Stronger than Yesterday: The Effect of Ethnic Conflict on Ethnic Identity

Abstract: 

Previous literature has limited discussion of the effect of ethnic conflict on ethnic identity. The conventional position claims that ethnic conflict hardens ethnic identity, but this claim has yet to be tested quantitatively (Kaufmann, 1996). Moreover, contrasting literature questions the effect of ethnic conflict on identity due to the existence of ethnic defection (Kalyvas, 2008; Lyall 2010). This study argues that ethnic conflict causes ethnic identity to harden because individuals seek out and identify with co-ethnics for survival (Toft, 2012). Using Afrobarometer survey data, a mixed effects logistic regression finds that the presence of ethnic conflict, as well as conflict duration and intensity, has a negative and significant relationship with ethnic identity. These findings indicate that incidents of ethnic conflict cause ethnic identity to weaken, and conflicts with longer duration and higher intensity are more likely to weaken ethnic identity.

Table of Contents: 

    Introduction

    Ethnicity affects societies and politics across the world. Despite several predictions regarding its decline in importance, the presence of ethnic consciousness has only increased with no country being immune, regardless of geographic location, regime type, or level of economic development (Connor, 1972). According to Horowitz (2000, p. 12), “in deeply divided societies, strong ethnic allegiances permeate organizations, activities, and roles to which they are formally unrelated.” The presence of ethnic consciousness and its permeation through different levels of society has caused ethnicity to become ever more important and subject to manipulation. As a mobilizing factor, ethnic identity can be harnessed and used for a wide variety of political actions.

    Among the variety of factors that cause variation in the salience of ethnic identity, a puzzle emerges when attempting to explain the effect of conflict on ethnic identity. This project seeks to answer the question: How does conflict affect ethnic identity? The conventional claim advanced by the literature is that conflict hardens ethnic identity, although this claim has never been tested quantitatively. This assumption of the state of ethnic identity in post-conflict societies creates simultaneous assumptions about policies that must be implemented to mediate this issue, such as territorial partition (Kaufmann, 1996). Recent scholarship on ethnic defection presents a challenge to this argument (Kalyvas, 2008; Lyall, 2010). When looking at these two arguments side-by-side, a puzzle emerges. If conflict hardens ethnic identity, then ethnic defection should not occur. The purpose of this paper is to address this debate in the literature and conduct a quantitative analysis of the effects of conflict on ethnic identity.

    Literature Review

    Ethnic identity has elicited a long running debate pertaining to its formation. Two specific schools of thought have emerged attempting to explain these origins. On one side of the debate sit the primordialists and on the other side are the constructivists. Primordialists argue that ethnic identity is innate and natural, and not a result of societal factors (Vick & Ishiyama, 2011). Due to its natural origins, primordialists argue that ethnic identity is fixed, and, as such, is a bond that is not subject to manipulation. In contrast to the primordialists, constructivists argue that ethnic identity is socially constructed (Vick & Ishiyama, 2011). Moreover, a more specific extension of the constructivist view is the instrumentalists’ argument. Instrumentalists argue that not only is ethnic identity socially constructed, but it is constructed specifically to be used as a tool to achieve an end goal (Blimes, 2006). From these two fundamental points on the nature of ethnic identity, there is a second debate in ethnic identity literature as to the factors affecting ethnic identity. While the primordialists focus primarily on birth and view ethnic identity as descent-based, constructivists, on the other hand, have produced a large body of literature discussing a variety of different factors that affect ethnic identity.

    One set of scholars point to the role of elites in affecting ethnic identity (Brass, 1997; Fearon & Laitin, 2000; Mueller, 2000; Toft, 2012). Politicians are viewed as political entrepreneurs who manipulate existing ethnic identities, creating further divisions between ethnic groups. Fueled by greed and self-desire, elites are able to achieve their goals more easily when the population is divided (Toft, 2012). Moreover, politicians manipulate ethnicity because it is a good mobilization and ordering instrument, which can later be used during periods of conflict (Mueller, 2000; Gubler, & Selway, 2012). As a result of ethnic manipulation, politicians create ethnic boundaries within countries that previously did not exist (Fearon & Laitin, 2000) and maintain these borders in different ways, such as not appealing outside of a given ethnic group (Simonsen, 2005). Similarly, scholars discuss the impact of collective action on ethnic identity. Ethnicity has the ability to facilitate collective action, which is when individuals act with and on behalf of the larger ethnic group. It is in the self-interest of an individual to identify with and act on behalf of some ethnic group as this produces benefits for that individual (Hardin, 1995, pp. 46-71). Collective action reinforces ethnic identity as individuals begin to identify themselves with a specific ethnic group and proceed to continuously act on their behalf.

