Who am I? : identity after violent ethnic conflict : Rwanda

Who Am I?

Identity after violent ethnic conflict:

Rwanda

Bachelor Thesis Rosa Douw 10001837 Tim Reeskens, Crossing Boundaries: nationalism and national identity January 20th 2013

Who Am I? Identity after violent ethnic conflict: Rwanda

How identities might form has often been described. Different theories on this topic are for example: a premordialist view (Smith, 1993:48), a modernist view (Gellner, 1981:766) and a social constructivist view (Tilly, 2006:209). A subject very often tackled in this field of research is how these ethnic identities can lead to or at least play a big role in the process towards ethnic conflict (for example: Gagnon, 1994). I would like to pose the question the other way around: why does ethnic identity not lead to conflict, especially shortly after violence? Do ethnic identities stay as important as they were? If so, then why are they not important enough to kill for anymore? Or, when arguing that a new national identity is created, how does this go about and what kind of national identity is it that emerges? How do people deal with the fact that who is your neighbour today was your enemy yesterday, who could have killed you or your family? A lot of research has been done on why conflict occurs in the first place, a lot less on why it does not occur. It is of course very challenging to research why something does not happen but studying cases of post-violent conflict might lead to an interesting insight that can later be tested on cases where such violent conflicts have not occurred yet and that might even help to prevent these violent conflicts in the future. Theories on the aftermath of violent conflict place a huge emphasis on the occurrence of national identity and on the importance of a sense of unity throughout a nation that prevent different ethnic groups from getting into violent conflict again. Social capital, a concept that evaluates social networks as an important aspect of unity in a nation, is another important value that is said to prevent violence from happening (Putnam, 2006:137). Therefore I would like to combine theories on different forms of national identity with the influence that these forms have on a country’s social capital and then further on link that to the likelihood of violence in a region.

In order to research this I will conduct a case study. The Republic of Rwanda is an interesting case because it remained a state after the extremely bloody conflict which occurred within the state and resulted in the genocide of 1994; the conflict that occurred between different ethnicities, namely the Hutus and the Tutsis. Ethnic identities still play a role today, despite the many efforts

made by the government and other organisations to create a national identity (Longman, 2005:207). Thus one might conclude that there are many different concurring ideas of identity that are present in Rwanda today as both local, ethnic and national affiliations are present.

Over the last decades a very large number of so-called ‘violent ethnic conflicts’ has occurred. What I am interested in mostly in this research is what happened to a society and to the national identity of its individuals after such a conflict. My central question therefore is: What makes that in post-conflict situations violence suddenly does not occur anymore? My descriptive questions in order to answer the main question would then be: if any, what form of identification is dominant in Rwanda; ethnic national identity, civic national identity or local affiliation? And what influence does this have on the social capital and likelihood of violence in Rwanda?    

2. Theory and hypotheses

To give a summary of theories that are relevant to this research I will first elaborate on two different understandings of national identity; ethnic and civic national identity. It is important to know which view on national identity is dominant in a country because this is what shapes the way people think about each other and their nation and it also plays an important role in the shaping of social capital. Social capital is the next concept elaborated on in this research. After all, a society with a large social capital is safer and less violent, thus it is an extremely relevant term in this research (Galea et al, 2002). Hereafter I will draw up some theories written on violent conflict and post-violent conflict development. These two terms I will firstly approach mainly theoretically and then afterwards apply them to the case Rwanda. Lastly I will elaborate on the issue of trust inside a nation and what influence this has on social capital. This is all in order to provide this research with a theoretical framework for the issue of post-violent conflict situations and the likelihood of violence in such conflicts.

2.1 Ethnic and civic national identity

A lot of research has been done on the difference between ethnic and civic national identity and the influence this has on societies (Brubaker, 1992;

Reeskens & Wright, 2013). Civic national identity is in its core a form of national identity that reaches out to outsiders; members of “the out-group” can become member of the group relatively easily as long as they obey the basic values and norms of the country (Brubaker. 1992:14). Ethnic national identity is then more exclusive and inward looking. Ancestors from the home country are extremely important when wanting to become member of the “in-group”. For immigrants it is therefore next to impossible to become ‘full members’ of society (Reeskens & Wright: 2013:157). These two different understandings of national identity have been studied a lot, especially in relation to societies where immigration levels have risen massively over the past few decades. The general thesis is now that people with a more civic understanding of national identity will exhibit more trust in people in general. This trust is widely concerned as one of the most important values of social capital (Reeskens & Wright, 2013:157).

Testing this thesis in Rwanda is interesting because, apart from refugees that returned to Rwanda after the conflict, there has not been a lot of immigration to the country. Both the ethnic groups could thus have an ‘equal claim’ on the national identity. In this context it is fascinating to see whether it is still the case that people with a more civic understanding of national identity have a more positive attitude to what they consider out-groups than people with a more ethnic understanding of national identity do. The extra dimension that I would like to add to this research is local identification because especially in Rwanda, due to the violent conflict, there might be a large group that strongly affiliates itself with its own smaller community, in which case the in-group may be even narrower and social capital levels may be even lower than in the case of ethnic national identity.

