Show Summary Details

Page of

PRINTED FROM the OXFORD RESEARCH ENCYCLOPEDIA, INTERNATIONAL STUDIES (internationalstudies.oxfordre.com). (c) Oxford University Press USA, 2017. All Rights Reserved. Personal use only; commercial use is strictly prohibited. Please see applicable Privacy Policy and Legal Notice (for details see Privacy Policy).

date: 20 February 2018

Methodological Developments in Nationalism, Ethnicity, and Migration Research

Summary and Keywords

In terms of methods, researchers working in nationalism, ethnicity, and migration have used everything from broad historical narratives to automated coding and event data analysis. Traditionally, narratives were the dominant methodological approach implemented to study these areas. The narrative approach allowed for explication of groundbreaking theoretical arguments generating testable hypotheses, the deep inspection of particular areas of the world or particular issues with richness of detail and process, and the investigation of a small number of cases. In addition, the use of formal theory to explore issues related to nationalism, ethnicity, and migration also has a long tradition. Formal theory allows for the construction of concise decision making models that force the researcher to be explicit about key assumptions made regarding preferences and the political structure involved. The formal theory approach has encouraged greater specificity from the arguments formed by scholars of nationalism, ethnicity, and immigration and has generated important theoretical insights. Finally, the most rapidly expanding approach to the study of nationalism, ethnicity, and immigration over the past two decades has been statistics. Statistical analyses offer the advantage of being able to bracket confidence intervals around the causal inferences one makes and to more formally control for a variety of competing factors. As statistical technology and training have become more common, the use of statistics has grown substantially.

Keywords: nationalism, ethnicity, migration, narratives, formal theory, statistics, forced migration

Introduction

The methodological directions of researchers in nationalism, ethnicity, and migration reflect to a large extent the methodological directions of political science as a whole. This should not be surprising given that studies related to the broad areas of nationalism, ethnicity, and migration can be found in the subfields of American politics (Olzak 1992; Oboler 1995; Nagel 1996; Rubenzer 2008), comparative politics (Horowitz 1985; 2003; Tamir 1995; Greenfeld 1996; Wimmer 1997; Varshney 2003), and international relations (Brown 1993; Koslowski 2000; Saideman 2001; Shain and Barth 2003).

In terms of methods, researchers working in these areas have used everything from broad historical narratives (Anderson 1993) to automated coding and event data analysis (Schrodt and Gerner 1994; Shellman and Stewart 2007a; 2007b). Laitin, in his overview of comparative politics, suggests a methodology driven by narratives (in which he includes comparative case studies), formal theory, and statistics (Laitin 2002a). Research in nationalism, ethnicity, and migration has employed all of these methods throughout the years. Recent research has identified important limitations and advantages that different methodologies have in exploring different questions and the need to focus on improving how these methodological approaches are used – and on what data (Horowitz 2008; Van Houten 2008). For this essay, we make a few choices in terms of what to highlight and the various advantages and limitations of research in this area. We first sketch several examples of the application of narrative, formal theory, and statistics within the context of nationalism, ethnicity, and migration. We then focus more closely on two areas. The first is an examination of basic methodological choices related to case selection, selection bias, and the operationalization of concepts related to ethnic conflict and grievance, specifically within the context of the widely used Minorities at Risk (MAR) database (Minorities at Risk Project 2008). Finally, we focus on the area of forced migration, where methodological sophistication has improved in recent years. We should note that given the scope of the literature we make no claims of comprehensibility or identifying the “key” works in the field of nationalism, ethnicity, and migration. Our examples are just that – examples.

Narratives

Traditionally, narratives were the dominant methodological approach implemented to study nationalism, ethnicity, and migration. The narrative approach allowed for explication of groundbreaking theoretical arguments generating testable hypotheses, the deep inspection of particular areas of the world or particular issues with richness of detail and process, and the investigation of a small number of cases (King et al. 1994; George and Bennett 2005).

Adopting a narrative and historical approach, Anderson, in his book Imagined Communities, builds an argument about how modernity and print capitalism helped construct and spread the imagined, socially constructed nature of nations by using a narrative approach to follow the historical spread of nationalist communities across the globe from the New World to Asia and beyond (Anderson 1993). Marx, in his book Making Race and Nation, takes a more comparative approach by examining the histories of Brazil, South Africa, and the United States to argue that, when controlling for other possible explanations, the variance in elite cohesion accounts for the differing development in the construction of national identity and race relations (Marx 1998). Horowitz, in his book Ethnic Groups in Conflict, uses narratives and comparisons across Asia, Africa, and the Caribbean to examine the behaviors of a number of ethnic groups and to argue that differences in ranked and unranked ethnic orders can have important implications for the level of violence in societies (Horowitz 1985).

Narrative research in nationalism, ethnicity, and migration has also begun to use newly developed or expanded approaches like process tracing and qualitative comparative analysis to tackle small N research. Process tracing allows for the “attempt to empirically establish the posited intervening variables and implications that should be true in a case if a particular explanation of that case is true” (Bennett and George 1997:147). Gurowitz has used process tracing to examine “changing policies toward both Koreans and recent migrant workers in Japan” (Gurowitz 1999:415) and what this treatment can tell us about the mobilization of international norms. Using this method, she finds that domestic actors active around the issue of migration turned to international norms when domestic resources were scarce (Gurowitz 1999). Varshney argues, “process-tracing is good for establishing short-run causality” (2002:15). He uses comparative process tracing to compare paired cities and identify the critical element that led some Indian cities toward ethnic conflict and others away from it. In his analysis, civil society connections that cut across ethnic identities appear to be a critical element in preventing the outbreak of large-scale ethnic riots (Varshney 2002).

Researchers have also begun to exploit new developments in qualitative research like the qualitative comparative analysis (QCA) approach (Ragin 1989; 2000), which is meant to allow the researcher to “systematically compare a limited number of cases” (Rihoux and Grimm 2006:2), drawing on Boolean logic (Bennett and Elman 2007). Rubenzer uses a QCA approach to examine what factors make ethnic identity groups in the United States more or less effective at influencing American foreign policy. He finds that organizational strength and activity are the key factors in determining success (Rubenzer 2008). Examining hate crimes in the United States by applying the QCA method, McVeigh et al. (2006) find that resourcefulness, heterogeneity, and funding are key elements in determining the likelihood that hate crimes will be reported.

Formal Theory

The use of formal theory to explore issues related to nationalism, ethnicity, and migration also has a long tradition. Formal theory allows for the construction of concise decision making models that force the researcher to be explicit about key assumptions made regarding preferences and the political structure involved (Amadae and Bueno de Mesquita 1999). Formal theory can also “help to clarify the microfoundations for […] intragroup processes” (Brubaker and Laitin 1998). The formal theory approach has encouraged greater specificity from the arguments formed by scholars of nationalism, ethnicity, and immigration and has generated important theoretical insights.

Posen (1993), borrowing from international relations theory, uses a formal theory approach to study the intrastate ethnic security dilemma. Posen argues that the security dilemma is applicable to the breakup of multiethnic states and that, in the absence of a hegemon, ethnic groups will be compelled to take actions for their own security which make other groups insecure – even if the original actor does not mean to threaten other groups. This internal security dilemma is particularly fierce when security is scarce (Posen 1993). Hollifield, in an attempt to integrate the impact of politics and economics, explicitly sets out to address a theoretical vacuum in the study of international migration in Western Europe. Focusing on international institutions, types of exchange, and authority relations, Hollifield builds a game theoretic model of migration (1992). His findings suggest that the interaction of economics and politics in the context of migration presents critical problems for civil rights and the sovereignty of Western European states (Hollifield 1992).

A growing trend in the use of formal theory is to marry the theory building and clarification of the approach with other methods that can then be used to qualitatively or quantitatively verify the findings of the formal model. One such example is Ethnicity and Electoral Politics by Birnir, where she applies formal theory, case studies, and statistical analysis to examine the relationship between ethnicity and electoral politics. Birnir builds her theory with a formal model approach that argues, contrary to much of the literature, that ethnic parties can act as a stabilizing force in new democracies given particular governmental institutional arrangements. Birnir then supports this argument with evidence from diverging ethnic parties in Spain and from cross-national quantitative analysis (Birnir 2006). Bapat’s research on negotiations between insurgents and governments provides another example of this trend. He applies game theory to argue that there exists a window for negotiations between governments and insurgents. This window occurs between the time when the insurgents have survived long enough for the government to seek noncoercive solutions and the point at which the insurrectionists have grown too powerful to back down. To Bapat, the ethnic nature of the insurgent group is a cohesive factor that is likely to increase the group’s chances of surviving over time. Like Birnir, Bapat proceeds to test the findings of his formal model with quantitative analysis (Bapat 2005).

