Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

The salience of nationalism and economic polarization : long term trends and electoral dynamics Lachance, Sarah 2016

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2016_november_lachance_sarah.pdf [ 1.05MB ]
Metadata
JSON: 24-1.0319320.json
JSON-LD: 24-1.0319320-ld.json
RDF/XML (Pretty): 24-1.0319320-rdf.xml
RDF/JSON: 24-1.0319320-rdf.json
Turtle: 24-1.0319320-turtle.txt
N-Triples: 24-1.0319320-rdf-ntriples.txt
Original Record: 24-1.0319320-source.json
Full Text
24-1.0319320-fulltext.txt
Citation
24-1.0319320.ris

Full Text

  THE SALIENCE OF NATIONALISM AND ECONOMIC POLARIZATION: LONG TERM TRENDS AND ELECTORAL DYNAMICS   by  Sarah Lachance  B.A., Université Laval, 2015   A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF  MASTER OF ARTS  in  The Faculty of Graduate and Postdoctoral Studies   (Political Science)   THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)  September 2016  © Sarah Lachance, 2016 ii  ABSTRACT  This research explores the relationship between ethnic cleavages and the class cleavage with data from the Comparative Manifesto Project (CMP) on liberal democracies from 1945 to 2015. According to the “freezing” thesis of Lipset & Rokkan (1967: 12), multinational countries have developed a weaker class cleavage because of anterior territorial-cultural cleavages. In the first part of the data analysis, multinational countries are compared against others on the two variables used to measure these cleavages: the salience of nationalism and economic polarizat ion. Multinational countries show a higher salience of nationalism, as expected, and lower economic polarization.  The second part of the data analysis addresses the relationship between the salience of nationalism and economic polarization in multinational countries. The regression analysis offers a dynamic account of the relationship between these variables across elections. The contextual model, also called “within-between random effects model” (REWB), explicitly models cross-country heterogeneity by separating them from within-country variance, which allows the disentanglement of long-term effects at the country level from short-term effects at the election level. The hypothesized negative relationship between the salience of nationalism and economic polarization holds at both hierarchical levels until the new democracies of the 90s are controlled for, which are confounding the effect of the country-level nationalism variable. However, the negative effect of nationalism on economic polarization remains stable at the election level. This thesis contributes to the scholarship stemming from Lipset & Rokkan’s (1967) by accounting for the short-term effect of the salience of ethnic cleavages on economic polarization at the election level.  iii  PREFACE  All parts of the research were conducted by the author of the thesis. There is no publication from the work presented in this thesis.    iv  TABLE OF CONTENTS  ABSTRACT ....................................................................................................................................... ii PREFACE ......................................................................................................................................... iii TABLE OF CONTENTS ..................................................................................................................... iv LIST OF TABLES ............................................................................................................................... v LIST OF FIGURES ............................................................................................................................ vi ACKNOWLEDGEMENTS ................................................................................................................ vii 1 INTRODUCTION ....................................................................................................................... 1 1.1 The multidimensionality of electoral competition: cleavages ..................................................1 1.2 The interplay between the Left-Right cleavage and ethnic cleavages ......................................3 1.3 The saliency theory of electoral campaigns ...........................................................................6 1.4 The Comparative Manifesto Project methodology .................................................................9 1.5 Polarization on the Left-Right economic dimension.............................................................. 11 1.6 Research Questions & Hypotheses ...................................................................................... 12 2 ARE MULTINATIONAL COUNTRIES DIFFERENT? .................................................................. 13 2.1.1 Key explanatory variable: nationalism ............................................................................. 14 2.1.2 Mean comparison: multinational countries are more nationalist ....................................... 15 2.1.3 Time trends: general increase in nationalism ................................................................... 16 2.2.1 Dependent Variable: polarization on the economic dimension .......................................... 17 2.2.2 Mean comparison: multinational countries are less polarized on the economic dimension . 18 2.2.3 Time trends: general decrease in polarization .................................................................. 18 3 ELECTORAL DYNAMICS ......................................................................................................... 19 3.1 Methods ........................................................................................................................... 19 3.2 Variables ........................................................................................................................... 21 3.3 Results: the negative effect of nationalism on economic polarization.................................... 22 3.4 Discussion ......................................................................................................................... 24 4 CONCLUSION ......................................................................................................................... 26 TABLES .......................................................................................................................................... 28 FIGURES ........................................................................................................................................ 30 REFERENCES .................................................................................................................................. 34       v  LIST OF TABLES  Table 1: Multinational Countries and Non-Multinational Countries...................................................... 28 Table 2: Regression of economic polarization ..................................................................................... 29                         vi  LIST OF FIGURES  Figure 1: The salience of nationalism in multinational and non-multinational countries, 1945-2015 ...... 30 Figure 2: The salience of nationalism in multinational and non-multinational countries by type of democracy (old or new), 1945-2015 ........................................................................................... 30 Figure 3: Economic polarization in multinational and non-multinational countries, 1945-2015.............. 31 Figure 4: Economic polarization in multinational and non-multinational countries by type of democracy (old or new), 1945-2015 ............................................................................................................ 31 Figure 5: The salience of nationalism in multinational and non-multinational countries ........................ 32 Figure 6: Economic polarization in multinational and non-multinational countries ............................... 32 Figure 7: Euroscepticism in multinational and non-multinational countries, 1945-2015 ........................ 33 Figure 8: Foreign population in multinational and non-multinational countries, 1945-2015 .................. 33                      vii  ACKNOWLEDGEMENTS  I thank my supervisor, Professor Richard Johnston, for his teaching and guidance throughout this research project. I also thank Professor Elizabeth Hirsh from the department of sociology of UBC for her help during the early stages of the research design, as well as Professor Marc André Bodet from Université Laval for his enduring support and availability.  Special thanks are owed to my parents, whose have supported me throughout my years of education, both morally and financially.  