverse “MEDW Germany”:

“MEDW 2013 Lower Saxony State Election Study”

“MEDW 2013 Lower Saxony Federal Election Study”

“MEDW 2014 Lower Saxony European Election Study”

“MEDW 2013-2014 Bavaria Panel Study”


In the (sub)dataverse  “MEDW Spain”:

“MEDW 2011 Spanish National Election Study”

“MEDW 2012 Catalan Regional Election Study”a

“MEDW 2014 Spanish European Election Study”

“MEDW 2015 Madrid Regional Election Study”


In the (sub)dataverse  “MEDW Switzerland”:

“MEDW 2011 Zurich Cantonal Election Study”

“MEDW 2011 Swiss Federal Election Study”

“MEDW 2011 Lucerne Cantonal Election Study”


In the (sub)dataverse “MEDW Special Datasets”:

“MEDW 2010 North Rhine-Westphalia State Election Study”

“MEDW 2010 Swedish National Election Study”

“MEDW 2012 French Presidential Election Study”

“MEDW 2014 Belgian National Election Study”

“MEDW 2016 Spanish National Election Study”


Posted in Blog | Comments closed

Duty to vote and political support in Asia

By Carol Galais, Universitat Oberta de Catalunya & André Blais, Université de Montréal 

What is the story

Although sense of civic duty is a well-known predictor of voting behavior, we do not know much about its foundations.  Hence the question: what are the psychological orientations that predispose citizens to believe that they do (or do not) have a duty to vote?

The moral and communitarian nature of the duty to vote have led several authors to establish links between this belief and attitudes towards democracy and one’s country. We posit that the belief that voting is a duty will be stronger among supporters of democracy and/or of their political community. We also posit that two contextual factors, degree of democracy and ethnic fractionalization, moderate the effect of political support on duty.

We test these expectations by means of two Asian Barometer Surveys conducted in 2004 and 2005, gathering 21 Asian countries, which provide a sufficiently diverse sample in terms of both democracy and heterogeneity.


Figure 1 shows the mean value of duty in each of the 21 countries. These values range from .99 (Bangladesh) to .67 (Turkmenistan). South Asian countries score higher, while East Asian countries (Japan, Korea and Mongolia) come second in line. Southeast and Central (former Soviet Socialist Republics) Asian countries tend to score lower.

Figure 1: Average duty to vote per country.

The estimations presented in Table 1 confirm the relationship between both aspects of political support and the duty to vote; support for one’s political community matters more than support for democracy. Finally, cross-level interactions are significant: the impact of support for democracy on duty is stronger in more democratic countries, and the effect of support for political community is more pronounced in ethnically fractionalized countries.

Table 1: Multilevel estimations of the duty to vote.


The belief that one has a moral obligation to vote stems more from attachment to one’s country than from the belief that democracy is a good thing.
The interaction between support for democracy and duty suggests that, under dictatorship, democracy supporters are more focused on basic rights than on their own duties. It is also possible that citizens develop beliefs about their duties as citizens only once the political system guarantees them political rights.
As for the interaction between fractionalization and duty, this suggests that societies that have overcome their divides can turn diversity into strength, building more encompassing national identities which bolster the civic duty to vote. Those who are proud of their diverse country adhere to the symbols, rules and institutions that keep the country together.

For more details, see: Galais, C., & Blais, A. (2016). “Duty to Vote and Political Support in Asia. International.” Journal of Public Opinion Research,

Posted in Blog | Comments closed

What do voters do when they like a local candidate from another party?

By André Blais, University of Montreal & Jean-François Daoust, University of Montreal 

What is the story?

In politics, what happens at the district level is still understudied compared to the national level. Paying attention to MPs and constituency politics seems to be in conflict with the traditional understanding of political representation. We focus on the local level and we address two questions:

  • How many voters particularly like a candidate from another party?
  • Do these voters vote for their preferred party or their preferred candidate?

 We use the Making Electoral Democracy Work data from the 2015 Canadian federal election.


Forty-eight per cent of the respondents expressed a preference for a local candidate. As could be expected, most of them (81%) mentioned the candidate associated with their preferred party. We thus have about 52 per cent with no preference for a local candidate, 39 per cent with a preference for the candidate of their preferred party, and 9 per cent who particularly like a candidate from a non-preferred party. Table 1 displays the descriptive statistics, where we first indicate the proportion of voters with no local candidate preference, then those with a local candidate preference which is congruent with the preferred party and finally those with a local candidate preference which is not congruent with the preferred party. This latter category is the one of interest.