2.2 Social Capital & Trust

Social capital is a quite complex concept that ought to be explained further. Robert Putnam, who has written a very large amount on social capital, defines the concept as follows: “social networks and the associated norms of reciprocity and trustworthiness” (Putnam, 2006:137). The core of Putnam’s social capital is that social networks have value both for the people that are in the networks as for their bystanders. Different forms of social capital create

different values. Networks of friends may create better health where civic groups can create a firmer democracy (Putnam: 2006:138). Social capital is also a relevant concept in this research because many studies have shown that in countries with a higher level of social capital children grow up more safely, better educated, healthier and that people have a higher life expectancy and their economy and democracy work better (Putnam, 2006:138). The lack of such social capital may thus be a good indicator for the absence or lack of all the things mentioned above and thus for the likelihood of violence in a region. The relationship between low levels of social capital and a rise in homicide and violent crime has often been proved already (Galea et al, 2002; Portes, 2000:60). There are different types of social capital; Putnam makes the distinction between bonding and bridging social capital: ‘bonding’ social capital (ties between people considered similar in some way) and ‘bridging’ social capital (ties between people who are unlike each other in some way)’ (Putnam, 2007:143). In this research I will focus on the bridging social capital and the level of general trust as indicators for social capital. In a society divided by ethnic borders I think it is especially the bridging capital that is relevant, because this indicates when social bonds between different groups emerge. I therefore find it interesting to see what level of social capital is found - both on an individual level and on a societal level - in a nation after a conflict as violent as in Rwanda.

Trust between groups and in other people in general can be great indicators for a less salient division between groups. Trust across and between groups is something that can only be created by a common sense of nationality, so some argue (Reeskens & Wright, 2013; Miller, 1995). Cross-group trust is, as said before, also seen as an indicator and an important value in social capital (Putnam, 2006:142). This cross-group trust is often associated with a more civic understanding of national identity in contrast to ethnic national identity. Trust, after a conflict like the one that took place in Rwanda, is a value that is extremely complicated to achieve and therefore it is all the more interesting to research what trust levels there are, 10 years after the conflict.

What I find is most interesting then and what I am trying to find out in this paper is what happens after a violent conflict. It is clear that the stereotypes

and ethnic distance that were created before and during the genocide still play a role in present day Rwanda and continue to shape people’s perceptions - despite the efforts from the Rwandan government (Wielenga, 2011:6; Longman, 2005:223). ‘Ethnic identity, although not primordial, is nevertheless a primary aspect of self-identity through the process of socialization. Thus ethnic identity cannot easily be ‘shaken off.’ (Wielenga, 2011:7). With this knowledge on the one hand, and on the other hand the general agreement that a sense of community - whether that be in the form of a national identity or any other form - is necessary for the reconciliation of the conflict and a sense of a common future makes it a very complex issue (Longman 2005:207). The extra dimension of whether a more civic or more ethnic understanding of national identity is dominant in the country makes the issue even more precarious. Just as salient ethnic identities and a large ethnic distance between groups makes violence more likely to occur, the same theory can be used the other way around. In a more complex social structure, characterised by different crosscutting categories, one person might belong to various different groups; ethnic, religious, occupational, with different in- and out-groups (Brewer, 1999:439). This causes social capital to be higher, more social networks to be present and attitudes towards the various out-groups are probably milder and more trusting while the same individuals might belong to an in-group in another category. Consequently, attitudes towards violence against other groups are more likely to be negative (Brewer, 1999:439). This sense of community or social capital that was mentioned before, seems very important for different ethnicities not to get caught up in a violent conflict (anymore).

2.3 Violent ethnic conflict

A lot of literature on ethnic conflict is about what role ethnic identities play in ethnic conflict and on different theories on how these ethnicities can be used by for example elites to mobilize people in a conflict (Gagnon, 1981). Less literature is written on why conflict does not occur in other societies where multiple ethnicities do live together. Many articles on post-violent conflict situations focus on policy and on development from a government perspective and thus give a more top-down approach (Junne & Verkoren, 2005; Hintjes, 2008). In this research I want to focus for the greater part on the

individual perception of ethnic identities and conflict in a post-violent conflict situation.

Before going into theories written on both policy and individual perception in the aftermath of violent conflict it is important to establish what is meant exactly by violent ethnic conflict and post conflict situations. The reason I want to use the term violent ethnic conflict rather than just ethnic conflict is that it would otherwise become very unclear when a conflict comes to an end. As Junne and Verkoren put it ‘there are few truly post-conflict situations’ (2005:1). For it is clear that there was a form of conflict in Rwanda long before the genocide in 1994 (Wielenga, 2011:3) and there is still some sort of tension present today, wherefore speaking of an ethnic conflict as such is not very helpful to the definitions in this paper. As I am interested in conflict situations after open warfare has ended I make the distinction between violent ethnic conflict and post conflict situations; this last term must always be accompanied by the side note that conflict has not really ended completely just yet. This discrepancy makes the topic all the more interesting as it makes it even harder to understand why people stop using violence when there is still clearly a conflict or at least a tension of some sort at hand.