    Another factor that is argued to affect ethnic identity is political institutions (Posner, 2005; Bakke & Wibbels, 2006; Toft, 2012). Scholars debate which political institutions are best to implement and the effects they will have on ethnic identities. Specifically, there is a structural debate between centralism and federalism. Some scholars claim that centralization causes an increase in ethnic cleavages due to increased competition amongst ethnic groups (Horowitz, 2000) and that decentralization of political institutions may prevent ethnic identity hardening, at least in the short-term (Brancati, 2006). Federalism is held up as an alternative path because it separates ethnic groups, prevents fears over resource distribution and threats to ethnic identity (Toft, 2012), and keeps those identities from becoming salient as a result of interaction (Rutherford et al., 2011). Moreover, federalism allows for a combination of shared rule and self-rule (Bakke & Wibbels, 2006). Others, however, state that federalism is not the solution. They claim that it causes ethnic groups to become isolated and, as a result, regional and ethnic parties emerge that can cause the hardening of ethnic identity. Ethnicity becomes institutionalized, mobilization occurs along ethnic lines, and ethnic groups continually vote along ethnic lines (Simonsen, 2005; Bakke & Wibbels, 2006; Brancati, 2006). These scholars point to failed federal states such as the Soviet Union and Yugoslavia as examples (Collier et al., 2004).

    Other scholars argue that elements of institutions outside of structure are important to understanding effects on ethnic identity. One such argument is that consociationalism, with a particular focus on the importance of proportional representation, is a means to keep an ethnically divided society peaceful (Lijphart, 1977, p. 39; Simonsen, 2005). With this approach, all groups would be represented proportionally in decision-making bodies and a greater sense of fairness would exist because all groups are able to influence a decision relative to their strength (Lijphart, 1977, p. 39). Posner (2004), on the other hand, argues that institutional rules shape identity choices as they affect the boundaries of political competition. In some countries, political institutions contribute to ethnic identities becoming politically salient when politicians identify ethnic groups large enough to yield significant influence in electoral outcomes (Posner, 2004).

    Furthermore, another factor that affects ethnic identity is the security dilemma (Posen, 1993; Rose, 2000; Toft, 2012; Gubler & Selway, 2012). The security dilemma is a situation when a state has collapsed or simply can no longer provide security to the population within its borders. Consequently, individuals proceed to identify themselves more closely with their ethnic group out of fear and self-defense (Toft, 2012). The security dilemma can also trigger other emotions and cause ethnic groups to arm themselves (Peterson, 2002). Economic development, moreover, increases social mobilization and communication, thus bringing groups into more constant contact with one another and helping to amplify the us versus them mentality (Connor, 1972). Finally, due to its importance to identity, territorial distribution is an additional factor that affects ethnic identity. Ethnic groups hold emotional and materialist claims to specific territory. (Wolff, 2006, pp. 42-46). As decisions are made on potential territorial changes – such as redefining an ethnic group’s relationship with the territory in which it resides – altering the territorial boundaries, or redistributing one’s ethnic group, members of different ethnic groups stand in opposition to one another. Ethnic identities become salient as a result (Coakley, 2003: 11; Collier et al., 2004).

    While significant scholarship exists arguing the effects of the above noted factors on ethnic identity, scholarship on the effect of conflict on ethnic identity is limited. Despite this, certain assumptions about the effect of conflict on ethnic identity have been made and taken as fact – influencing much subsequent work. Chaim Kaufmann (1996) boldly argues that ethnic identity is fixed and that conflict makes ethnic identities salient. He claims that ethnic identities are fixed by birth and so attempts to sway individual loyalties during ethnic conflict will yield no results. Accordingly, Kaufmann advocates for the partitioning of multiethnic states. He claims that due to these hardened ethnic identities, partition is the best way to prevent future violence as homogenous settlements result in peace. This claim, however, has not been substantiated with either qualitative or quantitative data.

    Kaufmann’s argument finds support from other scholars. It is argued that “civil conflict can deepen ethnic tensions and reinforce political ethnic identities” (Gurses & Rost, 2013, p. 483) or, to go a step further, conflicts create cleavages that did not exist in pre-conflict societies (Wolff, 2006, pp. 25-31). Due to the ever-increasing dominance of ethnic cleavages, civilians are polarized into identity camps and seek safety with co-ethnics (Collier et al., 2004; Urdal, 2006). Extremists will use violence in these situations in an attempt to force moderates to engage in violence and thus create sharper boundaries (Fearon & Laitin, 2000). This conflict elevates ethnicity to being the dominant social marker and creates deeper divisions in society (Simonsen, 2005). As a result, well-defined boundaries are needed for peace to exist – a well-integrated coexistence is impossible (Rutherford et al., 2011) and ethnic segregation places natural limits on the extent of violence (Weidmann & Salehyan, 2013).

    Later scholars have challenged Kaufmann’s position that civil war hardens ethnic identity. Stathis Kalyvas argues that civil wars are dynamic political and social contexts that can potentially shape the behavioral expressions of ethnic identity (2008). He claims that there can be multidirectional identity transformation due to conflict – coming in direct contention with Kaufmann who argued that ethnic identity only hardens due to conflict (1996). Kalyvas’ position is based on his analysis of ethnic defection. He, as other scholars before him, claims that ethnic identity is malleable because ethnic defection occurs during conflict (Abdelal et al., 2006; Chandra, 2006; Goddard 2006).