Finally, thanks to the Fonds de recherche du Québec – Société et culture (FRQSC) for funding.    1  1 INTRODUCTION  1.1 The multidimensionality of electoral competition: cleavages   The Left-Right dimension seldom contains the whole set of issues over which parties confront one another. As socio-historical structural forces, cleavages are a helpful analytical tool to make sense of multidimensional competition. Cleavages are deep social, normative and institutional oppositions that constrain political attitudes and structure party competition as a result. According to Bartolini & Mair (1990) and Deegan-Krause (2009), cleavages have a strong and lasting influence on political behaviour because they are entrenched in social structures, values and beliefs, and consolidated and reproduced through organizations, such as political parties. As a matter of fact, Deegan-Krause (2009) finds that the old cleavages identified by Lipset & Rokkan (1967) in their seminal work on European cleavages remain influential despite the appearance of ‘new politics’. Lipset & Rokkan (1967), in an analysis grounded in a historical and socio-polit ica l perspective, identified four cleavages that were structured by the National revolutions of the early 19th century and the Industrial revolutions triggered in the late 19th century. National revolut ions produced the cleavage between the “central nation-building culture” and the “ethnica l ly, linguistically or religious distinct subject populations”, and the cleavage opposing the nation-state to the Church. Industrial revolutions led to the cleavage between the primary and the secondary sectors of economy and to the cleavage opposing the working class to the owners and the employers (Lipset & Rokkan, 1967: 14). These revolutions produced different outcomes depending on their timing. For instance, the State-Church cleavage was initially less conflictual in Protestant countries because of the legacy of the Reformation and its support from the monarch, 2  as opposed to Catholic countries were the Church strongly mobilized against the secularization of the State prompted by the French Revolution (Lipset & Rokkan, 1967: 34). There has been much literature written about the decline of class-based politics. While a decline in class-based voting as structured by the manual/non-manual workers cleavage has been observed (Nieuwbeerta & de Graaf, 1999: 47), Knutsen (2009: 4) argues that the voting behaviour of the new socio-economic classes must be analyzed to account for the fact that economic structures have changed since the Lipset & Rokkan party system “freezing” thesis. The scholars that observe the persistence of class-voting use the Erikson/Goldthorpe class schema, which differentiates categories of employees on the basis of the employment relations they entail. This new categorization distinguishes between employees “involved in a service relationship with their employers” — salaried professional occupying higher technical, administrative, and manager ia l positions — and the wage-earning manual working class tied to their employers by a labour contract (Knutsen, 2009: 6).  Knutsen (2009: 5) also reports that the social makeup of the support for the new Left and the new Right is very different from their traditional counterparts; the less educated and the working class tend to vote for the new Right parties while the new Left parties find support among the more educated and the middle class voters.   In a chapter of The Impact of Values, Knutsen & Scarbrough (1998) elaborate a model of party choice that compares the ‘cleavages’ and the ‘values’ models using Western European elections from 1970 to 1990, a response to Inglehart (1981) who had declared the end of class voting and the advent of value-voting in 1987. They test three types of voting behaviour: structural voting, value voting and cleavage voting. Cleavage voting occurs when structural variables influence values, which in turn determine party choice. Their results show that value voting is more common in advanced industrial countries (Britain, Germany and Scandinavian countries) 3  while structural voting characterizes less advanced countries like Ireland and Portugal, and that cleavage voting is found mostly in culturally segmented countries like Netherlands and Belgium (Knutsen & Scarbrough, 1998: 515).  The general conclusion is that the basis of party choice is more stable than what the ‘new politics’ advocates suggest. Indeed, although value voting appears to be dominant in most countries, the impact of structural variables has not declined. They still influence voting behaviour mainly through their persistent influence on values, which makes cleavages still relevant in understanding voting behaviour. Deegan-Krause (2009: 16) reaches similar conclusions after finding that attitudes are predominant with regard to political choice, but that structural variables remain influential, especially those pertaining to language, religion and social class: “Even the shrinking role of class in western countries may reflect less the decline of structure than a shift in salient structural markers away from class hierarchy and toward occupational and sectoral categories”.   1.2 The interplay between the Left-Right cleavage and ethnic cleavages   Lipset and Rokkan have laid the foundation for the comparative literature on cleavages in western democracies. They were the first to analyze how territorial-cultural cleavages interact with other cleavages, to find that earlier cleavages born from national revolutions condition the salience of later cleavages. In countries where the nation-building enterprise launched by the dominant culture produced strong territorial-cultural cleavages opposing the center to the “subject cultures” of the peripheries, the polarization along class lines that characterized other countries following industrial revolutions was often constrained (Lipset & Rokkan, 1967: 46). In fact, the mobiliza t ion of the working class was hindered in societies divided by ethnic and religio us oppositions because 4  it required the transcendence of deep cross-cutting identity cleavages. Brodie & Jensons’ (1988) case study of Canada shows that this phenomenon is not restricted to Europe. They argue that brokerage politics – based on interregional and intercultural coalitions – have acted against the mobilization of a united working class at the national level. Indeed, it has historically served the interest of the two major parties to downplay divisive economic issues and bring to the forefront the management of cultural diversity in order to block the entry of a party on the Left. The timing of the enfranchisement of the masses also influenced the development of the Left, which was favored when it coincided with the consolidation of the socialist movement (Bartolini, 2000 :226). In countries where the extension of the suffrage was granted before major industrialization and urbanization, the party system was already consolidated around earlier cultural cleavages when the socialist movement emerged, thereby constraining the development of the Left.  Bartolini (2000) tests the effect of cultural heterogeneity (linguistic and religious) on the Left vote in Europe for a period spanning over one century (1860-1980) and finds that there is a strong negative correlation between the two. Because the cultural composition of a society is a stable dimension – cultural practices and values being transmitted from a generation to the next – it does not explain much of the variation in the Left vote over time. However, it is powerful in explaining variation across countries. The model he presents in chapter 4 also includes social mobilization, which is related to industrialization and urbanization, as a factor explaining variation in the Left vote. Bartolini’s results show that countries with high cultural heterogeneity and low social mobilization are associated with a weak Left vote (12,6%), thereby inhibiting polarizat ion on the Left-Right axis. Depending on the level of social mobilization, the Left vote in countries with a high level of cultural heterogeneity ranges from 12.6% to 31.1%, as compared to a range from 24.6% to 47.4% for countries having a low level of cultural heterogeneity (Bartolini, 2000: 5  203). However, the explanatory variables of this model measure social structures and not their political manifestation, which is an essential component of cleavages. Ethnic groups that are not politicized cannot be considered as the basis of an identity cleavage. While Bartolini’s results and historical analysis give us a clue of the existence of a negative relationship between ethnic cleavages and the class cleavage, the claim that it is identity politics that have constrained the development of the Left is still to be empirically demonstrated. In other words, the link between social structures and party competition is incomplete in Bartolini’s empirical analyses. The present research aims at filling this gap by using an independent variable that measures a politica l phenomenon, nationalism, which is the most appropriate indicator of the intensity of an ethnic cleavage. Similarly, the dependent variable, polarization on the Left-Right economic dimension, is rooted in the class cleavage, which is structured by organizations like labour unions, industr ia l lobbies and political parties.  