• No results found

What are the effects of being abducted into a rebel group? A case study in Uganda

N/A
N/A
Protected

Academic year: 2021

Share "What are the effects of being abducted into a rebel group? A case study in Uganda"

Copied!
65
0
0

Bezig met laden.... (Bekijk nu de volledige tekst)

Hele tekst

(1)

Graduate School of Social Sciences MSC Political Science – International Relations

What are the effects of being abducted into a rebel group?

A case study in Uganda

Author: Wendeline van Seventer Student number: 10538895 Supervisor: Abbey Steele Second reader: Sijeong Lim University of Amsterdam Master thesis Political Science – International Relations Submission date: June 20, 2018 Word count: 14,517 wendeline.vanseventer@student.uva.nl

(2)

Abstract

The purpose of this research is to investigate the effects of abduction into a rebel group, with a focus on social, economic, and political effects. This quantitative research is conducted using the Survey of War Affected Youth (SWAY) which consists of two phases, SWAY 1 (2005-2006) and SWAY 2 (2006-2007), which are merged in the dataset. Abducted people and non-abducted people are then selected and compared (N =1752) using the SWAY dataset. A potential gender difference regarding the effects of abduction is studied as well. From the regression analysis, it can be concluded that abduction has significant positive social effects for abducted people as well as significant negative economic effects for abducted people. It can also be concluded from the regression analysis that abduction has significant positive effects on the political participation of men, but not women. This describes the first gender difference finding regarding the effects of abduction, while the second difference is that abduction has a significant negative effect on feelings of family connectedness for men, but this does not hold for women.

Keywords: Civil war; abduction; effects of abduction into rebel groups; Uganda; LRA.

Acknowledgments

I would like to thank my supervisor, Dr. Abbey Steele, my family and my friends. These people have given me feedback and support during this whole process and I am grateful for that. This master thesis would have been unimaginable without them.

Wendeline van Seventer Amsterdam, June 2018

(3)

Table of Contents

1. Introduction ... 4

2. Theoretical framework ... 6

2.1 Definitions ... 6

2.2 Possible influence of abduction ... 7

2.3 Social effects ... 8

2.4 Economic effects ... 9

2.5 Political effects ... 11

2.6 Missing in the literature ... 12

3. Background of the civil war in Uganda ... 13

3.1 Case selection of Uganda ... 13

3.2 Background of the civil war in Uganda ... 14

4. Methodology ... 19

4.1 Data ... 19

4.2 Method ... 21

4.3 Operationalisation of the concepts ... 22

4.4 Testing of hypotheses ... 26

4.5 Transparency and replicability ... 27

4.6 Limitations ... 29

5. Analysis ... 31

5.1 Descriptive statistics ... 31

5.2 Hypotheses and falsification ... 32

5.3 Analysis of the social effects of being abducted ... 32

5.4 Analysis of the economic effects of being abducted ... 33

(4)

5.6 Tables ... 35

5.7 Summary and explanation of the effects ... 39

5.8 Gender ... 39 5.9 Tables ... 41 5.10 Interpretation gender ... 45 6. Discussion ... 46 6.1 Discussion ... 46 6.2 Limitations ... 47 7. Conclusion ... 49 7.1 Conclusion ... 49

7.2 Policy implications and future research ... 50

Bibliography ... 51

(5)

1. Introduction

Since the end of the Second World War in 1945, more than a hundred civil wars have erupted worldwide (website Uppsala Conflict Data Program, 2018). Unfortunately, various rebel groups have committed abductions over the course of these wars, often involving children as well. Examples of such groups include the “Lord’s Resistance Army in Uganda, the

Revolutionary United Front in Sierra Leone, and the Islamic State in Iraq and Syria” (Stanton, 2017, p. 2). Thousands of people continue to experience abduction into rebel groups each year, which grants this topic significant relevance to today’s world (website DoSomething, 2018).

After experiencing combat and returning home, it is not all peace and harmony

awaiting those abducted, as they can encounter problems such as rejection by family members, unemployment, psychological trauma, and hostility (Annan et al., 2011). Many scholars such as Pham et al. (2008) primarily focus on these negative effects of abduction into a rebel group. However, according to others, positive effects can also result from such a traumatic

experience, namely personal growth and greater political participation (Blattman, 2009, p. 231). Extensive research has been conducted regarding voluntary participation in rebel groups, such as the article by Collier & Hoeffler entitled “Greed and Grievance in Civil War”, which researched the motivations of such volunteers (2004). However, there is a limited amount of literature concerning those abducted into rebel groups, despite its high rate of occurrence each year. Notable exceptions however include research by Annan et al. (2011) and Blattman (2009), whose contributions are used in this thesis.

Blattman compared abducted youth to non-abducted youth in Uganda, by using a survey and semi-structured interviews. He focused on political participation and presented evidence of a link between past violence and increased political engagement among ex-combatants (Blattman, 2009, p. 231). Blattman only examined men, and therefore this current research contributes to Blattman’s by replicating his study for both men and women. Annan et al. (2011) studied the effects of war on abducted people in Uganda, as well as whether such effects vary by gender. They found low levels of post conflict hostility for abducted people, and they also found gender differences regarding the effects of war on abducted people, such as concerning the degree of family rejection. Annan et al. (2011) researched many different variables in their study, while this study entails a more specific focus on three categories of effects of abduction: social, economic, and political.

(6)

This thesis aims to answer the following research question:

What are the effects of being abducted into a rebel group?

As stated above, I focus on the social, economic and political effects of abduction into rebel groups, using Uganda as a case study in order to answer the research question. There are competing definitions of a case study; but the present research uses that proposed by Gerring: “A ‘case study’, … is best defined as an intensive study of a single unit with an aim to

generalise across a larger set of units” (Gerring, 2004, p. 341). Since the 1990’s, Uganda has experienced several civil wars, and “since the independence in 1962, Uganda is yet to witness a democratic handover of power from one leader to another” (website Peace Insights, 2018). Uganda provides an appropriate case to study regarding the research question, due to the thousands of people who have been abducted into the Lord’s Resistance Army (LRA). In addition, the LRA’s ongoing activity in countries other than Uganda makes this subject especially important to study. This research uses the dataset “Survey of War Affected Youth” (SWAY), which consists of data gathered in Uganda between 2005 and 2007. The

methodology section offers a more extensive description of this dataset.

Most literature examining civil war focuses on men in combat or on the victimhood of women in this context, such as the work of Brownmiller (1975). Nevertheless, women can also play a violent role in conflict (Wood & Thomas, 2017). For this reason, this study examines the roles played by both abducted men and women while in the LRA. The use of ‘role’ in this context refers to the activities undertaken by abducted people during their time with a rebel group, such as taking care of children or looting villages. Furthermore, this research examines whether these roles influence men and women differently and thereby contributes through investigating the effects of abduction into a rebel group while considering gender differences.