    Jason Lyall also challenges Kaufmann and builds off of Kalyvas’ work on defection. Analyzing the effect of the identity of counterinsurgents in Chechnya in the period 2000 to 2005, Lyall (2010) finds that pro-Russian Chechen counterinsurgents are more effective against Chechen insurgents, their co-ethnics, than Russian counterinsurgents. The existence of pro-Russian Chechen soldiers willing to fight against co-ethnics challenges Kaufmann’s key assumptions. If these soldiers are willing to fight co-ethnics that means ethnic identities are not fixed and that the competition to sway individual loyalties does in fact play a role in conflict, as shown by ethnic Chechens falling on both sides of the conflict.

    While both Kalyvas and Lyall’s work challenges Kaufmann’s fundamental position, neither of these scholars directly tackles the question regarding the relationship between conflict and identity. Kalyvas (2008) merely claims that there can be multidirectional identity transformation due to conflict, while Lyall (2010) clearly states that he is looking at how ethnicity affects the pattern of violence during conflict. As a result, there is still a hole in the literature where a quantitative analysis of the effect of conflict on ethnic identity is needed.

    Theory

    Building on the work of constructivist scholars, I argue that ethnic identity is not natural and innate, rather, ethnic identity is socially constructed (Anderson, 1991; Greenfeld, 1992; Fearon & Laitin, 2000; Posner, 2004; Toft, 2012). Taking ethnic identity as socially constructed, I claim that ethnic identity is malleable, not fixed – it can be changed and manipulated by a variety of factors. Within the political science literature, debates rage about what factors cause ethnic identity to change. These factors included institutions (Lijphart, 1977; Posner, 2004), elites (Brass 1997; Fearon & Laitin, 2000), the security dilemma (Posen, 1993; Toft, 2012), economic development (Connor, 1972), territory (Coakley, 2003; Wolff, 2006) and conflict (Kaufmann, 1996; Fearon & Laitin, 2000; Kalyvas, 2008; Lyall, 2010). I contend that conflict is crucial to understanding changes in ethnic identity. Following this line of research, I argue that ethnic conflict causes ethnic identity to change; specifically, I argue that ethnic conflict will harden ethnic identities.

    The societies that experience ethnic conflict are ethnically divided. When conflict erupts, the population is divided between combatants and civilians. In many instances, the majority of the population is civilians who are not engaged in ethnic conflict (Mueller, 2000).  Additionally, there is a division between moderates and extremists amongst the population. The ethnic identity of these extremists is already hardened. Extremists engage in ethnic conflict and attempt to use violence to push moderates to support extremism as well (Kaufmann, 1996). The conflict provides an avenue for the extremists to push for the hardening of ethnic identities (Fearon & Laitin, 2000). This is particularly the case if a security dilemma emerges and the state either collapses or is no longer able to provide security to the population within its borders.

    The security void that forms as a result causes fear and desire for self-defense amongst the population (Toft, 2012). As a result, individuals are driven to associate more closely with their ethnic group as they seek safety with co-ethnics (Collier et al., 2004; Urdal, 2006). With each individual seeking safety amongst their co-ethnics, the resulting polarization by way of division into identity camps undermines integration and heterogeneity amongst the population (Collier et al., 2004). Moreover, not only does the prospect of violence to an individual cause fear and desire for self-defense to the point of seeking safety amongst co-ethnics, the real atrocities committed during the conflict further deepen division along ethnic lines (Kaufmann, 1996). During World War II, the Croatian Ustasha terrorized Serbs, “in order to provoke a backlash that could then be used to mobilize Croats for defense against Serb retaliation” (Kaufmann, 1996, p. 142). This logic leads to this study’s first hypothesis:

    H1: The presence of ethnic conflict causes ethnic identity to harden.

    If H1 is supported, then the presence of ethnic conflict has a positive and significant relationship with ethnic identity. I argue that ethnic identity becomes more salient during ethnic conflict and hardens as a result. For example, during Sri Lankan ethnic riots in 1983 between Sinhalese and Tamils, the ethnic identity of the groups was hardened as Sinhalese came to view all Tamils as separatists (Kaufmann, 1996).

    There are elements of ethnic conflict that are aggregated together that have the potential to provide addition information about the effect of ethnic conflict on ethnic identity. More specifically, the duration and the intensity of ethnic conflict have the potential to cause variance that cannot be explained by H1 alone. As the security dilemma in a country experiencing ethnic conflict persists and atrocities continue to be committed, individuals will continue to seek safety amongst their co-ethnics. The perpetuation of the conflict will cause more individuals to identify along ethnic lines. This logic leads to the following additional hypotheses:

    H2: The longer the duration of ethnic conflict, the higher the likelihood that ethnic identity will harden.

