The EuroVotePlus experiment

Abstract

This paper reports on an online experiment that took place in several European countries during the three weeks before the 2014 elections for the European Parliament. We created a website where visitors could obtain information about the electoral rules used in different European Member States for this election. Participants were then invited to cast (simulated) ballots for the election according to three voting rules: closed list proportional representation, open list proportional representation with preferential voting, and open list proportional representation with cumulative voting and panachage. Participants were also invited to think about, and experiment with, the idea of electing some members of the European Parliament through pan-European party lists. The data gathered from this study enable researchers to consider the effects of electoral systems on outcomes in individual countries, and also to investigate the potential popularity and effects of Europe-wide European Parliament constituencies.

Keywords

European Parliament Election experiment open lists proportional representation voting rules

Introduction

Electoral systems have long been a topic of study for political scientists. A vast literature (see Cox, 1997; Duverger, 1951) considers how the rules of the game create incentives that affect the behavior of both parties and voters. Unfortunately, our understanding of the behavioral effects of electoral systems is limited by the available data. Most studies have used election outcomes, either at the national or district level, to compare results under different electoral systems (e.g., Clark and Golder, 2006; Singer and Stephenson, 2009). However, evidence of behavioural effects is most appropriately evaluated at the individual level. The difficulty is that the researcher needs to have some understanding of how voters would behave under a different electoral system. This type of information is not accessible in real-world election contexts.

Recently, researchers have tried to circumvent this problem with studies that invite voters to cast ballots under different electoral rules in the context of actual election contests. Building upon an experimental protocol first used in the field (Laslier and Van der Straeten, 2008), researchers have been able to calculate the psychological and mechanical effects of electoral systems (Blais et al., 2012; Van der Straeten et al., 2013).

In this article, we present information about a similar study that was conducted at the time of the 2014 European Union elections—EuroVotePlus. The data gathered from this study will enable researchers to consider the effects of electoral systems on individual voting behavior and on outcomes in individual countries, as well as to investigate the potential popularity and effects of European-wide European Parliament (EP) constituencies. In the following, we describe the objectives and implementation of the study, report some original descriptive findings, and conclude by considering its limitations.

Objectives of the study

In a context of decreasing identification with the European idea, and of disenchantment with its institutions, it is important to understand the relationship between European citizens and the EU as it is now, and as it could be, with alternative institutions (Habermas, 2012). In this spirit, recent studies have explored the possibility of facilitating political debates within a population as diverse as ‘the Europeans’ by experimenting with deliberative polls (see Fishkin et al., 2014; Isernia and Fishkin, 2014; and other articles in the special issue of this journal devoted to the EuroPolis experiment). The work presented here focuses on the election of Members of European Parliament (MEPs).

We concentrate on two aspects of how MEPs are elected: (a) that they are only elected from national lists and (b) that electoral systems are not equally open. We are interested in understanding how the voting behavior of European citizens would change if we altered these two factors. We have two main research questions:

Research question 1: National or European delegates? Each EU Member State has a certain number of seats (see the web appendix for details), and elects its own MEPs. Some have recently proposed adding additional members, elected through party lists in a single pan-European constituency.¹ Arguments in favor of this reform include the idea that it would help to foster a European identity–citizenship, better show voters how European politics actually works, and prevent MEPs from focusing on domestic interests only. However, MEPs elected in a constituency as large as the whole of Europe might also lose touch with local interests. Similarly, it might be very difficult for candidates to campaign among voters who do not speak the same language, and for voters to gather information on foreign candidates.²

Our objective with respect to this first issue is to test the reactions of European citizens to the idea of a pan-European constituency and to study how they would vote if given such an opportunity.

Research question 2: Voting for parties or for candidates? Research on attitudes toward voting rules has shown (Fournier et al., 2011) that citizens are strongly concerned about the nature of their vote and about the possibility of specifying their preferences more precisely. The voting rules used for EP elections vary considerably along this ‘voter choice’ dimension (see Table A1 in the web appendix for more details). As we are interested in understanding the effects of these differences, we chose to consider three real-world systems that differ along this dimension:

• France: Closed lists. Voters vote for one party list. Candidates on a list are ranked. The ranking is used to determine who on the list is elected: If the list wins two seats, the top two candidates on the list are elected. The party decides on the ranking of candidates and voters have no say over which candidates on a list are elected.