Due to the Lipset & Rokkan (1967) “freezing” paradigm, cleavages are mainly conceived as static forces that explain variation in the structure of party competition across countries. The literature on cleavages overlooks the dynamic character of cleavages, whose intensity varies across time and is perceivable through the time-varying quantity and mobilizing capacity of the civil and political organizations that represent them. In fact, the intensity of ethnic cleavages varies across generations and is punctuated by specific events (Lemieux, 2011). One can look at the Canadian case to find support for this claim: failed negotiations over the amendment of the constitution to include special rights for Quebec resulted in an upsurge in the support for independence, whereas the defeat of the “yes” at the 1980 and 1995 independence referenda led to its fading. Since the 1995 defeat, the Parti Québécois has moved the question of the independence to the background in its electoral platforms and discourse.  6  Some literature examines the interaction between cleavages at the election- level with panel data, but it focuses on the impact of cross-cutting and reinforcing cleavages on voter volatility (Bartolini & Mair, 1990; Toka, 1998; Bértoa, 2014), or the relative salience of various themes in electoral platforms without focusing specifically on the ethnic and class cleavages (Stoll, 2010; Albright, 2010). The present research addresses this gap by examining the dynamic interplay of the ethnic and class cleavages at the election level through political parties. The approach taken here diverges from Bartolini & Mair’s (1990), who assign parties to cleavage blocks. As the next section explains, parties emphasize or de-emphasize certain issues according to the electoral context, hence the difficulty in classifying and ordering parties along cleavage lines.  1.3 The saliency theory of electoral campaigns   The saliency theory was developed following the observation that parties are often vague about the means employed to overcome problems like unemployment – for instance, they do not mention if they would use free market incentives or state intervention – hence the impossibility to infer a position (Volkens, 2007: 117). Instead, saliency theory is concerned with “the degree of emphasis placed on certain broad policy areas, rather than each party’s support for, or opposition to, as specific policy within these areas” (Budge et al., 1987: 24). The saliency theory allows a dynamic analysis of party competition and is more sensitive to variation in the policy offer of parties across elections than analytical frameworks based uniquely on spatial theory and measures of position, which are rather constant throughout time for a same party. It has the advantage of allowing greater sensitivity to the particular electoral context of each election, especially for parties near the centre.    7  As the Manifesto Research Group lays out in its first major work (Budge et al., 1987, saliency theory is based on an alternative view of party competition where each party tries to push its champion issue to the forefront: “The picture of party competition in other words changes from the classical ‘great debate’ or direct argument over a common range of problems, to one where parties talk past each other, glossing over areas which might favor their rivals while emphasizing those on which they feel they have an advantage” (Budge et al., 1987: 24). In fact, Simon (2002) demonstrates with experiments and a game-theoretic model that dialogue is more often than not a losing strategy. From a rational point of view, it is better for parties to prime the issue on which they have an advantage than to engage in a dialogue from an unpopular position. The issues parties choose to prime are likely to be those over which they have an advantage because of a large support for their positions or because the public positively evaluates their handling of the issues, but they also depend on which issues are on the agenda in a particular election (Druckman, Jacobs & Ostermeier, 2004: 1182). The literature on candidate strategies to prime issues that are ranked high in terms of importance builds on the agenda-setting theory of McCombs & Shaw (1972). This theory claims that the public assigns importance to issues according to the relative emphasis placed on diverse issues by the media (Scheufele & Tewksbury, 2007: 11). Agenda-setting is closely related to priming, another political communication effect by which media content suggests the issues that should be taken as “benchmarks for evaluating the performance of leaders and governments” (Scheufele & Tewksbury, 2007: 11). These political communication effects were theorized from memory-based models of information processing that assume that people develop attitudes according to the pieces of information that are the most salient or, in other words, accessible from their memory (Zaller, 1991; Scheufele & Tewksbury, 2007: 11). As Graber & Holyk (2012:154) explain, voters’ attention and memory are limited resources, which makes 8  priming as a desirable strategy for parties during electoral campaigns. Following this theoretical framework, one should expect that ethnic politics, when highly politicized, will drive the agenda and force parties to reassess how they emphasize various issues. Under this scenario, one strategy is to downplay class politics in order to deliver a strong and clear message about policies to address ethnic issues – like greater decentralization or language rights – as voter attention and party-voter communication are limited resources. Following the same logic, a decrease in the salience of ethnic politics will bring a realignment of the party system around the traditional Left-Right axis.  Changes in emphasis are likely to have a concrete impact on public policy. As discussed above, the degree of emphasis that a party puts on an issue is influenced by the issues placed in the foreground by the media throughout the electoral cycle. In turn, the media are responsive to the form of the content that parties convey in their electoral platforms, and particularly to issue emphasis. The relative emphasis of various themes and issues shapes the media’s outline of the major issues of a campaign and, as a consequence, the party choice of voters and their expectations toward the governing party or coalition. As an example, a party putting great emphasis on environment will be expected to follow up in terms of legislation. In a similar fashion, two parties proposing a tax rise for corporations will not face the same pressure from citizens once in office depending on the emphasis they have put on the policy in their electoral platform and in the media. In sum, emphasis contributes to determining the mandate parties receive when elected into government.  This research takes a different approach to explain variation in issue emphasis than studies focusing on the divergences between position and emphasis or the impact of bad performance in past elections (Janda et al., 1995; Harmel et al. 2016). I am interested in how the class cleavage is affected by the varying salience of ethnic cleavages in multinational countries. Because Left-Right 9  economic politics are a basic dimension of electoral competition in all liberal democracies and,  consequently, their salience is quite steady throughout time (Stoll, 2010: 455), I add another indicator of change: polarization or depolarization, which involve a measure of position. More precisely, I am interested in examining whether the class cleavage has been affected by the increase in the salience of ethnic politics since the 1950s observed by Stoll (2010: 456) by observing the salience of Left-Right economic issues and the dispersion of the parties on the Left-Right continuum.  