Time: 28/08/2015, hours 17:50-19:30. Location: University of Montreal, Montreal, Canada, ECPR 2015 Building: Jean-Brillant Floor: 4 Room : B-4310

Transcription

1 Paper 5 (title revised): «Party families, ideological distinctiveness and cohesion: A strong test of the heuristics of the concept of familles spirituelles» Professor André Freire ISCTE-IUL Lisbon University Institute & CIES-IUL andre.freire@iscte.pt & andre.freire@meo.pt & Professor Emmanouil Tsatsanis ISCTE-IUL Lisbon University Institute & CIES-IUL etsats@gmail.com Panel P388: «The structure of political attitudes and values» Chair: André Freire, Professor at the Lisbon University Institute (ISCTE-IUL), Portugal, and Senior Researcher at CIES-IUL. andre.freire@iscte.pt Discussant: Oddbjørn Knutsen, Full Professor, Oslo University, Norway oddbjorn.knutsen@stv.uio.no in Section: «Public Opinion, Political Attitudes and Values» Time: 28/08/2015, hours 17:50-19:30. Location: University of Montreal, Montreal, Canada, ECPR 2015 Building: Jean-Brillant Floor: 4 Room : B

2 «Party families, ideological distinctiveness and cohesion: A strong test of the heuristics of the concept of familles spirituelles» (Title revised) Abstract (261 words) - revised Party family is a fundamental concept in comparative party research. The validity and heuristic capacity of the concept rests upon two fundamental elements: distinctiveness and cohesion. On the one hand, parties belonging to one party family should be significantly distinct from parties belonging to other families. On the other hand, there should be enough ideological policy cohesion among the parties belonging to each party family. However, extensive tests about the heuristic capacity of the concept on the ideological policy dimension are relatively limited and above all can be said to be relatively biased: they rely mainly on party manifestos, which overstate a unitary nature of political parties because they are a product of the national leadership and/or of the majority in charge of the party in each period; or they rely on expert surveys of political scientists and, thus, a «systematization bias» by this experts cannot be fully dismissed (i.e. probably they also tend to overstate the distinctiveness and the cohesion of the party families). However, parties are not unitary actors: they have factions, internal divisions, splits, etc. The use of survey data on party elites to test the heuristic capacity of the concept of familles spirituelles can be said to be both more realistic, vis-à-vis the non-unitary nature of political parties, and to provide a stronger test for the heuristics of the concept. The latter approach is used in the present paper by relying on survey data concerning prospective MPs from the Comparative Candidate Survey in 13 countries, and testing party families ideological distinctiveness and cohesion along to issue dimensions. 2

3 Introduction Party family, also called familles spirituelles (Beyme, 1985; Seiler, 1980), is a fundamental concept in comparative party research (Mair & Mudde, 1998). The validity and heuristic capacity of the concept rests upon two fundamental elements: distinctiveness and cohesion. On the one hand, parties belonging to one party family should be significantly distinct from parties belonging to other families. On the other hand, there should be enough ideological policy cohesion among the parties belonging to each party family. However, extensive tests about the heuristic capacity of the concept on the ideological policy dimension are relatively limited and above all can be said to be relatively biased: they rely mainly on party manifestos, which overstate a unitary nature of political parties because they are a product of the national leadership and/or of the majority in charge of the party in each period; or they rely on expert surveys of political scientists and, thus, a «systematization bias» by this experts cannot be fully dismissed (i.e. probably they also tend to overstate the distinctiveness and the cohesion of the party families). About these type of tests in the literature concerning comparative party research, see Camia & Caramani, 2012; Elff, 2012; Ennser, However, parties are not unitary actors: they have factions, internal divisions, splits, etc. The use of survey data on party elites to test the heuristic capacity of the concept of familles spirituelles can be said to be both more realistic, visà-vis the non-unitary nature of political parties, and to provide a stronger test for the heuristics of the concept (Laver, 2001; Leimbruger, Hangartner & Leeman, 2010). Moreover, congruence between voters and policy makers is a fundamental element of the quality in the democratic political representation process: the higher the level of policy and ideological congruence between citizens and legislators, especially if the latter behave in parliament in a way that matches their policy preferences, the more likely it is that government officials take the voters preferences into account in the policy making process and thus the higher the likelihood that electors preferences are reflected in public policies (APSA, 1950; Pitkin, 1967; Wessels, 2007; Leimbruger, Hangartner & Leeman, 2010). This is yet another reason to use survey data on party elites to study party representation. The latter approach is thus used in the present paper by relying on survey data concerning prospective MPs from the Comparative Candidate Survey (CCS) in 13 countries (Austria, Belgium, Denmark, Finland, Germany, Greece, Iceland, Ireland, Netherlands, Norway, Portugal, Sweden, Switzerland), fielded between 2005 and 2011, to test the ideological - policy distinctiveness and cohesion of different party families (new radical left & greens, orthodox communist, social democrats, liberals, Christian democrats, conservatives and radical right) across countries. This is achieved by measuring issue (and underlying value) orientations 3

