Measuring the electoral mobilization of ethnic parties

Abstract

Measurement of the electoral mobilization of ethnic parties has posed a considerable challenge to those performing comparative research on the political mobilization of ethnic groups. To address this issue, we propose indicators that estimate the electoral mobilization of ethnic parties by combining administrative and survey data. Specifically, we propose two measures: an absolute one, fully isolable from context, and a relative one, which corrects for turnouts. Furthermore, we show that a particular indicator based entirely on more widely available administrative data is valid when a narrow definition of ethnic parties is applied. Our indicators for the electoral mobilization of ethnic parties allow for valid comparisons across ethnic parties in different countries and regions at different points in time. We expect these new indicators to trigger further comparative studies on ethnic parties.

Keywords

Conceptual definition ethnicity/language/race ethnic parties measurement voters/citizens

Introduction

The mobilization of ethnic parties is highly relevant to several fields of political science. Conflict researchers, for instance, are concerned with the role ethnic parties play in the solidarity of multi-ethnic societies and in the process of democratization. Much of the debate in this literature is concerned with the question of whether the formation of ethnic parties should be facilitated or hindered. Consociationalist theory posits that facilitating the representation of ethnic groups through ethnic parties leads to the moderation of demands and to the stabilization of party systems – something that in turn allows for the formation of consociational arrangements (Birnir, 2007; Chandra, 2005; Lijphart, 1977, 2004). Others, however, argue that the presence of ethnic parties leads to ethnic conflict – noting that, since ethnic parties appeal exclusively to one or a few ethnic groups, they are likely to trigger ethnic outbidding and, consequently, confrontation (Horowitz, 1985; Rabushka and Shepsle, 1972; Reilly, 2006). Still others take an intermediate position, maintaining that ethnic parties have little effect on ethnic conflict and democracy (Ishiyama, 2009; van Amersfoort and van der Wusten, 1981), or that any tendency towards conflict will be dependent on the type of ethnic parties involved (Ishiyama and Breuning, 2011).

Apart from conflict and democratization research, ethnic parties are also highly relevant to the study of party systems. While scholars of ethnic conflict and democratization are primarily interested in the presence of ethnic parties as independent variables, scholars who focus primarily on party systems are interested in the presence of ethnic parties as (part of) their dependent variable. Such students of party systems have typically studied ethnic parties with an eye to the territorial dimension of electoral politics. And part of this interest stems from the fact that ethnic parties often mobilize a territorially concentrated constituency with a major impact on the number of parties in parliament (Sartori, 1997). Major points of debate today include whether ethnic parties might be considered a remnant of the freezing of party systems during democratization (Caramani, 2004; Rokkan, 1999) or whether they may instead be gaining ground as part of a new regionalist movement (Bartolini, 2005: 248–308; Keating, 2000).

Scholars of party systems are not solely interested in the existence of ethnic parties, as conflict researchers often are. They are also keenly interested in the degrees to which ethnic parties demonstrate electoral mobilization. Until the end of the last century, the study of ethnic parties’ electoral mobilization was dominated largely by single and comparative case studies (but see Allardt, 1979; De Winter, 1998; Urwin, 1983),¹ in large part because it was generally very difficult to operationalize the electoral mobilization of ethnic parties. Despite such methodological problems, the study of ethnic parties has more recently moved in a comparative direction, however (Bernauer and Bochsler, 2011; Birnir, 2007; Bochsler, 2010; Madrid, 2012; Tronconi, 2009). Still, these studies are held back by their lack of comparable indicators. This leads to a paradoxical situation in which scholars enter into sophisticated analyses of predictors of the number of votes of ethnic parties without being able to meaningfully rank ethnic parties according to their respective degrees of electoral mobilization.

To be sure, it is not disagreement over the definitions of ethnic groups and ethnic parties that is at issue here. There is widespread agreement over the definition of ethnic parties, which are generally seen as parties that primarily champion the interests of ‘their’ ethnic group. Furthermore, among those who have tried to conceptualize the electoral mobilization of ethnic parties, there is agreement that this concept should be understood as the degree to which ethnic parties are able to mobilize their targeted ethnic group.² Second, to the degree that divisions on definitional issues concerning ethnic parties remain – due to disagreement on the definition of ethnic groups, for instance – these divisions can help explain the absence of comparability across studies. They cannot, however, explain problems of comparability within studies. This is because, while different studies may apply different definitions, it is unlikely for there to be much variance in definitions for the same concepts within a single study.

The first comparative analysis that has attempted to measure the electoral mobilization of ethnic parties used its existence as an indicator (Allardt, 1979). Since it soon became clear that a more finely graded indicator was needed, the electoral results of ethnic parties became the subject of scholarly interest (Urwin, 1983). Most scholars working in this area have tried to ascertain ethnic parties’ degrees of electoral mobilization – understanding electoral mobilization as the degree to which an ethnic party is able to mobilize their targeted ethnic group – as calculated by regional vote shares (e.g. De Winter, 1998; Gordin, 2001; Sorens, 2005; Tronconi, 2005, 2009). As we will argue, this indicator is severely flawed because it makes unrealistic assumptions about the territorial concentration of ethnic groups. Furthermore, it assumes the absence of aggregation bias – a potentially problematic assumption that needs to be empirically tested.

In this article, we present new solutions to the problem of comparative indicators for electoral mobilization of ethnic parties. By combining survey and administrative data, we meaningfully measure the electoral mobilization of ethnic parties without making assumptions about the territorial concentration of the ethnic groups being studied, and also avoiding the ecological fallacy problem. Two indicators are presented: one absolute and one relative. While the absolute indicator proves more precise concerning the definition of the electoral mobilization of ethnic parties as ‘the mobilization of the targeted group’, the relative indicator is more sensitive to national turnouts.

As good survey data concerning the electoral behaviour of ethnic groups is scarce, our proposed measures might not be applicable to all research questions concerning the electoral mobilization of ethnic parties. To test if more widely available administrative data might be used for measurement of the electoral mobilization of ethnic parties, we compared the results of our proposed measures with an approximation constructed with administrative data. We find that for parties that score high on our index of party ethnicity, an indicator relying on administrative data can validly be used. Because of this, we come to the surprising conclusion that dividing the share of ethnic party votes by the share of ethnic minority group members – a measure that had been avoided due to alleged ecological fallacy problems – provides a valid indicator in most instances.

We present our argument in the following way: First, we deal with the conceptualization of ethnic parties as parties that mobilize a targeted ethnic group. Second, we discuss previous attempts to measure the electoral mobilization of ethnic parties and develop our own indicators. Third, we conduct validity tests for our indicators using survey and administrative data. We finish by summarizing our argument and discussing its potential meaning for future work on the electoral mobilization of ethnic parties.