    The assumption is that as the conflict continues, more individuals will be exposed to the conflict due to the longer period of time over which it occurs. As a result, ethnic identity will be more likely to harden. H2 looks at the overall hardening effect of ethnic identity in relation to the duration of the conflict.

    Similarly, another aspect of the nature of ethnic conflict is the intensity of the conflict. Just as ethnic conflicts vary in their duration, ethnic conflicts also vary in their intensity and just as the duration of the ethnic conflict affects change in ethnic identity, the intensity of the conflict also affects change in ethnic identity. Ethnic conflicts with higher levels of intensity are those that experience large quantities of deaths. These intense ethnic conflicts may include genocide or ethnic cleansing. As individuals come into contact with others attempting to kill them in an aim, “to create a society purified of undesirable classes or communal groups” (Harff, 2003, p. 70), the targeted individuals understanding of belonging to their specific ethnic group increases. This amplifies the already created boundaries that exist as a result of co-ethnic clustering out of fear and desire for self-defense (Collier et al., 2004; Urdal, 2006). This logic leads to the next set of hypotheses:

    H3: The greater the intensity of ethnic conflict, the higher the likelihood that ethnic identity will harden.

    The assumption is that intensity refers to the amount of deaths that occur during the ethnic conflict. As the amount of deaths during the conflict increases, or simply when there is a more intense conflict compared with a less intense conflict, then ethnic identity will be more likely to harden.

    These hypotheses focus on the nature of the conflict and the impact that differences in the nature of the conflict may have on the hardening of ethnic identity. If they are supported it shows that while ethnic conflict does cause the hardening of ethnic identity, the process has specific nuances. It is not, therefore, simply the presence of conflict that is necessary for ethnic identity to harden. Instead, the duration or the intensity of the conflict is crucial to whether or not ethnic identity hardens.

    Research Design

    This section of the paper will discuss the plan for testing out the hypotheses of the previous section. Specifically, this section will include the identification of the sample, operationalization, and a discussion of the analytical technique to be used to test the hypotheses.

    The data used for this study is cross sectional data from the Afrobarometer over several waves. The spatial domain of this project is restricted to six countries in Africa: Benin, Kenya, Malawi, Mali, Nigeria, and Zambia. All six of these countries are ethnically divided states. Kenya, Mali, and Nigeria have experienced ethnic conflict, while Benin, Malawi, and Zambia have not. As a result, Kenya, Mali, and Nigeria are the primary countries of interest whereas Benin, Malawi, and Zambia will serve as controls. The Uppsala Conflict Data Program (UCDP) Conflict Encyclopedia, which documents war and minor conflict and non-state conflict, determines the occurrence of ethnic conflict in Mali and Nigeria since 1989 and the Social Conflict in Africa Database (SCAD) determines the occurrence of conflict in Kenya. The conflict in Kenya was neither a war nor minor conflict from the labeling perspective of UCDP, but it was an intense ethnic conflict, which adheres to the parameters of this study. Overall, the temporal data spans the period 2001 to 2014, but due to the nature of this project, the temporal domain varies because of: 1) differences in years of ethnic conflict in each state; and 2) availability of data in each state. As a result, the spatial data in years for the six countries are as shown in Table 1.

    In Kenya, the observed ethnic conflict occurred in 2007-2008. The pre-conflict data years are 2003 and 2005, and the post-conflict data years are 2008 and 2011. In Mali, the observed ethnic conflict occurred in 2007-2009. The pre-conflict data years are 2001, 2002, and the post-conflict data years are 2008 and 2012. The survey data for Mali 2005 exists, but cannot be used due to its inaccessibility. In Nigeria, the observed ethnic conflict occurred in 2004. The pre-conflict data years are 1999, 2001, 2003, and the post-conflict data years are 2005, 2007, 2008 and 2013. Finally, looking at the data for the countries without conflict, the data for Benin is from the years 2005, 2008, and 2011, the data for Malawi is from the years 1999, 2003, 2005, 2008, and 2012, and the data for Zambia is from the years 1999, 2003, 2005, 2009, and 2012.

    This study has two units of analysis: 1) Individuals: each individual who responded to the survey question is looked at because his or her ethnic identity is unique and can potentially harden; and 2) Country: each country has country-level factors that impact the individual level of analysis. It is important to note, however, that this survey data has a major limitation. The survey responses from all years are not from the same individuals, thus the change of ethnic identity of the same individuals from pre-conflict to post-conflict cannot be measured. Instead, the survey respondents are randomly selected each year, but they are a representative sample of the population. As a result, this data can be used to analyze the overall change of ethnic identity within a given country.

    The dependent variable for this study is ethnic identity. Ethnic identity is “a subset of identity categories in which eligibility for membership is determined by attributes associated with, or believed to be associated with, descent” (Chandra, 2006, p. 398) which is socially constructed (Anderson, 1991; Greenfeld, 1992; Fearon & Laitin, 2000; Posner, 2004; Toft, 2012).