• Latvia: Open lists with preferential voting. Voters vote for one party list. On her chosen list, a voter can cross out the names of some candidates and give a ‘+’ to others. The score of a candidate is equal to the number of voters who chose the list minus the number of voters who crossed out his name, plus the number of voters who gave him a ‘+’. Within a list, candidates are elected according to their scores.

• Luxembourg: Open lists with cumulative voting and panachage. With k positions to fill, the voter has k votes available. She can freely distribute these votes to candidates of various lists, and she can give up to two votes to the same candidate. Votes are counted by lists, and candidates in a list are elected according to their scores.

Our objective with respect to this second issue is to compare voters’ behavior and electoral outcomes under different systems, and to consider the opinions of voters about these different systems. More flexible voting rules give citizens more control over politicians, but it may be difficult to gather information about candidates.

One of the advantages of this study is that, by focusing on different types of proportional representation (PR) rules, we open up new avenues of research. The literature on the impact of open and closed forms of PR with respect to voters is not well developed (for a review, see Colomer, 2011), but there are some relevant studies. On the basis of international comparisons, Karvonen (2004) shows that openness has few discernible effects at the aggregate level, except maybe on the central control of campaign finance and on party cohesion. Blumenau et al. (2014) use a survey experiment conducted in 2013 in the United Kingdom to examine the effect of open and closed ballots on party vote shares in a hypothetical EP election. They find that mainstream parties perform better under open than under closed rules (and vice versa for niche parties).³ Most of the other studies concentrate on the consequences of electoral systems for the behavior of the candidates and political outcomes: Are politicians more or less corrupt, campaigns more or less populist, minorities more or less favored in these variants of PR (Ames, 1995; Chang, 2005; Huber, 2012; Shugart et al., 2005)? Hix (2004), for instance, shows that the ballot structure (open or closed list) in a PR system determines whether MEPs will support their national party (closed list) or their parliamentary grouping (open list) in cases of conflict between the two. Thus, in contrast to most of the existing literature, our approach is focused on the behavior of individual voters in different systems.

The website

The interested reader will more easily follow this section by visiting the study’s website eurovoteplus.eu. To gather information on voter behavior under different electoral rules, we followed the procedure used in previous studies in Ontario, Quebec, and France (Blais et al., 2012; Van der Straeten et al., 2013). We developed a website that included a description of the three electoral systems under study, with short descriptions of the advantages and disadvantages of each set of institutions. It also contained a ‘practical’ part where visitors were invited to vote in simulated elections according to those rules. The study was open to individuals from any country. We advertised the website in the national and local media. All visitors, regardless of their home country, were invited to vote in an election for pan-European lists, and in some countries, we also offered the possibility of voting for local (national) lists.

The simulated national elections: Simulated national elections were organized in eight countries (the first set of countries in Table A1). In these countries, lists of local candidates (i.e. candidates who were running in the visitor’s district for the 2014 European election) were available. Except in Spain,⁴ we showed voters the whole list of candidates in their district. Each participant voted under the three different voting procedures in the following order: closed list, open list, and panachage. In these countries, the site was available in the national language (and also in Catalan in Spain).

The simulated national elections had a high degree of external validity because they were constructed to mimic the actual choices facing voters in the EP election.

The simulated pan-European elections: Visitors from all European countries (including those who also voted for national elections) were invited to take part in pan-European elections. The challenge was to set up elections that would mimic, from the point of view of the voter, realistic elections of European delegates through pan-European party lists. We also wanted the study design to be as neutral as possible.