4 of the candidates running for parliamentary office. We use survey items measuring both socioeconomic issue orientations and authoritarian-libertarian issue orientations to come up with a synthetic vision of the ideological space in these European countries. Then we test the distinctiveness and cohesion of the different party families in each of the two issue dimensions produced by Principal Component Analysis. On the one hand, the present paper is merely descriptive and exploratory. Descriptive because we are mainly interested in mapping the levels of ideological policy distinctiveness and cohesion among party families in the 13 countries under study, and not so much in explaining variations. Exploratory because due to the data availability we have only 13 European countries and thus inferences about probable characteristics of a European party system are necessarily limited and merely tentative. That said, this study is worth to be pursued not only because the type of data is both more realistic and provide a stronger test for the heuristic capacity of the party family concept but also because it can provide us with some indications of the existence of a European party system, and thus of the quality of political representation at the EU level. If for these 13 countries and with these more demanding tests we find European party families to be fundamentally distinctive and cohesive (like Camia and Caramani, 2012, and Elff, 2013, did with Party Manifesto data; and Ennser, 2012, did with expert survey data), these will be an even stronger indication that there are indeed good conditions for political representation at the EU level. The paper proceeds as follows. In the next section we review the relevant literature and present our basic theoretical framework. In the third section we present the data and methods. In the fourth and fifth sections we analyze the ideological policy distinctiveness and cohesion of party families in the 13 countries under study, respectively. The paper ends with concluding remarks. Literature review and theoretical framing According to the specialists on party studies, Peter Mair & Cas Mudde (1998, pp. 212), the concept of party family is a sine qua non in comparative party research «because it is only by identifying links and equivalences among parties in different polities that we can get a proper sense of what should and what should not be compared or of what is like and unlike». The division of parties in party families, or familles spirituelles, is pretty common in the study of these subjects (see for example Beyme, 1985; Seiler, 1980), but a fundamental issue is of course what is or what are criteria used to allocate different parties to different families (Mair & Mudde, 1998, pp ). In their review of the literature, Peter Mair and Cas Mudde report four criteria: first, origins or sociology of the parties; second, the international federations or other 4

5 transnational groups to which the individuals parties belong; third, party policy or party ideology; fourth, party name or party label. Of course, on the one hand, all the four criteria are relevant and can be used in combination. On the other hand, it is also obvious that each and every criterion have problems for the classification of parties. In our case, besides using the relevant literature to classify the parties in our sample, we also relied heavily on country specialists (see note 1 below) and/or CCS country directors. Since we will be using the policy and/or ideology criteria not to classify the parties but to test the distinctiveness and cohesion of party families, it is important to spend some time with that criterion. Peter Mair and Cas Mudde (1998, p. 217) assume that to classify the parties into familles spirituelles based on the congruence of the policies and/or ideologies that are professed or even pursued by the parties, is probably the most difficult to delineate. There are several sources of data that can be used for this purpose: expert surveys, policy statements, party manifestos, campaigns issues reported in the mass media, legislative behavior, party elite data, mass survey data (see Mair and Mudde, 1998, p. 217; Kriesi et al, 2008; Camia & Caramani,:2012; Leimbruger, Hangartner & Leeman, 2010; Ennser, 2012; Elff, 2013). The use of expert surveys is of course pertinent and one of its advantages is the diversity of comparative datasets that are available and that can be used to different purposes in extensive comparisons. The use of the same questions that the political specialists have to answer in order to classify the policy orientations of parties in different countries is yet another advantage. However, this advantage also encompasses a risk: Given that much of what individual parties do is constrained by the logistics of the system of competition in which they are located, what appears as a cross-national party similarity may therefore turn out to be a cross-national systemic similarity. ( ) The comparability of the data involved may therefore prove to have been an artifact of the method of comparison (Mair & Mudde, 1998, p. 218). Of course, the use of party elite surveys like we do in the present paper with the Comparative Candidate Survey data is a way to overcome this problem: the prospective MPs are asked to state their policy orientations using the same issue questions in each and every country. See the exact wording of the survey items in the CCS in the Appendix. Another problem with the use of expert survey data is the systematization bias : The homogeneity of party families possibly observed in expert surveys may then not be much more than a result of the systematic perspective that is typical of the view of political scientists (Ennser, 2012, p. 160). Again, of course this problem can be overcome by the party elite data we are using. Anyway, using the Benoit and Laver (2006) expert survey, fielded in , Laurenz Ennser (2012) found that party families are indeed distinctive in terms of their policy orientations but, contrary to the 5