Although the focus in this research paper is not primarily on the top-down policies implemented by the government in post conflict development it is important to mention both the theoretical aspect and the execution in Rwanda itself as it creates the context in which the individuals that did the survey live. Post-Conflict Peace building, ‘the crucial subsequent work of demobilizing ex-combatants, (re)building civil society institutions, creating conditions for economic and social development, or establishing political institutions to resolve and manage societal conflicts’, has been the focus of both NGOs, international organisations and national governments in recent years (Klause & Jütersonke, 2005:447). There has been quite a lot of critique on this development as there has not been a highly structural and systematic way of dealing with post-conflict situations (Klause & Jütersonke, 2005:448).

In Rwanda the government positions itself as an advocate against racial hate and forces of genocide. A new law introduced in 2003, the organic law,

introduced a new set of thought and speech crimes such as ‘divisionism, ethnic ideology and genocide mentality’ (Hintjens, 2008:10). Ethnicity of any form is illegal as is any form of identification or political expression. This all in order to work towards an enlightened Rwanda where racial mental maps do no longer play any role (Hintjens, 2008:10). This however at the same time leads to an extensive and somewhat exaggerated attitude towards anyone who thinks or acts differently than the government leaders at the moment. ‘Human rights NGOs, donors, ordinary individuals, politicians and former allies can find themselves declared enemies of the new Rwanda.’ (Hintjens, 2008:10). The Rwandan Patriotic Front (RPF) - in power since the end of the genocide - has been given the benefit of the doubt by the international community regarding their role in the genocide, and in the years after the genocide many organisations and countries - questionably out of guilt for not intervening during the genocide - worked together with the government in order to make sure that ‘never again’ such a thing would happen (Hintjens, 2008:11). The problem with this approach is that the RPF and the government have a monopoly on political correctness ‘since only the RPF can police divisionism, it follows that only the RPF can deliver real security for ordinary Rwandans in future.’ (Hintjens, 2008:11). It is very questionable whether this is the ‘right way’ to move forward from the genocide; there is at least a general consensus amongst researchers that the policies are far from perfect and that there are still a lot of things going on below the surface with regard to ethnicity (for further reading on ethnic tensions in present Rwanda read Hintjens, 2008).

2.4 In- and out-groups

In the field of social psychology a lot of research has been done on attitudes between different (ethnic) groups and the prejudices that are formed by in- and out-groups. Although it is not necessarily so that ‘in-group love means out-group hate’ the social differentiation between groups does provide an all-fertile ground for conflict to occur (Brewer, 1999: 435). This does not even mean that people have a negative attitude towards members of the out-group, though just that the more positive feelings as admiration sympathy and trust are reserved for members of the in-group (Brewer, 1999:438). There are however situations where the positive attitude towards the in-group is

directly related to a negative position towards the out-group. ‘A direct relationship between intense in-group favouritism and out-group antagonism might also be expected in highly segmented societies that are differentiated along a single primary categorization, such as ethnicity or religion.’ (Brewer, 1999:439). Especially when there is a dichotomous categorization, with just two groups, these situations give a high potential for conflict.

This dichotomous relationship between two ethnic groups is very clearly seen in Rwanda. There is a third minority, the Twa, but they have been largely marginalized in the conflict. ‘Rwanda in the 1990s was a harmonious country, but the dynamics of inferiority and superiority in identity politics continued’ (Wielenga, 2011:3). There was clearly a conflict before the genocide and the more salient ethnic identities became the more negative the attitudes between the two groups became. Negative stereotyping of ‘the other’ and superior stories of the in-group were very present in pre-genocide Rwanda. The Hutus claimed that they were the original inhabitants of Rwanda before the Tutsis came and took their land (Wielenga, 2011:5). These stories and stereotyping made the distance between the ethnic groups greater and ultimately this resulted in the extremely violent genocide in 1994. There were obviously a lot of other prominent factors that played a role in this development towards the genocide but the distance between the two groups and the prejudices that came with this were definitely a key in understanding what happened in Rwanda in 1994.

2.5 Hypotheses

In order to answer the main question and visualize my expectations I created a model (see figure 1). Previous research has shown that general ethnic national identity has a negative effect on social capital where civic national identity has a positive effect on social capital (Reeskens & Wright, 2013). The dimension of local affiliation has not yet been tested in this respect but my expectation would be that this would have a negative effect on social capital because people with a strong local affiliation tend to keep social networks very small and especially the bridging social capital and general trust I think would be quite low. The third step made in this model is the relation between social capital and likelihood of violence. Unfortunately data of the World

Value Survey could not provide empirical data for this as the questions on how people placed themselves in regard to violence to solve political and social conflict were not asked. But, given the theoretical arguments provided before I feel that it is safe to assume that a larger social capital has a negative effect on the likelihood of violence in a region.

3.0 Method

The following section will bring some clarity regarding the data from the World Value Survey and the operationalization of the abovementioned concepts. Firstly, I will exemplify the data collected during the World Value Survey (WVS) and the methods that were used for this survey. Secondly, I will explain each concept mentioned in fig. 1 and operationalization of these concepts in the WVS separately. Lastly I will elucidate the control variables added to the model to see whether there are no spurious relations. The conceptualisation of each concept is also schematically indicated in appendix 1. Each variable is also schematically mentioned with the range, mean and standard deviation in appendix 2.