Ethnic groups, ethnic parties and electoral mobilization

In studies of ethnic conflict and democratization there is an established tradition regarding the conceptualization of ethnic parties (van Amersfoort and van der Wusten, 1981; Horowitz, 1985; Lijphart, 1977; Rabushka and Shepsle, 1972). This literature has remained largely disinterested, however, in tracing variation in levels of electoral mobilization. By contrast, for the study of party systems quite the opposite is true. And while the comparative study of party families (which has a long-standing tradition in Western Europe) has always been interested in levels of electoral mobilization, ethnic parties constitute rather a new category (e.g. Camia and Caramani, 2012). This might be due to a strong Rokkanian tradition in comparative studies of party systems (e.g. Bartolini, 2000; Caramani, 2004; Kriesi et al., 2008; Urwin, 1980). As is well known, Rokkan (1999) categorized the Western European parties across urban–rural, capitalist–worker, church–state and centre–periphery cleavages. Ethnic parties constituted part of those parties that defended themselves against the hegemony of nation-builders in the political centre. In contrast to the other cleavages, Rokkan thought of this conflict between centre and periphery as one in which cultural and territorial opposition coincide.³ Consequently, regional and ethnic parties were perceived as belonging together. Such an approach can also be traced in the grouping of ethnic and regional parties under the umbrella label of ‘ethnoregionalists’ (e.g. Hechter and Levi, 2000; Müller-Rommel, 1998; Tronconi, 2005, 2009).

But as the experiences of Diasporas, Travellers and Migrants illustrate, ethnic identity does not necessarily go together with territorial concentration. Furthermore, ‘territorial defence’ does not have to be cultural. Parties can be defined as regionalistic for reasons other than ethnic ones. For instance, the German Party of Democratic Socialism owed its regional anchoring to its historical legacy as part of the former German Democratic Republic. What separates ethnic parties from all other parties is, of course, their attachment to one or a few ethnic groups. More specifically, ethnic parties primarily promote the interests of ‘their’ ethnic groups (and/or the exclusion of others) (Birnir, 2007: 35–36; Brancati, 2006; Chandra, 2004, 2011: 151; Gunther and Diamond, 2003: 184; Ishiyama, 2009: 64; Reilly, 2006: 11; Van Cott, 2005: 3; Wimmer, 1997: 647). This separates them from regionalist parties, which promote the interests of the entire population residing within a certain delimited territory. Ethnic parties, by contrast, can have goals that are entirely independent from territorial claims.

While there is broad agreement on the general definition of ethnic parties, this definition might be blurred somewhat by two potential sources of disagreement (see also Chandra, 2011). First, it remains unclear whether ethnic parties are restricted to ethnic minorities. While we focus on ethnic minority parties in this study, the difference between parties promoting the interests of an ethnic majority or minority is not as clear as it might seem. This is particularly true for multicultural societies, where, in numerical terms, one cannot speak of ethnic majorities. Here, the term ‘majority’ points to a relative majority, or in other words the largest ethnic group in a society. It is perhaps more appropriate, however, to understand an ethnic majority as the politically dominant group within a context. This is because the very politization of ethnicity is historically linked to the question of ‘who owns the state’, which is itself at the core of the processes of state and nation-building (Wimmer, 1997). Because of this, the distinction between majorities and minorities can be seen as resting on the division between state-framed ethnic groups and counter-state ethnic groups. This distinction Brubaker uses (1999) for ‘nations’ emphasizes ethnic groups’ historical ties to the states in which they reside. While state-framed ethnic groups associate themselves with their respective nation-builders, counter-state ethnic groups perceive themselves as not belonging to the founders of the ‘fatherland’.

Second, scholars remain divided on the definition of ethnic groups. According to the classical definition provided by Weber (1978[1922]: 389), ethnic groups are ‘those human groups that entertain a subjective belief in their common descent because of similarities of physical type or of customs or both, or because of memories of colonization and migration; this belief must be important for group formation; furthermore it does not matter whether an objective blood relationship exists’. Weber’s emphasis that ethnic groups share a subjective belief in common descent which must not be related to objective blood relationship has since remained central to discussions on ethnicity. While some have argued for a more encompassing or less voluntaristic definition of ethnicity, subjective belief in common descent has remained at the core of almost all definitions (Chandra, 2006).⁴ Nevertheless, disagreement on the potentially unsolvable question of how to define and classify ethnic groups (Fearon, 2002: 4; Hug, 2003: 269) render the definition of ethnic parties somewhat fuzzier than it might initially appear.

While disagreement over the definition of the term ‘ethnic groups’ might affect the categorization and selection of ethnic parties (Chandra, 2011), it should not, however, impact the measurement of the electoral mobilization of ethnic parties. What is important here is that there is broad agreement on the idea that ethnic parties are parties that primarily promote the interests of ‘their’ ethnic groups (and/or the exclusion of others) (see above). From this starting point, one can deduce that the electoral mobilization of ethnic parties should not be understood as the mobilization of the whole society, but of its targeted ethnic group(s) (Chandra, 2004: 5–6; De Winter, 1998: 210; Diamond and Gunther, 2001: 198). Especially when dealing with the ethnic parties of small ethnic groups, it is more important for these parties to be perceived as the legitimate political representatives of their respective ethnic groups than it is for them to achieve maximal representation in the national parliament. This is because these parties derive their power from the credibility of their ability to act unilaterally – for instance, via territorial or functional exit. Functional exit, for example, can take the form of not paying taxes, or establishing parallel institutions. Examples of territorial exits include separatism, irredentism and even mass emigration (Bartolini, 2005). In general, these constitute actions that the political representatives of an ethnic group can only credibly threaten to carry out if they can successfully mobilize among their ethnic community. Hence, to be in a powerful position, ethnic parties must convince others of their capacity for mobilization. This explains why ethnic parties want to mobilize the ethnic vote, and why the mobilization of an ethnic party should be measured in terms of the votes it receives from its targeted ethnic group (see also Chandra, 2004: 5-6; De Winter, 1998: 210).

The electoral mobilization of ethnic parties should not be confounded with ‘ethnic voting’, i.e. the degree to which vote choice is related to ethnicity. One crucial difference between the electoral mobilization of ethnic parties and ethnic voting comes in the fact that for the former the ethnic parties and their targeted ethnic groups constitute the relevant unit, whereas for the latter the entire electorate within a country constitutes the unit of analysis. Hence, levels of ethnic voting depend upon the country-wide number of ethnic parties and upon the size of the ethnic groups mobilized by them. For instance, Chandra and Wilkinson's (2008) EVOTE, a measure of the ‘practice of ethnicity’, is calculated as the ratio of the votes of all ethnic parties divided by the votes of all non-ethnic parties in each country. By contrast, since the electoral mobilization of ethnic parties is calculated at the level of the parties and not the entire country, the measure is independent of the number of ethnic parties and the ethnic group sizes per country.

Another crucial difference between the electoral mobilization of ethnic parties and ethnic voting is that the existence of ethnic parties is a necessary precondition for their electoral mobilization – something that is not the case for ethnic voting. Ethnicity can also be related to voting behaviour in the absence of ethnic parties where ethnicity is related to vote choice among non-ethnic parties. Consequently, scholars focusing on ethnic voting estimate the degree to which ethnicity is able to predict party choice at the individual level, independently of whether these parties are ethnic or not.⁵

Although the electoral mobilization of ethnic parties and ethnic voting are often correlated, this is not always the case (also Huber, 2012).⁶ An ethnic party might, for instance, receive a high share of votes from its targeted ethnic group, while at the state-wide level ethnic voting may be weak. Such a scenario would most likely occur in a country with only one or a few small ethnic minorities. If the members of those ethnic minorities voted en bloc for ethnic parties, while the members of the ethnic majority distributed their vote between several non-ethnic parties (or even between ethnic parties of the minorities), the electoral mobilization of the ethnic parties would be strong, and ethnic voting at the state-level weak. By comparison, if only a small share of the electorate cast votes for an ethnic party, while the vote choice of non-ethnic parties was strongly related to ethnic group membership, the electoral mobilization of ethnic parties would be weak and ethnic voting strong.