Table 1 Party and Local Candidate Perference 

We focus on the group that interests us the most, that is, those who like a candidate from a non-preferred party. Our (very simple) question is: Do they vote for their preferred party or their preferred candidate?

We find that 60 per cent of those with conflicting preferences voted for the preferred party and 40 per cent for the preferred candidate. Table 2 displays the results.


Table 2 Vote choice among Non-Strategic Voters with Incongruent Preferences

In general, those who prefer stronger parties (the Liberals and the Conservatives) tend to vote to a greater extent for their preferred party and those who prefer weaker parties tend to vote for their preferred local candidate. Quebeckers are somewhat more likely to stick with their preferred party. In short, in the three provinces, around one voter out of ten particularly liked a candidate from a party other than the one he or she preferred in the 2015 Canadian election. For two out of five of such voters, the preference for the local candidate trumped the party preference. Sticking with the local candidate is more frequent among those who prefer smaller parties, like the Bloc and the NDP.

For more details, see Blais, André and Jean-François Daoust. What do voters do when they like a local candidate from another party? Canadian Journal of Political Science. DOI:

Posted in Blog | Comments closed

Les législatives françaises de 2017 : Pourquoi la participation était-elle si faible et comment peut-on y remédier?

Par Filip Kostelka (cordinateur du projet Making Electoral Democracy Work)

Note: Version en anglais se trouve ici. / English version is here
Ce texte est une tribune publiée sur le site du journal Le Monde. 


Le fort taux d’abstention constitue le résultat le plus frappant du premier tour des élections législatives de dimanche dernier. La participation de 48,7 %, en déclin de 8,8 points par rapport à 2012, est la plus faible dans l’histoire des scrutins législatifs depuis 1945. L’une des causes principales est sans doute la mutation du système partisan: l’effondrement des partis du centre gauche et du centre droit, la décrédibilisation du l’extrême droite lors du dernier débat présidentiel, ainsi que le profil centriste et peu mobilisateur du vainqueur présumé de LREM.

Néanmoins, un autre facteur explicatif important est la fréquence record des élections. Le vote du dimanche a été le troisième en 2017 après les deux tours des élections présidentielles. Plus généralement, pendant les trois dernières années, certains des électeurs français ont pu voter – en fonction de la compétition partisane dans leurs circonscriptions – à 9 occasions : aux élections municipales de 2014 (deux tours), aux élections européennes de 2014 (un tour), aux élections départementales de 2015 (deux tours), aux élections régionales de 2015 (deux tours), et aux élections présidentielles de 2017 (deux tours). De surcroît, les Français ont également pu participer aux deux tours de chacune des élections primaires tenues avant les présidentielles par les Républicains, EELV et le Parti socialiste, entre novembre 2016 et janvier 2017.

Cette multiplication des scrutins est sans précédent dans l’histoire électorale française. Encore dans les années 1970, les citoyens étaient incomparablement moins sollicités. A titre d’exemple, pendant les trois années qui ont précédé l’élection législative de 1978, les Français sont allés voter au maximum quatre fois. Une moitié des électeurs a pu participer aux élections cantonales de 1976 (deux tours) et l’ensemble de l’électorat a été invité à renouveler les conseils municipaux en 1977 (deux tours).

Graphique 1 Les taux de participation au premier tour des élections législatives depuis 1958

En effet, la fréquence des élections s’est considérablement accrue depuis les années 1970. Cet accroissement provient d’une série de réformes institutionnelles : l’adoption des élections directes au Parlement européen (1979), la décentralisation et l’introduction des élections régionales (1986), et l’instauration du quinquennat présidentiel (2002). De plus, les réformes territoriales de 2010 et 2013 ont temporairement réduit la durée du mandat des élus régionaux et cantonaux de 6 à 5 et 4 ans respectivement. Enfin, il y a les primaires organisées de façon ouverte par le PS et l’EELV depuis 2012 et par les Républicains depuis 2017. Cette forte hausse de la fréquence électorale s’est accompagnée d’un déclin progressif de la participation électorale aux scrutins législatifs. Ayant commencé au début des années 1980, ce déclin a atteint un record dimanche (voir le Graphique 1).

Les études en science politique suggèrent qu’une fréquence élevée des élections réduit la participation en affectant à la fois les attitudes des citoyens et les capacités de mobilisation des partis politiques. Dans mes recherches, j’ai confirmé cet effet négatif de la fréquence des élections sur la participation électorale dans deux contextes différents : les démocraties postcommunistes en Europe centrale et orientale et deux démocraties fédérales en Europe occidentale (le Canada et l’Allemagne). Plus d’élections équivaut à moins de participation dans chacune de ces élections, et, en particulier, dans les élections de moindre importance.