This thesis is organised into seven sections, beginning with the introduction in the first section. The theoretical framework is presented in the second section as well as the

hypotheses of this research. In the third part, rationale is provided for the case selection of Uganda, along with a brief background of the conflict. The fourth section presents the

methodological approach and describes the methods and variables used to test the hypotheses. In the fifth section, the results of the research are analysed while the sixth discusses the limitations and disadvantages of this research. Finally, the conclusion provides a summary of the findings as well as recommendations for future studies on abduction into rebel groups.

(7)

2. Theoretical framework

In this section, the existing literature on civil war and abduction is presented.

2.1 Definitions

Ross states the following as a common definition of civil war: “A conflict between a government and an organized rebel movement that produces at least one thousand battle-related deaths” (Ross, 2004, pp. 47-48). This definition is used in this research, combined with the definition by Kalyvas: “… armed combat taking place within the boundaries of a recognized sovereign entity between parties that are subject to a common authority at the outset of hostilities” (Kalyvas, 2018, p. 1). It is important to know what is meant with civil war in this thesis, because abduction into rebel groups often happens during civil war.

Unfortunately, sexual violence is a frequent occurrence during civil war. Between 2000 and 2009, there were 177 armed groups involved in African civil wars and 41 per cent of these are known to have engaged in rape, according to reports (Nordås, 2011, p. 3). Besides rape, sexual violence also includes: “… sexual torture and mutilation, sexual slavery,

enforced prostitution, enforced sterilization, and forced pregnancy” (Wood, 2009, p. 133). In many civil wars, both non-state armed groups, such as rebel groups, and state militaries perpetrate sexual violence (Wood, 2009). In some conflicts, sexual violence has been committed on a large scale, while it has been very limited in others (Wood, 2006, pp. 309-318).

It is estimated at the present time there are 200,000-300,000 child soldiers in the world, of which 70-80% are boys and 20-30% are girls (website Human Rights Watch, 2018). Once part of a rebel group, children are often politically indoctrinated and some receive military training (Ibid.). In this thesis, the words ‘abducted people’ and ‘non-abducted people’ are used. Of the abducted people referred to in this research, some were children (under 18 years of age) at the time of the abduction, and others 18 years and older when they were abducted. This research does not only focus on child soldiers; therefore the term ‘abducted people’ is used to include both children and adults who have been abducted. In the case of Uganda, the terms ‘abducted person’ and ‘ex-combatant’ are used to identify a person abducted by the LRA, and who has been part of it. Considering the fact that only a very small number of people voluntarily join the LRA, it can be assumed that LRA ex-combatants are abducted

(8)

people. When this is not so, for example in the case of the leader of the LRA, then this will be explained clearly.

2.2 Possible influence of abduction

Rebel groups are often comprised of both people who have been abducted as well as people who voluntarily joined. Abduction likely results in similar effects for these groups, such as isolation from family and friends for a period of time or an inability to work or study while with the rebel group. However, abduction might entail specific outcomes beyond the influence of participation into a rebel group.

First, one difference is that abducted people are forced to join a rebel group and

thereby also have their freedom stolen. This might lead to traumatic experiences for abductees and I expect that this can influence their social lives in a negative way. For example, a family or community can feel ashamed due to an abduction, and when the abducted person returns this can result in family rejection or community rejection (Annan et al., 2011). Accordingly, I expect that abducted people might experience greater social difficulties than people who voluntarily joined a rebel group.

Second, if a person is afflicted with psychological problems due to abduction, it might be more difficult to find a job and provide for themselves. I therefore expect that abducted people might experience greater economic difficulties than people who voluntarily joined a rebel group.

Third, social cohesion within armed groups might be influenced by abduction. The use of press ganging or abduction as a recruitment strategy usually results in a low level of social cohesion in an armed group (Cohen, 2013, p. 464). The reason is that combatants do not know each other well and are thus less likely to help each other if needed (Ibid.).

Fourth, armed groups that use abduction or press ganging are more likely to engage in rape than armed groups who recruit through other means (Cohen, 2013). A reason is that armed groups with forced membership have low social cohesion and combatants who are therefore less socialised than combatants who voluntarily joined armed groups. According to Cohen (2017), public group violence can be used to socialise fighters since it can increase the social cohesion within a group, making it a potential strategy to be used by armed groups who use abduction or press ganging as a recruitment strategy.

Fifth, traumatic experiences might influence the political participation of abducted people (Blattman, 2009, p. 244). As a result of their traumatic experiences such as being forced to use violence, I expect that abductees might be more politically active than people

(9)

who voluntarily joined a rebel group. This theory of Blattman is further explained in Section 2.5.

Finally, due to the above reasons the consequences of being abducted into a rebel group might differ from those of voluntary participation.

2.3 Social effects

Being abducted into a rebel group can have social effects on abductees, according to Annan et al. (2011), who state the issues of rejection from family members, psychological trauma, and hostility amongst others (Annan et al., 2011, p. 877). According to Annan et al.: “… violence drives social and psychological problems …” (Ibid.). An example of a known psychological problem is the experiencing of nightmares about the rebel group by abducted people post-combat. Annan et al. also state that most women who return from armed group conflict reintegrate socially, but that there are those who do not. They state that men reintegrate less well, socially, than women (Ibid., p. 877).

Annan et al. use the following conceptual definition of reintegration, though with the proviso that “reintegration eludes easy definition” (Ibid., p. 879): “At a minimum, it implies some resumption of livelihoods and social relationships either to the life led before war or that of non-combatants peers” (Ibid., p. 879). The article of Annan et al. (2011) shows that there is likely a social difference post-combat between abducted people and non-abducted people, in that abducted people face problems with social acceptance such as that they possibly have a smaller chance to marry and there is stigmatization by communities (Ibid., p. 881). Annan et al. explain that there are scholars who state that “ex-combatants may pose a threat to peace because they are more likely to engage in violence”, and that this is one of the reasons why abducted people are thought to have social problems after abduction (Ibid.).

Another possible negative social effect of abduction is family rejection. Annan et al. (2011) state that women face this problem more than men (Annan et al., 2011, p. 878): “Women and girls are more likely to face family rejection, especially when they are sexually abused and/or have children with rebels”. There are families who reject the abducted people because they have committed atrocities while they were in a rebel group or because, and this is especially true for girls and women, they had sexual experiences without being married. Besides family rejection, abducted people can also experience community rejection, as this quote from a 15-year-old boy, abducted by a rebel group, illustrates:

(10)

“They like calling me names saying I am a rebel and this is because I was abducted and came

back after 1 year in captivity and some children with whom I was abducted have never

returned and their parents feel bad seeing me back home so they tell their children not to play together with me” (Annan et al., 2009, p. 652).

There is a relationship between levels of violence and problems reintegrating with families and communities, for those who have been abducted (Annan et al., 2009); both with regard to violence experienced and perpetrated. The more violence an abducted person has experienced or perpetrated, the more of a problem they will have reintegrating with family and community (Ibid.). Because of this, violence is used as a control variable in this research.

Annan et al. state that for some abducted people, it is helpful to undergo a cleansing ritual in order to be accepted again by their family and their community (Annan et al., 2009, p. 660). It has also been found to be helpful for the abducted to spend time both with

non-abducted people, and with other former abductees with whom they can share feelings and offer mutual support (Ibid.).