Statistics

Over the past two decades, the most rapidly expanding approach to the study of nationalism, ethnicity, and immigration has been statistics. Statistical analyses offer the advantage of being able to bracket confidence intervals around the causal inferences one makes and to more formally control for a variety of competing factors (Hechter and Okamoto 2001). As statistical technology and training have become more common, the use of statistics has grown substantially – either as a method practiced by itself or in conjunction with other methods. Another key factor underlying the proliferation of quantitative approaches is the growing availability of numerical data. Some of these data, like the MAR data project, which is thoroughly discussed below (Minorities at Risk Project 2008), have been generated by social scientists. Other expansions result from the increasing availability of datasets produced by foreign governments and institutions. These data have enabled the analysis of everything from the causes of nationalist sentiment (Sekulic et al. 1994; Bollen and Medrano 1998) to the motivations for prejudice against immigrants (Sniderman et al. 2000). In the latter example, Sniderman et al. use the results of computer assisted interviewing to examine if types of difference markers (skin color, language, etc.) influence residents’ attitudes toward immigrants. They find that it is the existence of difference as opposed to the particular type of difference that leads to prejudice in Italy (Sniderman et al. 2000).

In addition to the general upsurge in the statistical investigation of nationalism, ethnicity, and migration, there is also a growing sophistication in the techniques being practiced. For example, current research is more often confronting problems caused by the use of time-series cross-sectional data (Beck and Katz 1995). Salehyan, uses a time-series cross-sectional transition model to examine how international factors, and in particular the presence of refugees, impacts domestic conflict. He finds that, along with weak neighbors and rival neighbors, the existence of refugee diasporas increases the likelihood of a rebellion (Salehyan 2007).

There has also been an increase in automated coding and event data analysis, methods that allow for a much richer treatment of events and actor interactions (Gerner et al. 1994). For example, Schrodt and Gerner examine the impact of mediation efforts on ethnopolitical conflict in the Middle East and the Balkans. Using event data generated by automated coding, they find that mediators who direct conflictual behavior toward both sides and support the weaker antagonist are the most effective at reducing conflict (Schrodt and Gerner 2004). In addition to event analysis pertaining to conflict, other scholars have generated events data for use as independent variables to analyze and predict refugee flows.

The Impact of Basic Methodological Choices: Minorities at Risk and the Impact of Grievance

Minorities at Risk: Need for Data

The ability to apply quantitative methodology is circumscribed by the presence of numerical data, and this was a challenge in the area of ethnic conflict into the early 1990s. Most of the research on ethnic conflict focused on single cases or the comparative analysis of several cases (Gurr 1993b). Recognizing the need for data in order to bring to bear quantitative methodology, which would allow for “models and techniques of empirical conflict analysis to politically mobilized communal groups” (Horowitz 1985:162), Ted R. Gurr and James Scarritt began the MAR project in 1986 to identify groups subject to discrimination and at risk for conflict. The goal of the project was to fill the void that existed between the quantitative analysis that had begun to identify larger patterns related to conflict at the state level and the long tradition of studying ethnic conflict using small N analysis or quantitative analysis looking at data from a small number of well archived countries (for example see Horowitz 1985; Olzak 1992).

Building on previous work (see for example Gurr and Scarritt 1989), Gurr published a book (1993a) and article (1993b) in which he used the new data to do multivariate regressions on the likelihood that a MAR group would protest or rebel. He found that grievances related to autonomy and rights made it more likely that groups would engage in both types of behavior. Since 1993 over 65 books, articles, and book chapters incorporating the dataset compiled by Gurr and other researchers (Scarritt and McMillan 1995; Davis and Moore 1997; Khosla 1999; Saideman 2001; 2002; Scarritt et al. 2001; Saideman et al. 2002; Minorities at Risk Project 2004; Birnir 2006) have been published. (An extensive listing of works using this database can be found at the MAR Project website at www.cidcm.umd.edu/mar/resources.asp#links.) This literature has focused on themes as varied as factors that lead groups to rebel (for example see Gurr 2000; Toft 2003), factors that lead outside parties to intervene in internal conflicts (for example see Saideman 2001), ethnic conflict as a contagion (Fox 2004), and how internal discrimination may impact international behavior (see Caprioli and Trumbore 2003).

Research related to the MAR dataset has become central to a critical debate in the study of nationalism and ethnic conflict – whether or not grievances held by communal groups are key causes of violent ethnic conflict. Much of the work employing the MAR data has supported this claim (Gurr 1993a; 1994; 2000; Regan and Norton 2005; Wimmer and Min 2006). Despite this, the causal and explanatory power of grievance as a source of ethnic conflict remains seriously disputed. David Laitin, in his review of comparative politics for the Political Science: State of the Discipline compendium of the American Political Science Review, levels the critique that:

In the chapter “Why Minorities Rebel” [1993b] Gurr claims that level of group grievances and strength of the group’s sense of identity are the most important independent variables. Yet, oddly, no statistical model provided in the book demonstrates this reported finding. And considerable work using the data set in the wider research community (the second generation users) finds otherwise.

(Laitin 2002a)

Laitin goes on to argue that, contrary to the claims of Gurr (2000) and likeminded scholars, further research has found “[almost] no support in their statistical models for the hypothesis that ethnic discrimination or domination generates state failure” (Laitin 2002a). Laitin points particularly to the problem that MAR groups have been identified by selecting on the dependent variable – that is, groups were chosen because they were mobilized – thus making suspect the findings of research that uses the dataset (Laitin 2002a). Research on ethnic conflict, in particular research related to the issue of group grievance, both within the context of the MAR research project and outside of it, allows us to explore the deep impact that basic methodological choices can have on what researchers identify as key explanatory variables. We will specifically explore how the operationalization of variables and selection bias can result in different and at times contradictory findings.

Selection Bias and the Selection of Groups

Gary Goertz points out in his book Social Science Concepts that concepts and how they are operationalized (what he calls the data/indicator level) have a critical impact on the causal inferences one can make. By determining what is or is not an expression of the concept under investigation or posited as an explanatory variable, the choice of how to operationalize is critical for determining what the researcher will find in his or her analysis (Goertz 2006). In other words, as Adcock and Collier (2002) argue, “In sum, measurement is valid when the scores, derived from a given indicator […], can meaningfully be interpreted in terms of the systematized concept […] that the indicator seeks to operationalize.” The first challenge of creating the MAR dataset was conceptualizing exactly what a Minority at Risk was and operationalizing the concept.

In conceptualizing MAR groups Gurr took a fundamentally constructivist approach that held, “all group identities, both communal and national, are to a degree situational and subject to change” (Gurr 1993b:162; see also Gurr 1993a; 2000 for further discussion). MAR groups are defined as follows:

MAR focuses specifically on ethnopolitical groups, non-state communal groups that have “political significance” in the contemporary world because of their status and political actions. Political significance is determined by the following two criteria:

  • The group collectively suffers, or benefits from, systematic discriminatory treatment vis-a-vis other groups in a society; and,

  • The group is the basis for political mobilization and collective action in defense or promotion of its self-defined interests.

[…] Most communal identity groups also share a common history, or myths of shared experience, that often include their victimization by others. No one of these is essential to group identity. Fundamentally what matters is the belief – by people who share some such traits and by those with whom they interact – that the traits set them apart from others in ways that justify their separate treatment and status.

(Minorities at Risk Project 2008)

In addition, the following criteria apply:

  1. 1 They include groups only in countries with a population (within the year of interest) greater than 500,000;

  2. 2 They include groups only if in the year of interest they numbered at least 100,000 or, if fewer, exceeded 1 percent of the population of at least one country in which they resided;

  3. 3 They include groups separately in each country in which they meet the general criteria. For example, the Kurds are profiled separately in Turkey, Iraq, and Iran;

  4. 4 They include advantaged minorities like the Sunni Arabs of Iraq and the Overseas Chinese of Southeast Asia, but exclude advantaged majorities;

  5. 5 They exclude refugee and immigrant groups unless and until they are regarded by outside observers as permanent residents;

  6. 6 They count and code groups at the highest level within-country level of aggregation that is politically meaningful. For example, all Hispanics in the US are profiled as a single group because they are usually regarded and treated by Anglo-Americans as one collectivity; and,

  7. 7 They estimate membership in a group using the widest demographic definition, even though not all people who nominally are members of a group necessarily identify with it.

(Minorities at Risk Project 2008)

It is important to note that, according to the way the concept has been operationalized by the project, either of the two main criteria can be absent and the group will still be coded as a MAR group (Gurr 2000:7). If we examine, for example, the MAR groups identified by the project in the United States an important potential problem quickly emerges. Only the following four groups are identified: African Americans, Hispanics, Native Americans, and Native Hawaiians. While all of these groups clearly fit the criteria mentioned, there are obviously other groups that fit the criteria – especially the second criteria of political mobilization – as well but are not included.