1.4 The Comparative Manifesto Project methodology  The comparative Manifesto Project (CMP) methodology is based on the saliency theory of electoral competition, thus on measures of issue emphasis. The unit of the dataset is the party manifesto for each election. The “manifesto” is taken in its broader interpretation to encompass any electoral platform or clear statement that constitute the authoritative definition of a party’s policies for a given election. Despite the fact that electoral platforms as such are read by a small proportion of the electorate, they determine the campaign themes and the message that voters receive from the parties and the media. The CMP coders proceed by attributing to each quasi-sentence a category that refers to a policy issue or a position on an issue. The dataset is built of rows that correspond to each party for each election and columns that include the policy categories. The scores reflect each category’s share in percentage of the total codes of a party manifesto. Because the CMP data are measures of emphasis rather than position, its use to test propositions based on spatial theory is contested. As Harmel et al. (2016: 2) point out, “two parties may have the same position on an issue but […] 10  emphasize it to very different degrees”. Their central argument is that position and emphasis serve two purposes in a manifesto: position is taken primarily in function of the party’s identity and the will of its members, whereas emphasis vary from one election to another in function of the issues of the moment and the demands of the larger electorate. The link between policy emphasis, contemporaneous issues and the demands of the larger electorate makes this measure more reflective of short-term changes in the salience of cleavages than policy position. In line with the thinkers of the CMP, Harmel et al. show that following a bad electoral performance, the party will de-emphasize some issues rather than change its position on these issues in order to avoid alienating its membership (Harmel et al., 2016: 7). Nevertheless, Laver & Garry (2000) demonstrate that the economic Left-Right scale of the CMP – the one used in the present research – produce consistent results with computer-generated scales and scales based upon expert survey data. The coefficients of the Pearson correlations between pairs of items from the CMP scale and computer-generated scales range from 0.72 to 0.94, and from 0.94 to 0.99 for expert survey scales (Laver & Garry, 2000: 629-630). The cross-validation between CMP data and expert survey data is less conclusive in McDonald et al. (2007: 65), who use a much larger dataset that covers 79 parties in 17 countries and a factor analysis to find that one sixth of the variance of the CMP data “is distinctly different from that of the expert surveys”. A more interesting finding is that they attribute this “noise” to the greater sensitivity of the CMP data to cross-national variation and short-term changes in party positions, two characteristics that are particularly well-suited for the design of the present research. As McDonald et al. (2007: 73) point out, experts assess party positions on a large time frame by looking at general tendencies, thus describing a stationary process rather than a dynamic one. On the other hand, the nation-specific frame of reference of experts precludes cross-national comparison (McDonald et al., 2007: 74).  11  1.5 Polarization on the Left-Right economic dimension  The concept of polarization that will be used in this research is radicalization, which is a widening of the political spectrum led by more extreme positions and the increase of the distance between the parties’ positions (Baldassarri & Gelman: 2008). The standard deviation of the positions is the most common measure of this kind of polarization (Nuesser, Johnston & Bodet, 2014: 2). However, as mentioned in the previous section, the CMP Left-Right scale employed in this research is not a pure positional measure because it merges policy emphasis and policy position. The CMP defines broad positional categories (e.g. Protectionism: positive; Protectionism: negative), which encompass a set of themes under which statements, includ ing specific policies, are indexed during the coding process. The scores of the parties on that scale is measured by subtracting the proportion of statements associated to the Left from the proportion of statements associated to the Right. The particularity of the CMP Left-Right scale, compared to traditional scales based on expert surveys, is that it does not attribute a position to a party based on a substantive and holistic analysis of its policy proposals, rather, it infers a position from the frequency of statements relating to broad positional categories.  In sum, the placement of the parties on the CMP Left-Right scale is determined by two components: the general positional categories to which their statements refer and the frequency with which statements are made for each category. Under this measure, a move toward the center – that is, a decrease in the statements indicating a position on the Left-Right continuum – does not necessarily indicate a fundamental change in position, but rather a temporary de-emphasis that is often accompanied with the emphasis of other issues that are higher on the political agenda. This conceptualization of polarization is consistent with the assumptions of the saliency theory, who put in doubt the ability of parties to move freely along the Left-Right continuum (“leapfrogging”) 12  due to ideological constraints and voter expectations (Klingemann et al., 1994: 23). Instead, they argue that parties are “selectively emphasizing or de-emphasizing issues in their policy inventory” according to the configuration of the electoral competition (Klingemann et al., 1994: 24).  The inclusion of policy emphasis to measure polarization – which can apply to any politica l dimension – provides a line of explanation as to how extreme policy proposals influence actual policy-making. In many European countries, anti-immigrant politics have been present since, at least, the rise of the fascist movement after World War I, though with varying scope. One can imagine that the presence of anti-immigrant policy proposals has been constant throughout this period, but that its intensity varied in response to wars, economic conditions and migra t ion patterns. Put another way, the extremes of a political dimension may not move, but they may grow.  1.6 Research Questions & Hypotheses  The literature reviewed above suggests a negative relationship between the salience of ethnic cleavages and the salience of the class cleavage. Implicit in Lipset and Rokkan’s characteriza t ion of party systems where the class cleavage is not salient is an assessment of polarization; where the manual workers did not mobilize themselves at the organizational level, the Left did not develop itself at the electoral level, hence the limited scope of the Left-Right dimension of party competition. The first research question addresses the Lipset & Rokkan thesis: are the party systems of multinational democracies less polarized on the economic dimension than the party systems of the other democracies?  Hypothesis 1: Party systems in multinational democracies have a smaller standard deviation on the Left-Right economic dimension than the party systems of other democracies.  13  The second research question involves a more direct and dynamic measure of ethnic cleavages that taps into the electoral platforms of parties: does the salience of nationalism have an impact on economic polarization? As explained in the literature review, voters’ attention span is limited (Graber & Holyk, 2012: 154), which forces parties to adjust their message by emphasizing the particular issues that are in the foreground of each election at the cost of other issue dimensions.  Hypothesis 2: There is a negative relationship between the salience of nationalism and polarization on the economic dimension in multinational countries.   2 ARE MULTINATIONAL COUNTRIES DIFFERENT?  In order to justify the main analysis focused on multinational liberal democracies and electoral dynamics, it is necessary to measure multinational countries against others to see whether they really differentiate themselves on the key variables of interest. The data were taken from the CMP dataset (Volkens et al., 2015a) and covers 562 elections that are distributed as follows: 330 in 30 multinational countries and 229 in 15 non-multinational countries. To be coded as multinational, a country must comprise one or more national minorities that have been politic ized for most of the period covered, which stretches from 1945 to 2015. “Politicization” refers to the presence of ethnic parties or ethnic conflicts.1 Table 1 summarizes the distribution of countries according to these criteria. This section compares the means of multinational countries with the others on the key explanatory variable, the importance of nationalism, and on the dependent variable, polarizat ion                                                                  1 Kellas (2004) was the main reference work used for the selection of cases according to these criteria. 