Based on the theories discussed above, the following hypothesis is suggested:

Hypothesis 1: Abduction has negative social effects for those who have been abducted.

In this research, ‘social’ is measured using different factors to determine whether a respondent is part of a group, such as a sport group or a music group and to show the extent to which a respondent feels connected to their community and family. These factors give an impression of the respondent’s feelings of connectedness with other people and whether a respondent is part of a social group.

2.4 Economic effects

Another aspect of abduction is its negative economic impact on the victim. Abducted people are more likely to be unemployed than non-abducted people and this is especially true for men (Annan et al., 2011, p. 877). Abducted people cannot go to school or university when they are in a rebel group, and are also cut-off from the labour market. Consequently, an economic gap emerges between abducted people and non-abducted people. This situation is explained by Becker’s theory of human capital which confirms that one of the factors that has a positive influence on income is level of education (Becker, 1962). Becker’s theory

(11)

education costs time and money, and does not immediately lead to gain. However, in the long term, human capital is being accrued, which includes the knowledge and skills gained

through education and which can be used on the labour market. Becker states that with increasing levels of education, human capital also increases, and with that, so does income (Ibid.).

When people are abducted into a rebel group they cannot obtain an education during the abduction. Therefore, they potentially miss out on human capital in the long term. Consequently, with this likely loss of human capital, the abducted person will likely have a lower income than non-abducted peers. In the case of Uganda, females have fewer

opportunities as citizens than males, including those of school attendance and learning an occupation, and therefore the difference might be less pronounced in terms of the impact of being recruited by a rebel group (Annan et al., 2011, p. 877).

The following example illustrates why finding employment is harder for an abducted person than for a non-abducted person. There were displacement camps in Uganda, where citizens were forced to live by the government in order to protect them from the LRA. In these displacement camps, everyone feared the rebels, especially those who had escaped from the LRA, because if spotted, they could be killed by them (Annan et al., 2009, p. 649).

Consequently, there were restrictions for ex-abductees in terms of where they could be seen, which made it harder for them to make a living. Some people started their own businesses, but only the fortunate amongst these received financial support from family or friends to do this (Ibid.). In addition, some of the ex-abductees returned from combat with physical injuries, which made it harder or impossible for them to make a living, “especially because the majority of work in this region involves manual labor” (Ibid.).

Based on the theories explained above, the following hypothesis is suggested:

Hypothesis 2: Abduction has negative economic effects for those who have been abducted.

In this research, ‘economic’ is measured using different factors, to determine whether a respondent has a job, how many days per month they work on average as well as their daily earnings. In addition, ‘economic’ is measured regarding whether a respondent owns assets or any livestock. These indicators give an impression of a respondent’s economic conditions, by showing whether this person earns income or other ways forms of livelihood support such as livestock.

(12)

An opposite view to that discussed above is provided by Blattman, who states that there are also those who learn skills while with the rebel group, such as speaking in public (Blattman, 2009). After combat people can use such skills in their lives and also on the labour market, meaning they might not in fact miss out on human capital and income. The skills that can be learned by those abducted into rebel groups are discussed more extensively in the next part, which is about political effects.

2.5 Political effects

Abductees are often portrayed as dangerous people who can threaten the stability and growth of society (website BBC, 2007). This attitude is not always justified, as shown by Blattman (2009), who contrasted the political participation of Ugandan ex-combatants to non-abducted youths. As stated in the introduction, Blattman presents evidence for a link between past violence and increased political engagement among ex-combatants (Blattman, 2009, p. 231), his survey data suggest that “… abduction leads to substantial increases in voting and

community leadership” (Ibid.). The reason he gives for this finding is that these

ex-combatants have witnessed high levels of violence, and that their political participation and personal growth are positive effects springing from this trauma – though these patterns are not simply demonstrated by “… conventional theories of participation” (Ibid.). According to Blattman, ex-abductees are more likely to lead in their communities and to vote than non-abducted people (Blattman, 2009, p. 244).

Blattman suggests that violence can lead to personal growth, and that former

abductees have newfound self-control, confidence and ability (Ibid.). Annan et al. (2009, p. 649) also state that former abductees can play important roles in society: “Despite the limited economic and educational opportunities, some youths found ways to engage actively and establish meaningful roles – in business, dance groups, and even local government leadership”.

As confirmed by Blattman, this phenomenon of a strong link between traumatic events and political participation is also observable in other countries than Uganda. Similar patterns have been found in Indonesia, and also in Sierra Leone: “civilians whose household

experienced a killing, maiming or displacement are more likely to attend community meetings; more likely to join political groups; and more likely to vote than peers” (Blattman, 2004, p. 244).

(13)

Based on the above discussion, the following hypothesis is suggested:

Hypothesis 3: Abduction has positive effects on the political participation of those who have been abducted.

In this research, ‘political participation’ is measured using two factors, to determine whether the respondent voted for the referendum in 2005 and whether the respondent holds a political position. These factors show the extent of a respondent’s political participation by

demonstrating the respondent’s active political participation (holding a political position or not) as well as passive political participation (having voted or not).

2.6 Missing in the literature

In this research, the effects of being abducted into a rebel group are investigated. There is abundant research material on the motives for joining a rebel group voluntarily, but limited research on the effects of abduction into a rebel group, with the notable exceptions of the work of Annan et al., and Blattman. This research therefore aims to help fill the existing gap in the research. As stated in the introduction, Annan et al. (2011) researched many different variables in their study, while this study focuses on three categories of effects: social, economic, and political.

Furthermore, the focus in much of the research has been on the negative impacts on young combatants, but it is possible that there might also be positive impacts for abductees, which is shown by Blattman (2009). The research question leaves space for both negative and positive effects of being abducted into a rebel group.

The next contribution of this research is as follows. Most of this subject’s literature concentrates on the experiences of men in combat with little focus on female combatants, and when women are included in the literature, they are often portrayed as victims. This research shows that both men and women can play a role in combat. In addition, the current study replicates the research of Blattman (2009) but with the key difference of researching the experience of both boys and men as well as girls and women.

(14)

Figure 1. Unemployment rate

Uganda (% of total labor force)

3. Background of the civil war in Uganda

In this section, the rationale for the case selection of Uganda, and a brief background of the conflict, are provided. The case of Uganda is used to test the research hypotheses.

3.1 Case selection of Uganda

It is important to note that although the LRA in Uganda is a unique rebel group, it shares some characteristics with others in the world who also abduct people, use violence against a state, and use violence against civilians. Therefore, this research seeks conclusions that may be applied and generalised to other rebel groups in the world. If, for example, it is found that abduction into the LRA results in a reduced chance of employment, this would emphasise the importance of assisting ex-abductees in finding work post-conflict in other countries where rebel groups have abducted people as well.

There are also post-conflict conditions that could influence the outcomes of this research, such as the

unemployment rate. SWAY was conducted during a time when the unemployment rate increased in Uganda. In 2005, the unemployment rate of the total labour force was 2% and 8% in 2007 (Figure 1). This unemployment rate increase could influence the economic impacts on the lives of respondents.