In recent research on how much ethnic groups are able to impact foreign policy, Trevor Rubenzer adopts a qualitative comparative analysis to examine 10 ethnic groups for which extensive academic evidence suggests that identity is “the basis for political mobilization and collective action in defense or promotion of [their] self-defined interests” (Minorities at Risk Project 2008). Of these groups, four have been included and coded under the broadly defined MAR criteria, but six (Arab, Armenian, East European, Greek, Israeli, and Polish) have not (Rubenzer 2008). Thus groups that appear to fit the criteria are simply not being coded by MAR.

In addition to the problem that groups which warrant inclusion often fail to make the list, there exists the further problem of potential selection bias. This predicament is created by the dependent variable of interest being a main driver in the very definition of MAR groups (Laitin 2002b). The problem is compounded by the way MAR groups are operationalized because “the sample does not include the large number of ethnically defined groups that are small or that are not already marked by factors that might increase their odds of being engaged in violent conflict” (Fearon and Laitin 1999). Fearon and Laitin point out that this problem is not one confined to research using the MAR data. They argue that selection bias is a broad problem that extends across the study of ethnic conflict:

A great many analysts of interethnic relations, including ourselves, agree that ethnic tensions are pervasive and commonplace. The standard view in political science, however, goes farther by suggesting that ethnic violence and active conflict also are ubiquitous. Violence is assumed to follow ethnic tensions as night follows day. […] Donald Horowitz [1985:xi] writes: “By one reckoning, ethnic violence since World War II has claimed more than ten million lives, and in the last two decades ethnic conflict has become especially widespread.”

This widely accepted view seems to be based on a biased selection of cases. Scholars have focused their attention overwhelmingly on cases of significant ethnic violence – they “select on the dependent variable.”

(Fearon and Laitin 1996:716)

This problem, though, is particularly relevant to the MAR data because of MAR’s selection criteria (one of which is mobilization, and violent groups are going to attract more attention than their nonviolent counterparts; Hug 2003). The problem is made more severe in some geographic areas than in others because of the selection criteria. For example, 14 countries are not coded as having a MAR group and when data was analyzed correcting for these missing countries the results were substantially different (Birnir and Wilkenfeld 2007).

We should note though that interesting work done by Beger – while acknowledging the selection bias problem – points out that it may not invalidate the causal inferences made from the data in the way that Fearon and Laitin suggest because the particular mechanics of the MAR selection bias problem mean that:

empirical analyses are more likely to reject hypothesis that are empirically supported in the full population. This implies that any statistical relationships that do occur in the MAR sample of ethnic groups should be generalizable to the full population of ethnic or minority groups in the world. Of course it also implies that there may be some relationships that exist in the full population but fail to meet standard hypothesis tests within the MAR sample.

(Beger 2008:8–9)

Given the importance of having a reliable basic dataset on ethnicity and nationalism not beset by selection bias issues and given that the MAR dataset is the only largescale data of this kind (Birnir and Wilkenfeld 2007), two efforts have focused on addressing the present complications (Laitin and Gurr 1999; Birnir and Wilkenfeld 2007). Birnir and Wilkenfeld’s proposal plans to address the problem in several ways. First, as a preliminary measure, it plans to sample groups not in the dataset to identify the likelihood of mobilization for all groups – not just those selected through the MAR criteria – and “to explain the domain of the current MAR data and the nature, extent, and consequences of selection bias and omissions with respect to the universe of all minorities” (Birnir and Wilkenfeld 2007). The project then plans to identify a comprehensive list of cases that does not select on the dependent variable. Once this is done the project plans to identify a streamlined set of variables and back code them for a sample set of the new groups which can be used to deal with the problem of selection bias (Birnir and Wilkenfeld 2007).

The Problem of Operationalization and Proxies

The ethnic conflict literature is strongly divided over the impact of grievance on the likelihood of conflict, which is one of the reasons why resolving the selection bias problems in MAR are so important. The study of grievance, though, suffers from another problem that has, at its core, the basic methodological choice of how to operationalize or proxy grievance. Collier and Hoeffler argue that economic concerns and resource conditions are the keys to explaining the onset of civil wars. They find that while elements of ethnic dominance may increase the likelihood of civil war, grievance as a whole does not (Collier and Hoeffler 2004). Fearon and Laitin (2003) also find that ethnic grievance is not important. That is, none of the proxies for grievance that they employ indicate significance. However, the crucial difference between Fearon and Laitin (2003) and Collier and Hoeffler (2004) and scholars who believe grievance to be an important element of conflict are the proxies each uses to measure grievance (Gurr 1993a; 1994; 2000; Regan and Norton 2005; Wimmer and Min 2006).

Collier and Hoeffler (2004) “consider four objective measures of grievance: ethnic or religious hatred, political repression, political exclusion, and economic inequality.” They measure these four criteria of grievance in different ways. They measure hatred through fractionalization and political repression using the Polity III dataset. Political exclusion is measured using the concept of “ethnic dominance”:

Even in democracies a small group may fear permanent exclusion. A potentially important instance is if political allegiance is based on ethnicity and one ethnic group has a majority. The incentive to exploit the minority increases the larger is the minority, since there is more to extract. Hence, a minority may be most vulnerable if the largest ethnic group constitutes a small majority. We term this ethnic dominance. […] we define it as occurring if the largest ethnic group constitutes 45–90% of the population. On this definition it does not appear important: it is as common in peace episodes as in conflict episodes.

(Collier and Hoeffler 2004)

Finally, asset inequality is measured by the Gini coefficient of land ownership.

Fearon and Laitin estimate grievance in a similar fashion. They argue that grievances are hard to measure “but measures of average levels of discrimination are feasible” (Fearon and Laitin 2003:79). Along with Collier and Hoeffler (2004), they cite the quality of democracy, as measured by the Polity score, as a potential proxy for repression. And in an attempt to scrutinize the sources of ethnic nationalism and the possibility of discrimination, they too include a variety of ethno-fractionalization measures.

The problem with their findings is that these proxies are problematic measurements of ethnic grievance. None of the proxies used to counter the grievance argument, as laid out by Gurr (2000) and others, captures its key elements – that invidious distinctions between one identifiable group and another can help push groups to violence. Inequality by itself, regime type, and ethnic difference are not the same as ethnic discrimination because those indicators may be generic throughout the country and are not necessarily targeted against a particular identity that can serve as a basis for mobilization. We should note also that almost all the works that refute grievance not only proxy it in problematic ways (which we discuss below), but do so at the state level, which further muddies the ability of these proxies to effectively explain group action.

For example, asset inequality does measure grievance, but it does not tell us about ethnic grievance unless we can assess whether the inequality falls along ethnic lines. Indeed, the dynamics of grievance and violence may be very different when discrimination is along lines of wealth rather than lines of ethnicity. Some societies, the United States being a notable example, have a remarkable tolerance for classism even when they outlaw racism (Schuck 2003:213–14). As Sambanis points out:

One possible explanation for the non-significance of inequality in civil war cross country regressions is that most authors are using the wrong measure of inequality. First, the Gini index does not measure changes in the distribution of income across groups. […] While inter-personal economic inequality as measured by the Gini coefficient might be a relevant indicator to measure the average person’s proneness to join a class-based revolution against a government with failed redistributive policies, a more appropriate measure of inequality to analyze the onset of secessionist civil war may be inter-regional inequality (i.e. differences in average incomes across regions). Some regions are marginalized and poor because the state has historically not attempted to develop them and bring them closer to the center (e.g. Chiapas in Mexico, Azawad in Mali), or because of a history of resistance to integration (e.g. Chechnya in Russia).

(Sambanis 2004:10–11).

Likewise, gauging grievance by measuring democracy has its own set of difficulties. Fearon and Laitin argue, “Other things being equal, political democracy should be associated with less discrimination and repression along cultural or other lines, since democracy endows citizens with a political power (the vote) they do not have in dictatorships” (2003:79). But other things are not necessarily equal. While democracy may be associated with less discrimination, its presence does not ensure an absence of governmental discrimination. The state-sanctioned discrimination of certain ethnic groups may target the very rights by which people access the political process. The particular construction of two of the main datasets used to operationalize democracy exacerbates this oversight. The Polity data (Marshall and Jaggers 2004) catalogs the United States in 1809 as the world’s first democracy – a full 156 years before one might argue its regime became truly inclusive. In addition, many of the historic democracies held vast colonial empires built on political discrimination that favored residents of the metropole. This is not captured at all in Polity (Marshall and Jaggers 2004). Using Freedom House to proxy political discrimination is problematic as well (Freedom House 2006). Despite the fact that many European countries have regularly discriminated against their Roma populations both economically and politically (Fox 2001), these countries are still coded as free by Freedom House (Freedom House 2006). As a recent example, in 2008 the Italian government specifically targeted Roma for fingerprinting, a move that was subsequently condemned by the European Parliament as a “clear act of racial discrimination” (Sliva et al. 2008).