    After the interviewer establishes the ethnic group of the respondent, the survey data provides a measure for ethnic identity by asking the respondent how he or she feels about his or her ethnic group. There are multiple questions used across all rounds of Afrobarometer surveys. See Table 2 for detailed information regarding the questions used. Respondents are given one of two options for responses to questions in different surveys: 1) a binary option where 0 represents national identity and 1 represents group or ethnic identity; or 2) an ordinal spectrum of responses that provides an understanding of how strongly or weakly the respondent identifies with his or her ethnic group.

    Data from these survey questions will determine whether or not ethnic identity hardens. Hardening of identification with an ethnic group refers to the strengthening of identification with that ethnic group. In the context of this study, hardening of ethnic identity would be exemplified by an increase in the quantity of respondents identifying with their ethnic identity over the course of survey years. In the dataset for this project, identifying with one’s nationality will be coded as 0 and identifying with one’s ethnic group will be coded as 1 through the creation of a dummy variable. Only ordinal data must be recoded as the binary survey data already exists in the proper form. If the respondent identifies closer with their ethnic identity, then they are recoded as a 1. If the respondent identifies closer with their national identity or equally between national and ethnic identity, then these responses are recoded as a 0. If a survey respondent identifies equally between national and ethnic identity and his or her ethnic identity is not hardened, thus these responses are recoded as 0. With cross sectional data, not panel data, for both pre-conflict and post-conflict, this coding focuses on determining whether or not ethnic identities harden overall in a country.

    The primary independent variable for this study is ethnic conflict. Using the UCDP Conflict Encyclopedia and SCAD, the years of ethnic conflict were established for Kenya, Nigeria, and Mali. Ethnic conflict is a subset of civil war distinguished by being a dispute, “between communities which see themselves as having distinct heritages, over the power relationship between the communities” (Kaufmann, 1996, p. 138). Ethnic conflict is operationalized according to the parameters of the UCDP Conflict Encyclopedia and SCAD. For Kenya, Nigeria, and Mali, ethnic conflict did occur, but for Benin, Malawi, and Zambia, ethnic conflict did not occur. Moreover, the year of the ethnic conflict is necessary because it establishes the date in relation to which the pre-conflict and post-conflict identification with one’s ethnic group is analyzed in order to understand whether ethnic conflict causes ethnic identity to harden. A dummy variable will be included in order to distinguish the pre-conflict data from the post-conflict data.

    The additional independent variables are descriptive aspects of ethnic conflict: duration of ethnic conflict and intensity of ethnic conflict. These independent variables will be applied only to post-conflict observations. The duration of ethnic conflict and the intensity of the conflict operationalize according to the parameters of the UCDP Conflict Encyclopedia and SCAD. In this study, duration is defined as the number of months that the ethnic conflict lasted. Intensity is defined as the total number of deaths during the ethnic conflict. The UCDP Conflict Encyclopedia and SCAD will be used to determine both quantities.

    This study must control for individual-level and country-level variables in order to better isolate the effect of ethnic conflict on ethnic identity. There are several individual-level variables controlled for in this study. These variables correlate to the individual ethnic identity survey responses. The first variable is age of respondent. This variable affects an individual’s ethnic identity formation due to the connection with formative experiences and quantity of experiences. All respondents are at least eighteen years of age. Additionally, the education variable represents the highest level of education attained by the survey respondent. Higher levels of education are associated with more liberal ideologies. The third variable is gender of respondent. This variable is another form of identification that is affected by societal differences attributed to the male and female gender. These differences vary depending on the ethnic group to which one belongs and thus can affect ethnic identification. The interviewer provides this data. The living conditions variable represents the perceived economic situation of an individual. The presence or lack of contentment with the economic situation on the part of the individual can affect one’s desire to identify with one’s ethnic group. The final individual-level control variable is urban or rural. This variable represents the physical and demographic description of a respondent’s area of residence. This difference affects the type and amount of contact that individuals have with members of other ethnic groups and in turn can affect their own ethnic identification. This information is provided by the interviewer. All data is collected using survey questions from the Afrobarometer.

    Additionally, this study must control for several country-level variables. First, level of democracy is controlled for using the Polity2 variable from the Polity IV Project (Marshall, Gurr, & Jaggers, 2014). This variable represents the revised combined polity score for each country. Level of democracy can impact ethnic identity based on the presence or lack of political representation for ethnic groups. Second, gross domestic product (GDP) per capita is controlled for using data from The World Bank (World Bank, 2014). This variable estimates the wealth of a country. Countries with lower GDP per capita are more likely to experience hardened ethnic identities due to competition for resources amongst ethnic groups. Third, ethnic fractionalization is controlled for using the Alesina et al. 2003 Fractionalization Dataset. The amount of ethnic fractionalization can affect change in ethnic identity because countries with higher ethnic fractionalization are more likely to experience ethnic tension and hardened ethnic identities as a result. Fourth, previous ethnic conflict within ten years is controlled for using data from the UCDP Conflict Encyclopedia (Uppsala Conflict Data Program).