Pour réduire l’abstention électorale, la France devrait s’inspirer de pays qui connaissent des taux de participation plus élevés. Le meilleur exemple en est la Suède, qui est l’une des rares des démocraties occidentales où la participation électorale n’a pas diminué pendant au cours des derniers vingt années. Les suédois ne votent habituellement que deux fois tous les quatre ans car toutes les élections, sauf celles au Parlement européen, se tiennent simultanément. Il est vrai que la tenue simultanée des élections de différents types comporte le risque d’une « contamination ».

Par ce terme, les politistes désignent l’impact des enjeux politiques dans une arène électorale (par ex. législative) sur les résultats électoraux dans une autre arène (par ex. régionale). Cependant, la participation électorale en dessous des 50 % aux élections législatives est vraisemblablement bien pire que tout risque potentiel de contamination.

En combinant différents types d’élections, il est possible d’obtenir un nombre optimal d’échéances électorales qui maximise la participation sans exacerber les risques de contamination. Dans le contexte français, il serait logique de tenir simultanément les présidentielles et législatives d’un côté, et les différents scrutins locaux (élections régionales, départementales, et municipales) de l’autre. Cela ne ferait pas augmenter la participation seulement grâce à une fréquence des élections plus raisonnable mais aussi parce que les scrutins jugés moins importants (par ex. les législatives) bénéficieraient du caractère mobilisateur des scrutins jugés plus importants (par ex. les présidentielles).


Filip Kostelka a soutenu à Sciences Po, Paris une thèse intitulée «Mobiliser et démobiliser: le déclin énigmatique de la participation électorale dans les démocraties postcommunistes». Il est actuellement chercheur postdoctoral à la Chaire de recherche en études électorale de l’Université de Montréal et chercheur associé au Centre d’études européennes, Sciences Po, Paris.


Posted in Blog | Comments closed

The 2017 French Legislative Election: Why Was Voter Turnout So Low and What Can Be Done About It?

by Filip Kostelka, University of Montreal and Sciences Po, Paris

Note: French version is here. / Version en français se trouve ici. / 
This text was published in French as an op-ed  in the journal Le Monde. 


The most striking outcome of the first round of the 2017 election to the French National Assembly is that less than half of the registered voters came to the polls. The participation rate of 48.7 %, down by 8.5 percentage points from the last election in 2012, is the lowest in the history of the French legislative contests since 1945. Two factors are likely to have contributed to this particularly weak participation rate. The first and obvious factor is the recent transformation of the French party system: the collapse of the traditional parties on the centre left and centre right; the far right’s loss of credibility in the preceding presidential election; and the centrist profile of the anticipated winner, unlikely to generate strong positive or negative mobilization in the electorate.

Yet, there is another important culprit: high election frequency. Sunday’s election was the third round of voting in 2017 after two rounds of presidential elections. More generally, in the last three years, a French citizen could vote – depending on party competition in his or her electoral district – in up to 9 contests: municipal elections (2014, 2 rounds), European Parliament elections (2014), departmental elections (2015, 2 rounds), regional elections (2015, 2 rounds), and presidential elections (2017, 2 rounds). On top of that, French voters could also participate in two rounds of open presidential primaries organized in the run-up to the 2017 presidential elections by the main centre-right and centre-left parties as well as the Greens. This proliferation of elections is unprecedented in the French electoral history. Just a few decades ago, the number of participatory demands on French citizens was substantially lower. For instance, in the three years preceding the legislative election of 1978, there were at maximum 4 opportunities to vote: departmental elections (1976, 2 rounds but only half of the electorate was eligible to vote) and municipal elections (1977, 2 rounds).

Figure 1: Voter Turnout in the First Round of the French Legislative Elections since 1958

As a matter of fact, election frequency in France has strongly increased since the late 1970s. This is due to a host of institutional reforms: the introduction of direct elections to the European Parliament (1979), decentralization and the introduction of regional elections (1986), and the reduction of the presidential mandate from 7 to 5 years (2002). In addition, before last Sunday’s election, new territorial reforms (of 2010 and 2013) resulted in a temporary reduction of the term of the regional and some departmental representatives from 6 to 5 and 4 years respectively. Finally, mainstream French political parties have newly held open primaries before presidential elections: the centre left since 2012 and the centre right since 2017. This steep rise in election frequency coincides with the decline in voter turnout in the French legislative elections, which started in the early 1980s and reached its peak on Sunday (see Figure 1).