14  on the economic dimension. Time trends are also observable over the period covered by the data. The time scale’s units are years, but the data were measured on an electoral basis. Because elections are not held at the same moment across countries, the variables are not measured each year for each country. A moving average based on a mean calculated over 4 years (the mean electoral cycle covers 3.4 years) was used to prevent the distortion created by the irregular time measurement.  2.1.1 Key explanatory variable: nationalism  I use the data of the Comparative Manifesto Project (CMP), which codes the electoral platforms of parties to assess the salience of various themes.  Each party is given a score in percentage for each theme according to its salience in relation to all the themes coded in the manifesto or, put another way, its share of the codes. The salience of nationalism is measured by the salience of the items under the theme “National Way of Life”. Its positive version comprises “appeals to patriotism and/or nationalism; suspension of some freedoms in order to protect the state against subversion; support for established national ideas”.  Its negative version comprises statements “against patriotism and/or nationalism; opposition to the existing national state” and the negative of the items mentioned for the positive version. Depending on their allegiance, parties can put forth a position that represents the majority ethnic group or the national minorities. I calculated the salience of nationalism for each election by adding the scores of all the parties for each election on the positive and the negative versions of the National Way of Life theme, and divided the sum by the number of parties to account for the variation in the number of parties across elections. 15  2.1.2 Mean comparison: multinational countries are more nationalist  While the results confirm the intuition that the importance of nationalism is greater for parties in multinational countries, the difference between the means of the two groups is small. Multinational countries have a mean of 2.28, whereas others have a mean of 1.70.  Another particularity of the data is that the distribution of the nationalism variable is positively skewed for both groups. While the maximum value on the nationalism variable for multinational countries (max = 22) is substantially higher than for the others (max = 10), 99% of the observations for multinational countries have values of 10 or less. As Figure 5 evidences, outliers seem to drive the higher mean of multinational countries on the nationalism variable. However, the Student's t-test confirms that the difference between the means of the two groups on the nationalism variable is different from zero at a 99% confidence level, even when outliers are removed from the sample. In order to remove the bias caused by the skewness of the distribution, the cube root of the variable was used, with the same conclusive results. The cube root transformation gives distributions a nearly normal shape and, unlike logarithmic transformations, is defined for zero values (Ansolabehere, Snyder & Tripathi, 2002: 139).  The small difference in the importance of nationalism between multinational countries and the others can be explained by the fact that nationalism can arise from other causes than the presence of national minorities, like Euroscepticism or upsurges in immigration, which are investigated in the next subsection. Nonetheless, the higher occurrence of outliers in multinationa l countries suggests that these countries are more subject to extreme levels of nationalism.     16  2.1.3 Time trends: general increase in nationalism  The larger picture reveals a general increase in the importance of nationalism in electoral platforms since 1945. Parties in liberal democracies, multinational and others, are more nationalist in 2015 than they were in 1945. While non-multinational countries approximately follow multinational ones in the trend toward a greater importance of nationalism, multinational party systems are more nationalist overall. As Figure 1 shows, multinational countries did not experience a decline of nationalism following World War II like the others did, and underwent a surge of nationalism at the turn of the 70s. However, this upsurge quickly faded to give way to levels of nationalism comparable to those of non-multinational countries until 1990. From the 90s onward, both groups show a constant increase in the importance of nationalism, but the multinationa l countries’ trend is steeper and is driven by higher values overall. As Figure 7 suggests, the 90s increase in the importance of nationalism in non-multinational European countries could be related to the concomitant 90s increase in the salience of policies against the legislation of the European union in these countries. Since Euroscepticism is often associated with a reminiscence of nationalism, this trend could explain that non-multinational countries closely follow multinationa l countries throughout the 90s. However, values on this dimension are low overall, reaching their maximum at 1 percent of manifesto codes, which makes this lead a weak explanation. Figure 8 points at more convincing line of explanation: the proportion of foreign population. Right-wing nationalist parties often frame increasing immigration as a threat to the nation, as the Brexit referendum highlighted recently, hence the association between the increase in foreign population and the increase in the salience of nationalism. Non-multinational countries are clearly above multinational countries in terms of foreign population from the 60s onward (data is missing prior 17  to that year), distancing themselves from others of about five percentage points from the 90s onward.  While one could attribute the increase in nationalism of the 90s to the inclusion of the post-Soviet countries in the sample, Figure 2 shows that the general 90s trend toward a greater importance of nationalism is still observable among old democracies. However, the last decades show a blurrier picture with regard to the difference between multinational and the other old democracies. When new democracies are isolated in Figure 2, the lines of the multinational and the non-multinational groups of old democracies overlap. This suggests that the new multinationa l democracies of the 90s are driving the steep increase in nationalism that was observed for multinational countries in Figure 1.    2.2.1 Dependent Variable: polarization on the economic dimension  First, polarization on the economic dimension was measured for each election using the CMP data. The CMP Left-Right scale comprises items addressing social, economic and foreign policy issues and ranges from the extreme Left, -100, to the extreme Right, +100.  Some themes have a positive and a negative (i.e. for and against the military). The position of the parties is  measured by subtracting the total score of items associated to the Left from the total score of items associated to the Right, thereby associating the Left with negative scores and the Right, with positive scores. In order to construct an economic scale, I used the same method as the CMP for items pertaining to economy only. Then, I calculated the standard deviation of the scores of the parties on the economic scale for each election. I used an unweighted standard deviation, as it is the most common measure used to assess polarization (Nuesser, Johnston & Bodet, 2014: 2). 18  Moreover, weighting the scores of the parties on the Left-Right economic dimension in function of their share of the votes is not relevant with regard to the concept measured, since issue salience and party system polarization as defined here are not a function of electoral weight. In fact, the impact of issue salience occurs prior to the election day and is influenced by factors external to the relation between parties and voters, like the media and their agenda-setting effect.    