Parliamentary Election Turnout in Europe since 1990

POLITICAL STUDIES: 2002 VOL 50, 916 927 Parliamentary Election Turnout in Europe since 1990 Alan Siaroff University of Lethbridge John W. A. Merer Vanderbilt University This article examines the cross-national variations in turnout for parliamentary elections in Europe since 1990 a continent with a vast range in turnout levels and some clear subregional patterns, especially that of low turnout in East-Central Europe. A full range of socio-economic, mobilizational, party system, institutional, and contextual factors are examined for bivariate relationships with turnout. A multivariate model then indicates that cross-national turnout is higher where there is strictly enforced compulsory voting, in polarized two-party systems and countries with a high level of party membership, and where there are no relevant elected presidents or strong regional governments. Variances on these and other key factors are what accounts for the subregional pattern of East-Central Europe and the highest turnout case of Malta; however, Switzerland is confirmed to be a significant national dummy variable. Comparative levels of voter turnout have been analysed repeatedly in recent decades. 1 Although it was not their central innovation, in their recent analysis Blais and Dobrzynska (1998, p. 243) also included dummy variables for each of the continents, with Europe as the reference category. They found significantly lower turnout in North America and South America, and significantly higher turnout in Oceania, regardless of the model used (Blais and Dobrzynska, 1998, p. 247). Following from this, there is no reason why one cannot assess regional variations in turnout within a continent, provided it has enough cases. In particular, Europe seems the most logical continent for such an analysis. The number of European democracies, or at least electoral democracies, has basically doubled with the end of communism. Nevertheless, in East-Central Europe, whatever eagerness to vote there was after democratization has apparently dissipated. In Poland, perhaps the most extreme case, voter turnout was only 46.2 percent in its 2001 parliamentary elections, although this was even less, 43.2 percent, back in its 1991 elections. This article thus seeks to assess voter turnout in parliamentary elections in Europe since 1990, the year of the first free elections in East- Central Europe. The analysis will assess turnout in every system deemed an ongoing electoral democracy by Freedom House, thus looking at 38 different cases. 2 Limiting the analysis to the period since 1990 also has the advantage of not confounding the analysis with the phenomenon of the long-term decline in turnout found in most democracies (Lijphart, 1997, p. 5). Political Studies Association, 2002. Published by Blackwell Publishing, 108 Cowley Road, Oxford OX4 1JF, UK and 350 Main Street, Malden, MA 02148, USA

ELECTION TURNOUT IN EUROPE 917 Table 1 gives the data on turnout in European parliamentary elections since 1990 (through the end of 2001). 3 Malta, which is at the top, has in fact been the subject of individual case analysis (Hirczy, 1995). Switzerland, perhaps not surprisingly, remains at the bottom of the European rankings. Overall, the average turnout for all 118 elections analysed is 75.6 percent. The average turnout for the 51 elections held in post-communist Eastern and Central Europe (see Table 1) is indeed below average, at 71.0 percent. On the other hand, we can identify a region with aboveaverage turnout, that of Nordic Europe. For the 16 elections in the five Nordic countries (noted in Table 1), mean turnout is 80.8 percent. Clearly, then, there are some regional differences within Europe. Table 2 shows the results of bivariate regressions on turnout, starting with these regional dummies. As can be seen, the lower turnout found in East-Central European elections is statistically significant. However, this is not the case for the higher turnout in Nordic elections. The subsequent question, though, is whether the East-Central region is unique within Europe, or whether it simply reflects differences in various causal factors. In other words, is this dummy regional variable still significant after adding in other factors? In addition to these regional variables, one can likewise assign a dummy variable to the most extreme cases of Malta and Switzerland. These bivariate regressions are also given in Table 2. Not surprisingly, these dummy country variables prove significant, indeed highly so for Switzerland. Likewise, the issue here is whether these extreme cases reflect anything more than (extreme) differences in various causal factors. Causal Factors Based on the general literature on turnout, we thus test a range of variables in seven main areas, initially by running bivariate regressions. 4 These data are also given in Table 2. In rare cases lack of certain country data lessens the total N slightly. The first explanatory area is socio-economic. In this regard, factors such as level of development, small size, and high population density have been found, at least in some studies, to increase significantly turnout (Powell, 1982; Blais and Carty, 1990; Blais and Dobrzynska, 1998). Consequently we have included five variables here. First of all, there is total population, in millions for the late 1990s. Second, there is area, measured in 1,000km 2. Third, there is population density, which combines these two factors. We hypothesize that turnout will be higher in smaller countries, since citizens will feel more powerful as individuals versus the whole (see Dahl and Tufte, 1973, p. 43). Fourth, there is urbanization, which is somewhat of an alternate measure of population density. We hypothesize that turnout will be higher in more densely populated settings, as individuals will not only be but also feel closer to their neighbours as fellow citizens, and thus be more likely to take an interest in collective matters. Fifth and last, there is GNP per capita at purchasing power parities (PPP) for 1998. The use of purchasing power parities allows us to factor out exchange rates; however, Liechtenstein cannot be included here for lack of data. We hypothesize that in more developed countries people are more informed and have more resources (including time) to devote to politics, thus turnout will be higher.