Two indicators of electoral mobilization of ethnic parties

The goal of this section is to propose an indicator measuring the electoral mobilization of ethnic parties as the share of a given ethnic group that votes for ‘its’ ethnic party. This indicator should enable us to compare the mobilization of different ethnic groups across and within countries, in the same elections and at different points in time. The ideal measure of the electoral mobilization of ethnic parties – measuring the share of a targeted ethnic group that votes for the corresponding ethnic party – can be written as

I n d^{*} = \frac{a^{*}}{b^{*}}

where $a^{*}$ denotes the members of an ethnic group who vote for an ethnic party and $b^{*}$ the size of the ethnic group eligible to vote.

The problem of this measure, however, is that we can only rely on samples from surveys or on administrative data consisting of aggregates at the constituency level. Consequently, the data for which the indicator could be computed do not exist. This problem, of course, is not new to scholars of ethnic mobilization and some attempts have been made to solve it. Most approaches so far have tried to approximate the ideal type measure by relying on constituency-level data. They usually measure ethnic party support as the share of votes an ethnic party receives in the constituencies which it contests (e.g. De Winter, 1998; Gordin, 2001; Sorens, 2005).

This indicator, however, has many problems. One of these problems is that one runs the risk of ecological fallacy. The votes coming from other ethnic groups bias the values. Furthermore, and even more critically, this indicator is based on the assumption that all ethnic minorities are equally strongly concentrated across the electoral constituencies in which they participate. As such, to use it, one makes an assumption that if an ethnic party gets a low share of the votes in one electoral constituency, this is due to poor electoral mobilization among its targeted ethnic group. This assumption is too strong for several reasons.

First, it is simply not true that different ethnic minorities are equally concentrated territorially. For instance, the South-Tyrolean minorities are significantly more territorially concentrated than the Irish in Northern Ireland. Second, these groups’ electoral geography often accentuates these differences. Whereas the Tyrolean minority in Northern Italy is strongly concentrated in the electoral district of Alto-Adige and constitutes a vast majority there, the practice of Gerrymandering caused the Catholics in Northern Ireland to occupy a minority position in almost all constituencies (Rossiter et al., 1998). Third, if ethnic parties contest in constituencies where their ethnic group constitutes only a small minority, this might well be due to the electoral system and other context variables. For instance, due to the system of proportional representation adopted in Finland, the Swedish People’s party contests in many constituencies that have a low share of Swedish speakers, whereas in Westminster elections the Scottish Nationalist Party only runs in its strongholds. So, it is very possible that the share of the votes an ethnic party gets in one electoral constituency will be more strongly related to the share of members of the targeted ethnic group, than it will be to the degree that an ethnic group is able to mobilize its targeted ethnic group. Hence, measuring the share of votes an ethnic party receives in the constituencies it contests might not meaningfully measure the votes it receives from its targeted ethnic group.

Other indicators of the electoral mobilization of ethnic parties that have been used include the ‘index of electoral success’ proposed by Gordin (2001) and the maximal vote-share in an electoral constituency used by De Winter (1998). The former calculates for each constituency with the participation of an ethnic party its relative vote-share divided by the vote-share of the largest party in the constituency. The final score is the average calculated from these values for each constituency. This indicator is invalid, however, because it does not reduce the ecological fallacy problem, and strongly depends on implicit assumptions about the territorial concentration of the ethnic groups and the fragmentation of the entire party system. The latter takes the highest vote-share of an ethnic party in a constituency as the indicator for its success in an election. This measure reduces problems of ecological fallacy and territorial concentration only partially, and does not give a value that is representative across constituencies.

In the light of these problems, Chandra (2004) uses constituency-level data on party votes, turnouts and ethnic composition to reduce the problems of ecological fallacy and variance in the territorial concentration of the ethnic groups across constituencies. She uses these data to estimate the shares of members of Scheduled Castes voting for the Bahujan Samaj Party in Indian elections through application of King’s ecological inference method (King, 1997).⁷ The weakness of this method, however, is that it relies upon some potentially problematic assumptions (a truncated bivariate normal distribution, no aggregation bias in the data, and no spatial dependence between the territorial units (Chandra, 2004: 305)). More problematic, however, is that it requires data on the ethnic composition of the electorate on the constituency level, which is frequently unavailable.

In the light of the inapplicability of Chandra’s approach to many cases, it is necessary to find an indicator for which data are more often available, and that allows us to overcome the other previously mentioned problems. Since we want to measure the support of ethnic parties in terms of mobilization among targeted ethnic groups, it would be straightforward to compute this value as the number of the ethnic party votes divided by the size of the ethnic group. This indicator is simply the aggregate pendant to the ideal indicator. This first indicator can be written as:

I n d_{a 1} = \frac{a_{A}}{b_{A}}

where a_A denotes the number of votes for the ethnic party, and b_A the size of the ethnic group. Subscript _A signifies that the data are administrative aggregates. This indicator is a perfect measure if:

\begin{aligned} a_{A} = a^{*} \\ b_{A} = b^{*} \end{aligned}

A problem with a _A is that it implicates the assumption that the number of votes for the ethnic party equals the number of votes it receives from its targeted ethnic group. This means that it is assumed that all votes for an ethnic party have to come from its ethnic group. If this assumption is not held we will commit an ecological fallacy by obtaining $I n d_{a 1}$ in which the share of the targeted ethnic group voting for an ethnic party is overestimated. In other words, the index based on administrative data cannot account for cases in which voters other than members of the ethnic group in question vote for an ethnic party.

To overcome the ecological fallacy problem connected to the index based on administrative data, we propose another measure of the electoral mobilization of ethnic parties. In working towards this new indicator, however, we must first propose an Index of Party Ethnicity (IPE). This index describes how large an ethnic group’s vote-share is, out of the total number of votes received by a given ethnic party. The maximum value of IPE is 1, which reflects a situation in which all votes for an ethnic party are cast by members of its targeted ethnic group. The minimum value 0 would mean, by contrast, that not a single vote for an ethnic party came from the members of its targeted ethnic group. To calculate the IPE, it is necessary to have survey data in which, for each respondent, we have information on that respondent’s party choice and ethnicity. More formally we can write the IPE index as:

I P E = \frac{a_{s}}{v_{s}}

where $a_{s}$ denotes the number of people from the ethnic group who voted for the ethnic party, and v_s the number of persons who voted for the ethnic party. The subscript s denotes that these values are calculated with survey data.