While there is growing evidence that democracy may alleviate human rights abuses (Davenport 1999; Davenport and Armstrong 2004), a general pattern of human rights abuses is not the same thing as targeted discrimination. It is unsafe to assume that democratic gains will have the same effect on targeted discrimination – the impact on general repression may simply differ from that of targeted repression. Indeed, some researchers have found an “inverted U” relationship between repression and regime type (Muller and Weede 1990; Hegre et al. 2001; Regan and Henderson 2002 #1412). This suggests a complicated relationship between conflict and the institution of democracy that is likely to involve manifold institutions and grievances unrelated to specific issues of ethnic grievance. In addition, we should note that procedural democracy is an institution and not a policy or a behavior. While there is evidence that institutional arrangements can ameliorate the problems caused by ethnic fractionalization (Easterly 2001), these institutional arrangements are not necessarily the same thing as policies of discrimination. Procedural democracy carries a much larger load of possible causality based on institutional factors that, given the way Polity is measured, are much more likely to account for any possible correlation than any level of discrimination Polity may capture.

While recent evidence suggests that certain types of ethnic fractionalization may indeed lead to civil war (Reynal-Querol 2002), this should not be taken as a confirmation of the importance of discrimination and resulting grievance any more than should the opposing findings that ethnic fractionalization does not matter for civil war onset (Fearon and Laitin 2003). Difference alone is not necessarily invidious.

Those who do not use proxies but try to directly operationalize government discrimination, not surprisingly, have different findings than the research that has used democracy, ethnic fractionalization, and general levels of repression and human rights abuses. Regan and Norton, using the group as the level of analysis, find that both discrimination and ethnic fractionalization contribute to the likelihood that a MAR group will rebel. They find that when “a state is nondiscriminatory, the probability of observing rebellious activity drops by 69 percent to only 20 percent, holding all else at the baseline conditions” (Regan and Norton 2005). Wimmer and Min examine the question from a state level of analysis by creating a variable that identifies what percentage of a state’s population are MAR groups suffering from governmental political discrimination (making the assumption that non-MAR groups are unlikely to be discriminated against). Like Regan and Norton, they find that discrimination is positively and significantly related to the onset of civil wars (Wimmer and Min 2006).

The fact that cause of grievance, when measured directly through discrimination rather than proxied by more distant measures, is found to be positively and significantly related to violence lends credence to the grievance argument. However, despite the work of Beger (2008:8–9), the problems with the MAR data that form the basis for these findings still raise questions about the validity of the grievance argument. The problems of selection bias, operationalization, and selecting ambiguous proxies underline the importance of basic methodological choices and how they impact what causal inferences can be made. The efforts of a combined set of scholars to rectify the various selection problems that MAR suffers from (Birnir and Wilkenfeld 2007) hold the promise enabling the ethnic conflict research community to analyze the impact of grievance on conflict in a more rigorous and productive manner in the near future.

A Closer Look at Forced Migration

In this section we dive into the forced migration literature and examine the methodological improvements within this budding literature.

Forced Migration Data and Methods

Ideographic studies, such as descriptive case studies, advocacy and awareness pieces, and policy evaluations comprised primarily of books and monographs dominated the literature on forced migration until the mid to late 1990s. However, with agencies like the United Nations High Commissioner on Refugees (UNHCR) and the United States Committee on Refugees and Immigrants (USCRI) spearheading movements to collect data more efficiently and effectively, both large N and small N quantitative studies have emerged of late (for example see Stanley 1987; Gibney et al. 1996; Apodaca 1998; Faist 2000; Davenport et al. 2003; Moore and Shellman 2004; 2006; 2007; Neumayer 2004; 2005a; Shellman and Stewart 2007a; 2007b). More recently, entire special issues in journals such as Conflict Management and Peace Science (2007, vol. 24, no. 2) and Civil Wars (2007, vol. 9, no. 2) have published several quantitative articles on the topic. Below we review the major studies invoking large N and small N quantitative analyses of forced migration, highlighting the new methods, data, and modeling techniques that have emerged in the field.

Large N Studies

The early stages of the empirical analyses in this literature were not particularly strong, but the field has improved over time. Specifically, several of the studies suffer from selection bias as they only examine countries that produced refugees (e.g. Zolberg et al. 1989; Apodaca 1998). The two earliest comparative studies with strong empirics were Schmeidl (1997) and Davenport et al. (2003). Schmeidl focuses on refugee stocks in the third world from 1971 to 1990, whereas Davenport et al. focus on net forced migrant flows throughout the world from 1964 to 1989.

Schmeidl (2000:152) argues that “refugees and IDPs [internally-displaced persons] flee from similar root causes rather than responding to completely different occurrences.” Following work by Clark (1989), she contends that root causes, proximate conditions, and intervening factors influence forced migration. This structural approach to the problem fails to connect the behavior of actors (i.e. governments and dissident groups) to the behavior of other actors (i.e. forced migrants). Such a structural approach does not focus on the conflict processes that often cause individuals to flee their homes. Instead, countries are implicitly conceptualized as Eastonian (1965) systems in which inputs (violence, etc.) produce outputs (i.e. forced migrants).

Davenport et al. (2003) advance an alternative approach focusing on the choices of individual human beings. They argue that leaving one’s home is a choice people make and note that while many people leave in any given episode of forced migration, others choose to stay. They essentially argue that individuals abandon their homes when they fear for their liberty, physical person, and/or lives.

Davenport et al. (2003) also emphasize the competition for power by various actors in society. They argue that government and dissident (broadly conceptualized as rebels, separatists, terrorists, etc.) actions are the main source of threat to society at large. As a result, they contend that statistical models of forced migration should specify what behavior people will find threatening. Moore and Shellman (2004; 2006; 2007) expand this theoretical framework in their three studies on forced migrants, IDPs, and refugees, as do Neumayer (2004; 2005a; 2005b) and Salehyan and Rosenblum (2008) with their work on asylum seekers and asylum recognition.

Lately, others have improved the data quality, operationalized independent variables differently, examined more nuanced questions, and used new modeling techniques. For example, Moore and Shellman (2004) examined global flows of forced migrants from 1964 to 1995 using a zero-inflated negative binomial (ZINB) model, recognizing that the data were counts and inflated with two types of zeros. The negative binomial model allows one to estimate the magnitude of the flows that are produced as well as a parameter that represents the extent to which the events influence one another within each observation. The zero-inflated part of the model allows one to estimate the effects of the covariates on the probability a country produces zero refugees. See King (1989:764–9) for a detailed explanation of why the negative binomial model is useful for this sort of argument. The ZINB model assumes that some of the cases at risk for producing a positive count may produce a count of 0.

Other research has tackled the selection problem by modeling the selection processes at work, as well as distinguishing among refugees and IDPs. For example, Moore and Shellman (2006) tackle this problem by estimating two separate equations. First, they estimate a selection equation, which distinguishes those country-years that produced forced migration flows from those that did not. The second equation accounts for the proportion of forced migrants who sought refuge abroad, distinguishing the processes that generate refugees from those that generate IDPs. One of the significant findings is that while international wars and genocide tend to push people out and create refugee flows, civil wars in which the rebels and the state vie for the support of the local population tend to generate more IDPs relative to refugees.

Finally, Moore and Shellman (2007) combine a selection model and a ZINB model to examine the destination choices of refugees. As in their 2006 study, they employ a selection equation to model countries that produce refugees. In the second stage, they model where refugees go conditional on the premise that the country has produced positive flows. Rather than examine only contiguous dyads, their study incorporates all potential destination choices and models the variance across such choices using variables such as distance, contiguity, terrain, regime type, measures of the economy, and violent activities by the state and the dissidents. Given that all destination choices are included, the data are inflated with zeros. The ZINB is a useful statistical model for studying situations with a sample of two groups with distinct risks of experiencing a non-zero count and an integer count for the subgroup with greater risk. This study also examines monadic, dyadic, and directed-dyadic variables and takes into consideration the effects of spatial-heteroskedasticity (see Gleditsch and Ward 2001; Beck et al. 2006).

The Moore and Shellman (2007) study is the first large N empirical analysis of global directed-dyadic refugee flows. They find that while the violent behavior of states and dissidents, interactions between states and dissidents (civil wars), and the presence of foreign soldiers in the country of origin push people to seek refuge abroad, refugees do not discriminate among destinations on the basis of the behavior of these actors in (potential) countries of asylum. In contrast, they found that refugees were often produced in bad neighborhoods and often sought refuge in bordering countries experiencing civil war or at war with their origin country. The only exception was the presence of genocidal activity in potential asylum countries. Further, they found little support for the hypothesis that institutions that protect freedoms and economic opportunities play roles in destination choices. Culture and costs, on the other hand, played significant roles. Diaspora population had the largest substantive impact, and transaction costs, especially as measured by the presence of a border, had large substantive effects as well. Ultimately, Moore and Shellman find that people flee violence and that when they flee they have a strong tendency to go (1) where others have gone before them, and (2) to the nearest location where they can avoid high-level state-sponsored violence.