918 ALAN SIAROFF AND JOHN W. A. MERER Table 1: Turnout in Europe, 1990 2001 Number of Range of Country elections turnout (%) Mean turnout (%) Malta 3 95.4 96.2 95.9 Cyprus 3 90.5 94.3 92.6 Belgium 3 90.7 92.3 91.4 Luxembourg 2 85.8 88.3 87.1 Liechtenstein 4 85.3 87.5 86.6 Iceland 3 84.1 87.6 86.2 Turkey 3 83.9 86.9 85.3 Slovakia 4 75.7 95.4 84.9 Sweden 3 81.4 86.7 84.7 Denmark 4 81.4 87.0 84.6 Italy 4 81.4 87.3 84.4 Czech Republic 4 74.0 96.7 83.0 Austria 4 78.1 86.1 81.8 Germany 3 77.8 82.2 79.7 Bulgaria 5 66.6 90.6 78.1 Latvia 3 71.9 89.9 77.9 Greece 4 76.3 78.3 77.5 Norway 3 75.0 78.0 76.3 Netherlands 2 73.2 78.3 75.8 Macedonia 2 72.9 77.3 75.1 Spain 3 70.0 78.0 74.8 Croatia 3 68.8 78.0 74.1 Moldova 3 69.7 79.3 73.8 Slovenia 3 70.3 76.0 73.3 Ukraine 2 70.8 74.8 72.8 Romania 3 65.3 76.3 72.5 Finland 3 68.3 72.1 70.8 United Kingdom 3 59.3 77.7 69.5 Georgia 3 64.0 74.3 68.7 France 2 67.9 68.9 68.4 Ireland 2 65.9 68.5 67.2 Portugal 3 61.8 68.2 65.6 Estonia 3 54.9 69.6 64.1 Hungary 3 56.3 68.9 63.4 Lithuania 3 52.9 75.2 62.2 Russia 3 54.8 64.4 60.3 Poland 4 43.2 52.1 47.4 Switzerland 3 42.3 46.0 43.9

Table 2: Bivariate Regressions on Parliamentary Turnout in Europe Variable N Standardized B t-score Significance Adjusted R2 Root MSE East-Central Europe dummy 118-0.335-3.834 *** 0.105 11.565 Nordic dummy 118 0.167 1.829 0.020 12.103 Malta dummy 118 0.268 3.001 ** 0.064 11.826 Switzerland dummy 118-0.422-5.007 *** 0.171 11.132 population 118-0.208-2.288 * 0.035 12.008 area 118-0.208-2.288 * 0.035 12.008 population density 118 0.295 3.327 *** 0.079 11.729 urbanization 118 0.168 1.830 0.020 12.103 per capita GNP (at PPPs) 114 0.276 3.039 ** 0.068 11.836 party membership 118 0.489 6.036 *** 0.232 10.709 unionization 118 0.287 3.224 ** 0.074 11.761 electoral fragmentation 117-0.341-3.893 *** 0.109 11.589 parliamentary fragmentation 118-0.170-1.855 0.020 12.098 parties with 3% of seats 118-0.100-1.087 0.002 12.214 two-party vote concentration 117 0.319 3.606 *** 0.094 11.686 two-party seat concentration 118 0.192 2.112 * 0.029 12.047 vote lead first over second party 116-0.032-0.346-0.008 12.375 seat ratio first to second party 118-0.153-1.672 0.015 12.131 decisive election dummy 118 0.165 1.805 0.019 12.107 polarized party system dummy 118 0.235 2.610 ** 0.047 11.931 proportional representation 118 0.276 3.099 ** 0.068 11.798 strictly enforced compulsory voting dummy 118 0.335 3.827 *** 0.105 11.567 weekend voting dummy 118 0.003 0.035-0.009 12.276 Sunday voting dummy 118-0.249-2.770 ** 0.054 11.889 bicameralism dummy 118-0.252-2.801 ** 0.055 11.881 relevant elected president dummy 118-0.382-4.453 *** 0.139 11.345 strong regional governments dummy 118-0.303-3.425 *** 0.084 11.699 European Union dummy 118 0.125 1.359 0.007 12.180 * = significant at the 0.05 level, ** = significant at the 0.01 level, *** = significant at the 0.001 level. ELECTION TURNOUT IN EUROPE 919

920 ALAN SIAROFF AND JOHN W. A. MERER As the data in Table 2 show, turnout does in fact increase with smaller population, smaller area, greater population density, higher urbanization and higher GDP per capita. Except for urbanization, all of these are significant results, although density is the most significant one not surprisingly, since it combines population and area. The next area involves aspects of political and social mobilization, which has been noted to increase electoral participation (Crewe, 1981, pp. 251 3). Furthermore, Hirczy (1995, p. 268) concludes that strong and pervasive partisanship in the population at large is part of the explanation for the high turnout in Malta. Measuring partisanship cross-nationally is highly problematic; as such, we consider the best measure to be party membership as a share of the electorate. However, there are problems with accurate (not to mention up-to-date) information here, especially for Eastern Europe. In contrast, Mair and van Biezen (2001) give very precise data for Western European countries (and some Central European ones). Given the data limitations, we thus use values in multiples of five, that is, 0, 5, 10,... percent. Also in this regard we test the level of unionization as a percentage of