2.2.2 Mean comparison: multinational countries are less polarized on the economic dimension  The distribution of the two groups of countries on the polarization variable follows the first hypothesis of a weaker polarization on the economic dimension in multinational countries. The Student's t-test indicates that the difference between the means of the two groups on the polarization scale is different from zero at a 99% confidence level, even when outliers are removed from the sample. Multinational countries have a mean polarization score of 8 and a maximum score of 30, whereas others have a mean of 10 and a maximum of 36. Figure 6 illustrates the differences between the two distributions.  2.2.3 Time trends: general decrease in polarization  Figure 3 shows a general decrease in polarization on the economic dimension for both groups of countries, which were approximately 4 points lower on the polarization scale in 2015 than they were in 1945. While both groups closely follow each other throughout the years, party systems in multinational countries are substantially less polarized than their counterparts in the 50s and from the 90s onward.   19  At first glance, the general time trends on both variables seem to support the second hypothesis: as nationalism increases, polarization on the economic dimension decreases. However, it is new democracies that are driving the trend again, as Figure 4 shows. While a sharp decrease in polarization starting from the 90s is still observable for old democracies, since new democracies have substantially lower values on the polarization scale, Figure 4 reveals that they were stabilizing the overall scores of both multinational and non-multinational countries under their pre-1990 level in Figure 3.   3 ELECTORAL DYNAMICS   3.1 Methods   The regression analysis was conducted with panel data starting from 1945 to 2015 and comprising 28 multinational countries.1 The time unit is the election, which makes an unbalanced panel because electoral cycles vary across countries and because some countries had their first election in the late 70s, like Spain, or in the 90s. There is a total of 330 observations and a maximum of 27 elections by country.  A Hausman test was run on the data to determine whether fixed or random effects should be used. Random effects were rejected under the baseline model, suggesting that the error was correlated with unobserved individual country effects, but consistent under the fully specified model. Instead of using a fixed effects model that suppresses any effect related to cross-country variance, a hierarchical model that separates “between” effects from “within” effects was used in                                                                  1 These are the same multinational countries than the ones used in the previous section, except for Belarus and Malta, which were removed from the regression sample because the data covered too few elections . 20  order to account for heterogeneity bias in incorrectly specified models. The contextual model, also called “within-between random effects model” (REWB) explicitly models cross-sectional heterogeneity (“between” or “contextual” effects) by separating them from within-cluster variance (“within” or “individual” effects). Bell & Jones (2015: 145) demonstrate through simulations that the REWB model performs as well as the fixed effects model according to three measures: bias, efficiency – which accounts for the accuracy and the precision of the estimator – and optimism, that is, “how the standard errors compare to the true sampling variability of the simulations”. According to Bell & Jones (2015: 146), the REWB model is superior because it explicitly models variation across space – that is, across higher-level units, like countries – whereas the fixed effects models “assume[s] a priori that there is a single effect that affects all higher-level units in the same way”. If this assumption is false, a fixed effect model produces results that are misleading.  Raw scores were transformed into group mean centered values in order to avoid collinear ity with higher- level predictors referring to group means on the same variable. Thus the “within” coefficients express the change in the dependent variable for a one-unit within-group increase in the independent variables. Put another way, the value of the salience of nationalism for each election is relative only to its value in other elections within the same country. Substantively it makes sense to use country means instead of the grand mean for centering values because one should expect that the magnitude of relationship between the salience of nationalism and polarization on the economic dimension varies not only across elections, but also across countries, hence a “contextual effect”. More precisely, this methodological choice implies that the effect of a one-unit increase in the salience of nationalism on polarization varies across countries according to their historical level of nationalism. For instance, one could argue that countries having a high mean level of nationalism would show weaker negative effects because of an already low 21  polarization on the economic dimension.  In fact, the non-zero intra-class coefficient (rho) given by the regression in the baseline model indicates the presence of a contextual effect. Finally, robust standard errors were used to account for the clustering of errors by country.   3.2 Variables  1. Polarization on the economic dimension (DV). The dependent variable is the standard deviation of the parties on the economic Left-Right dimension. 2. Importance of nationalism (IV). The share of the items under the theme “National Way of Life” of the total number of codes in a party manifesto, in percentage. 3. Time. Since all elections are not held at the same moment, I control for time. However, with correct specification this variable might become irrelevant. 4. The importance of internationalism. This variable controls for a nationalism that would stem from an opposition to globalization rather than from the presence of national minorities. In a similar way to the nationalism variable, I calculated for each election the mean score on the internationalism variable (positive and negative versions combined) from the CMP dataset, which addresses issues like global governance, the need for world planning of resources, the need for international courts and the support for the United Nations, among others.  5. Foreign-born population. High levels of foreign population might enhance the appeal of ethnic discourses and Right policies, thus being correlated with the salience of nationalism and having an impact on the polarization of the party system. I use the data of the World  Bank on the percentage of foreign population (%). Since the data were collected on a five-22  year basis, I calculated an estimate of the foreign population for each missing year by interpolation, assuming constant increase or decrease between known values. 6. New democracy (90s). The pattern observed in the first section, whereby countries democratized in the 90s show higher levels of nationalism and lower levels of polarizat ion on the economic dimension, commands the inclusion of a variable to control for new democracies. This variable might prove to be a confounder that, when included to the model, will override the effect of the higher- level nationalism variable on economic polarization.    3.3 Results: the negative effect of nationalism on economic polarization  The baseline model shows a statistically significant coefficient for the higher- leve l nationalism variable. As hypothesized, there is a negative relationship between the importance of nationalism and polarization on the economic dimension. As the mean importance of nationalism increases of one percentage point, polarization on the economic dimension decreases of 0.64 standard deviation, which is substantial when we consider that 75% of the observations fall under 11 on this variable. Contextual effects – i.e. stemming from cross-country variation – accounts for 20% of the variance in polarization under this model. The regression results suggest that the general decrease in polarization in liberal democracies observed previously – as well as by Johnston, Nuesser and Bodet (2014) in European countries – is correlated with the increase in ethnic issues observed since 1945 in the descriptive analysis and by Stoll (2010: 456). These results can be interpreted as a validation of the socio-historical thesis of Lipset & Rokkan (1967) that identifies the strength of ethnic cleavages as a factor explaining cross-country variation in the 23  strength of the class cleavage. The presence of serial correlation and influential observations was tested, but proved to be negative. The second model introduces the time variable. This variable has a weak negative effect on polarization, but the results from the Wald test indicate that it improves the fit of the model at a 99% confidence level. More importantly, when time is controlled for, the within-effect of nationalism turns out statistically significant, which reveals the presence of dynamic effects at the election level. An increase in the importance of nationalism of one deviation from the country mean yields a decrease of 0.15 standard deviation in polarization. The addition of the time variable also reduces the magnitude of the between-effect of nationalism of 0.20 units. Now that within-effects are significant, between-effects explain only 7% of the variance in polarization.  