    The model used in this study is a mixed effects logistic regression. There are two units of analysis used in this model: individual and country. Individual is the lower level of analysis and country is the higher level of analysis. Thus, this model is used because elements at the country level influence the individual level, specifically ethnic identification. Moreover, two models will be run. The first will test H1The presence of ethnic conflict causes ethnic identity to harden – using all 47,533 observations. The second will test H2the longer the duration of ethnic conflict, the higher the likelihood that ethnic identity will harden – and H3the greater the intensity of ethnic conflict, the higher the likelihood that ethnic identity will harden – with only the 15,432 observations in post-conflict societies. The control variables in the first model are: age of respondent, education, gender of respondent, living conditions, urban or rural, ethnic fractionalization, log GDP per capita, polity2, previous conflict in ten years. The control variables in the second model are: age of respondent, education, gender of respondent, living conditions, urban or rural, ethnic fractionalization, log GDP per capita, polity2. The independent variable will differ based on the hypotheses being tested: H1 – ethnic conflict, H2 – duration of ethnic conflict, and H3 – intensity of ethnic conflict.  A mixed effects logistic regression will be used to test all three hypotheses.

    Results and Analysis

    The hypotheses of this study make two specific assertions. First, when comparing countries with ethnic conflict and countries without ethnic conflict, the presence of ethnic conflict hardens ethnic identity. Second, in countries where ethnic conflict occurs, longer and more intense ethnic conflicts are more likely to harden ethnic identity. Apart from the dependent variable of ethnic identity and the main independent variables of conflict, duration, and intensity, there are several country-level and individual-level controls, specified previously in the research design, that are included in the mixed effects logistic regression.

    The first model analyzes all observations in the dataset to test H1 the presence of ethnic conflict causes ethnic identity to harden. While Table 3 shows that the ethnic conflict variable is significant at a 99 percent confidence level, the results of the mixed effects logistic regression are not in the theorized direction. Rather, the presence of ethnic conflict causes ethnic identities to weaken. Specifically, the odds ratio shows that ethnic conflict causes a 68.9 percent decrease in ethnic identity. Moreover, education, previous conflict within 10 years and log GDP per capita are significant at the 99 percent confidence level, and age, ethnic fractionalization, and polity2 are significant at the 95 percent confidence level.

    The second model analyzed only those observations where the conflict variable was equal to one to test H2 the longer the duration of the ethnic conflict, the higher the likelihood that ethnic identity will harden – and H3the greater the intensity of the conflict, the higher the likelihood that ethnic identity will harden. Table 4 shows that the duration variable is significant at the 99 percent confidence level. The results, however, of the mixed effects logistical regression once again run counter to theoretical expectations. Rather than weakening ethnic identity, the odds ratio shows that an increase in duration of ethnic conflict causes ethnic identity to decrease, or weaken, at 86.9 percent. Furthermore, H3 was also not supported by the analysis. As shown in Table 4, the log of intensity is significant at the 99 percent confidence interval, but opposite of the hypothesized direction. The odds ratio shows that an increase in intensity of conflict causes a 99.5 percent decrease, or weakening, of ethnic identity. Additionally, education, ethnic fractionalization, polity2 and log GDP per capita are significant at the 99 percent confidence level.

    The results of the mixed effects logistical regression models provide evidence for the opposite of the posited relationship between ethnic identity and ethnic conflict, duration and intensity. Contrary to this study’s theoretical expectations, all three independent variables weaken ethnic identity. Previous literature, however, provides support for these results. The existence of ethnic defection challenges the argument that identities harden as a result of conflict (Lyall, 2010). Despite these findings, there are various limitations to this study and avenues for further research. The data used is survey data limited in scope to six countries in Africa (Benin, Kenya, Malawi, Mali, Nigeria and Zambia) and the survey responses come from different individuals every year. Thus, the effect of ethnic conflict on ethnic identity must be observed from a country-level and we cannot discuss the effects of conflict on individuals specifically. One way to alleviate this is to obtain data from the same group of individuals in pre-conflict and post-conflict environments, or develop a different methodology for measuring ethnic identity that does not rely on surveys. These developments would open the door for consideration of the magnitude of change in ethnic identity due to conflict and allow for the study of extremists.

    Conclusion

    This paper analyzed the effect of ethnic conflict on ethnic identity. The results are counter to the outlined theoretical expectations. This study’s findings provide support to the claim that ethnic conflict weakens, not hardens, ethnic identity. Additionally, this study provides support to the claim that both duration of conflict and intensity of conflict significantly impact ethnic identity in countries where conflict occurred. First, the longer the duration of an ethnic conflict, the more likely that ethnic identity will weaken; and second, the more intense an ethnic conflict, the more likely that ethnic identity will weaken.