The survey data used in this research was collected in the World Value Survey wave from 2005-2008. The World Value Survey is a cross-national survey that asks respondents various questions on a very large range of questions, from social status to the amount of trust you have in your neighbours. Between April and May in 2007 they conducted the survey in

Local  affiliation   _Ethnic   national   identity   Civic  national   identity   Social  capital     Likelihood   violence   _{Fig.  1}  

Rwanda amongst 1507 respondents. There were 1558 total contacts, but 1507 successful interviews (Website World Value Survey, 2013). Sampling for the interview was done through a random probability sampling technique. The country was divided in enumeration areas (EAs) and random sampling was used to select these EAs. Each province or region was taken into account and within each EA 6 to 8 interviews were conducted. The survey was translated into Kinyarwanda language and interviews were done face-to-face. Interviewers went through an intensive training course to make sure that everyone had the same understanding of every question. The operationalization of every concept was done by either a combination of different questions to create a larger scale or in some cases, if this was not possible, by a single question. In the following section I will explain every concept and its operationalization.

The attachment to the local community, or the local affiliation was measured through the statement ‘I see myself as a member of the local community’. This is one of the three independent variables. This question was asked opposed to statements like ‘I see myself as world citizen’ or ‘I see myself as citizen of the country’. The question had 4 options – 1, strongly disagree to 4, strongly agree – I reversed this variable in order to create a more intuitive scale, the higher the score the more local affiliation.

The division between ethnic and civic national identity, the other two independent variables in this research, was made by a range of questions asked in the survey. The main question was ‘in your opinion, how important should the following be as requirements for somebody seeking citizenship of your country? Specify for each requirement if you consider it as very important, rather important or not important’ (World Value Survey, 2005). People could then choose from the following answers ‘having ancestors from my country’ ’being born on my country’s soil’, ‘adopting the customs of my country’ and ‘abiding by my country’s laws’. The former two would indicate a more ethnic understanding of national identity and the latter two a more civic interpretation. The questions only had three options to choose an answer from – taking multiple questions to create a scale made the operationalization of the concept more reliable. Because it were two questions for each type of

national identity asked ina different phrasing this makes the scale even more reliable and therefore a quite convincing indicator of these concepts. The Cronbach’s alpha for the ethnic national identity scale was 0.735 (valid: 1484, missing: 23). This can be considered quite high and thus provides for a reliable scale. The civic national identity scale is a little less reliable; the Cronbach’s alpha indicated a reliability of 0,571 (valid: 1493, Missing: 14). This is not ideal but with the theoretical backup it is enough to at least give us an idea of what the influence of civic national identity is on social capital. Social capital, the dependent variable in this research, has a couple of different dimensions, as mentioned before. I will focus here on the bridging social capital and the generalised trust. Bridging social capital was measured through two different questions, both relating to trust. The question was ‘how much do you trust people from various groups. Could you tell me for each whether you trust people from this group completely, somewhat, not very much or not at all?’ (World Value Survey, 2005). People could choose from different options but for this concept I will use ‘people with another religion’ and ‘people of another nationality’ because these are clearly groups that are not the same as the respondent and therefore display bridging social capital best. For this matter it is important to know that most Hutus are catholic and most Tutsis are protestant (Foa, 7). The questions are thus very indicative for the tensions between these groups in contemporary Rwanda as well; this makes the data very reliable. The Cronbach’s alpha for the scale of these two questions is 0,679 (valid: 1464, missing: 43). It can thus be accepted that these two questions do indeed measure the same concept. Generalised trust is the other dimension of social capital that I want to mention in this research. This one is measured by the rather straightforward question ‘Generally speaking, would you say that most people can be trusted or that you need to be very careful in dealing with people?’ (World Value Survey, 2005). This question is dichotomous because it only has two options: ‘most people can be trusted’ or ‘need to be very careful’. To analyse this question, logistical regression would be the only real option, because the dependent variable would be dichotomous. But in this research the most relevant form of social capital is bridging social capital, as it is exactly the bonds between different ethnic groups that are interesting in this case. Generalised trust will only be

mentioned as supportive statements to give an indication of social capital. I will focus on the bridging social capital scale when it comes to the regression analysis.

My hypothesis stated that a firmer social capital would lead to a smaller likelihood of violence. Although the relationship between social capital and violence has been proven by a number of empirical studies before, it would have helped this study to have a specific operationalization of the concept of ‘likelihood of violence.’ Unfortunately this was not the case in the World Value Survey and this relationship has to remain theoretical for the moment, in the case of Rwanda. This and some of the ranges that are not ideal for regression analysis can be seen as slight weaknesses in the operationalization of the concepts but despite this I feel that with the data collected throughout the whole country and with most of the concepts firmly operationalised these can still provide a lot of insight in the situation in contemporary Rwanda regarding social capital and likelihood of violence. Especially because bridging capital in this specific case is an extremely important dependent variable as it places an emphasis on exactly that which makes Rwanda such an interesting case: the division between the two ethnic groups that was so violently made in the years leading up to and during the genocide.