6 theoretical expectations, the radical right is not the less homogeneous party family, the liberal family is; moreover, the radical right is even more homogenous than the conservatives. Additionally, Ennser (2012) also found that left-wing parties (social democrats and greens) are clearly more cohesive than the right (liberals, conservatives/christian democrats, radical right). The party manifestos data (Budge et al, 2001; Klingemann et al, 2006) is another fundamental source to classify individual parties into party families based on policy orientations and/or left-right ideology. Several advantages are behind the party manifestos data, not least the extensive availability of data for many countries (European, West and East, and non-european) and decades (for Western Europe since 1945) that allow systematic comparative research. Peter Mair and Cas Mudde (1998, p. 218) underline the two major advantages of this type of data: First, the classification and grouping of parties derived from what the parties themselves had chosen to emphasize in their election programs rather than from some position that had to be imputed to them by an external researcher. Second, by analyzing election programs over relatively long periods of time, the analysts hoped to tap into a more fundamental alignment than might be derived from simple snapshot data. The first advantage for the party manifesto data is also common to our own party elite data, although the second one is more or less unique to the textual data in electoral manifestos and coded by the Comparative Manifesto Project. Of course, there are also some disadvantages with the party manifesto data: first, some bias towards western parties, big parties and older parties; second, although the data allows for long term analysis the truth is that by its very nature it is strongly formatted by short-term influences (Mair and Mudde, 1998, p. 219). Of course, this type of bias is also more or less present in the CCS data. Additionally, estimations of parties left-right locations using party manifesto data produces significantly different results according to the estimation method used (Gabel & Huber, 2000; Dinas & Gemenis, 2010; Williams, 2015). More importantly, as we already stated above, the party manifesto data overstate the unity of party actors because the electoral manifestos are usually produced by the national party organs controlled by the majority in each period. However, the political parties are not unitary actors: the have factions, tendencies, internal tensions, sometimes even suffer from splits, etc. In this respect, the party elite data we are using in the present paper is both more realistic and more appropriate to account for the non-unitary nature of political parties. In any case, the extensive work of Camia and Caramani (2012) using party manifesto data, for parties, and mass survey data, for voters, for 30 countries and for a long period, for Western Europe and less for new European democracies, revealed several things. First, there is right from the beginning ( s) a high left-right ideological cohesion within party 6

7 families. Second, voters are usually more radicalized than parties. Third and fourth, both for parties and for party voters the convergence within families is low across time because intraparty family cohesion was already high in the beginning of the time series. Fifth, especially for the left parties intra-family cohesion is usually higher for economic left-right issues than for non-economic left-right issues. Overall, the more important conclusion of Camia and Caramani (2012, p. 74) is perhaps the one that first, the data presented above show that high levels of ideological cohesiveness for the most important party families making up for most of the party systems have been constant since the very beginning of the periods considered. ( ) Once Europeanised, there can be only little further Europeanisation of electorates and party systems. Of course, one of the issues that can be raised here is if these conclusions are, or are not, somehow data driven. Martin Elff s paper (2013) also relies on party manifesto data ( ) but for only 15 West European countries and using three dimensions (economic left-right, authoritarian libertarian and traditional-permissive), instead of two (economic and noneconomic left-right, in the Camia and Caramani paper), and different estimation methods arrives at somehow different conclusions: it turned out that there is at least as much variation between party families in terms of parties mean positions in the domains of economic policy and of social/moral policy than within party families, even though parties positions fluctuate or change to a degree that within party variation in terms of policy positions is at least as large as the between party variation within party families. That notwithstanding, it became clear that the average positions of party families are distinctive enough to warrant their further investigation (Elff, 2013, pp ). Thus, the questions of the level of distinctiveness between party families and of the level of cohesion within party families seem to be both controversial in the literature, even with the same type of data. Our purpose in the present paper is to explore inter-party family distinctiveness and intra-party family cohesion using data that we believe is more close to the non-unitary nature of political parties, something clearly overlooked by the nature of the party manifesto data. Data and methods For our analysis we use the integrated CCS cumulative dataset, which includes 13 European countries across one or more elections. Due to the lack of longitudinal data for most countries we have decided to examine data from the last available election from each country, with the earliest election being in 2006 and the latest in For detailed information on the countries and the parties under the banner of which candidates ran in the elections, consult 7