Other studies examine the geopolitics of forced migration. Iqbal (2007) illustrates how distance not only directly affects refugee flows, but also how it mediates the influence of conflict on refugees. To do so, her work employs a gravity model of refugee flows in Africa from 1992 to 2001 and uses newer conflict data from the Armed Conflict Dataset collected by the Peace Research Institute of Oslo.

Still other studies have reversed the causal arrow. Salehyan and Gleditsch (2006) argue that population movements are an important mechanism by which conflict spreads across regions. Specifically they contend that the presence of refugees and displaced populations can increase the risk of subsequent conflict in host and origin countries. They argue that such flows “may facilitate the transnational spread of arms, combatants, and ideologies conducive to conflict; they alter the ethnic composition of the state; and they can exacerbate economic competition” (Salehyan and Gleditsch 2006:335). Their analyses show that cross-border refugee flows are associated with a heightened risk of civil war in the country of asylum. This finding is consistent with Weiner’s (1996) contention that “bad neighborhoods” (e.g. those with high levels of civil war) tend to produce refugee flows.

Finally, Rubin and Moore (2007) take statistical modeling to the next step in that they use country-year data in the previous time period to examine whether or not such data can be useful for predicting the probability of forced migration in the following country-year. Their study shows that civil war, forced migration, and human rights violations in the previous time period are candidate “risk factors” for forced migration in the following country-year. Surprisingly, genocide is not such a candidate. The authors argue that genocide is not a good risk factor because, given the high death rates that are associated with genocidal episodes, there are fewer people who can flee the following year.

In sum, within the past decade the scholarly community has made great gains in systematically empirically modeling forced migration across a large number of countries and years. However, as Moore and Shellman (2004; 2006; 2007) point out, global sample pooled cross-sectional time-series studies produce average effects across space and time. While average effects are important, policy makers have understandably limited interest in such findings. That said, an understanding of the destination choices of forced migrants has policy implications: contingency planning could be well served by a model capable of producing serviceable out-of-sample forecasts. Yet global and large N analyses like the ones discussed above are poor candidates for such a model. With this point in mind, we now venture into discussing time-series case studies and the utility they can lend, as well as the improvements over time in data quality and modeling techniques.

Small N Quantitative Studies

While large N studies can provide information on the general patterns of forced migration, they can often mask the details of the migration process. Quantitative case studies can illuminate nuanced details lost in large N analyses. Moreover, systematic empirical case studies complement other global studies and serve as opportunities to evaluate the applicability of general global models to particular cases in clearly structured systematic ways.

Though there are obvious benefits to such studies, relatively few have been published to date. Morrison (1993) focuses on IDPs in Guatemala and looks at migration in response to both economic and political violence variables. He finds that economics, measured as per capita government tax receipts and spending, has greater positive effects on migration than violence, measured by politically motivated killings and per capita corpses found, but violence is still an important factor in determining IDP flows. His later study with May (Morrison and May 1994) uses a similar model to explain migration but includes unemployment figures in the model. Neither study really factors in time as the data are aggregated over a large period of time and the migration data come from the 1981 census. Arguably the most important element – time – is lacking from Morrison’s studies.

Longitudinal designs can capture “an empirically rich dynamic underlying the process tendencies” (Wood 1988:229). Dividing the temporal processes at work into smaller units can provide a closer look at causal mechanisms associated with forced migration. Moreover, time-series studies often grant the ability to perform quasi-experiments (Campbell and Stanley 1966; Wood 1988), which can trace the impacts of particular events on migration over time. Using such methods, we can more readily detect the sequences, magnitudes, and durations of key events – like policy shifts, coups d’état, and economic sanctions – on the migratory process. Time-series case studies provide more detail to the process of population displacement in that they focus on smaller units of time for a particular case (months and weeks), but suffer from the inability to generalize to additional cases and other limitations associated with case study research. Nevertheless, similar conceptual variables appear in both the cross-sectional studies and the case studies, and the results are similar across the designs. Though there are some conflicting results across the studies, on the whole, the results are consistent with the large N studies, suggesting that violence, economics, and cultural networks explain variations in forced migration.

Stanley (1987) charters this approach by analyzing migration to the US from El Salvador and Shellman and Stewart (2007a) follow by analyzing Haitian flight to the US. Stanley’s (1987) study shows that economic underperformance does not impact migration from El Salvador to the US. However, his measure is not a direct measure of the economic situation. Instead, he uses a counter variable to proxy the steady decline of the economy, which he observes using annual data. This is certainly not an optimal choice. The main problem with both the Stanley and Morrison studies is the dearth of available data across time on violence, economics, and migration.

While Shellman and Stewart (2007a) make many improvements with regard to data and temporal disaggregation, they are still left with a suboptimal measure of migration. Shellman and Stewart (2007a) focus on weekly migration flows from Haiti to the United States. While this is a very fine-grained study of migration, they use weekly interdiction counts by the US Coast Guard as a proxy of refugee flows. To demonstrate the indicator’s construct and external validity, they correlated the annual sums of interdictions with the available Moore and Shellman (2004) annual measure of Haiti to US refugee flows (obtained from the UNHCR) and found a 0.67 correlation between the two annually aggregated series. They then pointed out the limitations of the indicator in that it only captures those individuals who are caught trying to enter the US and ignores those who successfully enter the US illegally. It also only captures those individuals traveling to the US by boat (however, boats are the dominant form of transportation) and ignores individuals applying for refugee and asylum status in the US “in-country” office located in Port-Au-Prince. That said, given its relatively strong correlation to annual refugee flows, they contend that it serves as a good indicator of weekly migratory flows from Haiti to the US.

Shellman and Stewart are also able to incorporate fine-grained measures of the economy, violence, and foreign pressures into their time-series analysis. First, they include the Haitian monthly consumer price index (CPI) from the International Labour Organization in their model. Moreover, they measure monthly wages and CPI in the US. They also examine the impact of US economic sanctions over time. Finally, they are able to examine the impacts of US foreign policy, rebel violence, and government violence measured at the weekly level on Haitian migration to the US. These measures are constructed using events data – “day-by-day coded accounts of who did what to whom as reported in the open press” – which, according to Goldstein (1992:369), offer the most detailed record of interactions between and among actors. The data come from the KEDS project and Project Civil Strife, which are both automated events data coding projects. Both projects use Textual Analysis By Augmented Replacement Instructions (TABARI), developed by Phillip Schrodt, to generate political event data. (See http:/web.ku.edu/keds/index.html for information on the KEDS and TABARI projects.) TABARI uses a “sparse-parsing” technique to extract the subject, verb, and object from a sentence and performs pattern matching using actor and verb dictionaries. In short, TABARI matches words from an electronic text file (news story) to words contained in the actor and verb dictionaries, assigns a corresponding code to each actor and verb, and, finally, tags the event with a date. These particular data are coded from Associated Press reports available from Lexis-Nexis. The authors then create variables that reflect conflict event counts by Haitian rebels, the Haitian government, and the US government toward Haiti. In addition to the weekly event counts, the authors also examined the impacts of particular coups and hurricanes.

Shellman and Stewart (2007a) find that the Haitian economy tends to push people out as it worsens. Yet they failed to find evidence that short-run changes in the US economy attract Haitians. In contrast, they found that rebel violence has a large, substantive, and statistically significant impact on Haitian migration. Finally, economic sanctions, foreign pressures, and hurricanes like Georges also tended to increase migratory flows. This study is the most sophisticated longitudinal study to date given its focus on weeks, conflict processes between rebels and governments, the impact of foreign policy, and its ability to include direct measures of the economy that vary over time. Still, its weakness resides in a less than desirable dependent variable.

Such time-series case studies open the possibility for the construction of early warning models. The UNHCR Handbook for Emergencies defines early warning as “the collection, analysis, and use of information in order to better understand the current situation as well as likely future events. The particular focus is on events which might lead to population displacement” (United Nations High Commissioner for Refugees 2000). Schmeidl and Jenkins (1998, 36) contend that “improved analysis of temporal processes, automated event data development, the integration of case study, and quantitative methods, and greater clarity about units of analysis should create the capacity to provide timely and policy-relevant information.” Shellman and Stewart (2007b) take this charge seriously and attempt to incorporate all of these items in their early warning modeling approach by quantitatively analyzing weekly processes using automated event data generated for a particular case.