We decided that each participant would see seven lists of 10 candidates corresponding to the seven political groups that were actually registered at the EP at the time of the election. We randomly selected candidates for each list from among the MEPs registered in the corresponding group. The candidates in the simulated pan-European elections were thus real politicians. Randomization was done independently among participants to guarantee the statistical robustness of the analysis, meaning that any two participants saw different candidates. Each ballot provided the official photo of the candidate, his or her name, nationality, and of course, group affiliation. Further information was made available by having the name of each candidate linked to his or her official web page on the EP’s website.

Each participant voted under the three different voting procedures, in the following order: closed list, open list, and panachage. The set of candidates remained the same, for a given participant, under the three voting rules.

The site was available in the national language for visitors who voted in both national and pan-European elections. For countries where only the pan-European elections were available, two sets of countries have to be distinguished. In 11 countries, listed in the web appendix, the site was also available in the national language. In the remaining nine countries, no translation in the national language was available, so participants had to choose one of the 14 available languages.

After completing the ‘practical’ part of the website, that is, voting three or six times, participants were asked to complete a short questionnaire with standard demographic questions, questions about politics in general, and questions eliciting their opinions about the three voting rules that they just experienced as well as about the pan-European constituency.

Participation: The website opened on 4 May, 2014 and received around 22,000 visits before May 26. Many visitors declined to vote when they were offered the opportunity to participate in the election(s), and are thus not recorded as ‘participants’. We recorded the time spent by the participants consulting the information, filling out the ballots, and answering the questionnaire. The median time for the whole exercise was 642 s.

The online appendix indicates the number of participants in each country among the 3673 visitors who agreed to participate. The level of participation varied considerably across countries and is most likely related to our success (or lack thereof) in publicizing the website. For instance, press coverage was high in Italy and Germany, but nonexistent in Poland, Greece, and Bulgaria. Gathering participants in a free and open study such as ours can be challenging, but overall, taking into account the size of the targeted populations, participation was close to what was observed in a similar study in Ontario, Canada, in 2011 (Blais et al., 2012).

Another challenge with a free and open study is that the participants are often not representative of the general population. Not surprisingly, our sample is not representative of the whole population of Europe. For instance, we observe in our data that under the closed list system, 22% of the participants voted for the Social Democrat pan-European list and 10% for the Christian Democrats. In the real election, the Christian Democrat lists gathered more votes, throughout Europe, than the Social Democrats. It seems that Christian Democrat supporters were less attracted to the survey than Social Democrats. This lack of representativeness is not specific to our study.⁵ One solution is to weight the data using factual information such as the actual electoral results (we do this in our analysis of the Swedish data below). This corrects the overrepresentation of the support for some parties (for examples of similar weighting in similar experiments, see Blais et al., 2012; Van der Straeten et al., 2013).

Descriptive statistics: Some examples of research questions

This section provides some descriptive statistics, intended to illustrate the kind of questions one can address with the EuroVotePlus data. We discuss three potential research questions here: (a) the effect of the voting rule on who is elected, in a specific country (Sweden); (b) the effect of candidates’ gender in the pan-European election; and (c) the effect of candidates’ country of origin also in the pan-European election.

Effects of the voting rule on the results of the election

To give an example of how the EuroVotePlus data set can be used to compare electoral institutions, we use the national elections data in Sweden. In Sweden, 11 parties were competing for 20 seats in a single district; nine parties obtained seats. In all cases but one, the elected candidates of a party were the first-ranked candidates on their party’s list, despite the fact that Swedish electoral rules (which include preferential voting) can lead to changes in the order of party lists. The exception is the centrist party, where the single-elected candidate was third on the list.

In the data set, there are 499 Swedish respondents who completed the three national election ballots. Table 1 provides the results. Note that not all lists have the same number of candidates (column 2). Parties are ranked according to their official national score; column 3 provides the vote shares and column 4 the numbers of seats (20 MEPs were to be elected). Columns 5, 6, and 7 indicate the number of votes obtained by the parties in the simulated elections under open list, closed list, and panachage rules, respectively (under panachage, voters were allowed to cast up to 20 votes; on average, they cast 14.2 votes). Notice that the figures under open list and closed list rules are almost identical.

Table 1.

National elections results in Sweden.