Source: World Bank (2018)

(15)

3.2 Background of the civil war in Uganda

Uganda was a British Colony until it became independent in 1962, when Milton Obote was installed as president. In 1971, Idi Amin seized power in a military coup, and began a brutal regime characterised by acts of cruelty such as throwing his enemies to the crocodiles. His regime lasted for eight years and was held responsible for the deaths of 300,000 people. In 1979, Tanzanian troops drove Amin out and he fled the country (website Peace Insight, 2018).

Obote was reinstated at this point, but civil war broke out in 1981 and lasted till 1986. Yoweri Museveni led the National Resistance Army, now known as the Uganda People’s Defence Force (UPDF), and fought against the government troops of Obote. The battle was won by Museveni in 1986, and he was installed as the head of state thereafter. The regime of Museveni also proved to be ferocious, with numerous human rights violations. Northern Uganda in particular suffered from these violations, as a result of the many Obote followers who lived there (Ibid.).

In 1988, Joseph Kony, part of the ethnic group Acholi, started the rebel group the Lord’s Resistance Army (LRA) in Northern Uganda, in opposition to the Ugandan government (Annan et al., 2011, p. 641). At the time of its onset, there were various rebel groups active in Northern Uganda. ‘‘Most rebel groups were defeated by or settled with the new government by 1988, but Kony and a few remnants refused to surrender’’ (Ibid.). The Acholi in Northern Uganda suffered from different government for decades, and at the start of the LRA, there were some Acholi who joined the LRA voluntarily for this reason. However, the LRA became more violent against civilians and also against their own Acholi people, and therefore no Acholi people voluntarily joined the LRA anymore after a while, with a few exceptions (Ibid.). Because the LRA had few recruits, they started to abduct many people. During its time of operation in Uganda, the LRA abducted more than 60,000 people, especially children and young adolescents (website Peace Insight, 2018).

The LRA claimed to be based on the Ten Commandments of the Bible (Documentary “Kony 2012”). However, in an official LRA document, it is stated that the only aim of the LRA is to remain powerful and that they are not fighting for any other cause (Ibid.). The LRA was expelled from Uganda around the year 2006, after which they turned their attention to the Democratic Republic of Congo, the Central African Republic, and South-Sudan (website Volkskrant, 2016).

When the LRA had its strongest period, there were thousands of combatants, “… although the core of well-trained and experienced Acholi fighters may not have been more than a few hundred. Currently the LRA is estimated to have 150 to 300 armed fighters plus

(16)

hundreds of captive children and adults” (website Human Rights Watch, 2018). Other crimes committed by the LRA include:

- looting

- setting houses on fire

- letting people kill or hurt their own family and friends - raping

- killing - mutilating

- abusing people (physically, mentally and sexually) - exploiting communities and wildlife

- driving people away from their homes (Documentary “Kony 2012”).

Displacement camps

Unfortunately, many abducted people who went back to their families in Northern Uganda, could not go back to their original homes after combat (Annan et al., 2009, p. 643). “In 2002, the Ugandan government coerced the population into displacement camps, arguing that depopulation of the countryside would allow the government to provide adequate protection against the rebels. In 2006, there were 1.7 million people … in displacement camps, and most of the camps fell far below emergency standards for hygiene and availability of water” (Ibid.). Although the aim of the government was to protect the citizens in the displacement camps from the rebels, fear of the rebels was prevalent in the camps (Ibid., p. 649). This was especially true for those who had escaped from the LRA, in that if spotted, they could be killed (Ibid.).

International Criminal Court

President Museveni requested the International Criminal Court (ICC) to convict the leaders of the LRA in light of the atrocities that they had committed. The ICC, in response, issued arrest warrants for five LRA leaders, among them Joseph Kony and Dominic Ongwen. Kony was accused on 8 July 2005 of 12 crimes against humanity and 21 war crimes (ICC, 2018).

However, he was never arrested, and it is uncertain whether he still lives (website Volkskrant, 2016).

Dominic Ongwen, a commander of the LRA who deserted, was abducted by the LRA when he was a child. He is accused by the ICC of war crimes and crimes against humanity in

(17)

Northern Uganda (De Morgen, 2015). Additionally, he is accused of being responsible for slavery, mutilation, murder, looting, and deliberately attacking citizens (Ibid.). Ongwen surrendered in the Central African Republic in 2015, after being a fugitive for ten years. He is currently in ICC custody and his prosecution continues (website ICC, 2018). To date, a special court in East-Timor is the only international tribunal by which a child soldier, (a 14-year-old boy) has been convicted and was imprisoned. The ICC only has jurisdiction over those of 18 years or older, and is therefore unable to prosecute child soldiers (website ICC, 2018).

Other LRA leaders against whom arrest warrants were served by the ICC included Vincent Otti, Raska Lukwiya and Okot Odhiambo. The indictments against Lukwiya were withdrawn after the rumors of his alleged death were confirmed. It is suspected that Otti is also dead, but this has not been confirmed. Odhiambo has to date not been found. However, it can prosecute people who were child soldiers, when they reach the age of 18.

The conflict with the LRA in Uganda is Africa’s longest running conflict (website Invisible Children, 2018). Unfortunately, the media and the United Nations (UN) have paid little attention to this conflict in recent years, though some media attention was created by, amongst others, the “Kony 2012” campaign.

“Kony 2012”

The “Kony 2012” campaign was started by an American named Jason Russell, who could not understand why the conflict in Uganda had been going on for so many years without foreign intervention. Neither could he understand how Kony could abduct thousands of children without attempts being made by the international community to stop this. Russell believed more attention should be paid to this problem, both by the media and by governments. His “Kony 2012” campaign is a clear example of a civil society movement: “We could not wait for governments and institutions to step in” (Documentary “Kony 2012”). He wanted to make Kony famous, not as a tribute, but to show the world the kinds of atrocity he had committed.

The “Kony 2012” documentary went viral. Within six days of being uploaded onto You-Tube, it had been watched more than 100 million times. Russell went on to start a non-governmental organisation (NGO) called Invisible Children, which rebuilt schools, created jobs, and provided an early warning radio service to protect citizens from rebel attacks in Uganda. Invisible Children also tried to convince LRA soldiers to desert, offering instructions on how to do this safely. This is very important, because if combatants surrender peacefully, the LRA will weaken and these combatants can then be helped with reintegrating in society.

(18)

Russell, in addition, talked with the then American government to convince policy-makers to invest in the search for Kony, with the result that in 2011 President Barack Obama sent 150 American advisors to Uganda to help the soldiers of the African Union find him. In 2014, the Americans doubled their number of advisors deployed to this mission (website NOS, 2014). Although Kony has still not been found, it is hoped that this attention by the media and policy-makers to the conflict in Uganda will have been helpful in the fight against the LRA.