8 Tables 1.1 and 1.2 below. Since the sample size from each country varies markedly, we have constructed a weight which harmonizes the impact of each country on the construction of a common ideological space based on the input of candidates across all 13 countries. Given that many of the major European countries are not included in the CCS dataset (France, Italy, Spain, UK), we can hardly claim that we are able to reconstruct a pan-european ideological space onto which we can plot the ideological position of party families and individual candidates from the various countries. However, the inclusion of 13 European countries allows us to examine ideological space beyond the confines of single nations or even specific regions of the continent. For the analysis of a common ideological space across the 13 European countries we employ a principal components analysis (PCA) to identify both the number as well as the content of latent dimensions that are prevalent across all countries. This approach certainly misses national particularities and nuances in the structure and content of party competition in each country, but it is necessary in order to be able to compare and contrast the ideological position of candidates, parties and party families on a set of universal dimensions of political conflict and ideological orientation. After the extraction of the main dimensions, we employ regression factor scores on the extracted components to map the location of the candidates within this common space. Our classification of the different parties along the main party families rests to a large extent on a previous elaboration and classification by country experts 1. 1 This coding was elaborated by several country experts that helped us in this task: Sara Hobolt (Denmark); Ian McAlister and Rachel Gibson (Australia); Eva Onnudotti (Iceland); Gail McElroy (Ireland); Marcelo Jenny (Austria); Åsa Bengtsson and Oddbjørn Knutsen (Sweden), Jean Benoit-Pilet and Lieven De Winter (Belgium) and Claes de Vreese (Netherlands). We also retrieved some data from a dataset on the comparative selection of party leaders, led by Jean-Benoit-Pilet and William Cross. 8

9 Table 1.1 Parties and respondents (candidates that answered the survey) by country (I) Country Election Year Party Family Political Party Sample Size Austria 2008 Communists KPOE, Linke 64 New Left/Greens Gruene 215 Socialists/Social Democrats SPOE 170 Liberals LIF 34 Conservatives OEVP, FRITZ, DC 248 Radical Right FPOE, BZOE, RETTOE, KHK 235 Total 966 Belgium 2007 New Left/Greens ECOLO, GROEN! 111 Socialists/Social Democrats PS, sp.a. 81 Liberals MR, Open VLD, Ljist 141 Dedecker Conservatives CDH, CD&V, N-VA 105 Radical Right FN, Vlams Belang 61 Other Spirit 10 Total 509 Denmark 2011 New Left/Greens Red/Green Alliance, Socialist 105 People s Party Socialists/Social Democrats Social Democrats 35 Liberals Social Liberals, Liberal Alliance, Liberals 111 Conservatives Christian Democrats, Conservative People s Party 82 Radical Right Danish People s Party 42 Total 375 Finland 2011 Communists Communist Party 62 New Left/Greens The Left Alliance, Green 191 League Socialists/Social Democrats Social Democrats 86 Liberals Centre Party, Swedish 128 People s Party in Finland Conservatives National Coalition, Christian 144 Democrats Radical Right Finns Party 96 Other Pirate Party, Change 2011, Liberty Party, Workers Party, Seniors Party 204 Total 908 Germany 2009 Communists Die Linke 141 New Left/Greens Bündnis 90/Die Grünen 157 Socialists/Social Democrats SPD 158 Liberals FDP 148 Conservatives CDU, CSU 174 Other No party 6 Total 784 Source: CCS (2013), Comparative Candidates Survey [Dataset - cumulative file], distributed by FORS, Lausanne, For further information about CCS, see Note: for the classification of parties into party families we relied not only on the literature of this subject but also on a personal survey with specialists from each country, namely the CCS country directors. 9