Party	Official			Observed No. of votes			Weight	Vote share, weighted		No. of seats, weighted
Party	No. of candidates	Vote share	Seats	Open	Closed	Panachage		Closed	Pana chage	Closed	Panachage
Socialdemo.	30	24.19	5	75	75	836	1.609	23.89	17.27	5	4
Miljöpartiet	40	15.41	4	60	61	949	1.281	15.57	15.8	4	3
Moderaterna	30	13.65	3	51	50	733	1.336	13.26	12.31	3	3
Folkpartiet	42	9.91	2	71	68	942	0.697	9.72	11.62	2	2
Sverigedem.	20	9.67	2	72	72	912	0.670	10.21	9.45	2	2
Centerpart.	44	6.49	1	18	17	334	1.798	5.60	6.29	1	1
Vänsterpart.	39	6.30	1	62	61	955	0.507	6.42	9.15	1	2
Kristdemokr.	43	5.93	1	19	20	336	1.559	5.90	6.46	1	1
Feministiskt	10	5.49	1	31	33	424	0.883	5.77	5.81	1	1
Piratpartiet	20	2.23	0	35	36	564	0.317	2.66	4.69	0	1
Junilistan	9	0.73	0	5	6	93	0.729	1,00	1,15	0	0
	327	100	20	499	499	7078		100	100	20	20

Note: Votes and seats obtained by each party according to official results (columns 3 and 4) and in the EuroVotePlus experiment under alternative electoral rules (columns 9 to 12).

The sample of respondents is not representative (in terms of votes) of Sweden as a whole. For example, in our sample, the Miljöpartiet (Greens) and the Vänsterpartiet (Left Party) receive (almost) the same numbers of votes, whereas in the official election, the former received 15% of the votes and the latter 6%. In order to correct for these biases, we weight our observations so that we correctly predict the official results. Since Sweden uses a variant of the open list system, we use the open list results to weight the observations (using the closed list results would make almost no difference). Column 8 shows the weights, computed at the party level.

The weighted scores under the open list system are equal to the real scores (column 3) and are therefore not shown here. They can be compared with the scores of the parties provided in column 9 for the closed list system and column 10 for the panachage system. Party scores under closed and open lists are similar. There are more differences when one switches to panachage. In particular, the Socialdemokraterna (the Social Democrats) lose votes (17% instead of 24%), and the Pirapartiet (Pirate Party) more than doubles its support (4.7% instead of 2.2%).⁶

To compute the number of seats shown in the last two columns, we use the vote shares from columns 9 and 10.⁷ The number of seats obtained under closed list (column 11) and open list (column 4) rules is the same, but there are differences when one switches to panachage (column 12). Here the main two parties, Socialdemokraterna and the Miljöpartiet, each lose one seat, the Vänsterpartiet obtains one more seat, and the Pirapartiet obtains a seat. The panachage rule thus seems to be less favourable to the main parties and more favorable to small parties.

While electoral institutions make some clear but small differences in terms of seats, they have a larger impact on which candidates are elected. We computed the composition of the Swedish delegation that would have been elected with the various rules and compared it to the official results (remember that with the official rule, elected candidates are, with the exception of the Centerpartiet, the top-ranked candidates on the party lists). Out of 20 delegates, with the Luxembourg rule (which allows for panachage and cumulative voting) three are replaced, a 15% change in the composition of the delegation. With the Latvian open rule (which allows for negative voting), 5 candidates out of 20 candidates are replaced, a 25% change. These results suggest that giving more say to voters may substantially alter which candidates get elected.