The LRA still poses a threat

In 2016, the American NGOs Invisible Children and The Resolve demonstrated that the LRA still poses a threat in the Central African Republic, despite reports of its demise (website Volkskrant, 2016). The LRA abducted 217 people between January and March 2016 in the South East of the Central African Republic; of these, 54 were children. This was despite the presence of a peacekeeping force comprising the UN, an anti-LRA unit of the African Union, Ugandan troops, and American advisors. People were also abducted by the LRA in the northeast of the Democratic Republic of Congo. Additionally, the LRA is known to have poached elephants in the Democratic Republic of Congo and to have traded the ivory with the Sudanese army for weapons and provisions in 2016 (Ibid.).

Summarising, the LRA is still active and poses a threat in the Democratic Republic of Congo, the Central African Republic, and South Sudan although they are not as active and lethal as they have been in the past. In those countries the LRA abducts people, poaches elephants, and commits robberies amongst other things. Therefore, it is highly relevant that the LRA is not forgotten, and that UN peacekeeping forces and African soldiers remain operational against this rebel group. Another way of weakening the LRA is by issuing an active call for a peaceful surrender of its combatants (Ibid.).

Recruitment method and different activities within the LRA

Abduction is the sole recruitment method used by the LRA, and the majority of abductees were abducted after 1996 and originate from the Acholi districts of Gulu, Kitgum and Pader in Northern Uganda (Blattman & Annan, 2010, p. 883). The LRA primarily abducts children through violent removal from their families, where roving groups are sent to rural homes to abduct children during night raids (Ibid.). Children are primarily abducted as they are easier to manipulate than adults (website Human Rights Watch, 2018). The small number of abducted adults are generally used as porters and rarely become fighters (Ibid.). Boys and men are primarily used as fighters and some have chores such as carrying objects or attending to

(19)

supplies. Girls and women are generally used as sex slaves and some are forced to marry a combatant and bear his children. Girls and women also perform chores such as cleaning, cooking and carrying objects.

These different roles for men and women in the LRA might entail different effects on the abducted people. An abducted girl experiencing excessive sexual violence in the LRA, might face more negative effects as a result, compared to an abducted boy who only had to carry objects and perform chores. This is a simplified example of gender differences within the LRA, but it is also possible for an abducted boy to experience more horrible incidents in the LRA compared to an abducted girl. However, the point is that there is a gender difference regarding people’s assigned activities while in the LRA. Consequently, the gender difference in these activities might lead to different effects on the abduction outcomes for men and women.

Some children who have attempted escape from the LRA have been murdered or mutilated by the LRA (website Human Rights Watch, 2018). Fortunately, there were also children who escaped from the LRA without getting caught. “Eighty-four percent of

abductees eventually escaped, usually in an unsupervised moment such as the heat of battle” (Blattman & Annan, 2010, p. 883). A portion of the remaining abductees died and it is estimated that there are hundreds of abductees still with the LRA (website Human Rights Watch, 2018).

(20)

4. Methodology

In this section, the methodological approach is described, and the methods, data, and variables used in this research are explained. In addition, the choices made in this research are justified in the operationalisation of the concepts.

4.1 Data

This study uses large-N quantitative analysis to test the hypotheses, which is a suitable approach since the aim of this research is to test causal effects of being abducted into a rebel group. The dataset used for this research originates from the Survey of War Affected Youth (SWAY). This was a research programme undertaken in Uganda, which gathered data between 2005 and 2007 in order to explain “the causes and consequences of civil war violence and child soldiering” (website Chris Blattman, 2018). Both abducted and non-abducted people were surveyed, which importantly allowed for comparison between these groups.

SWAY first conducted household surveys which were randomly selected from World Food Programme lists. There were 1200 households selected, some of which were

inaccessible or could not be located and therefore 1,018 households were surveyed. At the time of the survey, household heads were requested to list all the people that lived in their household in 1996, “a year commonly remembered as the year of the first national election since 1980”. Second, youth were randomly selected from these retrospective household rosters to be interviewed for an individual survey (SWAY Codebook, 2013).

SWAY consisted of two phases, SWAY 1 and SWAY 2, which are merged in the dataset of this research. The phases were conducted from July 2005 to March 2006 and between October 2006 and August 2007 respectively. SWAY was the conjunct project of American and Ugandan researchers as well as Ugandan volunteers among others (SWAY Codebook, 2013). “Both phases of SWAY were funded principally by the United Nations International Children’s Emergency Fund (UNICEF)” (Ibid.).

The respondents in SWAY 1 were between 14 and 30 years of age when surveyed, and only included boys and men. The respondents in SWAY 2 were between 14 and 35 years of age when surveyed, and regarded girls and women. The term ‘youth’ is used because the respondents were children or young adults at the time of their abduction. The surveyed respondents who had not been abducted were also children or young adults at the time when

(21)

the LRA was active in their region. The period of LRA abduction reported by the respondents was between one day and 12 years (Ibid.). In most cases, when rebel groups abduct people, they abduct children or young adults (Beber & Blattman, 2013). In this regard the LRA’s activity in Uganda is representative of other rebel groups, as the LRA primarily abducted children and young adults.

SWAY was conducted in two of the four Acholi districts principally affected by the war, Kitgum and Pader (SWAY Codebook, 2013), which are districts in Northern Uganda (Figure 2) where the LRA committed many atrocities. “The districts of study were chosen because they are two of the most affected in the region, with relatively little research … compared to similar neighboring districts” (Annan et al., 2009, p. 643).

The original SWAY dataset with SWAY 1 and SWAY 2 combined consists of 2260 people, with 1216 males and 1044 females, and includes abducted and non-abducted people. 508 abducted people from the original dataset never returned or had died and could

consequently not be selected for this research and therefore the number of respondents in the final dataset is 2260 – 508 = 1752.

(22)

Figure 2. Map of Uganda

Source: Codebook SWAY (2013)

4.2 Method

In this research, the unit of analysis regards the individual. The data are analysed using bivariate and multivariate regression analysis. The ‘SWAY’ dataset is used to test several hypotheses and is statistically analysed through the SPSS computer program. First, the descriptive statistics are analysed and an overview of the variables is provided (Table 1). Second, the bivariate regressions are presented, with the independent variable (X) ‘being

(23)

abducted’ and the dependent variables (Y’s) ‘social’, ‘economic’, and ‘political’. W is a vector of the control variables. The formulas are as follows:

Y social = X + W + error Y economic = X + W + error Y political = X + W + error

The expectation was that being abducted would entail a negative influence on the dependent variables ‘social’ and ‘economic’ and a positive influence on ‘political’. In the next step, control variables are added to the regressions since they are related to the dependent variables and therefore might influence them. I am not interested in the control variables in particular, but they are discussed if found to have a significant effect on one of the dependent variables. The expectation was that the independent variable ‘being abducted’ would still have an influence on the dependent variables ‘social’, ‘economic’, and ‘political’, including the control variables.

4.3 Operationalisation of the concepts

Dependent variables

‘Social’ is a term which can mean many things. However, in order to make this research more specific, there will be a focus on particular indicators of the social life of the respondents. In this research, what is meant with social is a combination of the following variables:

-Y social

Y social consisted of:

Ysoc1 = member of at least one group

The respondents were asked whether they are a member of at least one of the following groups:

- peace club

- drama, music, or dance club - farmers’ group or cooperative - water committee

(24)

- sports team

- member of church, prayer or Bible study group - any other community or church group

The respondent could answer with ‘yes’ or ‘no’.