the labour force. Given similar problems in cross-national data here, we have made estimates for certain cases, and rounded the available values to the nearest ten percent, that is, 0, 10, 20,... percent. We hypothesize that turnout will be higher both where political parties and where unions have more members to mobilize. Table 2 shows the bivariate regressions for mobilization factors. Turnout does increase with higher party membership, and this relationship is highly significant. Indeed, it alone explains over 20 percent of the total variance. Turnout also increases with higher unionization, although not quite in as significant a way. The next three areas involve separate aspects of the party system. The first of these is party system fragmentation. As Blais and Dobrzynska (1998, p. 248) hypothesize, the greater the number of parties, the more choice electors are offered, and the higher the turnout. We thus test a range of measures of party system fragmentation: electoral fragmentation, parliamentary fragmentation, the number of parties winning at least 3 percent of the seats (Ware, 1996), two-party vote concentration, and two-party seat concentration. The latter two measures obviously aim to assess whether there are two, and only two, serious rivals in an election. Table 2 shows the bivariate regressions for measures of party system fragmentation. Turnout does increase with less fragmentation (or higher concentration). The strongest measures of party system fragmentation are clearly electoral fragmentation (a negative relationship) and two-party vote concentration (a positive relationship). In other words, as a single variable, electoral fragmentation/concentration matters more than parliamentary fragmentation/concentration for turnout. That said, there is obviously a trade-off between higher electoral fragmentation and higher two-party vote concentration. It seems that for higher turnout one needs to have reasonable fragmentation (in other words, choice) but still two dominant parties as in Germany. Related to aspects of party system fragmentation are aspects of party system competitiveness. Here the issue, or debate, is whether or not turnout is higher in close elections. Closeness can be measured in more than one way though. Blais and Dobrzynska (1998, p. 249) use the gap in votes between the first place and second

ELECTION TURNOUT IN EUROPE 921 place parties (in terms of votes). We use this too, but we also measure the ratio of seats won by the first party (in terms of seats) to the second party (in terms of seats). Finally, like Blais and Dobrzynska (1998, p. 249), we also use a dummy variable for a single party majority government, which we call electoral decisiveness. We hypothesize that greater gaps between the top two parties, in other words less competition, will lead to lower turnout as the results seem somewhat of a foregone conclusion. Likewise, we hypothesize that turnout will be lower where one party wins an outright majority, since the result is predictable. Table 2 shows the bivariate regressions for measures of party system competitiveness. None of these variables are actually significant just by themselves. It is interesting to note, though, that whereas both the vote lead of the first over the second party and the seat ratio of the first to the second party have a negative coefficient that is, the greater the ratio (and thus the lead of the first party) the lower the turnout electoral decisiveness itself (in the sense of a single party majority) has a positive coefficient. Finally, the third party system area is that of party system polarization, which has been found by Crepaz (1990) to significantly increase turnout. However, he measures polarization by the distance between the most left-wing and the most rightwing parties, regardless of size. We have limited our analysis to parties with at least 10 percent of the vote, our cut-off for relevance here. Given the lack of scholarly left-right placements for some Central and Eastern European countries, we have used a dummy variable. This is calculated as follows: for parties with at least 10 percent of the vote (as noted), we measure the distance between the two most extreme parties on the 10-point left-right scale given in Huber and Inglehart (1995), supplemented by our estimates. Where this gap is more than 3.75, we consider the party system to be polarized. 