We now turn to the pan-European elections. We restrict attention to the 2052 participants who cast ballots under the three different electoral rules. To provide examples of the kind of analysis that can be done with such data, we concentrate on two topics: (a) the effects of the openness of the list on the choice of voters to vote for male versus female candidates and (b) the question of national and transnational votes.⁸

Preferential voting and panachage: Gender effects

Existing cross-national work provides mixed conclusions about whether open or closed electoral rules are more likely to result in more women being elected (Fortin-Rittberger and Rittberger, 2014: 502). This might be because in these studies electoral rules vary along with other country characteristics that affect voters’ propensity to vote for women. We are able to avoid this problem because each respondent in our data set voted under several electoral arrangements. Table 2 depicts how the electoral success of female candidates varies with the voting rule. Under the closed list system, 38.2% of votes (indirectly) go to female candidates; given that voters vote for lists, this is the percentage of female candidates on the lists when lists are weighted on the basis of the votes they obtained (in this section, we do not weight observations to correct for potential biases in our sample). Under the open list and panachage systems, the figures, 40.2% and 43.1%, respectively, are simply the percentages of votes received by female candidates, taking into account the possibilities of negative and double votes. These three figures can be compared with the overall proportion of female candidates on the ballots proposed to the participants, which was 32%. One can see that more flexible procedures clearly favor female candidates.

Table 2.

Percentage of votes given to female candidates in the pan-European election under alternative voting rules.

	Total	Gender
	Total	Men	Women
Closed list	38.2%	38.6%	38.5%
Open list	40.2%	39.7%	43.4%
Panachage	43.1%	41.5%	50.7%
N	2052	1331	486

Making the same computations but distinguishing voters’ gender, one can see that the effect holds true for male and female voters, but is considerably more important for female voters (columns 3 and 4 in Table 2; note the gender bias in participation among the 1817 participants who answered the gender question in the questionnaire). Under the closed list system, we observe no difference between male and female voters: on average, they vote for lists with similar female representation. However, differences appear when we consider more open electoral systems: when women can cast individual votes, as allowed under our panachage rule, the percentage of votes received by female candidates shifts to 50.7% (41.5% for men). These differences call for further explanation and more sophisticated analyses.

National and transnational votes in the pan-European election

Finally, we consider what our study can reveal about national and transnational votes. In Table 3, a row corresponds to the country of the participant, and a column corresponds to the nationality of a candidate. We only report the 10 countries with at least 25 participants (the number of participants is indicated in the column labeled N). The figures are based on the 2052 participants who cast votes under the panachage system. Two examples will show how the figures were computed.

Table 3.

Vote transfers across countries in the pan-European election under the panachage rule.

	Germany	France	UK	Italy	Spain	Poland	Romania	Belgium	Portugal	Hungary	Sweden	Finland	N	Votes cast
Germany	2.63	1.29	0.49	0.36	1.06	0.23	0.70	0.83	1.00	0.51	1.85	0.95	318	2931
France	1.38	2.50	0.45	0.37	1.07	0.24	0.76	1.15	1.57	0.34	1.53	0.94	428	3957
UK	1.27	0.98	1.33	0.85	1.00	0.35	1.20	1.64	0.58	0.29	0.83	1.29	58	523
Italy	1.09	0.85	0.59	2.59	1.22	0.32	0.84	1.02	1.66	0.54	1.28	1.29	427	3041
Spain	1.65	1.43	0.51	0.41	2.40	0.23	0.40	0.95	2.00	0.00	1.85	1.19	87	742
Poland	0.81	0.72	0.74	0.97	0.56	3.05	1.40	0.91	1.12	1.06	1.55	1.48	25	210
Romania	1.25	0.90	0.66	0.72	1.04	0.95	4.25	1.17	1.17	0.34	1.08	2.39	58	433
Belgium	1.35	1.72	0.41	0.34	1.04	0.37	0.17	3.05	1.34	0.59	2.01	0.41	66	251
Portugal	0.79	0.74	0.61	0.72	1.16	0.48	0.59	0.33	8.25	0.78	1.26	0.72	34	286
Hungary	1.49	1.03	0.50	0.45	0.78	0.68	1.26	0.96	0.82	3.95	2.66	1.90	80	698
Sweden	1.40	1.08	1.07	0.40	0.70	0.30	0.41	0.89	1.28	0.40	3.99	1.80	382	2986
Finland	1.32	1.08	0.88	0.43	0.76	0.55	0.33	0.38	0.94	0.00	3.74	3.71	30	251
Votes received	3241	2408	1713	1175	758	398	335	551	738	130	959	430

Note: Table reports the observed relative propensity of a participant from a Member State (row) to vote for a candidate from a Member State (column). Diagonal entries correspond to voters voting for candidates of their own Member State.