In order to control whether the theoretical model created really does explain the variance in the empirical data I will add a couple of control variables to see whether they might explain levels of social capital a lot better and whether there might be spurious relations in the models. The first control variable is age. Especially in Rwanda this may make a big difference as there is a generation that experienced the genocide very vividly and a younger generation that did not. This could mean that the older generation would show less trust in people in general and in people with another religion and nationality specifically. In the World Value Survey, age has been recoded into different age groups. The first age group is 15-24; this group might show less of a trauma due to the genocide than the other groups, as they did not really experience the genocide.

Patriotism is another control variable I would like to research. The operationalization in the World Value Survey is through the question: ‘How proud are you to be Rwandan?’ Adding patriotism in general is a good control variable as it might show that the difference between ethnic and civic national identity is not as important as a general pride. A slightly different concept is a ‘sense of belonging’ to the country. This is operationalised as ‘ I see myself as a citizen of my country’ as opposed to ‘I see myself as a member of the local community’ that was mentioned earlier for local affiliation. Lastly, I want to control for subjective social class, this question asks of what social class respondents consider themselves to be part. The question was asked as follows: ‘People sometimes describe themselves as belonging to the working class, the middle class, or the upper or lower class. Would you describe yourself as belonging to the: upper class, upper middle class, lower middle class, working class or lower class?’ (World Value Survey, 2005). The social class respondents considered themselves to be a part of, may influence the level of social capital they exhibit. These three variables were operationalized through questions where people could choose from options 1, very proud (resp. strongly agree, or upper class) to option 4, not at all proud (resp. strongly disagree, lower class). Because this would give a slightly unclear relation when analysed in the regression model I reversed these variables in a way that it now means that the more people feel patriotic the higher the score will be. In this way the models read more naturally and intuitively. The same goes for sense of belonging and social class.

Lastly, I made sure that for each regression the same group of respondents was used in order to create a reliable regression in which no missing values could blur the results. When controlling for these missing values 1328 respondents remained that were applicable for this research and responded to all the questions used in this research.

4. Data Analysis

In this paragraph I will run through the different analyses performed. Firstly, I will look at the level of generalised trust in Rwanda compared to other countries in order to give a first indication of the level of social capital in Rwanda. Then in order to answer the first research question of what type of

identification is dominant in Rwanda, I will aggregate a frequency table and explain this. After this, the theoretical model as mentioned before will be empirically tested; do local affiliation, ethnic and civic national identity have an effect on the bridging social capital in Rwanda? First I will look at the model in total and then explain each variable separately. Lastly, I will add the control variables and see what influence these have on the model and what explanations we might have for those.

4.1 Trust

The data show an incredibly low level of generalised trust in Rwanda. 95% of the respondents indicated that you cannot be careful enough when dealing with other people. This is extremely high, in comparison: in the same wave of the World Value Survey 73% of the respondents in Russia and only 56% of the respondents in the Netherlands said you cannot be careful enough. This is obviously not an outcome that is extremely surprising, however it does show that the aftermath of the genocide still plays a very significant role in contemporary Rwanda. This also implies that the level of social capital in Rwanda is quite low, as generalised trust is an important indicator of this social capital, this gives some indication and expectation regarding bridging social capital. You would expect that this is not all too high either.

4.2 dominance of identification

To see what kind of identification is dominant in Rwanda I aggregated frequency tables of the different kinds of identification. Because civic and ethnic national identity were recoded into scales consisting of two questions they now have a range from 2 to 6. I determined that 2 to 3 would be the lower category, 4 the middle category and 5 and 6 the high category. This means that if respondents had 5 or 6 on the scale, I would consider them as having strong feelings of that form of national identity. I recoded these variables into these three categories for this table specifically. Later, for the regression, I will use the 6 point scale again, as this then can be used as an interval variable. Interestingly, only 2.4% of the respondents indicated to feel very strongly about civic national identity. This might bring forth interesting results later on when linking this to social capital. From this very low percentage the premature assumption that social capital might be quite low

can be made, but this will obviously need to be proven more firmly in further analyses. 10.7% of respondents found ethnic national identity a very important understanding of national identity. This is considerably higher than civic national identity but compared to one’s local affiliation still very low as 51,9% of the respondents said that they strongly agreed with the statement ‘I see myself as a member of the local community’ and another 47% agreed with this statement. This shows that the local affiliation of people is still quite strong in Rwanda, chiefly when compared to ethnic and civic understandings of national identity.