Shellman and Stewart (2007b) build on their former study (2007a) using the same data but adding an important twist. They first model each of the independent variables (Haitian CPI, US CPI and wages, US foreign policy, Haitian rebel violence, and Haitian government violence) in separate equations, use those models to forecast the economic and security variables that affect forced migration, and then use the predicted values to predict Haitian flight to the US (United Nations High Commissioner for Refugees 2000). They assert that in order to predict Haitian migration to the US, it is necessary to predict the variables that affect those flows (e.g. increases in violent behavior). Therefore they first develop a model that predicts the risk factors for migration and use those predictions to predict migration itself. Such a multi-equation modeling approach opens doors for additional early warning models to be developed along the same lines.

In sum, while there are few quantitative time-series cases studies focusing on refugee flows, the analyses have gone from aggregate models (e.g. Morrison 1993) to weekly forecasting models (Shellman and Stewart 2007b) in a relatively short time period. With new automated event data extraction tools and better data produced from the UNHCR in monthly and quarterly time periods, we hope future scholars will make additional contributions in this domain of inquiry.

Ways To Move Ahead

Studies of nationalism, ethnicity, and migration as well as forced migration have benefited from new statistical techniques and more widely available data on cogent variables. However, statistical methodology has not simply replaced the older forms of narrative and theoretical analysis. On the contrary, the new data and techniques have made it possible to create hybrid arguments that blend formal theory, narrative evidence, and quantitative analysis to illuminate root causes from a multiplicity of angles. Birnir’s (2006) aforementioned work on ethnicity and electoral politics is one example of this type of varied approach.

Yet statistical methods are dependent upon the quality and availability of data, and methodology when it comes to data collection, operationalization, and application is crucial. As the old adage goes, “computers don’t make mistakes, people do.” The tools of statistical analysis are only as good as their inputs. Producing and identifying sound data sources should be a high priority in the field. Additionally, choosing the proper variables to approximate ideals must be done with great care.

And yet as the availability of data, its quality, and the statistical methods available continue to expand, so too do the opportunities to intertwine the study of ethnicity, nationalism, and migration with that of forced migration. Perhaps studies such as Sniderman et al. (2000), which identifies the differences that will result in antiimmigrant discrimination within certain countries, might be combined with more dynamic models that predict the destination of refugee groups, as in Moore and Shellman’s (2007) work. A great number of possible combinations with great potential exist.

Additionally, more time-series case studies will help ascertain the dynamics of ethnic conflict and cooperation and potential resulting migratory flows. Time-series work allows researchers to track sequences, magnitudes, and durations of key events and examine how they affect one another. Small N studies offer rich details on the micro-processes at work, precisely because they focus on small amalgamations of time (weeks or months). The future abilities and power of time-series methods, however, are especially driven by newer and better disaggregated data becoming available. At present, researchers readily admit to using whatever they can get their hands on – like Shellman and Stewart’s (2007a) undesirable dependent variable: Coast Guard interdictions.

This methodological arena, and particularly this subject matter, would greatly benefit from a heightened geospatial component. Take for example Sambanis’s (2004) work on why Gini coefficients often fail to predict conflict. He finds that a far greater indicator in predicting civil war is a measure of interregional inequality, which can signify that a region has been marginalized by the state. Measures such as these, however, are not often available. In specific application to the forced migration literature, information on regional displacement origins by time might be used to scrutinize the individual decision to flee. Furthermore, greater geospatial information would provide invaluable information on IDPs, who are not counted in refugee flows. Along this vein, advances in, partnerships with, and diffusion from the field of geographic information systems (GIS) will surely inspire heretofore unknown types of analyses.

We specifically call for the need to disaggregate data collection across space and time. Presently available datasets have provided a springboard for the first generation of disaggregated research techniques, but newer and more complete sets will be required if the full potential of such techniques is to be realized. This void is beginning to be filled with more sophisticated automated coding technology, which is enabling disaggregation to the daily and city level. New disaggregated data should encourage the development of still more powerful methodology as researchers attempt to seek out all that they can from them.

When combined, the efforts of sound data collection and methodological advancement might lay the groundwork for fully functioning early warning, forecasting, and contingency planning models. Such tools would prove an invaluable asset to policy makers at the national and multinational levels.

References

Adcock, R., and Collier, D. (2002) Measurement Validity: A Shared Standard for Qualitative and Quantitative Research. American Political Science Review 95 (3), 529–46.Find this resource:

Amadae, S.M., and Bueno de Mesquita, B. (1999) The Rochester School: The Origins of Positive Political Theory. Annual Reviews in Political Science 2 (1), 269–95.Find this resource:

Anderson, B. (1993) Imagined Communities. London: Verso.Find this resource:

Apodaca, C. (1998) Human Rights Abuses: Precursor to Refugee Flight? Journal of Refugee Studies 11 (1), 80–93.Find this resource:

Bapat, N.A. (2005) Insurgency and the Opening of Peace Processes. Journal of Peace Research 42 (6), 699–717.Find this resource:

Beck, N., and Katz, J.N. (1995) What To Do (and Not To Do) with Time-Series Cross-Section Data. American Political Science Review 89 (3), 634–47.Find this resource:

Beck, N., Gleditsch, K.S., and Beardsley, K. (2006) Space Is More than Geography: Using Spatial Econometrics in the Study of Political Economy. International Studies Quarterly 50, 27–44.Find this resource:

Beger, A. (2008) Simulating the Effects of Selection Bias in the Minorities at Risk Project, Tallahassee, Florida State University.Find this resource:

Bennett, A., and Elman, C. (2007) Qualitative Methods: The View from the Subfields. Comparative Political Studies 40 (2), 111–21.Find this resource:

Bennett, A., and George, A.L. (1997) Process Tracing in Case Study Research. Paper presented at the MacArthur Foundation Workshop on Case Study Methods, Belfer Center for Science and International Affairs (BCSIA), Harvard University, Oct. 17–19.Find this resource:

Birnir, J.K. (2006) Ethnicity and Electoral Politics. Cambridge: Cambridge University Press.Find this resource:

Birnir, J., and Wilkenfeld, J. (2007) Minorities at Risk: Addressing Selection Bias Issues and Group Inclusion Criteria for Ethnopolitical Research, NSF grant proposal, University of Maryland.Find this resource:

Bollen, K., and Medrano, J.D. (1998) Who are the Spaniards? Nationalism and Identification in Spain. Social Forces 77 (2), 587–621.Find this resource:

Brown, M.E. (1993) Ethnic Conflict and International Security. Princeton: Princeton University Press.Find this resource:

Brubaker, R., and Laitin, D.D. (1998) Ethnic and Nationalist Violence. Annual Review of Sociology 24 (1), 423–52.Find this resource:

Campbell, D.T., and Stanley, J.C. (1966) Experimental and Quasi-Experimental Designs for Research. Chicago: Rand McNally.Find this resource:

Caprioli, M., and Trumbore, P.F. (2003) Ethnic Discrimination and Interstate Violence: Testing the International Impact of Domestic Behavior. Journal of Peace Research 40 (1), 5–23.Find this resource:

Clark, L. (1989) Early Warning of Refugee Flows. Washington: Refugee Policy Group.Find this resource:

Collier, P., and Hoeffler, A. (2004) Greed and Grievance in Civil Wars. Oxford Economic Papers 56 (4), 563–95. At http:/oep.oxfordjournals.org/cgi/content/full/56/4/563, accessed Mar. 10, 2009.Find this resource:

Davenport, C. (1999) Human Rights and the Democratic Proposition. Journal of Conflict Resolution 43 (1), 92–116.Find this resource:

Davenport, C., and Armstrong, D.A. (2004) Democracy and the Violation of Human Rights: A Statistical Analysis from 1976 to 1996. American Journal of Political Science 48 (3), 538–54.Find this resource:

Davenport, C., Moore, W.H., and Poe, S.C. (2003) Sometimes You Just Have to Leave: Domestic Threats and Forced Migration, 1964–1989. International Interactions 29 (1), 27–55.Find this resource:

Davis, D.R., and Moore, W.H. (1997) Ethnicity Matters: Transnational Ethnic Alliances and Foreign Policy Behavior. International Studies Quarterly 41 (1), 171–84.Find this resource:

Easterly, W. (2001) Can Institutions Resolve Ethnic Conflict? Economic Development and Cultural Change 49 (4), 687–706.Find this resource:

Easton, D. (1965) A Framework for Political Analysis. Englewood Cliffs: Prentice Hall.Find this resource:

Faist, T. (2000) Transnationalization in International Migration: Implications for the Study of Citizenship and Culture. Ethnic and Racial Studies 23 (2), 189–222.Find this resource:

Fearon, J.D., and Laitin, D.D. (1996) Explaining Interethnic Cooperation. American Political Science Review 90 (4), 715–35.Find this resource:

Fearon, J.D. and Laitin, D.D. (1999) Collaborative Project: “Minorities at Risk” Data Base and Explaining Ethnic Violence, NSF grant proposal. At www.stanford.edu/group/ethnic/DLJFNSF.doc, accessed Mar. 10, 2009.

