Sport for Peace: exploring the contribution of sport to cohabitation in a divided society

Sport for Peace

Exploring the contribution of sport to cohabitation in a divided society

Tim Driessen Master's Thesis

Spring 2021

Department of Peace and Conflict Research, Uppsala University Supervisor: Kristine Höglund

Word Count: 18343

Abstract

Sport-for-peace programs rely on sport and contact in order to contribute to intergroup reconciliation efforts and a peaceful society. This thesis asks ‘how contact in a sport-for-peace program influences the willingness to live together in a deeply divided society’? Specifically, this thesis poses that more years of intergroup contact (hypothesis 1) and a better quality of intergroup contact (hypothesis 2) will lead to more willingness to live together. This research is based on a unique theoretical framework that relies on contact theory and is grounded in a relational peace framework. It is suggested that (quantity and quality) of contact improve the intergroup relationships within a sport-for-peace program which, in turn, improves participants’ willingness for cohabitation. A mixed method design is used in this study, drawing on unique survey data for logistic regression and the analysis of semi-structured interviews. Results give little support for hypothesis 1, but support hypothesis 2. This study found that especially the quality of contact in a sport-for-peace program is a good predictor for willingness to live together, and that this relationship is primarily driven by the extent to which the intergroup contact is considered pleasant by participants.

Key words: Sport for Peace, sport-for-peace programs, intergroup contact, cohabitation, contact

hypothesis, relational peace, Cyprus, divided societies.

Acknowledgments

There are some people that contributed significantly to the content of this thesis. I would like to thank Kristine for her overall guidance as well as sharing her specific expertise on conducting a qualitative analysis through semi-structured interviews. I am also grateful to Lisa Hultman and Annekatrin Deglow for their help with the quantitative analysis in R. I would have not been able to pull off a mixed method design in this thesis without the help of the three of you. I also want to thank all my friends at PPCY for their contribution to this thesis as a result of the many conversations that I have had with you in Cyprus as well as after I have left. A special thanks goes out to those that I could interview for this thesis.

Then there are also some people that have contributed by providing me with very needed

distractions, thereby making the overall thesis writing process a lot more fun. I would like to thank

Clark and his dog Halloumi for the countless walks in which we discussed the underlying topics

for this thesis while enjoying a cold beer or two. I am also grateful for the places from where I

wrote this thesis. I secured my internship at PeacePlayers Cyprus while traveling through southern

Europe with Tesse and our van ‘Herbie’ and surfboards. After my internship in Cyprus, Tess and

I lived in Austria for a couple of months where mountains and snow provided a welcome

distraction from thesis writing. I’m thankful for a girlfriend who not only gives me academic

support and proofread my final version, but also teaches me my first park moves on a snowboard

or takes me on off-piste adventures. Lastly, I’m thankful for my family. My mom’s ability to detect

every little bit of significant progress, my dad’s love for event-planning and special beers and my

sister’s skills in surprise visits ensured that my final month of thesis writing has also been a month

of celebrations.

Table of Contents

List of figures and tables ... 5

1. INTRODUCTION ... 6

2. THEORY ... 8

2.1 P

... 8

2.2 T

... 10

3. RESEARCH DESIGN ... 15

3.1 P

... 15

3.2 H

... 17

3.3 D

& M

... 19

3.3.1 Quantitative methods ... 19

3.3.2 Qualitative methods ... 20

3.4 E

... 22

3.5 O

... 23

3.5.1 The dependent variable ‘willingness to live together’ ... 23

3.5.2 The independent variable ‘quantity of contact’ ... 24

3.5.3 The independent variable ‘quality of contact’ ... 25

3.6 V

... 28

4. THE RESULTS OF THE QUANTITATIVE ANALYSIS ... 29

4.1 D

... 29

4.2 R

... 31

5. ANALYSIS ... 33

5.1 Q

... 33

5.2 Q

... 35

5.3 Q

... 38

5.4 A

:

... 41

5.5 L

... 44

5.6 E

... 46

5.7 A

... 48

6. CONCLUSION ... 50

7. WORKS CITED ... 52

8. APPENDIX ... 56

8.1 T

-

... 56

8.2 C

... 59

List of figures and tables

Figure 1: Theoretical argument 14

Figure 2: Causal mechanism 19

Figure 3: General information of interview participants 21 Figure 4: The distribution of answers to the survey questions for the dependent variable 23 Figure 5: The distribution of answers to years in the PPCY program 25 Figure 6: Formation of the quality components genuine interaction and sociable interaction 26 Figure 7: Distribution of answers for genuine interaction 26 Figure 8: Distribution of answers for natural interaction 26 Figure 9: Distribution of answers for pleasant interaction 27 Figure 10: Distribution of answers for cooperative interaction 27 Figure 11: Distribution of answers for sociable interaction 27 Figure 12: Distribution of answers for the independent variable ‘quality of interaction’ 27