5 We hypothesize that in such a case turnout will be higher since more is at stake policy-wise. Table 2 shows the bivariate regression for party system polarization. This is significant in itself. The next area involves institutional aspects of the vote itself. The first aspect here is whether the electoral system is one of proportional representation, the central question of Blais and Carty (1990). Here we code countries 0 for a single member system, 2 for a fully compensatory proportional system, and 1 for a mixed system combining single member with proportional seats but without full compensation (such as in Italy from 1994, Lithuania, or Russia). Following Blais and Carty (1990, p. 167), we hypothesize that turnout will be higher under proportional systems since voters will not feel that their vote is wasted, more parties will likely run, and parties will have an incentive to campaign everywhere. The next variable is compulsory voting, perhaps the central solution to increasing turnout (Lijphart, 1997, pp. 8 9). Here we follow the analysis of Maria Gratschew of IDEA (2001) and distinguish between the strict enforcement of compulsory voting found in Belgium, Cyprus and Luxembourg, and the weak enforcement of compulsory voting found in Greece, Italy (if even that), Liechtenstein and Turkey. We hypothesize that it is only strict enforcement of compulsory voting that will make a significant difference, due to the monetary fines levied. 6 The other variable in this area is weekend voting, which was found by Franklin (1996, pp. 226 30) to increase turnout by 5 6 percent. We test also for Sunday (only) voting. In both of the latter cases, we

922 ALAN SIAROFF AND JOHN W. A. MERER hypothesize that turnout will be higher when voting is held on a day when (most) people do not work, since they will have less demands on their time. Table 2 also shows the bivariate regressions for institutional factors relating to the vote. As is found in other analyses, proportional representation is a significant positive factor in turnout. As for strictly enforced compulsory voting, not surprisingly this is even more significant. Weekend voting is not significant but Sunday voting is so, although here the coefficient is negative, contrary to the argument. The seventh, and last, area of causal factors involves the importance of the national parliament for a country s political system. As scholars such as Blais and Dobrzynska (1998, p. 245) argue, other elected bodies could weaken the power of the national parliament. The absence of such bodies in Malta has been noted by Hirczy (1995, p. 260). Other elections also affect parties ability and will to mobilize. As Cox (1999, p. 408) notes, more elections in which important offices are separately elected means that the value of getting out the vote in any one election is diluted. Full control of the polity will in some cases not even be attainable and, again, parties will have to choose how to allocate their finite mobilizational resources across elections. We tested various ways of assessing other bodies, including multiple categories of importance. However, for most of the consequent comparison categories the N proved to be too small to produce meaningful results. Consequently, we construct four different dummy variables here. The first dummy variable is for those systems that are bicameral as opposed to unicameral. Here we hypothesize that the potential divided control of bicameralism will depress turnout for the lower house. The second dummy variable is for those cases where there is a relevant elected president, that is, those presidencies listed as either very strong or moderately strong in Siaroff (2000, pp. 10 11), which are Bulgaria, Croatia, Finland, France, Macedonia, Moldova (through 2000), Poland, Portugal, Romania, Russia and Ukraine. 