4.3 ethnic national identity, civic national identity and local affiliation

From the data we can see that, as expected, ethnic national identity has a negative effect on bridging social capital. This is however only a very slight negative effect and is also only significant in the first two models (see: table 2). When local affiliation is added to the equation, the model is not significant anymore. Interestingly, civic national identity seems to have no effect at all and is also not significant. Local affiliation seems to have only a very slight and insignificant positive effect on bridging social capital. The fact that the effect is positive is unexpected, but this effect is so little and insignificant that I cannot give any real meaning to this result. Apart from this, there is also

Table 2: influence of ethnic national identity, civic national identity and local affiliation on bridging social capital

Model 1 Model 2 Model 3

Ethnic national identity -0.073*

(0.048)

-0.065* (0.054)

-0.058 (0.054)

Civic national identity -0.018

(0.069) -0.011 (0.070) Local affiliation 0,046 (0.110) Constant 6.295 (0.163) 6.367 (0.205) 5.647 (0.485) R2 0.005 0.006 0.008 N 1328 1328 1328

Notes: All betas are standardized. Standard deviation indicated between brackets. *Significant for 0.05

hardly any variation in the data that are explained by this model as stated in the R-square. 0.005 or 0.008 is such a small number that it can be said that there is not really any variance that is actually explained in this model.

The only variable that seems to explain something and is significant is ethnic national identity. This means that the more ethnic someone’s understanding of the concept of national identity is, the less that person will have a feeling of social capital. This outcome is quite forward and in accordance with earlier research to this relation (Reeskens & Wright, 2013). More interesting then is why civic national identity seems to have no effect at all. One would expect that these two concepts are related and the exact opposite of each other and would thus give contrary results, but the positive effect of civic national identity seems to be completely absent in Rwanda. This could be because the level of general trust is so low that one could hardly speak of an existing social capital that people can build on. This is a generous assumption that needs further empirical research with maybe a broader operationalization of the concept of social capital and a greater emphasis on the active creation of social capital such as membership of organisations and the participation in voluntary work (Reeskens & Wright, 2013). The addition to the theory I suggested during this research, the influence of local affiliation on social capital, does not seem to explain anything according to this model. Feeling part of your local community, as it seems thus does not have any influence on the extent to which you trust people from another religion or nationality. The answer to the second descriptive question ‘And what influence does this [the dominance of either local affiliation, ethnic or civic national identity] have on the likelihood of violence in Rwanda?’ seems to be very unsatisfactory, namely that ethnic national identity has a slight negative effect and local affiliation and civic national identity do not have any effect at all. The question remains then, what does? If theory falls short and Rwanda seems to be very atypical when compared to other countries, what does provide a more satisfactory answer to how social capital and violence comes about? So far I only found a very slight negative effect of ethnic national identity on social capital but there must be a positive effect as well.

4.4 control variables

At this point in the analysis it might be of interest to see what the control variables add to the model and to see whether they might be able to better explain social capital and thereby likelihood of violence. Because civic national identity and local affiliation were not significant and did not have any positive nor negative effect I did not keep them in the analysis when controlling variables, so I did a regression analysis with bridging social capital as dependent variable and ethnic national identity, sense of belonging, patriotism, subjective social class and age as independent variables.

The first thing that stands out when looking at table 3 is that the models are a lot more significant. Apart from that we can see that the R-squared jumps up when adding sense of belonging to the model. Another jump is made when patriotism is added. But the R-squared is still relatively low in these models as well so it appears as if another explanation for social capital should be found in another direction than that of patriotism, but the effect of ethnic national identity stays significant in all the models, which shows that it does

Table 3: influence of ethnic national identity, sense of belonging, patriotism, subjective social class and age on social capital

Model 1 Model 2 Model 3 Model 4 Model 5 Ethnic national identity -0.073*

(0.048) -0.054* (0.048) -0.063* (0.048) -0.066* (0.048) -0.066* (0.048) Sense of belonging 0.125* (0.126) 0.133* (0.127) 0.133* (0.127) 0.133* (0.127) Patriotism -0.068* (0.130) -0.068* (0.130) -0.068* (0.130)

Subjective Social Class 0.037

(0.056) 0.037 (0.056) Age -0.004 (0.041) Constant 6.295 (0.163) 4.037 (0.524) 5.150 (0.689) 5.023 (0.695) 5.037 (0.703) R2 0.005 0.021 0.025 0.026 0.026 N 1328 1328 1328 1328 1328

actually have an effect and there is not a spurious relationship. So withstanding, this part of the hypothesis can be considered proven.

When taking a closer look at the variable ‘sense of belonging’ it looks like it has a positive effect on social capital. The less people feel as though they belong to the country the less social capital they exhibit, but interestingly the effect of patriotism is contrary. The less proud respondents are to be Rwandan the more social capital they show. This is a most interesting outcome, as it seems that a very mild form of patriotism such as ‘a sense of belonging’ has a positive effect on social capital where too much patriotism seems to work counterproductive. This is contrary to what theories on patriotism normally claim. It would therefore be very useful and relevant to do further research on the social capital in Rwanda as it seems to work very atypical from theories on social capital so far.

Age and subjective social class, as is shown in model 5, do surprisingly not have any significant effect and the effect hardly has any direction either, so it is safe to say that these two control variables do not change or add anything to the model.

When looking at the main research question and the descriptive questions, it looks like local affiliation is the most dominant form of identification in Rwanda but this does not have any effect on the sense of social capital of individuals. Ethnic national identity does have a negative effect on social capital as expected but this effect is only very slight and does not provide a satisfactory base for explanation of the variation. Sense of belonging to the nation was a control variable in this research but does have quite a large and significant positive effect on social capital, but patriotism on the other hand has a slight negative effect which poses the question how to unite this fact with theories written on patriotism so far. Further research on Rwanda and the social capital there is definitely necessary in order to provide for a more satisfactory explanation of the creation and existence of social capital.