The IPE can now be used to correct for the ecological fallacy problem of $I n d_{a 1}$ . By multiplying the number of ethnic party votes by the share of the votes from the targeted ethnic group (i.e. the IPE), the number of votes from an ethnic party being cast by members of the ethnic group can be calculated. Ind_a2 reads as:

I n d_{a 2} = \frac{a^{'}}{b^{'}} = \frac{a_{A} \times I P E_{}}{b_{A}} = \frac{a_{A} \times a_{S}}{b_{A} \times v_{S}}

where IPE denotes the index of party ethnicity, $a^{'}$ is the estimated number of the votes for an ethnic party coming from members of the targeted ethnic group, and $b^{'}$ is the size of the ethnic group.⁸

In the case of the IPE receiving the score of 1, $I n d_{a 1}$ does not contain any problems of ecological fallacy, which implies that

I n d_{a 2} = I n d_{a 1} i f I P E = 1

So far, we have operationalized the electoral mobilization of ethnic parties – defined as the share of the targeted ethnic group that votes for an ethnic party – as the number of people from an ethnic group who vote for the corresponding ethnic party divided by the size of the ethnic group in question (see Equation 1). Ind_a2 combines the strengths of administrative and survey data. In sum, we use highly reliable administrative data to measure the absolute number of votes and the sizes of the ethnic groups in question, we use survey data to correct for potential problems with ecological fallacy. This is an elegant and quite simple operationalization and it is the adequate operationalization if one is interested in absolute differences of electoral mobilization of ethnic parties.

The electoral mobilization of ethnic parties in national elections, however, takes place within an important context – one involving given parties’ relative ease of mobilization within a certain country. In particular, it is easier to mobilize people to vote if turnout is generally high in a country, and harder if the turnout is usually low. Since our absolute measure of the electoral mobilization of ethnic parties is insensitive to the context, factors influencing the turnout will quite likely also influence the mobilization of ethnic parties. In national elections, turnout is mainly a context variable at the national level. It is affected, for instance, by factors related to electoral systems and electoral culture (e.g. Franklin, 1996). Explanations of variance in the electoral mobilization of ethnic parties that use the absolute indicator of the electoral mobilization of ethnic parties should therefore include these factors as explanatory variables.

However, the factors determining national turnouts might not be perfectly modelled, and investigators might want to concentrate on factors at a different level of analysis (e.g. the level of the ethnic groups) – so, an indicator related to national turnouts might be more meaningful. The relative measure we propose can be written as:

I n d_{r}^{*} = \frac{a^{*}}{b^{*} \times t}

where t denotes the turnout in the examined elections. The difference between $I n d_{a}^{*}$ and $I n d_{r}^{*}$ is in the denominator: the number of votes from the ethnic group for the ethnic party is not divided by the size of the ethnic group in question, but by the size of the ethnic group multiplied by the national turnout. The indicator can be interpreted as the number of people from the ethnic group – who should take part in elections if there were to be the same turnout within the ethnic group as within the whole population – voting for an ethnic party. While this indicator measures the electoral mobilization of ethnic parties independently of factors determining national turnouts, it is still sensitive to differences in ethnic parties’ abilities to bring their voters to the polls. This explains why an ethnic party can score higher than the value 1 if it reaches a higher turnout among the members of its ethnic community than the other parties manage to achieve. On the basis of $I n d_{r}^{*}$ one can – with the addition of the turnout in the denominator – further calculate Ind_r1 and Ind_r2 in the same way as Ind_a1 and Ind_a2.

Our relational indexes have two important features. First, their construction does not assume that turnout rates across different ethnic groups in the same country are equal. These turnout rates can be different and, if they are, our measures are sensitive to that fact in the desired way. Suppose, for instance, that we have two ethnic groups, A and B, in a country, and that these groups are equal in terms of the number of citizens belonging to them. If the turnout rate in group A is higher than it is in group B, and all voters in group A who turned out voted for their ethnic party, then the value of our indexes will be higher for group A than they will be for group B, no matter how the members of group B voted. However, if the turnout rate for group A is higher than that of group B but not related to higher mobilization for their ethnic party (i.e. the share of voters from group A voting for their ethnic party is not higher than the share of voters from group B voting for their ethnic party), then the values of our indexes for group A might even be lower than those for group B. This would be the case if the number of votes from the targeted ethnic group were lower for group A than for group B (given the equal size of both groups). On the other hand, our indexes are not sensitive to the causes of mobilization (or demobilization). For example, if members of group A and B usually vote for their ethnic parties (and not for other parties), but turnout among group B is lower, then, no matter what the reason, the values of our indexes for group B will be lower than they are for group A.

Second, a possible difference between the national turnout and the turnout among an ethnic group will not be problematic for our indexes. If there is such a difference, then the indexes will behave in the desired way. This means, if the turnout among an ethnic group is higher than the national turnout, and the citizens from this ethnic group voted for their ethnic party rather than for other parties, then the indexes will show high levels of electoral mobilization for the ethnic party of the group in question. As was mentioned earlier, in this case the values of the relational indexes could be higher than 1. By contrast, if the turnout of the ethnic group in question is lower than the national turnout, then the relational indexes will indicate lower levels of electoral mobilization and the values of the relational indexes will be lower than 1. In case of the same turnout of the ethnic group and at the national level (again with all members of the ethnic group voting for their ethnic party), the relational indexes will be equal to 1. This is an important feature of the relational indexes from a comparative perspective, because it enables comparisons between electoral mobilization of ethnic parties in different contexts in a substantially meaningful way.

Empirical validation

The aim of this section is twofold. First, we will empirically test how serious the problem of ecological fallacy is for the administrative indicator (Ind_a1 ). This is of keen interest, since the indicator that relies solely on administrative data will be applicable for far more elections than the indicator with correction for ecological fallacy (Ind_a2 ), for which survey data must also be available. Second, we will compare the values of the absolute and the relative indicator (Ind_r2 ). This will give clues to the extent to which our results depend on the specific definition of levels of mobilization as either absolute or relative.

For our empirical analysis we are interested in all ethnic minorities that are mobilized by ethnic parties in a democratic context. In order to decide which groups can be considered ethnic minorities, we rely on datasets that are frequently used in the study of ethnic mobilization.⁹ While so doing, we apply a rather loose means of operationalizing ethnic groups. Out of the available cases, we have selected all cases in which at least one party tries to mobilize an ethnic minority, for which survey data of good quality are available.

In order to categorize parties as ethnic (or not), we have assessed how prominently the parties in question made reference to the interests of their targeted ethnic group in their political offers. In doing so, we followed a long-standing tradition in the comparative study of party families.¹⁰ As proxies for the ethnic character of the political offer of the parties, we relied on party names, categorization in previous studies (in particular Bochsler, 2010; Sorens, 2005; Tronconi, 2009), and – in cases of uncertainty – on the opinions of country experts. As a result, we were left with only a few boundary cases. Among these, we decided to include the Catalan Convergence and Union (CiU) in Spain, but to exclude the post-communist parties in the Basque Country and Catalonia.

Our dataset contains information on 14 ethnic minorities, from 13 countries, for 31 elections (see Table 1). Because of our regional expertise and our model’s need for high quality administrative and survey data, we focus only on elections in industrialized democracies. Our dataset, however, covers a broad range of cases in very dissimilar settings. In addition, it includes information on elections at the state-wide and regional levels. For all these cases, we gathered administrative data on the electoral results achieved by ethnic parties and on the sizes of their respective ethnic groups. We relied on census or register data with ethnic self-classification at its basis. In the cases of the Swedish speaking minority in Finland and the Catalans and Gallicians in Spain, we used data on language ability. In the case of the Irish in Northern Ireland we relied on denomination.¹¹ For all of these cases we were able to apply the same categories in the survey data – circumstances that rendered our indicators derived from survey data and aggregate data comparable.¹²

Table 1.