7 The third dummy variable is where there are strong elected regional legislatures and governments, that is, both elected and strong in terms of controlling at least 20 percent of total government spending. Thus defined, strong elected regional legislatures and governments (in short form, strong regional governments) exist only in Germany, Russia, Spain and Switzerland. We hypothesize that both relevant elected presidents and strong regional governments will divide power sufficiently to make the national parliament less important, and thus turnout for it lower. Indeed, in an extreme case of a very strong presidency, the parliamentary election may well be seen as second-order to that of the president. The final dummy variable is more straightforward: European Union membership or not. We hypothesize that membership of the European Union, to which member states have transferred sovereignty and various powers, would lower national turnout since the national government would not be as relevant in as many areas as in non-eu countries. Table 2 shows the bivariate regressions for other important elected actors, all of which except European Union membership serve to depress turnout. The rela-

ELECTION TURNOUT IN EUROPE 923 tionship here is highly significant for both a relevant elected president and strong (elected) regional governments. The high significance of the latter is perhaps more striking since it is based on only four countries. Multivariate Analysis and Overall Conclusions We tested several models excluding any regional or national factors, and of these the best one appears to be Model One as given in Table 3. In this multivariate analysis, electoral decisiveness now has a negative coefficient, whereas the seat ratio of the first to second parties now has a positive coefficient. That said, according to this initial model, the most important factors affecting turnout (in terms of the significance levels) are strictly enforced compulsory voting and a relevant elected president, followed by party system polarization and strong regional governments. We would particularly stress the (negative) effect of a relevant elected president, since compulsory voting is a rather self-evident factor. In total this model explains (only) just over half of the total variance. Finally, we tested whether adding in specific national or European regional dummies improves the model. In terms of the extreme cases, Maltese elections are not significant as a dummy (Model Two in Table 3), and indeed have a negative coefficient. In contrast, however, Swiss elections (Model Three) are highly significant statistically although these cause the two-party vote concentration variable to lose some of its significance and, more relevantly, the strong regional governments dummy to lose all of its significance; a Swiss dummy also changes electoral decisiveness back to a weak positive effect. Moreover, in this Model Three, unionization, high party membership and especially the absence of a relevant elected president become more significant with the last factor now being the most significant factor overall. Turning to the regional dummy variables, East-Central Europe is no longer significant within a multiple regression analysis (Model Four). More expectedly, neither Nordic Europe (Model Five) nor the European Union (Model Six) is significant as an additional regional factor. Likewise, none of these regional dummies changes the total variance explained. Model Three strikingly increases the variance explained however, thus let us choose it to be our preferred final model. To return to our initial question, it appears that the regional variations within Europe occur as a result of the distribution of independent variables. For East-Central Europe, we would stress the common occurrence of relevant elected presidents (which in Model Three depress parliamentary turnout by 11 percent), the absence of any strictly enforced compulsory voting (which perhaps would be seen in the region as too dictatorial ), the rarity of a polarized party system, and the low party membership in the region. Party membership should certainly be emphasized since this is often ignored or downplayed as a political variable. For example, after pointing out that party membership levels are clearly lower in Eastern Europe compared with Western Europe, Lewis (2000, p. 99) goes on to state that [n]ot too much should be made of [party] membership ratios in themselves. We would disagree, since this factor does have broader political ramifications. Specifically, in Model