The third model include variables pertaining to the importance of internationalism. While the mean level of internationalism has no effect on polarization, the variable has a statistica l ly significant negative effect at the election level and leads to an improvement of the model fit at a 98% confidence level. As the importance of internationalism increases of one deviation from the country mean, polarization on the economic dimension decreases of 0.23 standard deviation. The coefficients of the nationalism variables are robust to the inclusion of the internationalism variables and remain stable, which strengthens the assumption that the effect of the nationalism variables on polarization stems from nationalism(s) concerned with national minorities.  Model 4 includes the foreign population variables, which are not statistically significant. The coefficients of both nationalism variables increase in magnitude, but more substantially for the country-level variable. Now, an increase of one percentage point in the mean importance of nationalism is associated with a decrease of 0.52 standard deviation in polarization on the economic dimension. Differences in foreign population across countries were thus suppressing 24  part of the effect of the nationalism variable at the country level. Contextual effects now explain 19% of the variance in polarization. As expected, the time variable becomes statistically non-significant with greater specification. However, including the foreign population variables does not result in a statistically significant improvement of the model fit. Note also that missing data for these variables caused the removal all observations before 1960 and brought the total number of observations to 275.  The last model includes the new democracy variable, which identifies countries that were democratized in the 90s. As anticipated, the between-coefficient of the nationalism variable shrinks and loses its statistical significance after the inclusion of this variable. The between-effect of the nationalism variable was thus driven by new democracies in previous models. Indeed, the new democracy variable has a strong negative effect on polarization and improves the fit of the model at a 95% confidence level. Concretely, new democracies are less polarized than the old democracies of 2.3 standard deviations. However, the effect of the election-level nationalism variable remains stable and statistically significant, revealing the short-term influence of the salience of nationalism on economic polarization. As in the baseline model, contextual effects explain 20% of the variance in polarization on the economic dimension, but they are now defined by the new democracy variable.  3.4 Discussion  The impact of the inclusion of the new democracy variable in the last model casts a shadow on the relationship between the salience of nationalism and polarization on the economic dimension at the country-level. In fact, new democracies are confounding the effect of nationalism 25  on economic polarization. One can wonder whether the higher level of nationalism found in the parties of the post-Soviet democracies is due to the novelty of independence. In the same vein, their low level of economic polarization might be attributable to the early stage of democratiza t ion of these countries and to the time and resources required for the political mobilization of classes. It will be interesting to see whether this effect lasts in time. These results do not allow the validation of the Lipset & Rokkan’s (1967) thesis associating strong ethnic cleavages and weak class cleavages at a cross-national level. The limited scope of the variable employed in this research to measure the strength of ethnic cleavages might be responsible for this inadequacy, as the “National Way of Life” CMP category does not include references to minority rights and claims for greater autonomy, as language rights (Volkens et al., 2015b: 17-18). They might be more appropriate to measure extreme ethnic claims, like independence, and framed as an opposition to the nation-state, as the description of the “National Way of Life: Negative” suggests  (Volkens et al., 2015b: 18). Including references to moderate minority claims under the nationalism theme could increase the mean of multinational countries on this variable while reducing the weight of outliers.  Nevertheless, the election- level nationalism variable proved to be a robust predictor of polarization, which negative effect validates the main hypothesis tested here, that is, there is a negative relationship between the salience of nationalism and polarization on the economic dimension.   The party manifesto as a unit of analysis is certainly a limit when assessing the relative salience of dimensions of electoral competition, as this document ought to cover a large set of policy domains, of which many are not at the front and center of the “real” political campaign carried out in the field. The necessity to offer an exhaustive policy proposal might have deflated the relative salience of nationalism in party manifestos, as the Left-Right economic domain is a 26  fundamental dimension of electoral competition in liberal democracies and covers a broad range of issues. In fact, media content and advertising might be more representative of the relative importance of the effective dimensions of competition during a given election, as they are the medium through which the campaign unfolds and party competition takes shape. Another potential limit is that manifesto data is not the most appropriate material to measure polarization. Gabel & Huber (2000) demonstrate that manifesto Left-Right scores tend to locate extreme parties closer to the center than survey-based approaches (expert and mass surveys). Nevertheless, while these potential shortcomings might have affected the magnitude of the coefficients reported here, it is unlikely that they would have biased them as they point toward conservative measures for both variables of interest.   4 CONCLUSION  While the data used in this research offer little support for Lipset & Rokkan’s (1967) thesis on the negative effect of the presence of ethnic cleavages on the development of the Left, a negative relationship between the salience of nationalism and polarization on the economic dimension is observable at the election level. These results give an insight into the dynamic relationship between ethnic cleavages and economic polarization in the electoral context and provide a line of explanation for the depolarizing trends observed in Europe by Johnston, Nuesser and Boder (2014). Another contribution is the discussion on the concept and the measure of polarization used in this research. The addition of a measure of policy emphasis to the measure of policy position makes the polarization variable more sensitive to the particular competitive context of each elections, and thus to short-term trends. While parties are reluctant to substantially change their 27  position on central issues, fearing to lose their clientele alongside their identity, they can reduce the emphasis they put on economic issues when issues related to nationalism are ranked higher in the campaign agenda. In this context, economic depolarization can be thought as a decrease in the intensity of the issue positions associated to its poles.  As mentioned earlier, the measurement of the key concepts of this research is limited by the source of the data from which the independent variable and the dependent variable are derived, party manifestos, and the content to which the CMP nationalism category refers. However, the CMP data have the advantage of allowing cross-country comparisons across time and is the most complete dataset of its kind. Moreover, the policy emphasis measure from which the data is produced is particularly well suited to the analysis of short-term trends in electoral competition, which was done in this research by using a contextual model (REWB) that separates within-country effects from between-country effects. Potential future research could give a better account of the electoral dynamics of the relationship between nationalism and economic polarization by using data from media and advertising content, which would improve the accuracy of the estimates of the electoral salience of nationalism and economic issues.       