Ethnic groups, elections and ethnic parties.

Country	Ethnic minority	Election	Census category	Relative group size¹	Ethnic parties
Belgium	Flemish	National 2003	Language	59%	Flemish Bloc; New Flemish Alliance
Bulgaria	Turkish	National 1991	Ethnicity	9%	Movement for Rights and Freedom
Bulgaria	Turkish	National 2001	Ethnicity	9%	Movement for Rights and Freedom
Canada	Quebecois	National 1997	Ethnicity	20%	Bloc Quebecois
Canada	Quebecois	National 2004	Language	22%	Bloc Quebecois
Estonia	Russians	National 1995	Ethnicity	13%	Our Home is Estonia!
Finland	Swedish speakers	National 2007	Language	6%	Swedish People's Party in Finland
Great Britain	Scottish	National 1997	Ethnicity	9%	Scotish National Party
Great Britain	Welsh	National 1997	Ethnicity	67%²	Plaid Cymru
Italy	South Tyroleans	Regional 1988	Language	70%	Südtiroler Volkspartei; Südtiroler Heimatbund; Freiheitliche Partei Südtirols
Italy	South Tyroleans	Regional 1993	Language	70%	Südtiroler Volkspartei; Südtiroler Heimatbund; Die Freiheitlichen; Ladins
Italy	South Tyroleans	Regional 1998	Language	68%	Südtiroler Volkspartei; Südtiroler Heimatbund; Die Freiheitlichen; Ladins
Italy	South Tyroleans	Regional 2003	Language	68%	Südtiroler Volkspartei; Südtiroler Heimatbund; Die Freiheitlichen; Ladins
Latvia	Russians	National 1993	Ethnicity	14%³	Party of Russian Citizens; Equal Rights
Latvia	Russians	National 2002	Ethnicity	15%	Human Rights; Russianàs party
New Zealand	Maori	National 1996	Ethnicity	15%	Mana Maori
New Zealand	Maori	National 2002	Ethnicity	15%	Mana Maori
Northern Ireland	Irish	Regional 2003	Religion	44%	Social Democratic and Labour Party; Sinn Fein
Romania	Hungarian	National 1996	Ethnicity	7%	Democratic Alliance of Hungarians in Romania
Romania	Hungarian	National 2004	Ethnicity	7%	Democratic Alliance of Hungarians in Romania
Slovakia	Hungarian	National 1998	Ethnicity	10%	Party of Hungarian Coalition
Spain	Basque	National 2000	Ethnicity	66%⁴	Basque Nationalist Party; Eusko Alkartasuna
Spain	Basque	National 2004	Ethnicity	66%⁴	Basque Nationalist Party; Eusko Alkartasuna
Spain	Basque	National 2008	Ethnicity	66%⁴	Basque Nationalist Party; Eusko Alkartasuna; Aralar
Spain	Basque	Regional 2001	Ethnicity	74%	Basque Nationalist Party; Euskal Herritarrok
Spain	Basque	Regional 2005	Ethnicity	74%	Basque Nationalist Party; Eusko Alkartasuna; Aralar; Partido Communista de las Tierras Vascas
Spain	Basque	Regional 2009	Ethnicity	74%	Basque Nationalist Party; Eusko Alkartasuna; Aralar
Spain	Catalan	National 1996	Language	17%	Republican Left of Catalonia; Convergencia y Unio
Spain	Catalan	National 2000	Language	17%	Republican Left of Catalonia; Convergencia y Unio
Spain	Catalan	Regional 2003	Language	75%	Republican Left of Catalonia; Convergencia y Unio
Spain	Gallician	National 2004	Language	6%	Nationalist Block of Gallicia

Notes: ¹At the level of the polity in which the election took place. ²In Wales. ³Corrected for an estimated 60% of disenfranchisement among Russians. ⁴In the Basque country and Navarra.

Table 2 gives the values of our indicators for the electoral mobilization of ethnic party blocs.¹³ Taking the absolute indicator with correction for ecological fallacy (Ind _a2 ) as the reference, it emerges that the South Tyrolean ethnic parties demonstrate the highest level of electoral mobilization out of the ethnic parties in question. Around 80 percent of the South Tyroleans voted for one of the German speaking parties. Hence, the German ethnic parties in South Tyrol prove to be not only hegemonic among their ethnic community here (Holzer and Schwegler, 1998), but they also demonstrate high turnout among their group. The latter is not simply due to the fact that there is high overall turnout in Italy, either. Indeed, this observation is also confirmed by the fact that, even when corrected for national turnout (Ind_r2), the South Tyrolean ethnic parties demonstrate very high values of electoral mobilization. The only ethnic party that achieves similarly high values is the Party of the Hungarian Coalition in Slovakia.

Table 2.

Electoral mobilization of ethnic parties.

Country	Ethnic minority	Election	Turnout	Ind_a2 (absolute indicator)	Ind_a1 with administrative data	Ind_r2 (relative indicator)	Ind_a1 with survey data
Italy	South Tyroleans	Regional 1998	0.86	0.83	0.84	0.97	0.94
Italy	South Tyroleans	Regional 1988	0.90	0.81	0.83	0.90	0.90
Italy	South Tyroleans	Regional 2003	0.82	0.79	0.80	0.97	0.87
Italy	South Tyroleans	Regional 1993	0.86	0.78	0.80	0.91	0.89
Slovakia	Hungarian	National 1998	0.83	0.76	0.79	0.91	0.81
Latvia	Russians	National 2002	0.71	0.68	0.95	0.95	0.76
Romania	Hungarian	National 1996	0.71	0.66	0.72	0.93	(0.62)¹
Finland	Swedish speakers	National 2007	0.65	0.55	0.61	0.84	(0.60)
Spain	Basque	Regional 2001	0.79	0.54	0.56	0.68	0.43
Bulgaria	Turkish	National 1991	0.82	0.54	0.65	0.66	(0.66)
Northern Ireland	Irish	Regional 2003	0.63	0.53	0.58	0.84	0.57
Romania	Hungarian	National 2004	0.59	0.49	0.52	0.84	(0.77)
Spain	Basque	Regional 2005	0.68	0.47	0.49	0.69	0.37
Spain	Basque	National 2004	0.75	0.45	0.45	0.60	0.44
Bulgaria	Turkish	National 2001	0.67	0.43	0.53	0.64	0.68
Latvia	Russians	National 1993	0.90	0.39	0.46	0.43	0.37
Spain	Catalan	Regional 2003	0.62	0.37	0.40	0.60	0.39
Spain	Basque	National 2000	0.64	0.37	0.38	0.58	0.51
Spain	Basque	Regional 2009	0.65	0.36	0.38	0.56	0.33
Spain	Basque	National 2008	0.64	0.33	0.33	0.51	0.34
Canada	Quebecois	National 2004	0.61	0.32	0.34	0.53	0.15
Canada	Quebecois	National 1997	0.66	0.31	0.36	0.47	0.29
Estonia	Russians	National 1995	0.69	0.26	0.31	0.38	(0.35)
Spain	Catalan	National 1996	0.77	0.22	0.27	0.28	0.37
Belgium	Flemish	National 2003	0.92	0.21	0.22	0.23	0.21
Great Britain	Scottish	National 1997	0.71	0.16	0.17	0.22	0.16
Spain	Catalan	National 2000	0.69	0.14	0.20	0.21	(0.61)
Spain	Gallician	National 2004	0.76	0.09	0.09	0.13	(0.05)
Great Britain	Welsh	National 1997	0.74	0.09	0.11	0.13	0.10
New Zealand	Maori	National 2002	0.76	0.01	0.01	0.02	0.02
New Zealand	Maori	National 1996	0.63	0.01	0.01	0.01	0.02

Sources: CSES I and II, Election Studies from the Centro de Investigaciones Sociologicas, British Election Studies, Northern Ireland Life & Times Survey 2003; Atz 2007; Bulgarian Post-Election Study 1991; Choice '95 Parliamentary Election in Estonia; Finnish National Election Study 2007; Post-election Survey in Latvia 2002; New Zealand Election Studies.