924 ALAN SIAROFF AND JOHN W. A. MERER Table 3: Multiple Regressions on Parliamentary Turnout in Europe Model One Model Two Model Three B error t B error t B error t Population 0.002 0.005 0.345 0.015 0.012 1.276 0.006 0.005 1.164 density Urbanization -0.004 0.068-0.060-0.002 0.067-0.029-0.099 0.062-1.605 Parliamentary 17.394 19.069 0.912 16.313 19.035 0.857 11.637 16.823 0.692 fragmentation Two-party vote 0.306 0.127 2.406* 0.298 0.127 2.345* 0.142 0.116 1.223 concentration Decisive -0.588 2.600-0.226 0.045 2.641 0.017 0.984 2.307 0.427 election dummy Seat ratio first 1.848 1.201 1.539 1.805 1.198 1.507 1.458 1.06 1.376 to second parties Strictly 15.227 3.498 4.353*** 13.359 3.791 3.524*** 14.776 3.081 4.795*** enforced compulsory voting dummy Strong regional -9.600 2.948-3.257** -9.327 2.948-3.164** -0.465 3.069-0.152 governments dummy Relevant -8.189 1.969-4.158*** -7.691 2.004-3.838*** -10.713 1.792-5.977*** elected president dummy Party 0.436 0.184 2.364* 0.503 0.192 2.628** 0.554 0.164 3.382*** membership Unionization 0.089 0.050 1.776 0.114 0.054 2.127* 0.089 0.044 2.026* Polarized party 2.516 0.742 3.393*** 2.638 0.746 3.537*** 2.238 0.655 3.418*** system dummy Geographic [none] [Malta] [Switzerland] dummy -18.193 14.464-1.258-30.321 5.434-5.580*** Constant 28.580 21.518 1.328 26.113 21.547 1.212 49.367 19.31 2.557* N 117 117 117 Adjusted 0.545 0.548 0.647 R 2 Note:* significant at the 0.05 level, ** significant at the 0.01 level, *** significant at the 0.001 level Three a 10 percent difference in total party membership as a share of the electorate leads to almost a 6 percent difference in turnout. For Malta, as the country with the highest turnout in Europe (and this without compulsory voting), we would stress its polarized party system, its extremely high level of party membership (also the highest in Europe), and the total concentration of power in its (unicameral) legislature, exclusive of any elected president or regional governments at all. We would argue further that Hirczy s key findings in

ELECTION TURNOUT IN EUROPE 925 Model Four Model Five Model Six B (error) t B (error) t B (error) t 0.002 0.005 0.349 0.001 0.006 0.139 0.002 0.005 0.418-0.005 0.069-0.078 0.005 0.070 0.070-0.029 0.072-0.402 17.343 19.167 0.905 18.934 19.398 0.976 18.077 19.079 0.948 0.304 0.130 2.333* 0.321 0.131 2.445* 0.283 0.129 2.191* -0.566 2.622-0.216-0.681 2.616-0.260-0.056 2.652-0.021 1.873 1.232 1.520 1.818 1.207 1.506 2.034 1.215 1.674 15.146 3.605 4.202*** 14.778 3.630 4.071*** 14.500 3.571 4.060*** -9.611 2.964-3.243** -9.876 3.012-3.278*** -9.333 2.959-3.154** -8.156 2.004-4.070*** -8.188 1.977-4.142*** -8.227 1.97-4.177*** 0.433 0.188 2.308* 0.435 0.185 2.351* 0.476 0.189 2.525* 0.089 0.051 1.762 0.104 0.059 1.757 0.100 0.051 1.959 2.504 0.753 3.325*** 2.579 0.755 3.414*** 2.515 0.741 3.391*** [East-Central Europe] [Nordic Europe] [European Union] -0.211 2.068-0.102-1.718 3.524-0.488 1.993 1.971 1.011 28.910 21.863 1.322 25.581 22.456 1.139 29.167 21.523 1.355 117 117 117 0.541 0.542 0.545 his (1995) case study of Malta can in fact be largely generalized across Europe. Moreover, though, it is the extent of these factors in Malta that matter, not anything unique about Malta itself. In contrast, Switzerland which is idiosyncratic in terms of its direct democracy remains not only a useful but indeed the only useful additional regional/national variable. Adding a Swiss dummy to the model means that almost two-thirds of the