10 Table 1.2 Parties and respondents (candidates that answered the survey) by country (II) Country Election Year Party Family Political Party Sample Size Greece 2007 Socialists/Social Democrats PASOK 146 Conservatives New Democracy 95 Total 241 Iceland 2009 New Left/Greens Left Green Movement 94 Socialists/Social Democrats Social Democratic Alliance 97 Liberals Liberal Party 33 Conservatives Independence Party 92 Other Progressive Party, Civic Movement 180 Total 496 Ireland 2007 New Left/Greens Greens 20 Socialists/Social Democrats Labour 30 Liberals Fiana Fáil 42 Conservatives Fine Gael 38 Other Sinn Fein, Progressive Democrats, Independents 45 Total 175 Netherlands 2006 New Left/Greens Groen Links, Socialist Party 37 Socialists/Social Democrats PvdA 22 Liberals VVD, D66 44 Conservatives CDA 32 Other SGP, Pvd Dieren, Christian Union 35 Total 170 Norway 2009 New Left/Greens Socialist Left Party 155 Socialists/Social Democrats Labour, 150 Liberals Venstre 136 Conservatives 10 Hoyre, Christian Democratic Party 271 Radical Right Progress Party 147 Other Centre Party 147 Total 1006 Portugal 2011 Communists Communist Party, PEV 42 New Left/Greens Left Bloc 52 Socialists/Social Democrats Socialist Party 45 Conservatives Social Democratic Party, CDS-PP 118 Total 257 Sweden 2010 New Left/Greens Feminist Party, Green Party, Left Party 486 Socialists/Social Democrats Social Democrats 303 Liberals Centre Party, Liberal Party 401 Conservatives Christian Democrats, Conservatives 523 Radical Right Sweden Democrats 27 Total 1740 Switzerland 2011 Communists PdA/PdT 43 New Left/Greens GPS/PES 214 Socialists/Social Democrats SP/PS, 272 Liberals FDP/PRD, PLS/LPS, 190 Conservatives CVP/PDC, BDP 261 Radical Right SVP/UDC, MCG, SD/DS, FPS/PSL 192 Other LdU, EVP/PEP, CSP/PCS, GLP, EDU/UDF, Lega 470 Total 1642 Source: see Table 1.1. Note: see Table 1.1.

11 We explore the patterns of distinctiveness and cohesion among the main six party families (Communist, New Left/Green, Socialist/Social Democrat, Liberals, Conservative, New Radical Right) by focusing on the mean location and the patterns of dispersion of candidates on the common ideological space across but also within the party families. Tracing a common ideological space Questions related to the structure and dimensionality of ideological space and political competition in contemporary democracies has been one of the most enduring fixtures in political science literature. The identification of structural transformations in western societies in the post-war era has prompted new analyses of the major divisions in contemporary political systems, with a tendency to move away from assumptions of unidimensionality and to acknowledge that ideological space in most countries has a more complex structure. Technological changes, the reorganization of economic production and the increased significance of the ever-expanding middle class have supposedly dulled the classic capital-labour conflict, which has long thought to underlie the left-right universal dimension. In addition, economic growth, increased affluence levels and the rapid expansion of education have created a secularising dynamic and a purported generational value shift (Inglehart 1977). The implicit or explicit assumption that informs most studies on the structure of ideological space is that all conflicts, old and new, can be captured in two broad dimensions: a conflict over distributional preferences and the economic role of government, reflecting a divergence of objective material interests; and a cultural dimension formed by fundamental value divides. What differs from study to study is the identification of the specific content of these two dimensions and their respective alignment onto the two-dimensional space (Tsatsanis et al. 2014: 520). For example, Herbert Kitschelt (1994) has argued that socio-economic changes during the 1970s and 1980s have rotated the main axis of ideological demand and, consequently, of political supply in western Europe, from a simple dilemma between 'left-socialist' and 'rightcapitalist' (i.e. exclusively economic-distributive) alternatives, to a more complex contraposition between 'left-libertarian' and 'right-authoritarian' poles. The latest structural transformations that are supposed to rearrange the configuration of national cleavage structures are associated with the process of globalization as argued by Hanspeter Kriesi and his colleagues (Kriesi et al. 2008). The assumption underlying the thesis is that as the impact of globalization in its various aspects (economic, cultural and political) can assume a diversity of forms for the different members of a national community, new disparities and new forms of conflict are created. Citizens tend to perceive these differences in terms of winners and losers of globalization 11

12 and that the aspirations or grievances of these competing groups will be articulated by political parties (Kriesi et al. 2008, p. 3). Since all these divisions are supposed to be common in all contemporary European democracies (even if articulated and aligned in varying modes in different national contexts), we have selected to apply our data reduction method to items that are included in the dataset and are conceptually related to these broader dimensions at a lower level of abstraction. Our initial exploration included items on the questions of European integration and of globalization, as it was our initial expectation that the items related to the demarcation-integration divide (European integration, globalization and immigration) proposed by Kriesi would either form a distinct axis or align with the classic cultural axis (authoritarianism-libertarianism; see Flanagan, 1987; Flanagan and Lee, 2003). Table 2. Ideological Dimensions among European Candidates Libertarianism- Authoritarianism Economic Left- Right People who break the law should be given stiffer.709 sentences Same-sex marriages should be prohibited by law.636 Immigrants should be required to adjust to the.634 customs of [country] Immigrants are good for the economy (recoded).605 Women should be able to decide about abortion.526 (recoded) Income and wealth should be redistributed towards.805 ordinary people Providing a stable network of social security should.800 be the prime goal of government Stronger measures should be taken to protect the.648 environment Eigenvalues Variance (%) N Note: Dimensions were extracted by performing a Principal Components Analysis on a weighted 13-country sample and selecting dimensions with an initial eigenvalue over 1. The reported loadings and % of variance explained were obtained following a Varimax rotation. 12