    Despite the fact that all three of these results run counter to theoretical implications, these results nonetheless prove interesting in the context of the debate surrounding the effect of ethnic conflict on ethnic identity. Kaufmann (1996) argues that conflict hardens ethnic identity and several scholars support his position, including (Fearon & Laitin, 2000; Collier et al., 2004; Simonsen, 2005; Urdal, 2006; Wolff, 2006; Gurses & Rost, 2013). On the other hand, Kalyvas’ (2008) makes the claim that conflict can cause identity to experience a multidirectional transformation rather than a simple hardening of ethnic identity. Furthermore, Lyall (2010) argues that the presence of ethnic defection means that ethnic identities are not fixed and do not always harden due to conflict. This study’s main finding that ethnic conflict weakens ethnic identity helps us to understand the phenomenon of ethnic defection. If identity hardened during conflict, then individuals would feel bound to their ethnic groups and ethnic defection would not occur. But, because ethnic identity weakens during conflict, individuals do not feel bound to their groups, making ethnic defection possible.

    The support that this study provides for the claim that ethnic conflict weakens ethnic identity has potentially serious policy implications. If ethnic identity weakens rather than hardens as a result of ethnic conflict, then post-conflict policies must take this into consideration. Policies that advocate separation within a country or partitioning of countries on the assumption that ethnic conflict hardens ethnic identity (Kaufmann, 1996) may need to be reconsidered. Moreover, there is a body of literature that argues for the effect of institutions on shaping ethnic identity (Posner, 2005; Bakke & Wibbels, 2006; Toft, 2012). The implementation of post-conflict polices and institutions that are built upon the presumption that ethnic identity hardens due to conflict may contribute to ethnic hardening where ethnic conflict did not.

    This paper provides a quantitative analysis of the relationship between ethnic conflict and ethnic identity. The findings of this study provide support for the claim that the debate surrounding the relationship between ethnic conflict and ethnic identity is more complex than previously anticipated. Further, broader research is necessary to better understand this relationship. The additional clarity that would come with additional research is needed as policy is shaped around the conventional understanding of the relationship between ethnic conflict and ethnic identity. If further research provides added support for this study’s findings that ethnic conflict weakens ethnic identity, serious policy reformation would need to follow.