Notes: ¹less than 100 cases.

It is interesting to see that there are quite a few ethnic party blocs with similar levels of electoral mobilization in the midfield. In these cases, the variance in levels of electoral mobilization is often stronger across time or across the level at which the election takes place than it is across ethnic groups. This can be seen in the case of the Basque minority in Spain, which mobilized between 33 percent and 54 percent of its targeted electorate in the elections under analysis.¹⁴ Only slight variance in levels of electoral mobilization can be found for the Mana Maori in New Zealand, however; this party is able to mobilize only a very small share of its targeted electorate, and remains by far the most unsuccessful ethnic party in our sample.

Having described our data, we can now empirically test our indicators. In order to test how serious problems of ecological fallacy may be for the indicator that relies solely on administrative data (Ind_a1 ), we have performed an analysis of voting behaviour at the individual level. What we are interested in is whether it is primarily the members of the ethnic minorities that vote for ethnic parties, or whether these parties also receive a significant share of their votes from members of other ethnic groups. To this effect, Table 3 gives the vote-share ethnic parties received from members of their targeted ethnic groups, referred to as Index of Party Ethnicity (IPE). In most cases the values on this index show that an overwhelming majority of the ethnic party votes do indeed come from members of their targeted ethnic groups. In 21 out of 31 cases the value for the index of party ethnicity is higher than 90 percent, and in an additional eight cases it is higher than 80 percent. In two cases the index of party ethnicity shows values higher than 70 percent. In these cases, the data allow room for ambiguity as to the ethnic character of the parties’ voters.

Table 3.

Index of party ethnicity (IPE).

Country	Ethnic minority	Election	IPE
New Zealand	Maori	National 2002	100%
Spain	Gallician	National 2004	100%
Spain	Basque	National 2004	100%
Spain	Basque	National 2008	99%
Italy	South Tyroleans	Regional 1998	99%
Italy	South Tyroleans	Regional 2003	99%
Italy	South Tyroleans	Regional 1988	98%
Belgium	Flemish	National 2003	98%
Italy	South Tyroleans	Regional 1993	97%
Spain	Basque	National 2000	97%
Slovakia	Hungarian	National 1998	97%
Spain	Basque	Regional 2001	97%
Spain	Basque	Regional 2005	96%
Canada	Quebecois	National 2004	95%
Spain	Basque	Regional 2009	95%
Romania	Hungarian	National 2004	95%
Great Britain	Scottish	National 1997	94%
Spain	Catalan	Regional 2003	94%
Romania	Hungarian	National 1996	93%
Northern Ireland	Irish	Regional 2003	91%
Finland	Swedish speakers	National 2007	90%
New Zealand	Maori	National 1996	88%
Canada	Quebecois	National 1997	86%
Estonia	Russians	National 1995	85%
Great Britain	Welsh	National 1997	85%
Latvia	Russians	National 1993	84%
Bulgaria	Turkish	National 1991	82%
Bulgaria	Turkish	National 2001	82%
Spain	Catalan	National 1996	81%
Latvia	Russians	National 2002	71%
Spain	Catalan	National 2000	70%

Sources: See Table 2.

Overall, the results concerning the ethnic composition of the selected electorates offer strong support for the notion that these parties can be characterized as primarily ethnic. The fact that some parties at the bottom of the table do not receive an overwhelming majority of their vote from a specific ethnic group might be explained by the inclusion of parties with an ambivalent profile vis-à-vis the targeted electorate. In Catalonia, Convergence and Union (Convergència i Unió) applied a broad understanding of Catalan identity that went beyond ethnic origin, with the aim of attracting voters who had migrated from other parts of Spain (Hierro, 2011). In the case of Latvia, the For Human Rights Party in United Latvia refrained from focusing exclusively on the Russian minority vote, and instead targeted all Slavic ethnic minorities (Ikstens, 2004: 1057).

Despite these exceptions, ecological fallacy does not seem to be a serious problem for the measurement of the electoral mobilization of ethnic parties. This assertion can be further supported by comparing the values achieved for Ind_a1 with those for Ind_a2, since the degree to which variance in party ethnicity triggers the values on Ind_a1 causes its deviation from the values on Ind_a2 (Ind_a2 contains a correction term for ethnic party votes that do not come from the targeted ethnic group). Figure 1 shows that, with the exception of the Russian minority party in Latvia, values for Ind_a1 come very close to those for Ind_a2 . This convincingly suggests that the electoral mobilization of ethnic parties can indeed be measured in situations where it is necessary to rely on aggregate administrative data – so long as the parties being analysed have a clear focus on a single ethnic group. This is good news for scholars who wish to comparatively analyse the electoral mobilization of ethnic parties across a number of ethnic groups and elections. This also shows, however, that one must be very careful in the process of case selection. When relying on aggregate administrative data, one cannot compare the electoral mobilizations of ethnic parties with those of parties which are not clearly ethnic.

In our next step, we compare the values dependant on our absolute and relative measures. In this manner, we assess the importance of the part of the national or regional context that has an impact on turnouts in the given elections. Comparing Ind_r2 with Ind_a2 , we note a surprisingly strong correlation between the two types of measures. While values for Ind_a2 are limited to a range between 0 and 1, values for Ind_r2 cannot be lower than 0 – though they can be higher than 1 if a party mobilizes more strongly among the members of its ethnic community than among others. The strong fit between Ind_r2 and Ind_a2 presented in Figure 2 shows that different turnouts in different countries do not strongly affect the resulting estimations concerning the electoral mobilization of ethnic parties across ethnic groups.

So far our analysis has rested upon the assumption that high quality administrative data are available for all of the cases under consideration. However, while this has been true for our sample it might not always be the case – especially for countries in the developing world. In such instances it might be more meaningful to use only survey data. However, one must still ask: Is the variable Ind_a1 as valid when it is calculated on the basis of administrative data as it is when calculated using survey data? Figure 3 compares the values of Ind_a2 with those of Ind_a1 – this time calculated using survey data only.¹⁵ It clearly shows larger discrepancies for Ind_a1 when it is calculated with survey data, than when it is calculated without (see Figure 1). One important reason for this could be sampling error. An obvious reason for why administrative data are generally superior to survey data comes in the fact that the electoral results are not estimated, but are actual counts. This fact might weigh particularly heavily when a sample is particularly small – which is often the case when subpopulations like ethnic minorities are investigated. Furthermore, by relying on survey data turnout might be overestimated. Since our Ind_a1 measure is sensitive to turnout this might be an additional source of error for the measurement of the electoral mobilization of ethnic parties when survey data are taken. So, we propose this model with the caveat that using surveys as the only data source for the aforementioned calculation of indices of the electoral mobilization of ethnic parties should be approached as a last resort.