Fearon, J.D., and Laitin, D.D. (2003) Ethnicity, Insurgency, and Civil War. American Political Science Review 97 (1), 75–90.Find this resource:

Fox, J. (2001) Patterns of Discrimination, Grievances and Political Activity among Europe’s Roma. Journal on Ethnopolitics and Minority Issues in Europe (Winter 2001/2). At www.ecmi.de/jemie/download/Focus11-2001Fox.pdf, accessed Mar. 9, 2009.Find this resource:

Fox, J. (2004) Is Ethnoreligious Conflict a Contagious Disease? Studies in Conflict and Terrorism 27 (2), 89–106.Find this resource:

Freedom House (2006) Freedom in the World 2006. Selected Data from Freedom House’s Annual Global Survey of Political Rights and Civil Liberties. Washington: Freedom House.Find this resource:

George, A.L., and Bennett, A. (2005) Case Studies and Theory Development in the Social Sciences. Cambridge: MIT Press.Find this resource:

Gerner, D.J., Schrodt, P.A., Francisco, R., and Weddle, J.L. (1994) The Analysis of Political Events Using Machine Coded Data. International Studies Quarterly 38 (1), 91–119.Find this resource:

Gibney, M., Apodaca, C., and McCann, J. (1996) Refugee Flows, the Internally Displaced, and Political Violence (1980–1993): An Exploratory Analysis. In A.P. Schmid (ed.) Whither Refugees? The Refugee Crisis: Problems and Solutions. Leiden: PIOOM, pp. 45–61.Find this resource:

Gleditsch, K.S., and Ward, M.D. (2001) Measuring Space: A Minimum Distance Database and Applications to International Studies. Journal of Peace Research 38 (6), 739–58.Find this resource:

Goertz, G. (2006) Social Science Concepts: A User’s Guide. Princeton: Princeton University Press.Find this resource:

Goldstein, J.S. (1992) A Conflict–Cooperation Scale for WEIS Events Data. Journal of Conflict Resolution 36 (2), 369–85.Find this resource:

Greenfeld, L. (1996) Nationalism and Modernity. Social Research 63 (1), 3–40.Find this resource:

Gurowitz, A. (1999) Mobilizing International Norms: Domestic Actors, Immigrants, and the Japanese State. World Politics 51 (3), 413–45.Find this resource:

Gurr, T.R. (1993a) Minorities at Risk: A Global View of Ethnopolitical Conflict. Washington: Institute of Peace Press.Find this resource:

Gurr, T.R. (1993b) Why Minorities Rebel: A Global Analysis of Communal Mobilization and Conflict since 1945. International Political Science Review 14 (2), 161–201.Find this resource:

Gurr, T.R. (1994) Peoples against States: Ethnopolitical Conflict and the Changing World System: 1994 Presidential Address. International Studies Quarterly 38 (3), 347–77.Find this resource:

Gurr, T.R. (2000) People vs. States. Washington: Institute of Peace.Find this resource:

Gurr, T.R., and Scarritt, J.R. (1989) Minorities Rights at Risk: A Global Survey. Human Rights Quarterly 11 (4), 375–405.Find this resource:

Hechter, M., and Okamoto, D. (2001) Political Consequences of Minority Group Formation. Annual Review of Political Science 4 (1), 189–215.Find this resource:

Hegre, H., Ellingsen, T., Gates, S., and Gleditsch, N.P. (2001) Toward a Democratic Civil Peace? Democracy, Political Change, and Civil War, 1816–1992. American Political Science Review 95 (1), 33–48.Find this resource:

Hollifield, J.F. (1992) Migration and International Relations: Cooperation and Control in the European Community. International Migration Review 26 (2), 568–95.Find this resource:

Horowitz, D.L. (1985) Ethnic Groups in Conflict. Berkeley: University of California Press.Find this resource:

Horowitz, D.L. (2003) The Deadly Ethnic Riot. Berkeley: University of California Press.Find this resource:

Horowitz, S. (2008) Mapping Pathways of Ethnic Conflict Onset: Preferences and Enabling Conditions. Ethnopolitics 7 (2), 307–20.Find this resource:

Hug, S. (2003) Selection Bias in Comparative Research: The Case of Incomplete Data Sets. Political Analysis 11 (3), 255–74.Find this resource:

Iqbal, Z. (2007) The Geo-Politics of Forced Migration in Africa, 1992–2001. Conflict Management and Peace Science 24 (2), 105–19.Find this resource:

Khosla, D. (1999) Third World States as Intervenors in Ethnic Conflicts: Implications for Regional and International Security. Third World Quarterly 20 (6), 1143–56.Find this resource:

King, G. (1989) Variance Specification in Event Count Models: From Restrictive Assumptions to a Generalized Estimator. American Journal of Political Science 33 (3), 762–84.Find this resource:

King, G., Keohane, R.O., and Verba, S. (1994) Designing Social Inquiry: Scientific Inference in Qualitative Research. Princeton: Princeton University Press.Find this resource:

Koslowski, R. (2000) Migrants and Citizens: Demographic Change in the European State System. Ithaca: Cornell University Press.Find this resource:

Laitin, D.D. (2002a) Comparative Politics: The State of the Subdiscipline. In I. Katznelson, and H. Milner (eds.) Political Science: State of the Discipline. New York: Norton.Find this resource:

Laitin, D.D. (2002b) Culture and National Identity: “The East” and European Integration. West European Politics 25 (2), 55–80.Find this resource:

Laitin, D., and Gurr, T.R. (1999) Collaborative Project: “Minorities at Risk” Database and Explaining Ethnic Violence. NSF grant proposal, Stanford University.Find this resource:

Marshall, M.G., and Jaggers, K. (2004) Polity IV Project. Center for International Development and Conflict Management 2003. At www.systemicpeace.org/polity/polity4.htm, accessed Mar. 10, 2009.

Marx, A.W. (1998) Making Race and Nation: A Comparison of the United States, South Africa, and Brazil. Cambridge: Cambridge University Press.Find this resource:

McVeigh, R., Neblett, C., and Shafiq, S. (2006) Explaining Social Movement Outcomes: Multiorganizational Fields and Hate Crime Reporting. Mobilization: An International Quarterly 11 (1), 23–49.Find this resource:

Minorities at Risk Project (2004) Publications. At www.cidcm.umd.edu/mar/publications.asp, accessed July 1, 2008.

Minorities at Risk Project (2008) Defining a Minority at Risk, Sept. 18, 2008. At www.cidcm.umd.edu/mar/definition.asp, accessed Mar. 10, 2009.

Moore, W.H., and Shellman, S.M. (2004) Fear of Persecution. Journal of Conflict Resolution 48 (5), 723–45.Find this resource:

Moore, W.H., and Shellman, S.M. (2006) Refugee or Internally Displaced Person? To Where Should One Flee? Comparative Political Studies 39 (5), 599–622.Find this resource:

Moore, W.H., and Shellman, S.M. (2007) Whither Will They Go? A Global Analysis of Refugee Flows, 1955–95. International Studies Quarterly 51(4), 811–34.Find this resource:

Morrison, A.R. (1993) Violence or Economics? What Drives Internal Migration in Guatemala? Economic Development and Social Change 4, 817–31.Find this resource:

Morrison, A.R., and May, R. (1994) Escape from Terror: Violence and Migration in Post-Revolutionary Guatemala. Latin American Research Review 29 (2), 111–32.Find this resource:

Muller, E.N., and Weede, E. (1990) Cross-National Variation in Political Violence: A Rational Action Approach. Journal of Conflict Resolution 34 (4), 624–51.Find this resource:

Nagel, J. (1996) American Indian Ethnic Renewal: Red Power and the Resurgence of Identity and Culture. New York: Oxford University Press.Find this resource:

Neumayer, E. (2004) Asylum Destination Choice: What Makes Some European Countries More Attractive than Others? European Union Politics 5 (2), 155–80.Find this resource:

Neumayer, E. (2005a) Bogus Refugees? The Determinants of Asylum Migration to Western Europe. International Studies Quarterly 49 (3), 389–410.Find this resource:

Neumayer, E. (2005b) Influences on Asylum Recognition Rates in Western Europe. Journal of Conflict Resolution 49 (1), 43–66.Find this resource:

Oboler, S. (1995) Ethnic Labels, Latino Lives: Identity and the Politics of (Re)Presentation in the United States. Minneapolis: University of Minnesota Press.Find this resource:

Olzak, S. (1992) The Dynamics of Ethnic Competition and Conflict. Palo Alto: Stanford University Press.Find this resource:

Posen, B. (1993) The Security Dilemma and Ethnic Conflict. Survival 35 (1), 27–47.Find this resource:

Ragin, C.C. (1989) The Comparative Method: Moving beyond Qualitative and Quantitative Strategies. Berkeley: University of California Press.Find this resource:

Ragin, C.C. (2000) Fuzzy-Set Social Science. Chicago: University of Chicago Press.Find this resource:

Regan, P.M., and Henderson, E. (2002) Democracy, Threats, and Political Repression: Are Democracies Internally Less Violent? Third World Quarterly 23 (1), 119–36.Find this resource:

Regan, P.M., and Norton, D. (2005) Greed, Grievance, and Mobilization in Civil Wars. Journal of Conflict Resolution 49 (3), 319–36.Find this resource:

Reynal-Querol, M. (2002) Ethnicity, Political Systems, and Civil Wars. Journal of Conflict Resolution 46 (1), 29–54.Find this resource:

Rihoux, B., and Grimm, H. (2006) Innovative Comparative Methods for Policy Analysis. New York: Springer.Find this resource:

Rubenzer, T. (2008) Ethnic Minority Interest Group Attributes and US Foreign Policy Influence: A Qualitative Comparative Analysis. Foreign Policy Analysis 4 (2), 169–85.Find this resource:

Rubin, J.H., and Moore, W.H. (2007) Risk Factors for Forced Migrant Flight. Conflict Management and Peace Science 24 (2), 85–104.Find this resource:

Saideman, S.M. (2001) The Ties that Divide: Ethnic Politics, Foreign Policy, and International Conflict. New York: Columbia University Press.Find this resource:

Saideman, S.M. (2002) Discrimination in International Relations: Analyzing External Support for Ethnic Groups. Journal of Peace Research 39 (1), 27–50.Find this resource:

Saideman, S.M., Lanoue, D.J., Campenni, M., and Stanton, S. (2002) Democratization, Political Institutions, and Ethnic Conflict: A Pooled Time-Series Analysis, 1985–1998. Comparative Political Studies 35 (1), 103–29.Find this resource:

Salehyan, I. (2007) Transnational Rebels: Neighboring States as Sanctuary for Rebel Groups. World Politics 59 (2), 217–42.Find this resource:

Salehyan, I., and Gleditsch, K. (2006) Refugees and the Spread of Civil War. International Organization 60 (2), 335–66.Find this resource:

Salehyan, I., and Rosenblum, M.R. (2008) International Relations, Domestic Politics, and Asylum Admissions in the United States. Political Research Quarterly 61 (1), 104–21.Find this resource:

Sambanis, N. (2004) Poverty and the Organization of Political Violence. Brookings Trade Forum 2004, 165–211.Find this resource:

Scarritt, J.R., and McMillan, S.M. (1995) Protest and Rebellion in Africa: Explaining Conflicts between Ethnic Minorities and the State in the 1980s. Comparative Political Studies 28 (3), 323–49.Find this resource:

Scarritt, J.R., McMillan, S.M., and Mozaffar, S. (2001) The Interaction between Democracy and Ethnopolitical Protest and Rebellion in Africa. Comparative Political Studies 34 (7), 800–27.Find this resource:

Schmeidl, S. (1997) Exploring the Causes of Forced Migration: A Pooled Time-Series Analysis, 1971–1990. Social Science Quarterly 78 (2), 284–308.Find this resource:

Schmeidl, S. (2000) The Quest for Accuracy in the Estimation of Forced Migration. In S.C. Lubkemann, L. Minear, and T.G. Weis (eds.) Humanitarian Action: Social Science Connections, occasional paper no. 37, Watson Institute for International Studies, Brown University, pp. 127–57.Find this resource:

Schmeidl, S., and Jenkins, J.C. (1998) The Early Warning of Humanitarian Disasters: Problems in Building an Early Warning System. International Migration Review, 32 (2), 471–86.Find this resource:

Schrodt, P.A., and Gerner, D.J. (1994) Validity Assessment of a Machine-Coded Event Data Set for the Middle East, 1982–92. American Journal of Political Science 38 (3), 825–54.Find this resource:

Schrodt, P.A., and Gerner, D.J. (2004) An Event Data Analysis of Third-Party Mediation in the Middle East and Balkans. Journal of Conflict Resolution 48 (3), 310–30.Find this resource:

Schuck, P.H. (2003) Diversity in America: Keeping Government at a Safe Distance. Harvard: Belknap Press.Find this resource:

Sekulic, D., Massey, G., and Hodson, R. (1994) Who Were the Yugoslavs? Failed Sources of a Common Identity in the Former Yugoslavia. American Sociological Review 59 (1), 83–97.Find this resource:

Shain, Y., and Barth, A. (2003) Diasporas and International Relations Theory. International Organization 57 (3), 449–79.Find this resource:

Shellman, S.M., and Stewart, B.M. (2007a) Political Persecution or Economic Deprivation? A Time-Series Analysis of Haitian Exodus, 1990–2004. Conflict Management and Peace Science 24 (2), 121–37.Find this resource:

Shellman, S.M., and Stewart, B.M. (2007b) Predicting Risk Factors Associated with Forced Migration: An Early Warning Model of Haitian Flight. Civil Wars 9 (2), 174–99.Find this resource:

Sliva, J., David, A., and Oleksyn, V. (2008) Italy’s Fingerprinting of Roma Is Racial Discrimination, European Parliament Says. International Herald Tribune (July 10).Find this resource:

Sniderman, P.M., Peri, P., and Piazza, T. (2000) The Outsider: Prejudice and Politics in Italy. Princeton: Princeton University Press.Find this resource:

Stanley, W.D. (1987) Economic Migrants or Refugees from Violence? A Time-Series Analysis of Salvadoran Migration to the United States. Latin American Research Review 22, 132–55.Find this resource:

Tamir, Y. (1995) The Enigma of Nationalism. World Politics 47 (3), 418–40.Find this resource:

Toft, M.D. (2003) The Geography of Ethnic Violence: Identity, Interests, and the Indivisibility of Territory. Princeton: Princeton University Press.Find this resource:

United Nations High Commissioner for Refugees (2000) Handbook for Emergencies. At www.unhcr.org/cgi-bin/texis/vtx/publ/opendoc.pdf?tbl=PUBL&id=3bb2fa26b, accessed Mar. 10, 2009.

Van Houten, P. (2008) Response: The Promise and Limitations of Quantitative Studies of Ethnic Conflict. Ethnopolitics 7 (2), 321–4.Find this resource:

Varshney, A. (2002) Civic Life and Ethnic Conflict: Hindus and Muslims in India. New Haven: Yale University Press.Find this resource:

Varshney, A. (2003) Nationalism, Ethnic Conflict, and Rationality. Perspectives on Politics 1 (1), 85–99.Find this resource:

Weiner, M. (1996) Bad Neighbors, Bad Neighborhoods: An Inquiry into the Causes of Refugee Flows. International Security 21 (1), 5–42.Find this resource:

Wimmer, A. (1997) Who Owns the State? Understanding Ethnic Conflict in Post-Colonial Societies. Nations and Nationalism 3 (4), 631–66.Find this resource:

Wimmer, A., and Min, B. (2006) From Empire to Nation-State: Explaining Wars in the Modern World, 1816–2001. American Sociological Review 71 (6), 867–97.Find this resource:

Wood, D. (1988) Principals, Bureaucrats, and Responsiveness in Clean Air Enforcements. American Political Science Review 82 (1), 213–34.Find this resource:

Zolberg, A.R., Suhrke, A., and Aguayo, S. (1989) Escape from Violence: Conflict and the Refugee Crisis in the Developing World. New York: Oxford University Press.Find this resource:

Minorities at Risk (MAR) Project. At www.cidcm.umd.edu/mar, accessed Mar. 10, 2009. The MAR Project monitors MAR groups and collects quantitative data for 284 groups. It has also collected data on 118 ethnopolitical organizations claiming to represent these groups in the Middle East and North Africa.

Minorities at Risk Library Page, University of Maryland. At www.lib.umd.edu/MCK/TUTOR/mar.html, accessed Mar. 10, 2009. This page was created by the University of Maryland library to provide easy access to many resources that can be useful for researching minority groups.

Kansas Event Data System (KEDS) Project. At http:/web.ku.edu/keds/index.html, accessed Mar. 10, 2009. The KEDS project “uses automated coding of English-language news reports to generate political event data focusing on the Middle East, Balkans, and West Africa.” In addition this is a great resource for learning about and using automated coding.

MIRIS: Minority Rights Information System. At http:/dev.eurac.edu:8085/mugs2/index.jsp?TopBarItem=Home, accessed Mar. 10, 2009. The MIRIS database provides access to legal texts relating to minority rights as well as other documents related to minority groups, including international treaties and agreements.

Religion and State (RAS) Project, Bar Ilan University. At www.biu.ac.il/soc/po/ras, accessed Mar. 10, 2009. The RAS project is built on the MAR model and provides similar data on the relations between religious groups and governments.

Acknowledgments

We would like to thank our anonymous reviewers for their helpful feedback.