13 However, the factor loadings of items on globalization and European integration demonstrated more complex patterns of alignment with other items, making the task of identifying meaningful latent dimensions almost impossible. When removed, the principal components analysis reveals a much more conventional ideological structure comprised of the two classic dimensions: : an economic conflict over distributional preferences, reflecting a divergence of objective material interests; and a cultural-political conflict informed by fundamental value divides (Teperoglou and Tsatsanis, 2011: 1208). The only surprising result is the high loading of the new politics item on the economic materialist dimension. However, the directionality is less surprising, given that postmaterial pro-environment positions go typically hand-in-hand with leftist economic preferences in electorates, a pattern that appears to be replicated at the level of party candidates as well. The two components give us a rather conventional two-dimensional ideological space which we use in our following section in order to map the location of the respondents using their regression factor scores for each of the two dimensions. Ideological distinctiveness of European Party Families Our first task is to examine the location of entire party families on the ideological space by creating groups of respondents based on the classification of the parties with which the candidate decided to run on the selected election. Immediately a familiar pattern emerges. The location of the candidates according to party family follow the expected sequence along the two dimensions, however the level of distinctiveness appears to be far from sharp, given the huge overlaps that the relatively large standard deviations reveal. Indeed, there are generally high levels of overlapping, more pronounced for some party families rather than others. For example, candidates that belong to left-of-center party families (communist, New Left, socialist) appear to occupy the same space along the economic axis, despite the subtle mean differentiations. The same picture emerges among the right-of-center party families, even though the degree of dispersion, as evidenced by the space occupied along the space of two standard deviations (both visualized with error bars and numerically represented in Figure 1). The Liberal party family appears to be most heterogeneous of all on both dimensions. If that is somewhat to be expected on the economic axis, it is more surprising when it comes to the cultural dimension, where one would expect higher homogeneity for liberal political parties. 13

14 Figure 1. Mapping European Party Families on the Common Ideological Space L-R Mean -1 st. dev. +1 st. dev. Lib-Auth Mean -1 st. dev. +1 st. dev. Radical Right Conservative Liberal Soc.Dem./Socialist New Left / Greens Communist Other

15 On the contrary, the communist party family reveals the highest levels of homogeneity, which is however also an artefact of the comparatively low number of candidates included in this party family (it is also the reason why we did not include a figure for the communist party family in our discussion below). The clustering that is observed on the economic dimension, with the formation of two distinct ideological blocs, is not observed on the cultural dimension. Here is the distances between the party families are much more symmetrical, with the radical right party family adopting extreme positions towards the authoritarian end of the axis. Somewhat surprisingly, the positions of the communist and the New Left party families overlap on this dimension, even though one would expect that the New Left party candidates would adopt more extreme positions towards the libertarian end of the axis in comparison to their materialist colleagues belonging to the old Left parties. The fact that the new politics item on the environment is included on the economic instead of the cultural axis has probably contributed to the emergence of this unexpected finding. Overall, even though the mean location of candidates follows more or less expected patterns, we cannot easily talk about a high level of distinctiveness among party families. The policy preferences of candidates on the economic axis especially lead us to identify a left and a right political family, rather than particular party families. Value orientations appear to be much more important in the differentiation between the various party families. Exploring the patterns within each party family perhaps will give us some answer as to why this is the case. Ideological Cohesion of European Party Families The exploration of individual party families reveals a common thread that runs across most party families, despite the absence of candidates from important countries in our sample. Region appears to often trump party family, undermining both ideological distinctiveness between European party families and ideological cohesion within them. Beginning with the Conservative party family (see Figure 2), which constitutes the dominant party family in contemporary (and especially post-crisis) European politics, we can observe that regional differences undercut ideological cohesion. The southern parties in Greece and Portugal tend to be clearly to the left of conservative parties in the other European countries. In fact, the position of candidates of New Democracy, PSD and CDS-PP is closer to that of candidates from left-of-center parties of some Northern and Central European countries. Differences along the cultural axis are less pronounced, but even here it seems that conservative candidates 15