Under panachage, we observe that German participants gave a total of 2931 votes (column 15), of which 376 (about 12.8%) were given to French candidates. In the proposed ballots, the proportion of French candidates is 14,344/143,640; that is, about 10%. The ratio of the former percentage to the latter is 1.29, and is shown in the cell (Germany, France). The ratio would be close to 1 if German voters voted randomly, ignoring the nationality of the candidates; that the ratio is larger than 1 means that German voters were attracted to French candidates. Likewise, we observe that 972 of 2931 German votes (about 32.6%) were given to German candidates. In the ballots proposed to the voters, 17,752 candidates out of 143,640 were German (about 12.4%). The ratio is thus 2.63.

The figures in Table 3 reveal the relative propensity of voters from a country to vote preferentially for the candidates of the various countries, taking into account the size of the pools of candidates from the different countries. One can see immediately that voters tend to vote for candidates of their own country (the diagonal of the matrix, where all the numbers are larger than 1), and that this effect seems larger for small countries.⁹ The information in the table suggests an intriguing structure of ‘political trust’ among countries. In particular, some countries’ candidates seem to attract the votes of most European citizens (Sweden, Germany, Portugal, Finland), whereas other countries’ candidates systematically receive fewer votes (Poland, Hungary, Italy, Romania). Further work is required to understand to what extent this pattern is in accordance with independent findings about reciprocal trust among Europeans (Bornhorst et al., 2010).

Strengths and limitations of the EuroVotePlus data set

The EuroVotePlus data set is a rich resource for those interested in understanding individual-level voting behavior in the context of EP elections. It will also be of interest to scholars interested in analyzing the effects of electoral systems on voter choice. Having data on votes in different contexts and under different electoral rules allows researchers to evaluate how these rules affect vote choices. The external validity of the study contributes to the value of the data set. Individuals were not casting ballots in simulated elections with artificial candidates—they were voting for real people in the context of a real election campaign. The results we have presented above give an idea of the type of research that can be conducted with this data set.

Nonetheless, it is important to point out the limitations of the data. Most obviously, as noted above, the representativeness of the sample is weak. Weighting is the most obvious solution (an example of such weighting is provided in the case of the Swedish national elections), but careful analyses need to be performed to ascertain how this bias may affect conclusions about the impact of voting rules. In particular, using the question on preferences about European integration to weight the observations may also be warranted to correct for a potential pro-European bias in our sample.

The fact that all participants used the three voting procedures is a strength of the protocol (to trace the effects of the voting rule at the individual level), but at the same time a drawback. Indeed, participants may have wanted to avoid inconsistencies across their answers, thus biasing their behavior. The natural remedy for this problem would have been to randomize the order in which the voting rules were presented (and study whether we find an order effect). We chose not to do so to make our website as user-friendly as possible. For this reason, we followed an internal logic of increasing complexity of the voting act: participants first voted with closed list PR, second with open list PR, and then panachage. Note that any overconsistency effect should make our participants less prone to vote for different parties across voting systems, which could lead to an underestimation of the impact of voting rules on electoral behavior. In other words, our estimates are likely to be conservative.

Despite these drawbacks, we believe that this new data set can significantly contribute to our understanding of how the rules of the game affect voters’ behavior in EP elections.

Footnotes

Authors’ note

This experiment was conducted as part of the Making Electoral Democracy Work project (). Replication data can be found there and at the homepage of the journal.

Notes

References

Ames

(1995) Electoral strategies under open-list proportional representation. American Journal of Political Science 39(2): 406–433.

Blais

Héroux-Legault

Stephenson

(2012) Assessing the psychological and mechanical impact of electoral rules: A quasi-experiment. Electoral Studies 31(4): 829–837.