926 ALAN SIAROFF AND JOHN W. A. MERER variance can be explained, even if this largely eliminates the importance of the strong regional governments variable. This level of 65 percent thus appears to be about the limit of a multivariate explanation for cross-national variations in contemporary parliamentary election turnout in Europe. (Accepted: 11 June 2002) About the Authors Alan Siaroff, Department of Political Science, University of Lethbridge, 4401 University Drive, Lethbridge, Alberta, T1K 3M4, Canada; email: alan.siaroff@uleth.ca John W. A. Merer, Department of Political Science, Vanderbilt University, Box 1817 Station B, Nashville, TN, 37235, USA; email: john.w.merer@vanderbilt.edu Notes 1 For example, Powell, 1980, 1982, 1986; Crewe, 1981; Jackman, 1987; Blais and Carty, 1990; Jackman and Miller, 1995; Franklin, 1996; Lijphart, 1997; Blais and Dobrzynska, 1998. 2 Armenia, which has only recently become an electoral democracy, is not included in this analysis. 3 For the Czech Republic and Slovakia, data are included for their regional elections within Czechoslovakia in 1990 and 1992. All data are from Siaroff (2000), updated by the authors. 4 To emphasize, what we are analysing here are thus overall national turnout levels, not the tendency to vote of any individual citizen in a European democracy. 5 Of course, with exact numbers on each relevant party in all cases, one could make a much more precise measure of polarization, as is done in Hazan (1997) for selected Western European nations. 6 As is the case in Australia, perhaps globally the best-known case of strict enforcement of compulsory voting. 7 In fact, Lithuania was listed in Siaroff (2000) as having a weak president, but on further analysis it has been upgraded to moderately strong. References Blais, A. and Carty, K. (1990) Does proportional representation foster voter turnout?, European Journal of Political Research, 18, 167 81. Blais, A. and Dobrzynska, A. (1998) Turnout in electoral democracies, European Journal of Political Research, 33, 239 61. Cox, G. W. (1999) Electoral rules and the calculus of mobilization, Legislative Studies Quarterly, 24, 387 419. Crepaz, M. M. L. (1990) The impact of party polarization and postmaterialism on voter turnout: a comparative study of 16 industrial democracies, European Journal of Political Research, 18, 183 205. Crewe, I. (1981) Electoral Participation, in D. Butler, H. R. Penniman and A. Ranney (eds), Democracy at the Polls: a Comparative Study of Competitive National Elections. Washington DC: American Enterprise Institute. Dahl, R. A. and Tufte, E. R. (1973) Size and Democracy. Stanford CA: Stanford University Press. Franklin, M. N. (1996) Electoral Participation, in L. LeDuc, R. G. Niemi and P. Norris (eds), Comparing Democracies: Elections and Voting in Global Perspective. Thousand Oaks CA: Sage. Gratschew, M. (2001) Compulsory voting. Stockholm: International IDEA <www.idea.int/voter_ turnout/compulsory_voting.htm>. Hazan, R. Y. (1997) Centre Parties: Polarization and Competition in European Parliamentary Democracies. London: Pinter. Hirczy, W. (1995) Explaining near-universal turnout: the case of Malta, European Journal of Political Research, 27, 255 72. Huber, J. and Inglehart, R. (1995) Expert interpretations of party space and party locations in 42 societies, Party Politics, 1, 1, 73 111.

ELECTION TURNOUT IN EUROPE 927 Jackman, R. (1987) Political institutions and voter turnout in the industrial democracies, American Political Science Review, 81, 405 23. Jackman, R. and Miller, R. A. (1995) Voter turnout in the industrial democracies during the 1980s, Comparative Political Studies, 27, 467 92. Lewis, P. G. (2000) Political Parties in Post-Communist Eastern Europe. London: Routledge. Lijphart, A. (1997) Unequal participation: democracy s unresolved dilemma, American Political Science Review, 91, 1 14. Mair, P. and van Biezen, I. (2001) Party membership in 20 European democracies, 1980 2000, Party Politics, 7, 5 21. Powell Jr., G. B. (1980) Voter Turnout in Thirty Democracies, in R. Rose (ed.), Electoral Participation. Beverly Hills CA: Sage. Powell Jr., G. B. (1982) Contemporary Democracies: Participation, Stability and Violence. Cambridge MA: Harvard University Press. Powell Jr., G. B. (1986) American voter turnout in comparative perspective, American Political Science Review, 80, 17 45. Siaroff, A. (2000) Comparative European Party Systems: an Analysis of Parliamentary Elections since 1945. New York: Garland. Ware, A. (1996) Political Parties and Party Systems. Oxford: Oxford University Press.