Ysoc2 = number of groups the respondent belongs to

This variable indicates the number of groups, as indicated in the list above, a respondent belongs to. The respondents could give an answer from 0 to 7.

Ysoc3 = index of community connectedness

This variable indicates to what extent the respondent feels connected to his or her community. It rates from 0-6, 0 means low connected and 6 means highly connected.

Ysoc4 = index of family connectedness

This variable indicates to what extent the respondent feels connected to his or her family. It rates from 0-6, 0 means low connected and 6 means highly connected.

The way Y social is measured is appropriate because it shows how connected a respondent feels to both family and their community, which are indicators of someone’s social life. The question is asked to how many groups a respondent belongs, and the benefit of this is that it gives an idea how socially active a respondent is.

-Y economic

Y economic consisted of:

Yeco1 = did you work in the last 4 weeks Yeco2 = how many days work

Yeco3 = daily earnings

Yeco4 = index of asset ownership Yeco5 = if household has livestock

The respondents were asked: ‘‘Did you work in the last four weeks?’’ If the answer was ‘yes’, this counted in this research as ‘being employed’. If this question was answered in the

affirmative, the respondent was then asked: ‘How many days did you work in the last four weeks?’, which served as an indicator of normal working patterns. The respondents were also

(25)

asked about their daily earnings, and they could answer how many Ugandan Shilling they make on an average working day. The index of asset ownership is a standard normal

distribution, and it shows a relative index of asset ownership, because it compares the assets of a respondent with other respondents. For example: if a respondent scores 3.08 on index of asset ownership, this respondent has more assets than a respondent who scores -1.7 on this index. If a respondent scores 0,00 on this index, this person has an average amount of assets, compared to other respondents. Whether a respondents household has any livestock, was deemed a valid question to ask, since livestock is an important indicator of economic life in a developing country such as Uganda.

-Y political

Y political consisted of: Ypol1 = did you vote in 2005 Ypol2 = do you have a political job

The respondents were asked if they voted in 2005, which refers to a referendum that was held in Uganda on restoring multi-party politics. This is an appropriate way to measure political participation, considering it was the most recent event where a respondent could vote and it shows their passive political participation. Additionally, the respondents were asked if they hold a political position, which shows their active political participation. Both questions could be answered with ‘yes’ or ‘no’. There were also questions about political opinions, such as: “Do you have anti-democratic beliefs about forms of government?” Considering the fact that the effects of being abducted on political participation are researched in this thesis, and not the effects of being abducted on political opinions, the questions about the political opinions of the respondents were not selected. Because of this, more profound research can be carried out but the downside of this is that the research is less broad.

Independent variable

X1 = being abducted into a rebel group

Respondents were asked if they had ever been abducted, and had to answer with ‘yes’ or ‘no’. Consequently, a distinction could be made between abducted and non-abducted people. Whether the respondents answered this question honestly, will be discussed in Section 4.5.

(26)

Control variables

X2 = age at the time of the survey

This variable is used as a control variable, since age potentially plays a role regarding how certain events are experienced. Abduction could entail a greater effect on children than on adults, as the personality of adults is more formed by life experience compared to that of children before they were abducted. This variable is measured in years.

X3 = years abducted

This variable is used as a control variable, because it is assumed that the longer the period of abduction, the greater the post-abduction effects.

X4 = forced to be a wife in the LRA

This variable represents one of the roles a person could have in the LRA, although it is also possible for roles to overlap, such as a women who is forced to be a wife as well as forced to perform labour. Variables X5, X6, and X7 are used as control variables as they are potential indicators of the dependent variables, since someone’s role in the LRA might influence the types of activities they experienced while in the LRA.

X5 = forced to be a fighter in the LRA

This variable represents one of the roles a person could have in the LRA.

X6 = forced to do labour in the LRA

This variable represents one of the roles a person could have in the LRA.

I expect that the control variables X5, X6 and X7 have different effects on the outcome, since for example a fighter in the LRA has different experiences compared to someone who was forced to perform labour in the LRA. A fighter might experience more violence than a carrier for example.

X7 = violence

This variable regards all violence a respondent has suffered, witnessed or perpetrated. There were 25 questions asking about violence, such as ‘Did you experience someone being killed?’ If someone experienced 25 violent acts, this counts as 25, while if a person did not experience violence at all, this counts as 0.

(27)

This variable (as well as variable X8) are used as control variables because the more violence an abducted person has experienced or perpetrated, the greater their difficulty reintegrating with family and community (Annan et al., 2009). Therefore, X8 and X9 are potentially important indicators of Y social. It is also possible that non-abducted people have experienced violence and therefore this question is posed to all respondents.

X8 = experienced sexual violence

The respondents are asked whether they were sexually abused and whether they were sexually exploited. If they experienced both kinds of sexual violence, this counts as 2, while if they experienced only one kind, this counts as 1. If they did not experience sexual violence, it counts as 0.

X9 = has ever married

It is assumed that having been married has an effect on the social life of respondents, and therefore this variable is used as a control variable.

X10 = had children from forced marriage in the LRA

Some respondents replied ‘yes’ to this question, but there were no respondents who answered the question ‘Do you have children?’ in the affirmative. Therefore, this variable is used to indicate whether abducted people had children from forced marriage in the LRA. This is potentially an important indicator of Y economic, because it is assumed that having children has a negative effect on the economic life of respondents, since children are expensive.

X11 = years of education

As explained in the theoretical framework, education has a positive influence on income, and therefore education is used as a control variable. A respondent’s level of attained education is measured in years.

4.4 Testing of hypotheses

Quantitative analysis was used to test the hypotheses, and the data analysed by means of linear regressions.

(28)

Hypothesis 1

For this hypothesis, a bivariate regression was carried out with ‘being abducted’ as the independent variable and ‘social’ as the dependent variable. Hereafter, the control variables were added to the regression and a multivariate regression was carried out.

Hypothesis 2

In order to test this hypothesis, a bivariate regression was executed, with ‘being abducted’ as the independent variable and ‘economic’ as the dependent variable. The control variables were then added to the regression and a multivariate regression executed.

Hypothesis 3

For this hypothesis, a bivariate regression was conducted, with ‘being abducted’ as the independent variable and ‘political’ as the dependent variable. In the next step, the control variables were added to the regression and a multivariate regression performed.

4.5 Transparency and replicability

The sample of SWAY and the surveys are executed in an organised way. First, households were randomly selected for the household surveys in Kitgum and Pader in Northern-Uganda as these regions were principally affected by the war. Of the 1200 households sampled, 1018 were able to participate in the survey, which is 85 percent and therefore, the decline rate is 15 percent. Second, from these households youth were randomly selected for the individual surveys. “Each household had a probability of selection

proportional to the number of people listed in its household” (SWAY Codebook, 2013). The average time required to complete the survey was 90 minutes.