Political science literature shows that high election frequency depresses voter turnout through several channels, affecting both citizens’ attitudes and political parties’ mobilization capacities. In my research, I found support for the negative effect of election frequency on voter turnout in two very different contexts. First, in my PhD dissertation defended at Sciences Po, Paris in 2015, I demonstrate that election frequency substantively contributes to the strong decline in voter turnout that has been observed in post-communist democracies since the 1990s. Second, in a paper presented at the 2017 Canadian Political Science Association meeting, my co-author Alexander Wüttke (University of Manheim) and I observe a robust relationship between election frequency and voter turnout in Canada and Germany. The more frequent elections are the lower voter turnout in every single election, particularly in less important elections.

As low voter tumour is normatively undesirable, French policy-makers should take lessons from other countries that record (much) higher voting rates. The best example is Sweden, one of the rare Western democracies in which voter turnout even increased since the early 1990s. Swedes typically vote twice every four years as all elections but those to the European parliament are held simultaneously. Of course, the simultaneity of different election types entails the risk of contamination (i.e. the political developments in one electoral arena may affect the results in another arena). Nonetheless, an abstention rate of more than 50 % is perhaps worse than any realistic degree of contamination.

Combining various types of electoral contests could achieve a Pareto-optimum number of elections in terms of high turnout and low contamination effects across different electoral arenas. In the French context, it seems logical to combine presidential and legislative elections on the one hand; and municipal, departmental, and regional elections on the other. This would boost voter turnout not only because of lower election frequency but also because the less important election type (e.g. legislative elections) would benefit from the mobilization effect of the more important type (e.g. presidential elections). Such a measure would probably not solve the issue of the decline in voter turnout altogether but it could largely offset the negative trend.

Posted in Blog | Comments closed

What is more important: winning or winning in a fair election?

By Jessica Fortin-Rittberger, Philipp Harfst, and Sarah C. Dingler (University of Salzburg)

What is the story?

Perceptions of electoral fairness have an impact on voters’ attitudes and behaviour. Consequently, we believe that electoral malpractice will negatively affect citizens’ satisfaction with democracy. In a recently published article in the Journal of Elections, Public Opinion and Parties we argue that election tinkering shapes citizens’ feelings about democracy.


The first hypothesis we examine draws on the direct link between electoral fraud and voters’ attitudes towards democracy. We expect that a high amount of fraud reported in elections should be related to low levels of satisfaction with the way democracy works.

Winning an election could change the nature of this relationship and positively affect voters’ attitudes irrespective of the degree of electoral misconduct. Thus, we also hypothesise that voters’ status as winners or losers will mediate the effect of electoral fraud on satisfaction with democracy.


Using survey data from CSES, we explore 48 elections in 29 countries in the timespan between 1998 and 2006. Rather unsurprisingly, we can show that high levels of electoral fraud correspond to a lower degree of satisfaction with democracy (see Figure 1).

Figure 1: Overall magnitude of problems in elections in relation to average satisfaction with democracy.

Notes: Data taken from CSES and QED. Figure contains 57 elections. Pearson’s r = −0.55.

However, this relationship is not straightforward. We find that citizens’ attitudes also depend on the outcomes of the election. While citizens who have voted for the winning party are more satisfied, this relationship only holds when elections are free and fair. As soon as elections are fraught with manipulation and malpractice, winning and losing no longer exert different effects on voters’ evaluation of the way democracy works. Election fraud thus affects the perceptions of citizens in the same way, no matter if they are on the winning or losing side.


Our findings have a far-reaching impact: If satisfaction with democracy is anchored on citizens’ evaluation of the performance of governments, the cost of electoral malpractice is high. Fraudulent practices are likely to negatively affect citizens’ evaluations of government and, ultimately, could undermine regime stability, especially in emerging or fragile democracies. We know that broad support for democratic values is an underlying condition for democratic consolidation. Widespread electoral fraud could therefore result in particularly inauspicious climates for the survival of new democracies. Yet our findings offer a glimmer of hope: citizens’ levels of satisfaction in third wave democracies remains higher than in older established democracies in spite of electoral malpractice.

For more details, see Jessica Fortin-Rittberger, Philipp Harfst & Sarah C. Dingler. 2017. “The costs of electoral fraud: establishing the link between electoral integrity, winning an election, and satisfaction with democracy”. Journal of Elections, Public Opinion and Parties, DOI: 10.1080/17457289.2017.1310111.



Posted in Blog | Comments closed