5.0 Discussion

In this section of my research I will critically look at my own research and see what could have been done differently and better but was not done due time limits and data restrictions. Principally I will look at the concepts used in this research and see whether they could have been operationalised better. Secondly, I will take a look at the operationalization of the research question, as it was not ideal and I will also explain that due to some data that I unfortunately did not receive I could not conduct the research and analysis exactly as I had planned and would have wished to do.

Firstly, I will discuss the concept of generalised trust. It would have benefitted this research if another logistical regression with generalised trust as dependent variable was conducted. The simpler analysis that I did gave an indication and was informative and relevant for this research but it would have helped the operationalization of social capital if this logistical regression was conducted.

Social capital as a concept was quite narrowly operationalised as only bridging social capital because at first sight that was the most relevant for this research question. But when taking the outcome of the data analysis into account it would be interesting to see a broader definition with more emphasis on the creation of active social capital such as membership of organisations, churches, voluntary work etc. Especially when considering that the level of social capital is incredibly low in Rwanda at the moment and trust levels are low as well, it might be a good idea to see what creation of social capital has on these levels.

Local affiliation was operationalised through only one question that was not as suitable for the purpose of the research; it would have been better to have a distinction between Hutus and Tutsis and the extent respondents felt close to their ethnic group as opposed to their nation. This was impossible due to the lack of questions asked on ethnicity in the World Value Survey in Rwanda. This is not very surprising as it is a very sensitive topic in Rwanda at the moment and people do not feel all that comfortable talking about it. This was thus a very logical decision of the survey; not to ask this question directly but

through questions such as ‘trust in people with another religion’, which to some extent is a euphemism for the difference between Hutus and Tutsis. However for this research to be precise it would have been a pre to have the exact ethnical status of the respondents.

In the research question the focus is on the likelihood of violence in Rwanda when considering different aspects of national identity and the influence this has on social capital. The link between social capital and violence remains theoretical throughout this research, which is definitely one of the biggest weaknesses in this research as it is an essential part of the causal mechanism that is now not empirically tested.

The weaknesses in this research and especially in the general operationalization of the research question are largely due to a certain data set that would have made the research much better suited for the research question. The data collected by Timothy Longman et al. would have been ideal for the operationalization of the research question. They conducted a survey among 2091 respondents for their article ‘Connecting Justice to Human Experience: Attitudes toward Accountability and Reconciliation in Rwanda’ (2004). They aggregated a scale of ethnic distance that would have been very helpful. Ethnic distance is a scale aggregated from responses individuals gave to a series of questions on how comfortable they were with the presence of members of another ethnic groups in different situations (Longman et al, 2005:207). My expectation would have been that one’s affiliation with one’s own ethnic identity and ethnic group makes the feeling of ethnic distance greater. Another question that was specifically asked in this survey and that would have made the empirical link between social capital and violence a lot more explicit is whether it is appropriate for authorities to use arms in different circumstances. The larger the feelings of ethnic distance the more likely people might be to give a positive answer to this question (Longman et al, 2005: 220). It was a shame that even though I repeatedly contacted Longman and his authors no one responded apart from the co-editor of the book ‘My Neighbor, My Enemy’. He unfortunately could not help me any further as he was not in the possession of the raw data himself.

6.0 Conclusion

To conclude this paper I will give a short summation of each section and elaborate on the conclusions drawn from the data collected. I will look back at the research question and see whether a satisfying answer can be and if not how to proceed with further research to do so.

The theory of ethnic and civic national identity state that people with a more ethnic idea of national identity are probably more likely to show less social capital where people with a more civic nationalistic idea are more likely to show social capital (Putnam, 2007). The same goes for societies: societies in which civic national identity is dominant will have a firmer social capital than in societies in which ethnic national identity is the main source of national identity. Social capital is a very important concept that is often linked to violence; societies with a larger social capital show less violent crimes and homicides (Portes, 2000:60). It was therefore interesting to test this theory in Rwanda, besides that the concept of local affiliation was added in case people did not feel a national identity at all. Rwanda is an interesting case because after the very violent conflict and genocide in 1994 society had to be rebuilt almost from scratch. To see how social capital and violence relate to society today is therefore a very relevant and interesting thing to do that might provide us with insight in future ethnic conflicts as well.

The first descriptive question asked was whether local affiliation, ethnic or civic national identity was dominant in contemporary Rwanda. Civic national identity was shown to be very marginally present. Ethnic national identity was present but local affiliation was definitely and by far the most dominant identification-strategy in Rwanda with 51% feeling member of the local community. The second question was whether this influenced the social capital in Rwanda. Regression analysis showed that only ethnic national identity had a significant negative effect on social capital but this effect only accounted for very little of the variance in the data. When adding the control variables; sense of belonging to the nation, patriotism, subjective social class and age it was very interesting to see that especially sense of belonging to the nation and patriotism had a significant effect. Sense of belonging to the nation had a positive effect on social capital. This is in accordance with what theories

describe on the relation between patriotism and social cohesion and social capital. But interestingly enough, the stronger concept of patriotism, being proud to be Rwandan, had a negative effect on social capital; it was only very slight but still asks for further research on this topic.