    References

    • Abdelal, R., Herrera, Y. M., Johnston, A. I., & McDermott, R. (2006). Identity as a variable. Perspectives on Politics, 4(4), 695-711. Retrieved from http://www.jstor.org/stable/20446277
    • Anderson, B. (1991). Imagined communities: Reflections on the origin and spread of nationalism. London: Verso.
    • Bakke, K. M. & Wibbels, E. (2006). Diversity, disparity, and civil conflict in federal states. World Politics, 59(1), 1-50. Retrieved from http://muse.jhu.edu/journals/world_politics/v059/59.1bakke.html
    • Blimes, R. T. (2006). The indirect effect of ethnic heterogeneity on the likelihood of civil war onset. The Journal of Conflict Resolution, 50(4), 536-547. Retrieved from http://www.jstor.org/stable/27638505
    • Brancati, D. (2006). Decentralization: Fueling the fire or dampening the flames of ethnic conflict and succession? International Organization, 60(3), 651-685.
    • Brass, P. R. (1997). Theft of an idol: Text and context in the representation of collective violence. Princeton: Princeton University Press.
    • Chandra, K. (2006). What is ethnic identity and does it matter? Annual Review of Political Science, 9, 397-424. doi: 10.1146/annurev.polisci.9.062404.170715
    • Coakley, J. (2003). The challenge. In John Coakley (Ed.), The territorial management of ethnic conflict (pp. 1-22). Portland: Frank Cass.
    • Collier, P. & Hoeffler, A. (2004). Greed and grievance in civil war. Oxford Economic Papers, 56(4), 563–595. Retrieved from http://www.jstor.org/stable/3488799
    • Connor, W. (1972). Nation-building or nation-destroying? World Politics, 24(3), 319-355. Retrieved from http://www.jstor.org/stable/2009753
    • Fearon, J. D. & Laitin, D.D.. (2003). Ethnicity, insurgency and civil war. American Political Science Review, 97(1), 75–90. Retrieved from http://www.jstor.org/stable/3118222
    • Fearon, J. D. & Laitin, D. D. (2000). Violence and the social construction of ethnic identity. International Organization, 54(4), 845–877. doi: http://dx.doi.org/10.1162/002081800551398
    • Goddard, S. E. (2006). Uncommon ground: Indivisible territory and the politics of legitimacy. International Organization, 60(1), 35-68. doi: http://dx.doi.org/10.1017/S0020818306060024
    • Greenfeld, L. (1992). Nationalism: Five roads to modernity. Cambridge: Harvard University Press.
    • Gubler, J. R. & Selway, J. S. (2012). Horizontal inequality, crosscutting cleavages, and civil war. Journal of Conflict Resolution, 56(2), 206-232. doi: 10.1177/0022002711431416
    • Gurses, M. & Rost, N. (2013). Sustaining the peace after ethnic civic wars. Conflict Management and Peace Science, 30(5), 469-491.
    • Hardin, R. (1995). One for all: The logic of group conflict. Princeton: Princeton University Press.
    • Harff, B. (2003). No lessons learned from the holocaust? Assessing risks of genocide and political mass murder since 1955. American Political Science Review, 97(1), 57-73. Retrieved from: http://jstor.org/stable/3118221
    • Hendrix, C.S. and I. Salehyan. (2009). "Social Conflict in Africa Database (SCAD)." Retrieved from www.scaddata.org.
    • Horowitz, D. (2000). Ethnic groups in conflict. Berkeley: University of California Press.
    • Kalyvas, S. (2008). Ethnic defection in civil war. Comparative Political Science, 41(8), 1043-1068. doi: 10.1177/0010414008317949
    • Kaufmann, C. (1996). Possible and impossible solutions to ethnic civil wars. International Security, 20(4), 136–175. Retrieved from http://www.jstor.org/stable/2539045
    • Laitin, D. D. (1998). Identity in formation: The Russian-speaking populations in the near abroad. Ithaca: Cornell University Press.
    • Lijphart, A. (1977). Democracy in plural societies: A comparative exploration. New Haven: Yale University Press.
    • Lyall, J. (2010). Are coethnics more effective counterinsurgents? Evidence from second Chechen war. American Political Science Review, 104(1), 1-20. doi: http://dx.doi.org/10.1017/S0003055409990323
    • Marshall, M.G., Gurr, T.R., & Jaggers, K. (2014). Polity IV Project: Political regime characteristics and transitions, 1800-2013. Retrieved from www.systemicpeace.org
    • Mueller, J. (2000). The banality of “ethnic war.” International Security, 25(1), 42–70. Retrieved from http://www.jstor.org/stable/2626773
    • Petersen, R. (2002). Understanding ethnic violence. Cambridge: Cambridge University Press.
    • Posen, B.R. (1993). The security dilemma and ethnic conflict. In Michael E. Brown (ed), Ethnic conflict and international security (pp. 103-124). Princeton: Princeton University Press.
    • Posner, D. (2004). The political salience of cultural difference: Why chewas and tumbukas are allies in Zambia and adversaries in Malawi. American Political Science Review, 98(4), 529-545. Retrieved from http://www.jstor.org/stable/4145323
    • Posner, D. (2005). Institutions and ethnic politics in Africa. Cambridge: Cambridge University Press.
    • Rose, W. (2000). The security dilemma and ethnic conflict: Some new hypotheses. Security Studies, 9(4), 1–51.
    • Rutherford, A., Harmon, D., Werfel, J., Bar-Yam, S., Gard-Murray, A.,  Gros, A.,Xulvi-Brunet, R. & Bar-Yam, Y. (2011). Good fences: The importance of setting boundaries for peaceful coexistence. PLoS ONE, 9(5), 1-15. doi: http://dx.doi.org/10.1371/journal.pone.0095660
    • Simonsen, S. G. (2005). Addressing ethnic divisions in post-conflict institution-building: Lessons from Recent Cases. Security Dialogue, 36(3), 297-318. doi: 10.1177/0967010605057017
    • Toff, M. D. (2012). The origins of ethnic war: A historical and critical account. In Manus I. Midlarsky (Ed), Handbook of war studies III: The intrastate dimension (pp. 229-259). Ann Arbor: University of Michigan Press.
    • Uppsala Conflict Data Program (Date of retrieval: 2014/07/03) UCDP conflict encyclopedia (UCDP database). Uppsala University. Retrieved from www.ucdp.uu.se/database.
    • Urdal, H. (2006). A clash of generations? Youth bulges and political violence. International Studies Quarterly, 50(3), 607-629. doi: http://www.jstor.org/stable/4092795
    • Vick, D. & Ishiyama, J. T. (2011). Ethnic and identity politics. In John T. Ishiyama and Marijke Breuning (Eds), 21st century political science: A reference handbook (pp. 217-224). Los Angelos: SAGE Publications, Inc.
    • Weidmann, N.B., & Salehyan, I. (2013). Violence and ethnic segregation: A computational model applied to Baghdad. International Studies Quarterly, 57(1), 52-64. doi: 10.1111/isqu.12059
    • Wolff, S. (2006). Ethnic conflict: A global perspective. Oxford: Oxford University Press.
    • World Bank, 2014. GDP per capita (current US$). Retrieved from World DataBank: World Development Indicators database.

    Table 1: Survey Rounds and Country Years

    Table 2: Operationalization of Variables using Survey Questions with Sources

    Table 3: The Effects of Conflict on Ethnic Identity

    Robust seeform in parentheses
    *** p<0.01, ** p<0.05, * p<0.1

    Table 4: The Effects of Conflict Duration and Intensity on Ethnic Identity

    Robust seeform in parentheses
    *** p<0.01, ** p<0.05, * p<0.1