16 from the Nordic countries tend to adopt much more libertarian positions in comparison to their colleagues in the continent. Figure 2. Mapping Conservative European Political Parties on the Common Ideological Space This regional pattern seems to be less clear-cut in the social democratic/socialist party family (see Figure 3). There is a bit more homogeneity along the economic dimension in relation to the conservative party family, whereas there is less clustering along the cultural dimension. Regional patterns here are a bit less pronounced. For example PASOK appears to be on the left of everyone but the Portuguese socialist party appears to adopt much more centrist positions. Scandinavian socialist candidates appear to be more progressive than their counterparts in the 16

17 continent but there are several exceptions. Overall, the socialist party family demonstrates higher levels of ideological coherence in relation to the conservative party family. Figure 3. Mapping Social Democratic European Political Parties on the Common Ideological Space As already seen in Figure 1, the Liberal party family is by far the most heterogeneous of all the European party families (see Figure 4). The dispersion on the economic dimension runs from centrist to rightist (neoliberal) economic orientations, reflecting the fact that in some countries liberal political parties tend to be on the left of the conservatives and in some countries to the right. Perhaps even more surprisingly, there is a high level of dispersion along the authoritarianlibertarian value orientation dimension, with parties in Catholic regions tending to move closer 17

18 to more conservative/authoritarian positions in relation to their counterparts in Scandinavia. More so than the pattern of distribution along the economic dimension, region and religious cultural context appear to have a significant impact on the lack of ideological cohesion on the value dimension. 18

19 Figure 5. Mapping Liberal European Political Parties on the Common Ideological Space The New Left/Green party family (see Figure 5) is by far the most cohesive of all the party families, perhaps aided by the fact that the sample includes mainly parties from Scandinavia and North/Central Europe. There is a clear clustering of candidates of all parties on the Left- Libertarian quadrant of the common ideological space but, even so, Scandinavian candidates tend to hold the most libertarian positions on questions related to value orientations. 19

20 Figure 5. Mapping New Left/Green European Political Parties on the Common Ideological Space Finally, the candidates from the New Radical Right family (see Figure 6) reveal once again some known traits of the heterogeneity of this party family, especially when it comes to economic orientations. Despite the fact that there are a lot of important parties that are missing from our sample (e.g. FN, PVV, Lega Nord, etc.) the few remaining parties display high levels of dispersion along the economic dimension (which is of secondary importance to this party family) coupled with high levels of cohesion along the authoritarian-libertarian value dimension. The small sample which is the product of missing countries, refusal of candidates to be 20

21 interviewed in participating countries, as well as the fact that in some party systems parties belonging to this party family simply did not exist before 2011, do not permit us to examine regional variations. However, given what we know about the dual origin of this party family (more anti-tax in t Scandinavia, more statist in continental Europe), we would expect to see in a larger sample the candidates from Scandinavia to adopt consistently more right-wing orientations on the economic axis in relation to their counterparts in continental Europe. 21

22 Figure 6. Mapping New Radical Right European Political Parties on the Common Ideological Space 22

23 Concluding remarks This exploratory study of Comparative Candidate Survey data revealed some interesting general patterns and points to some directions concerning future research. While European party families have managed to carve out to a large extent their own distinct ideological niche as evidenced by the patterns of responses of candidates from 13 European countries, the level of inter-family ideological distinctiveness and intra-family ideological cohesion is undercut to a large extent by national and regional particularities. Whereas a conservative candidate, for example, in Portugal will tend to be to the right on economic issues of most of his socialist colleagues in his own country, he is almost certainly to the left of most of his counterparts in Northern European countries. In other words, national and regional context still often trumps political and ideological affiliation at the European level, underlying the barriers in any attempt to move away from the loose confederations that are today s European political parties and towards the creation of more centralized political party organizations at the European level. The lack of longitudinal data prevent us from examining the degree in which there has been any degree of Europeanization and intra-family convergence. However, no matter what the longterm trajectory has been, any attempt to move towards more classic organizational forms at the European party level will run the danger of factionalism, and not just because of different national priorities but also due to substantive differences in ideological orientations, especially within the three mainstream party families. 23