Conclusion

Scholars of ethnic parties agree that the mobilization of an ethnic party should be measured as the share of the targeted ethnic group voting for the ethnic party. Its measurement, however, has been a major problem for research on ethnic parties so far. Attempts to measure the electoral mobilization of ethnic parties have typically rested on potentially problematic assumptions or – in the case of Chandra (2004) – on data that are usually not available.

In this article, we have shown how the electoral mobilization of ethnic parties can ideally be measured with the existing data sources. To do so, we demonstrated that more ideal measures estimate the electoral mobilization of ethnic parties with administrative data and a correction term for party ethnicity derived from survey data. We also proposed two ideal measures: an absolute and a relative one. While the absolute measure is independent of every context, the relative measure corrects for national turnouts. The idea behind the relative measure is that the electoral mobilization of ethnic parties might be evaluated within the national electoral context, which generally determines the ease of electoral mobilization in a given country. However, as we have also shown, the resulting value for the electoral mobilization of ethnic parties does not change much depending on the choice of an absolute or relative measure, even if the variance in national turnouts is high.

Surprisingly, our empirical results also showed that a major, potentially problematic assumption of previous measures – ecological fallacy – is hardly violated. Hence, the measure calculated exclusively with administrative data (and therefore without a correction term) is likely a good estimator of our ideal measure of the electoral mobilization of ethnic parties. This, however, seems only to be true if the parties being compared are clear-cut instances of ethnic parties. Furthermore, to measure the electoral mobilization of ethnic parties with administrative data alone, the share of votes coming from a party’s targeted ethnic group (party ethnicity) should not vary over time. If there are good reasons to assume that the party ethnicity of all parties being studied is high at every point in time, an indicator based entirely on administrative data might be used productively for comparative historical analysis. In this manner, the indicator we have derived for the electoral mobilization of ethnic parties provides a tool for making valid comparisons across ethnic parties in different countries at different points in time. In such, it will not only enable scholars to evaluate previous work on the electoral mobilization of ethnic parties – but it might also trigger comparative studies that reach beyond the actual state of the art in this field.

Footnotes

Acknowledgements

For helpful comments we thank the attendees at a presentation on an earlier version of this article at the Polish Public Opinion Research Centre and 5th ECPR General Conference. Potsdam, 10–12 September 2009. In particular we thank Maria José Hierro and Oddbjørn Knutsen, Rafael Leonisio, and Olga Gyarfasova.

Funding

This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors.

Notes

References

Allardt

(1979) Implications of the ethnic revival in modern, industrialized society: A comparative study of the linguistic minorities in Western Europe. Commentationes Scientiarum Socialum 12: 1–81.

Barreto

(2007) Sí Se Puede! Latino candidates and the mobilization of Latino voters. American Political Science Review 101(3): 425–441.

Bartolini

(2000) The Political Mobilisation of the European Left, 1860–1980: The Class Cleavage. Cambridge: Cambridge University Press.

Bartolini

(2005) Restructuring Europe. Centre Formation, System Building, and Political Structuring between the Nation State and the European Union. Oxford: Oxford University Press.

Basedeau

Erdmann

Lay

Stroh

(2011) Ethnicity and party preference in sub-Saharan Africa. Democratization 18(2): 462–489.

Bernauer

Bochsler

(2011) Electoral entry and success of ethnic minority parties in central and eastern Europe: A hierarchical selection model. Electoral Studies 30(4): 738–755.

Bird

Saalfeld

Wüst

(eds) (2009) The Political Representation of Immigrants and Minorities: Voters, Parties and Parliaments in Liberal Democracies. Abingdon: Taylor & Francis.

Birnir

(2007) Ethnicity and Electoral Politics. Cambridge: Cambridge University Press.

Bochsler

(2010) Territory and Electoral Rules in Post-Communist Democracies. Houndsmills: Palgrave.

10.

Brancati

(2006) Decentralization: Fueling the fire or dampening the flames of ethnic conflict and secessionism? International Organization 60: 651–685.

11.

Brown

(2005) Playing the (non)ethnic card: The electoral system and ethnic voting patterns in Malaysia. Ethnopolitics 4(4): 429–445.

12.

Brubaker

(1999) The Manichean myth: Rethinking the distinction between ‘civic’ and ‘ethnic’ nationalism. In: Kriesi

Armingeon

Siegrist

(eds) Nation and National Identity: The European Experience in Perspective. Zürich: Rüegger, 55–71.

13.

Cain

Kiewiet

Uhlaner

(1991) The acquisition of partisanship by Latinos and Asian Americans. American Journal of Political Science 35(2): 390–422.

14.

Camia

Caramani

(2012) Family meetings: Ideological convergence within party families across Europe, 1945–2009. Comparative European Politics 10: 48–85.

15.

Caramani

(2004) The Nationalization of Politics: The Formation of National Electorates and Party Systems in Western Europe. Cambridge: Cambridge University Press.

16.

Cederman

Girardin

Wimmer

(2006) Getting ethnicity right: An expert survey on power distributions among ethnic groups. Paper prepared for presentation at the Annual Convention of the American Political Science Association, Philadelphia, 30 August to 3 September 2006.

17.

Chandra

(2004) Why Ethnic Parties Succeed: Patronage and Ethnic Head Counts in India. Cambridge University Press.

18.

Chandra

(2005) Ethnic parties and democratic stability. Perspectives on Politics 3(2): 235–252.

19.

Chandra

(2006) What is ethnicity and does it matter? Annual Review of Political Science 9: 397–424.

20.

Chandra

(2011) What is an ethnic party? Party Politics 17(2): 151–169.

21.

Chandra

Wilkinson

(2008) Measuring the effect of ‘ethnicity’. Comparative Political Studies 41(4–5): 515–563.

22.

De la Garza

(2004) Latino politics. Annual Review of Political Science 7: 91–123.

23.

De Winter

(1998) Conclusion: A comparative analysis of the electoral, office and policy success of ethnoregionalist parties. In: De Winter

Türsan

(eds) Regionalist Parties in Western Europe. London and New York: Routledge, 204–247.

24.

De Winter

Gomez-Reino Cachafeiro

(2002) European integration and ethnoregionalist parties. Party Politics 8, 483–503.

25.

Diamond

Gunther

(2001) Political Parties and Democracy. Baltimore, MD: Johns Hopkins University Press.

26.

Evans

(1999) The End of Class Politics? Class Voting in Comparative Context. Oxford: Oxford University Press.

27.

Fearon

(2002) Ethnic Structure and Cultural Diversity around the World: A Cross-National Data Set on Ethnic Groups. Unpublished, Stanford University (August).

28.

Ferree

(2006) Explaining South Africa’s racial census. Journal of Politics 68(4): 803–815.

29.

Franklin

(1996) Electoral participation In: Leduc

Niemi

Norris

(eds) Elections and Voting in Global Perspective. London: Sage, 216–235.

30.