The household survey asked questions regarding pre-displacement characteristics, including holdings of various household assets, land, and occupation of the house’s head. Second, the survey consisted of questions concerning household members’ age, mortality, and abduction history as well as each youth’s present occupation, location, and level of education (Ibid.). War experiences were assessed in the individual surveys in two main ways. “First, detailed information was collected about any time spent with the LRA, including: time, place, duration, nature of the abduction; experiences as a forced wife or mother to children in the bush; initiation ceremonies; the respondent’s roles and responsibilities, rank and promotion; methods of control and discipline; and, finally, experiences with family, community, and NGOs upon return. Second, respondents completed a checklist of war experiences with an

(29)

interviewer based on a version of the Harvard Trauma Questionnaire that was adapted to local circumstances by the researchers (website Harvard Program in Refugee Trauma, 1998). In assessing well-being, the survey took a multidimensional approach, covering economic activity, physical and mental health, community participation, social support, and risky behaviour” (SWAY Codebook, 2013).

Furthermore, “all youth that were interviewed received a small gift of household items worth approximately a day to a half-day of wages” (Ibid.). The small gift is important to note as wealthy people possibly have more available time for the survey compared to poorer people, since the latter must often work many hours a day. Accordingly, the researchers gave the respondents a gift, which abates the worry of potentially oversampling wealthy people.

The respondents were reassured by the researchers of SWAY that the information they provided would remain confidential and anonymous, and it is highly likely that the majority of the respondents answered the questions honestly as a result. It is however unsure whether all respondents answered the questions honestly, since for example a respondent could have lied about the fact that he or she was abducted by the LRA due to the stigma of abduction. However, even if people are not honest I cannot conceive a systematic bias that would occur. For example, it is not that the people who lied about abduction are wealthier than those who were honest, which would influence the dependent variable ‘Y economic’ of the respondents. I am therefore unconcerned that there are potentially respondents who lied about their

answers in the survey.

I can conclude that the samples of SWAY are representative, since the households were randomly selected from World Food Programme lists, which is a reliable organisation. A further reason is because the World Food Programme lists, created in 2002 and 2003 represent “the earliest and most complete and accurate sample frame of households available in the region” (SWAY Codebook, 2013).

Finally, an important question to ask is the following: Would the effects of abduction into the LRA be similar or different in other rebel groups? The effects of abduction by the LRA might be similar for other rebel groups, such as the Revolutionary United Front in Sierra Leone or the Islamic State in Iraq and Syria, as these groups have also used violence against the state in which they are active and have also abducted people into their rebel group. However, there are also some rebel groups who do not abduct people, and consequently the results of this research are not applicable to such groups.

(30)

4.6 Limitations

The respondents were asked: ‘Did you work in the last four weeks?’ If the answer was ‘yes’, this counted as ‘being employed’ and a further question was asked: ‘How many days did you work in the last four weeks?’ This second answer indicated how often the respondents usually worked. The question: ‘Do you have a job?’ was deemed to be too direct in that it could have discomforted the respondent, given that many people in developing countries such as Uganda work in the informal sector. This informal sector often involves irregular and erratic working patterns, and therefore the question was formulated as: ‘How many days did you work the last four weeks?’

This question entails a downside, which is that, by specifically asking about the previous four weeks, a respondent who normally works full-time, but who had recently taken time off due to ill-health, for instance, would give an answer that failed to reflect their usual working pattern. However, this was deemed unlikely to be the case for the majority of

respondents. Furthermore, this is a standard question in developing countries, and therefore an appropriate one to ask.

(31)

Table 1. Descriptive statistics of the used variables N Minimum Maximum Mean SE Daily earnings

(Ugandan Shillings)

1752 0 83333.34 2427.39 5173.96 Days worked in last 4

weeks

1752 0 28 8.18 9.59

If household has any livestock 1752 0 1 .55 .498 Index of asset ownership 1752 -1.7 3.08 .000 1.000 Violence 1752 0 24 7.66 5.003 Number of groups 1752 0 7 .75 1.113 Index of community connectedness 1752 0 6 5.10 1.112 Index of family connectedness 1752 0 6 4.62 1.506

Age at the time of the survey 1752 14 38 22.52 5.86 Vote 1752 0 1 .47 .499 Political job 1752 0 1 .00 .068 Years abducted 1752 0 12 .43 1.21 Years of education 1752 0 16 6.04 3.41 Abducted (ref. =not) 1752 0 1 .491 .500 Gender (ref. = male) 1752 0 1 .503 .500 Worked in last 4

weeks (ref. = not) 1752 0 1 .714 .452 Had child from forced

marriage in the LRA (ref. = not)

1752 0 1 .017 .128

Ever married (ref. = not)

1752 0 1 .412 .492

Did experience sexual violence

(ref. = not)

1752 0 2 .107 .553

Was forced to be a wife in the LRA (ref. = not)

1752 0 1 .034 .180

Was forced to be a fighter in the LRA (ref. = not)

1752 0 1 .018 .134

Was forced to do labour in the LRA (ref. = not)

1752 0 1 .469 .499

Member of at least one group (ref. =not)

1752 0 1 .390 .487

(32)

5. Analysis

In this section, the results of this research are reported and analysed.

5.1 Descriptive statistics

The dataset includes 1752 people, of which 870 (49.7%) are men and 882 (50.3%) women. Of the 870 men, 530 were abducted, while of the 882 women, 330 were abducted (Table 2). This means that 860 (49.1%) respondents were abducted and 892 (50.9%) respondents were not. When the respondents were asked whether they had worked during the last four weeks, 71.4% answered in the affirmative. The daily earnings of the respondents were 2427.39 Ugandan Shilling on average, with a standard error of 5173.96. 55% of the respondents households owned livestock. The average age of the respondents at the time of the survey was 22.52 years with a standard error of 5.86. The respondents were abducted for .43 years on average with a standard error of 1.21. The average period of education of respondents was 6.04 years with a standard error of 3.41. 1.7% of respondents had a child from a forced marriage while 41.2% had been married. 1.8% of respondents were forced to be a fighter in the LRA, 46.9% were forced to perform labour, and 3.4% were forced to be a wife. On a scale from 0 to 6,

respondents scored on average 5.1 on the index of community connectedness and 4.62 on the index of family connectedness. 39% of respondents are a member of at least one social group. 47% of respondents voted in the referendum of 2005.

Table 2. Abduction and gender

Male Female Total Not abducted 340 552 892

Abducted 530 330 860

(33)

5.2 Hypotheses and falsification

If my hypotheses are correct, then I expect to observe: - A negative influence of abduction on Y social. - A negative influence of abduction on Y economic. - A positive influence of abduction on Y political.

5.3 Analysis of the social effects of being abducted

Ysoc1 = member of at least one group

A bivariate regression is conducted and shows the following results (Table 3):

Being abducted does not have a significant effect on being a member of at least one group. A multivariate regression is conducted out and shows the following results (Table 4):

Including the control variables, being abducted does not have a significant effect on being a member of at least one group. The control variable education has a significant positive effect on being a member of at least one group (P<.05).