24 References APSA - American Political Science Association (1950), «Toward a More Responsible Two-Party System», American Political Science Review, 22, pp Benoit, Kenneth, and Laver, Michael (2006), Party Policy in Modern Democracies, London, Routledge. Budge, Ian, et al (2001), Mapping Policy Preferences. Estimates for Parties, Electors and Governments , Oxford, Oxford University Press. Camia, Valeria & Caramani, Daniele (2012), «Family meetings: Ideological convergence within party families across Europe, », Comparative European Politics, vol. 10, pp CCS Comparative Candidates Survey Module I [Dataset - cumulative file]. Distributed by FORS, Lausanne, Dinas, Elias, & Gemenis, Kostas (2010), «Measuring Parties Ideological Positions With Manifesto Data: A Critical Evaluation of the Competing Methods», Party Politics, vol. 16, nº 4, pp Elff, Martin (2013), «On the Distinctiveness of Party Families», Prepared for delivery at the 71th Annual Conference of the Midwest Political Science Association, Palmer House Hilton, Chicago, Illinois, April 11-14, Ennser, Laurenz (2012), «The homogeneity of West European party families The radical right in comparative perspective», Party Politics, vol. 18, no. 2, pp Flanagan, S. C. (1987), Value change in industrial societies, American Political Science Review, vol. 81, no. 4, pp Flanagan, S. C. and Lee, A.-R. (2003), The new politics, culture wars, and the authoritan-libertarian value change in advanced industrial democracies, Comparative Political Studies, vol. 36, no. 3, pp Freire, André (2015), Left-Right Ideology as a Dimension of Identification and as a Dimension of Competition, Journal of Political Ideologies, Volume 20, Nº1, pp Gabel, Matthew J., & Huber, John D. (2000), «Putting Parties in Their Place: Inferring Party Left-Right Ideological Positions from Party Manifestos Data», American Journal of Political Science, Vol. 44, No. 1 (Jan., 2000), pp Hooghe, Liesbet, Gary Marks & Carole J. Wilson (2004), Does Left/Right Structure Party Positions on European Integration?, in Gary Marks & Marco R. Steenbergen (eds), European Integration and Political Conflict. Cambridge University Press, Cambridge, pp

25 Inglehart, R. (1977) The silent revolution: Changing Values and political styles among western publics, Princeton University Press, Princeton. Inglehart, Ronald (1987), «Value change in industrial societies», American Political Science Review, vol. 81, nº. 4, pp e Kitschelt, Herbert (1994, 1996), The Transformation of European Social Democracy, Cambridge, Cambridge University Press. Klingemann, Hans-Dieter, et al (2006), Mapping Policy Preferences II. Estimates for Parties, Electors and Governments in Eastern Europe, European Union and OECD , Oxford, Oxford University Press. Kriesi, Hanspeter, et al (2008), West European Politics in the Age of Globalization, Cambridge, Cambridge Universtiy Press. Laver, Michael (editor) (2001), Estimating the Policy Position of Political Actors, London, Routledge. Leimbruger, Philipe, Hangartner, Dominique & Leeman, Lucas (2010), «Comparing Candidates and Citizens in the Ideological Space», Swiss Political Science Review, Volume 16 (3), pp Mair, Peter, & Cas Mudde, «The Party Family and Its Study», Annual Review of Political Science, 1998,1,pp Miller, Warren, e Stokes, Donald (1963), «Constituency influence in Congress», in American Science Review, 57, nº1, pp Pitkin, Hanna F. (1967), The Concept of Representation, Berkley, Los Angeles e Londres, University of California Press. Powell, G. B. jr. (2004), Political representation in comparative politics, Annual Review of Political Science 7: Seiler, Daniel Louis (1980), Partis et Familles Politiques, Paris, Presses Univ. de France. Teperoglou, Eftichia, and Tsatsanis, Emmanouil (2011), «A new divide? The impact of globalization on national party Systems», West European Politics, vol. 34, no. 6, pp Tsatsanis, Emmanouil, Freire, André and Tsirbas, Yannis (2014), The impact of the economic crisis on the ideological space in Portugal and Greece: A comparison of elites and voters, South European Society and Politics, vol. 19, no. 4, pp von Beyme, Klaus (1985), Political Parties in Western Democracies, Aldershot, UK: Gower. Wessels, Bernhard (2007), Political representation and democracy in Dalton, Russell J., and Klingemann, Hans-Dieter (eds.), The Oxford Handbook of Political Behavior, Oxford, Oxford University Press, pp

26 Williams, Laron K. (2015), «It s all relative: Spatial positioning of parties and ideological shifts», European Journal of Political Research, vol. 54, nº??, pp

27 Appendix survey questions on issue orientations C: Opinions and Policies C2. People hold different views on political issues. What do you think of the following? C2.1 C2.3 C2.4 C2.6 C2.7 C2.8 C2.10 C2.11 Immigrants should be required to adapt to the customs of Portugal. Stronger measures should be taken to protect the environment. Same-sex marriages should be prohibited by law. People who break the law should be given stiffer sentences. Providing a stable network of social security should be the prime goal of government. Income and wealth should be redistributed towards ordinary people. Immigrants are good for Portuguese economy. Women should be free to decide on matters of abortion. Completely disagree Do not agree Neither agree, nor disagree Agree Completely agree 27