Gordin

(2001) The electoral fate of ethnoregionalist parties in Western Europe: A Boolean test of extant explanations. Scandinavian Political Studies 24: 149–170.

31.

Gunther

Diamond

(2000) Species of political parties: A new typology. Party Politics 9: 167–199.

32.

Hechter

Levi

(2000) The comparative analysis of ethnoregional movements. Ethnic and Racial Studies 2: 262–274.

33.

Hierro

(2011) The ascriptive and voluntary elements of the identity in the vote for Convergència y Unió. Unpublished Working Paper.

34.

Holzer

Schwegler

(1998) The Südtiroler Volkspartei: A hegemonic ethnoregionalist. In: De Winter

Türsan

(eds) Regionalist Parties in Western Europe. London and New York: Routledge, 158–173.

35.

Horowitz

(2000[1985]). Ethnic Groups in Conflict. Berkeley and Los Angeles, CA and London: University of California Press.

36.

Huber

(2012) Measuring ethnic voting: Do proportional electoral laws politicize ethnicity? American Journal of Political Science, online first publication: 1–16.

37.

Hug

(2003) Selection bias in comparative research: The case of incomplete data sets. Political Analysis 11(3): 255–274.

38.

Ikstens

(2004) Latvia. European Journal of Political Research 43: 1054–1058.

39.

Ishiyama

Breuning

(2011) What’s in a name? Ethnic party identity and democratic development in post-communist politics. Party Politics 17(2): 223–241.

40.

Ishiyama

(2009) Do ethnic parties promote minority ethnic conflict? Nationalism and Ethnic Politics 15(1): 56–83.

41.

Keating

(2000) The New Regionalism in Western Europe: Territorial Restructuring and Political Change. Cheltenham: Edward Elgar Publishing.

42.

Kidd

Diggs

Farooq

Murray

(2007) Black voters, black candidates, and social issues: Does party identification matter? Social Science Quarterly 88(1): 165–176.

43.

King

(1997) A Solution to the Ecological Inference Problem: Reconstructing Individual Behaviour from Aggregate Data. Princeton, NJ: Princeton University Press.

44.

Knutsen

(2006) Class Voting in Western Europe: A Comparative Longitudinal Study. Mahwah, NJ: Lexington Books.

45.

Kriesi

Grande

Lachat

Dolezal

Bornschier

Frey

(2008) West European Politics in the Age of Globalization. Cambridge: Cambridge University Press.

46.

Leal

Barreto

Lee

de la Garza

(2005) The Latino vote in the 2004 election. PS: Political Science & Politics 38(1): 41–49.

47.

Leonisio

(2012) Parliament on the centre-right, government on the left: Explaining Basque exceptionalism. Regional & Federal Studies 22(1): 45–60.

48.

Lijphart

(1977) Democracy in plural societies: A comparative exploration. New Haven, CT: Yale University Press.

49.

Lijphart

(2004) Constitutional design for divided societies. Journal of Democracy 15(2): 96–109.

50.

Lindberg

Morrison

(2008) Are African voters really ethnic or clientelistic? Survey evidence from Ghana. Political Science Quarterly 123(1): 95–122.

51.

Madrid

(2012) The Rise of Ethnic Politics in Latin America. Cambridge: Cambridge University Press.

52.

McAllister

Kelley

(1983) Changes in the ethnic vote in Australia, 1967–79. Politics 18(1): 98–107.

53.

McLaughlin

(2007) Beyond the racial census: The political salience of ethnolinguistic cleavages in South Africa. Comparative Political Studies 40(4): 435–456.

54.

Messina

(2007) The Logics and Politics of Post-WWII Migration to Western Europe. Cambridge: Cambridge University Press.

55.

Minorities at Risk (MAR) (2009) Codebook Version 2. Downloaded from: www.cidcm.umd.edu/mar/.

56.

Müller-Rommel

(1998) Ethnoregionalistic parties in Western Europe: Theoretical considerations and framework of analysis. In: De Winter

Türsan

(eds) Regionalist Parties in Western Europe. London and New York: Routledge, 17–27.

57.

Oesch

(2008) The changing shape of class voting. European Societies 10(3): 329–355.

58.

Pan

Pfeil

(2003) Die Volksgruppen in Europa: Ein Handbuch. Wien: Braunmüller.

59.

Rabushka

Shepsle

(1972) Politics in Plural Societies: A Theory of Democratic Instability. Columbus, OH: Merrill.

60.

Reilly

(2006) Political engineering and party politics in conflict-prone societies. Democratization 13, 811–827.

61.

Rokkan

(1999) State Formation, Nation-Building, and Mass Politics in Europe: The Theory of Stein Rokkan ( Flora

Kuhnle

Urwin

(eds). Oxford: Oxford University Press.

62.

Rossiter

Johnston

Pattie

(1998) The partisan impacts of non-partisan redistricting: Northern Ireland 1993–95. Transactions of the Institute of British Geographers 23(4): 455–480.

63.

Sartori

(1997) Comparative Constitutional Engineering: An Inquiry into Structures, Incentives, and Ooutcomes. New York: New York University Press.

64.

Smith

(1986) The Ethnic Origins of Nations. Oxford: Blackwell.

65.

Smith

(1991) National Identity. Reno, CA: University of Nevada Press.

66.

Sorens

(2005) The cross-sectional determinants of secessionism in advanced democracies. Comparative Political Studies 38(3): 304–326.

67.

Teney

Jacobs

Rea

Delwit

(2010) Ethnic voting in Brussels: Voting patterns among ethnic minorities in Brussels (Belgium) during the 2006 local elections. Acta Politica 45: 273–297.

68.

Togeby

(1999) Migrants at the polls: An analysis of immigrant and refugee participation in Danish local elections. Journal of Ethnic and Migration Studies 25(4): 665–684.

69.

Tronconi

(2005) Identità etnica e competizione politica: unànalisi del voto ai partiti etnoregionalisti in Europa occidentale. Rivista Italiana di Scienza Politica 35: 77–106.

70.

Tronconi

(2009) I partiti ethnoregionalisti; La politica dell'identita territoriale in Europa occidentale. Bologna: il Mulino.

71.

Urwin

(1980) From ploughshare to ballotbox: The politics of agrarian defence in Europe. Oslo: Universitetsforlaget.

72.

Urwin

(1983) Harbinger, fossil or fleabite? ‘Regionalism’ and the West European party mosaic. In: Daalder

Mair

(eds) Western European Party Systems: Continuity and Change. London: Sage.

73.

van Amersfoort

van der Wusten

(1981) Democratic stability and ethnic parties. Ethnic and Racial Studies 4(4): 476–485.

74.

Van Cott

(2005) From Movements to Parties in Latin America: The Evolution of Ethnic Politics. Cambridge: Cambridge University Press.

75.

Van den Berghe

(1981) The Ethnic Phenomenon. New York: Elsevier.

76.

Weber

(1978[1922]) Economy and Society, vol. 2 ( Roth

Wittich

Fischof

(eds)). Berkeley, CA: University of California Press.

77.

Wimmer

(1997) Who owns the state? Understanding ethnic conflict in post-colonial societies. Nations and Nationalism 3(4): 631–665.

78.

Wüst

(2004) Naturalised citizens as voters: Behaviour and impact. German Politics 13(2): 341–359.