The central question of how come violence does not occur any more in post violent conflict situations can definitely not be answered completely with the research that was done here. Ethnic national identity surely does have a negative effect on the social capital and this is thus a type of national identity that a nation in a post violent conflict situation wants to avoid. But the assumption that civic national identity has a positive effect on social capital is not one that can easily be made. There are a lot of other factors such as sense of belonging that play a part but still an utterly satisfying answer cannot be given so far.

The operationalization of the research question was not ideal either, mainly because the link between social capital and violence was a theoretical link that was not backed by any empirical data in this research. An additional data set collected by Longman et al (2004) would have made the operationalization a lot easier, especially considering the link between social capital and violence. Regardless, some interesting findings were made in this research and what is more is that these provide a lot of new insights that require further research to be done in this very area.

Appendix 1:

Concept Variable Question

National identity

Local community G020 I see myself as member of the local community

Ethnic national identity

G028 How important should having ancestors from your country be as requirements for somebody seeking citizenship of your county?

G029 How important should being born on my country’s soil be as requirements for somebody seeking citizenship of your country?

Civic national identity

G030 How important should adopting the customs of your country be as requirement for somebody seeking citizenship of your country?

G031 How important should abiding by your country’s laws be as requirement for somebody seeking citizenship of your country?

Social Capital

Generalised trust A165

Generally speaking, would you say that most people can be trusted or that you need to be very careful in dealing with people?

Bridging social capital

G007_35 I’d like to ask you how much you trust people from various groups. Could you tell me for each whether you trust people from this group completely, somewhat, not very much or not at all? People of another religion

G007_36 Same question as above with: people from another nationality

Appendix 2:

Variable Range Mean Standard dev.

Bridging social capital 2-6 5.885 2.031 Ethnic nationalism 2-6 3.212 1.167 Civic nationalism 2-6 2.9533 0.907 Local affiliation 1-4 3.516 0.522 Sense of belonging 1-4 3.761 0.443 Patriotism 1-4 3.775 0.432 Social Class 1-5 1.695 0.981 Age 2.49 1.362

Literature:

Brewer, M. B. (1999). The psychology of prejudice: Ingroup love and outgroup hate? Journal of social issues, 55(3), 429-444.

Brubaker, R. (1992). Citizenship and Nationhood in France and Germany. Cambridge: Harvard University Press.

Fao, Robert, (ahead of print), Social Capital and the rule of law: lessons from contemporary Rwanda,

Gagnon, V. P., Jr. (1994). Ethnic Nationalism and International Conflict. The Case of Serbia. International Security, 19(3), pp. 130-166.

Galea, S., Karpati, A., & Kennedy, B. (2002). Social capital and violence in the United States, 1974–1993. Social science & medicine, 55(8), 1373-1383.

Gellner, Ernest (1981) Nationalism, Theory and Society, 10:6, 753-776

Hintjens, H. (2008). Post-genocide identity politics in Rwanda. Ethnicities,8(1), 5-41

Junne, G., & Verkoren, W. (Eds.). (2005). Postconflict development: meeting new challenges. Lynne Rienner Publishers

Krause, K., & Jütersonke, O. (2005). Peace, security and development in post-conflict environments. Security dialogue, 36(4), 447-462

Longman, T., Pham, P., & Weinstein, H. M. (2004). Connecting justice to human experience: Attitudes toward accountability and reconciliation in Rwanda. My neighbor, my enemy: Justice and community in the aftermath of mass atrocity, 206-25

Miller, D. (1995). On nationality. Oxford University Press

Portes, A. (2000). Social capital: Its origins and applications in modern sociology. LESSER, Eric L. Knowledge and Social Capital. Boston: Butterworth-Heinemann, 43-67

Putnam, R. D. (2007). E pluribus unum: Diversity and community in the twenty‐first century the 2006 Johan Skytte Prize Lecture. Scandinavian political studies, 30(2), 137-174

Reeskens, T., & Wright, M. (2013). Nationalism and the Cohesive Society A Multilevel Analysis of the Interplay Among Diversity, National Identity, and Social Capital Across 27 European Societies. Comparative Political Studies, 46:2, 153-181

Smith, Anthony (1993) The Ethnic Sources of Nationalism, Survival, 35:1, 48-61 Stover, E., & Weinstein, H. M. (Eds.). (2004). My neighbor, my enemy: Justice and community in the aftermath of mass atrocity. Cambridge University Press

Tilly, Charles (2006) ‘Political Identities in Changing Polities’, Identities boundaries and social ties

Wielenga, C. Genocide and identity: Stereotyping,‘othering’ and violence in Rwanda

Data sets:

WORLD VALUES SURVEY 2005 OFFICIAL DATA FILE v.20090901, 2009. World Values Survey Association, www.worldvaluessurvey.org