Blumenau J, Eggers AC, Hangartner D, et al. (2014) Open/closed list and party choice: Experimental evidence from the UK. Working paper, London School of Economics, London. Available at: http://personal.lse.ac.uk/hix/Working_Papers/open_closed_list_experiment_v11.pdf (accessed 10 March 2015).

Bol

Labbé St-Vincent

Lavoie

(forthcoming) Recruiting for laboratory voting experiments: Exploring the (potential) sampling bias. In: Blais

Laslier

Van der Straeten

(eds) Voting Experiments, New York: Springer.

Bornhorst

Ichino

Kirchkamp

(2010) Similarities and differences when building trust: The role of cultures. Experimental Economics 13(3): 260–283.

Bright J, Garzia D, Lacey J, et al. (2015) Europe's voting space and the problem of secondorder elections: A transnational proposal. European Union Politics . Epub ahead of print 22 June 2015. DOI: 10.1177/1465116515588965

Chang

(2005) Electoral incentives for political corruption under open-list proportional representation. The Journal of Politics 67(3): 716–730.

Clark

Golder

(2006) Rehabilitating Duverger’s theory: Testing the mechanical and strategic modifying effects of electoral laws. Comparative Political Studies 39(6): 679–708.

Colomer

(2011) Personal Representation: The Neglected Dimension of Electoral Systems, Colchester: ECPR Press.

10.

Cox

(1997) Making Votes Count. Strategic Coordination in the World’s Electoral Systems, Cambridge: Cambridge University Press.

11.

Duverger

(1951) Les partis politiques, Paris: Armand Colin.

12.

European Parliament (2012) Second Report on a proposal for a modification of the Act concerning the election of the members of the European Parliament by direct universal suffrage of 20 September 1976 (2009/2134(INI)). Committee on Constitutional affairs. Available at: http://www.europarl.europa.eu/sides/getDoc.do?pubRef=-//EP//TEXT+REPORT+A7-2012-0027+0+DOC+XML+V0//EN (accessed 6 April 2015).

13.

Fishkin

Luskin

Siu

(2014) EuroPolis and the European Public Sphere: Empirical explorations of a counterfactual idea. European Union Politics 15(3): 328–351.

14.

Fortin-Rittberger

Rittberger

(2014) Do electoral rules matter? Explaining national differences in women’s representation in the European Parliament. European Union Politics 15(4): 496–520.

15.

Fournier

van der Klock

Blais

(2011) When Citizens Decide: Lessons from Citizen Assemblies on Electoral Reform, Oxford: Oxford University Press.

16.

Habermas

(2012) The Crisis of the European Union: A Response, Cambridge: Polity.

17.

Hix

(2004) Electoral institutions and legislative behaviour. Explaining voting defection in the European Parliament. World Politics 56(2): 194–223.

18.

Huber

(2012) Measuring ethnic voting: Do proportional electoral laws politicize ethnicity? American Journal of Political Science 56(4): 986–1001.

19.

Isernia

Fishkin

(2014) The EuroPolis deliberative poll. European Union Politics 15(3): 311–327.

20.

Karvonen

(2004) Preferential voting: Incidence and effects. International Political Science Review 25(2): 203–226.

21.

Lacey

(2014) Must Europe be Swiss? On the idea of a voting space and the possibility of a multilingual demo. British Journal of Political Sciences 44(1): 61–82.

22.

Laslier

J-F

Van der Straeten

(2008) Approval voting in the French 2002 presidential election: A live experiment. Experimental Economics 11(1): 97–105.

23.

Pukelsheim

(2014) Proportional Representation. Apportionment Methods and Their Applications, Heidelberg: Springer.

24.

Shugart

Valdini

Suominen

(2005) Looking for locals: Voter information demands and personal vote-earning attributes of legislators under proportional representation. American Journal of Political Science 49(2): 437–449.

25.

Singer

Stephenson

(2009) The political context and Duverger’s theory: Evidence at the district level. Electoral Studies 28(3): 480–491.

26.

Van der Straeten

Laslier

J-F

Blais

(2013) Vote Au Pluriel: How people vote when offered to vote under different rules? PS: Political Science and Politics 46(2): 324–328.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.25 MB