Ysoc2 = number of groups to which you belong

A bivariate regression is conducted and shows the following results (Table 3):

Being abducted does not have a significant effect on the number of groups to which a person belongs. A multivariate regression is conducted and shows the following results (Table 4): Including the control variables, being abducted does not have a significant effect on the number of groups to which a respondent belongs. The control variable education has a significant effect on number of groups (P<.05).

Ysoc3 = index of community connectedness

A bivariate regression is conducted and shows the following results (Table 3):

Being abducted does not have a significant effect on the index of community connectedness. A multivariate regression is conducted and shows the following results (Table 4):

Including the control variables, being abducted has a significant positive effect on the index of community connectedness (P<.05). The control variable age has a significant positive effect on the index of community connectedness (P<.05). The control variable violence has a significant negative effect on the index of community connectedness (P<.05).

Ysoc4 = index of family connectedness

(34)

Being abducted has a positive effect on the index of family connectedness, an effect which is significant (P<.05). A multivariate regression is conducted and shows the following results: (Table 4): Including the control variables, being abducted does not have a significant effect on the index of family connectedness. The control variable years abducted has a significant positive effect on index of family connectedness (P<.05). The control variable violence has a significant negative effect on the index of family connectedness (P<.05).

5.4 Analysis of the economic effects of being abducted

Yeco1 = did you work in the last 4 weeks

The people that were abducted, worked 758 days in total over the last 4 weeks, which means 0.88 days per person on average. The people that were not abducted, have worked 731 days in total in the last 4 weeks, which means 0.82 days per person on average. It is remarkable that the abducted people have worked more days per person on average than the non-abducted, since the hypothesis suggests that being abducted brings negative economic effects. An independent sample T-test is conducted to compare the means of the two groups and it shows that the difference is not statistically significant.

A bivariate regression is conducted and shows the following results (Table 5): Being abducted does not have a significant effect on being employed. A multivariate regression is conducted and shows the following results (Table 6): Including the control variables, being abducted does not have a significant effect on employment. The control variable age has a significant positive effect (P<.05) on being employed. The control variable education has a significant negative effect on being employed (P<.05).

Yeco2 = how many days worked

A bivariate regression is conducted and shows the following results (Table 5): Being abducted does not have a significant effect on days worked during the last four weeks.

A multivariate regression is conducted and shows the following results (Table 6): Including the control variables, being abducted does not have a significant effect on

employment. The control variables age and violence have a significant positive effect on days worked during the last four weeks (P<.05).

(35)

Yeco3 = daily earnings

A bivariate regression is conducted and shows the following results (Table 5):

Being abducted does not have a significant effect on daily earnings. A multivariate regression is conducted and shows the following results (Table 6):

Including the control variables, being abducted does not have a significant effect on daily earnings. The control variables age, married, and education have a significant positive effect on daily earnings (P<.05).

Yeco4 = index of asset ownership

A bivariate regression is conducted and shows the following results (Table 5): Being abducted has a significant negative effect on the index of asset ownership (P<.05),

which is consistent with hypothesis 2. A multivariate regression is conducted and shows the

following results (Table 6): Including the control variables, being abducted has a significant negative effect on the index

of asset ownership (P<.05), which is consistent with hypothesis 2. The control variable education has a significant positive effect (P<.05) on the index of asset ownership.

Yeco5 = if household has any livestock

A bivariate regression is conducted and shows the following results (Table 5): Being abducted does not have a significant effect on owning livestock. A multivariate

regression is conducted and shows the following results (Table 6): Including the control variables, being abducted does not have a significant effect on owning

livestock.

5.5 Analysis of the political effects of being abducted

Ypol 1 = vote

A bivariate regression is conducted and shows the following results (Table 7):

Being abducted does not have a significant effect on voting. A multivariate regression is conducted and shows the following results (Table 8):

Including the control variables, being abducted does not have a significant effect on voting, while the control variables age and married have a significant positive effect (P<.05) on voting.

(36)

Ypol 2 = do you have a political job

A bivariate regression is conducted and shows the following results (Table 7): Being abducted does not have a significant effect on holding a political job. Unfortunately, it was not possible to conduct a multivariate regression with ypol2 as a dependent variable, and therefore, a conclusion cannot be drawn from this variable including the control variables.

5.6 Tables

Hypothesis 1 is tested in Tables 3 and 4, with the expectation that abduction has negative social effects. Hypothesis 2 is tested in Tables 5 and 6, with the expectation that abduction has negative economic effects. Hypothesis 3 is tested in Tables 7 and 8, with the expectation that abduction has positive political effects.

Table 3. Social effects of abduction

*p<.05 Social Group Number of groups Community connectedness Family connectedness Abducted -.014 (.023) -.076 (.055) .166 (.093) .186* (.078) Constant .395 (.016) .787 (.055) 5.036 (.056) 4.524 (.056) N 1752 1752 1752 1752

(37)

Social Group Number of groups Community connectedness Family connectedness Abducted .012 (.059) -.099 (.154) .297* (.131) .131 (.171) Age .005 (.004) .017 (0.101) .032* (.009) -.001 (.011) Years abducted -.018 (.027) -.070 (.010) .104 (.061) .175* (.079) Was forced wife -.053

(.144) .216 (.376) .293 (.319) -.024 (.416) Fighter -.068 (.139) .289 (.361) -.212 (.307) -.233 (.400) Labour -.068 (.068) -.020 (.177) .065 (.150) .001 (.196) Violence -.006 (.007) -.015 (.018) -.045* (.015) -.048* (.020) Sexual violence .018 (.070) .069 (.183) -.127 (.155) -.274 (.202) If ever married -.012 (.045) .158 (.118) -.106 (.100) .045 (.131) Had child from forced

marriage .123 (.155) -.122 (.403) -.367 (.342) .393 (.447) Education .023* (.006) .065* (.017) .013 (.014) .008 (.019) Constant .387 (.109) .396 (.285) 4.440 (.242) 3.909 (.316) N 1752 1752 1752 1752

Table 4. Social effects of abduction, including control variables

*p<.05

Referenties

GERELATEERDE DOCUMENTEN

De klas zag in dat dit niet het geval was, en zo viel het kwartje dat het parkeren van de auto’s blijkbaar op meer manieren kan dan het uitkiezen van de leerlingen: bij het

Based on previous literature and their own results, these authors 117 dened four possibilities for increasing the energy efficiency: (i) developing active high-surface area

In this thesis we present three studies, in which we employ a variety of methods to shed light on the neurophysiology of affect in the context of human media interaction as measured

This discussion of key outcomes and themes emerging from the study results in terms of four focuses: overall percentages of visible cadastral boundaries, similarities and

This allows us to establish a natural relation between the quantum mechanics Berry phase and previous results on operator mixing in conformal perturbation theory.. We will discuss

In addition, this pilot study showed that the AMSS questionnaire is feasible to advise GPs on diagnosis and management of allergic patients in primary care and in that sense the

9 shows the RMSE ¯ (top) and the number of features selected (middle) for all the evaluated combinations of λ and γ for the three biophysical parameters. As in the case of

The modeled thicknesses were exploited to evaluate the e ffectiveness of the probe anchoring and consecutive DNA hybridization, indicating an increase in DNA probes immobilized