28  TABLES  Table 1: Multinational Countries and Non-Multinational Countries Table 1 Multinational countries Non-multinational countries Albania Australia Austria Estonia Belarus France Belgium German Democratic Republic Bosnia-Herzegovina           Germany Bulgaria Greece Canada Hungary Croatia Iceland Cyprus Luxembourg Czech Republic New Zealand Denmark Norway Finland Portugal Ireland Slovenia Italy Sweden Latvia United States Lithuania  Macedonia  Malta  Moldova  Montenegro  Netherlands  Poland  Romania  Russia  Serbia  Slovakia  Spain  Switzerland  Ukraine  United Kingdom      29  Table 2: Regression of economic polarization TABLE 2: REGRESSION OF ECONOMIC POLARIZATION, CLUSTER-ROBUST SE  Model 1 Model 2 Model 3 Model 4 Model 5  β (SE) β (SE) β (SE) β (SE) β (SE) Nationalism (B) -0.640*** -0.444** -0.450** -0.523*** -0.141  (0.191) (0.192) (0.198) (0.172) (0.193) Nationalism (W) -0.142 -0.148* -0.135* -0.165** -0.169**  (0.088) (0.070) (0.069) (0.070) (0.072) Time  -0.059*** -0.057*** -0.041 -0.027   (0.018) (0.020) (0.027) (0.028) International. (B)   0.007 -0.037 -0.242    (0.363) (0.413) (0.412) International. (W)   -0.234** -0.348** -0.352**    (0.100) (0.155) (0.156) Foreign Pop. (B)    0.109 0.049     (0.062) (0.069) Foreign Pop. (W)    -0.020 -0.072     (0.146) (0.153) New Democracy      -2.318*      (1.121)       Constant 9.318*** 11.612*** 11.498*** 10.173*** 10.636***  (0.794) (0.931) (1.758) (2.474) (2.466)       rho 0.199 0.073 0.080 0.194 0.206 N 330 330 330 275 275 * p<0.05; ** p<0.02; *** p<0.01        30  FIGURES   Figure 1: The salience of nationalism in multinational and non-multinational countries, 1945-2015  Figure 2: The salience of nationalism in multinational and non-multinational countries by type of democracy (old or new), 1945-2015  31  Figure 3: Economic polarization in multinational and non-multinational countries, 1945-2015  Figure 4: Economic polarization in multinational and non-multinational countries by type of democracy (old or new), 1945-2015   32  Figure 5: The salience of nationalism in multinational and non-multinational countries  Figure 6: Economic polarization in multinational and non-multinational countries   33  Figure 7: Euroscepticism in multinational and non-multinational countries, 1945-2015  Figure 8: Foreign population in multinational and non-multinational countries, 1945-2015    34  REFERENCES  Albright, Jeremy J. 2010. The multidimensional nature of party competition. Party Politics. 16:6, 699-719.  Ansolabehere, Stephen, James M. Snyder Jr. & Micky Tripathi. 2002. Are PAC Contributions and Lobbying Linked? New Evidence from the 1995 Lobby Disclosure Act. Business and Politics 4:2, 131-155.   Baldassarri, D. & A. Gelman. 2008. Partisans without constraint: Political polarization and trends in American public opinion. American Journal of Sociology, 114:2, 408-446.  Bartolini, Stefano. 2000. Chapter 4 – Cultural Heterogenity. In The political mobilization of the european Left, 1860-1980: The class cleavage. Cambridge University Press.  Bartolini, Stefano & Peter Mair. 1990. Identity, Competition, and Electoral Availability: The Stabilisation of European Electorates 1885-1985. Cambridge: Cambridge University Press.  Bell, Andrew & Kelvyn Jones. 2015. Explaining fixed effects: Random effects modeling of time-series cross-sectional and panel data. Political Science Research and Methods 3:1, 133-53.  Bértoa, Fernando Casal. 2014. Party systems and cleavage structures revisited: A sociologica l explanation of party system institutionalization in East Central Europe. Party Politics 20:1, 16-36.  Brodie, M. J. & J. Jenson. 1988. Crisis, challenge and change: Party and class in Canada revisited. Carleton University Press.  Budge, Ian, David Robertson & Derek Hearl. 1987. Ideology, Strategy and Party Change: Spatial  Analyses of Post-War Election Programmes in 19 Democracies. Cambridge University Press.  Deegan‐Krause, K. 2009. New Dimensions of Political Cleavage. Oxford Handbooks Online, p.16 Retrieved 26 Apr. 2016, from http://www.oxfordhandbooks.com.ezproxy.library.ubc.ca/view/10.1093/oxfordhb/9780199270125.001.0001/oxfordhb-9780199270125-e-028.  Gabel, Matthew J. & John D. Huber. 2000. Putting Parties in Their Place: Inferring Party Left-Right Ideological Positions from Party Manifestos Data. American Journal of Political Science 44:1, 94-103.  Graber, Doris A. & Gregory G. Holyk. 2012. Chapter 12: Civic Knowledge and Audiovisua l Learning. In Semetko, Holli A., Margaret Scammell (eds). 2012. The SAGE handbook of political communication. GB: Sage Publications Ltd.  Harmel et al. 2016. Manifestos and the ‘‘two faces’’ of parties: Addressing both members and voters with one document. Party Politics, 1-11. 35   Inglehart, Ronald. 1987. From Class-based to Value-based Politics. In Peter Mair (ed.). 1990. The West European Party System. Oxford: Oxford University Press.  Inglehart, Ronald. 1971. The silent revolution in Europe: Intergenerational change in post-industrial societies. American Political Science Review 65: 4, 991-1017.  Janda, Kenneth, Robert Harmel, Christine Edens & Patricia Goff. 1995. Changes in Party Identity: Evidence from Party Manifestos. Party Politics 1, 171-196.  Kellas, James G. 2004. Nationalist Politics in Europe: The Constitutional and Electoral Dimensions. Palgrave Macmillan.  Klingemann, Hans-Dieter, Richard I. Hofferbert & Ian Budge. 1994. Parties, Policies and Democracy. Westview Press.   Knutsen, O. 2009. The Decline of Social Class? Oxford Handbooks Online. Retrieved 26 Apr. 2016, from http://www.oxfordhandbooks.com.ezproxy.library.ubc.ca/view/10.1093/oxfordhb/9780199270125.001.0001/oxfordhb-9780199270125-e-024.  Knutsen, O. & E. Scarbrough. 1998. Cleavage Politics. In The Impact of Values. Oxford Univers ity Press.  Laver, Michael & John Garry. 2000. Estimating Policy Positions from Political Texts. American Journal of Political Science 44:3, 619-634.  Lemieux, Vincent. 2011. Les partis générationnels au Québec : Passé, présent, avenir. Les Presses de l'Université Laval.  Lipset, Seymour Martin & Stein Rokkan. 1967. Cleavage Structures, Party Systems, and Voter Alignments. In Peter Mair (ed.). 1990. The West European Party System. Oxford: Oxford University Press.  McCombs, M.E. & D.L. Shaw. 1972. The Agenda-Setting Function of Mass Media. Public Opinion Quarterly 36, 176-187.  McDonald, Michael D., Silvia M. Mendes & Myunghee Kim. 2007. Cross-temporal and cross-national comparisons of party Left-Right positions. Electoral Studies 26, 62-75.  Nieuwbeerta, Paul & Nan Dirk de Graaf. 1999. Chapter 2 – Traditional Class Voting in Twenty Postwar Societies. In G. Evans (ed.). End of Class Politics?: Class Voting in Comparative Context. Oxford University Press.  Nuesser, Andrea, Richard Johnston & Marc André Bodet. 2014. The Dynamics of Polarizat ion and Depolarization: Methodological Considerations and European Evidence. Manuscript. 36   Raudenbush, Stephen W. 1989. Centering Predictors in Multilevel Analysis: Choices and Consequences. Multilevel Modelling Newsletter 2:1, 10–2.  Scheufele, Dietram A. & David Tewksbury. 2007. Framing, Agenda Setting, and Priming: The Evolution of Three Media Effects Models. Journal of Communication 57, 9-20.  Simon, Adam F. 2002. The winning message: Candidate behavior, campaign discourse, and democracy. Cambridge University Press.  Stoll, Heather. 2010. Elite-Level Conflict Salience and Dimensionality in Western Europe: Concepts and Empirical Findings. West European Politics 33:3, 445-473.  Toka, Gabor. 1998. Party Appeals and Voter Loyalty in New Democracies. Political Studies XLVI, 589-610.  Volkens, Andrea. 2007. Strengths and weaknesses of approaches to measuring policy positions of parties. Electoral Studies 26, 108-120.  Volkens, Andrea, Pola Lehmann, Theres Matthieß, Nicolas Merz, Sven Regel & Annika Werner. 2015a. The Manifesto Data Collection. Manifesto Project (MRG / CMP / MARPOR). Version 2015a. Berlin: Wissenschaftszentrum Berlin für Sozialforschung (WZB).  Volkens, Andrea, Pola Lehmann, Theres Matthieß, Nicolas Merz, Sven Regel and Annika Werner. 2015b. The Manifesto Project Dataset - Codebook. Manifesto Project (MRG / CMP / MARPOR). Version 2015a. Berlin: Wissenschaftszentrum Berlin für Sozialforschung (WZB).  Zaller, John. 1992.  The Nature and Origins of Mass Attitudes.  New York:  Cambridge Univers ity Press.          

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.24.1-0319320/manifest

Comment

Related Items