Figure 13: Causal mechanism (repeated) 39

Figure 14: Answers for willingness to live together for each population group 42 Figure 15: Answers for willingness to live together aggregated by how frequently crossing the

border 43

Figure 16: Comparison between distribution of answers to ‘quality of contact’ and ‘willingness

to live together’ 46

Table 1: Overview of the presence/absence of scope conditions in the PPCY program 17

Table 2: General information of interview participants 21

Table 3: Survey questions for intergroup interaction 25 Table 4: Summary statistics for the control variables 28

Table 5: Contingency table for hypothesis 1 30

Table 6: Contingency table for hypothesis 2 30

Table 7: Logistic regression results 31

Table 8: Logistic regression results for robustness check 32

Table 9: Logistic regression results for each component of quality of contact 37

1. Introduction

The field of Sport for Development and Peace (SDP) has grown rapidly after it appeared on the agenda of the United Nations in the early 2000s. The number of organisations affiliated with SDP grew from 118 in 2008 (Kidd, 2008) to 1026 at the start of 2021 (The International Platform on Sport and Development, 2021). Concomitantly, the academic community became interested in SDP and peer-reviewed publications have increased exponentially in the past two decades (Schulenkorf, Sherry, and Rowe 2016).

Nonetheless, the ‘Peace’ in SDP remains relatively understudied. As Coalter (2010) points out, most SDP organizations are pre-occupied with individual and community development and do not consider their contribution to the creation of peaceful societies. Similarly, there has been little academic research on the effects of SDP programs on peace outcomes. Consequently, there are almost no studies that base their research in existing conflict resolution or peace theories, causing the Sport for Peace field to be rather undertheorized. Specifically, there is a large gap in the literature on how sport contributes to improving ethnic and social relationships in and building bridges between divided communities in post-conflict societies (Krasniqi and Krasniqi 2019). In order to address this gap, it is crucial to consider theories within parental disciplines of Sport for Peace, such as peace and conflict studies. The analysis of sport-for-peace programs is important for the field of peace and conflict because the programs contribute to intergroup reconciliation and a peaceful society through a relational approach.

While researching SDP topics within parental disciplines is critical, the type of research conducted is equally important (Darnell et al. 2018). Currently, SDP research has been dominated by instrumental and qualitative research investigating the appropriateness of SDP for certain groups of society, future avenues for program design and constrains for SDP programs to achieve their goals (Schulenkorf, Sherry, and Rowe 2016). While this is undoubtedly important, there is a call within the SDP community for novel, nuanced research that quantitatively investigates some of the key posited connections between sports on the one side and development and peace on the other.

This study contributes both theoretically and empirically to the Sport for Peace field as well as the larger SDP field. First, this study introduces a combination of contact theory and relational peace theory to explore the connection between sport-for-peace programs and peace outcomes.

Intergroup contact theory posits that positive effects of intergroup contact occur in contact

situations characterized by four key conditions: equal status, cooperation, common goals, and authority support (Allport 1958). The process that leads from intergroup contact to positive outcomes is through establishing positive relationships (T. F. Pettigrew 1998). However, contact theory is predominantly used for studying the effect of contact on prejudice reduction while this study is interested in the effect of contact on peace-related outcomes. Relational peace theory offers a solution because it posits that peace-related outcomes are obtained through the construction of positive relationships and therefore directly links contact to peace-related outcomes. Second, this study analyzes the relationships between these different variables with a mixed methods approach. A quantitative analysis of a novel survey dataset of a large sport-for- peace program is conducted in combination with a qualitative analysis of semi-structured interviews with staff, coaches and alumni of the same sport-for-peace program. As a result, this research addresses two of the existing research gaps within the Sport for Peace field by (1) grounding the theory in an existing peace theory and (2) providing the field with more quantitative research.

The research question that this research aims to answer is as follows: “How does contact in a sport-for-peace program influence the willingness to live together in a deeply divided society”

Specifically, this paper looks at the relationship between two independent variables, the quantity of contact and the quality of contact in a sport-for-peace program, and the dependent variable

‘willingness to live together’. This is explored in the context of PeacePlayers Cyprus, a sport-for- peace organization that targets Greek-Cypriot and Turkish-Cypriot youth in Cyprus. The expected relationship between quantity of contact and willingness to live together is captured in the following hypothesis.

Hypothesis 1: More years of intergroup contact in a sport-for-peace program will lead to more willingness to live together in a deeply divided society.

This research considers both quantity and quality of contact because it allows for a comparison.

The comparison in the effect of quantity and quality of contact on participants’ willingness to live

together provides interesting insights for the design and implementation of sport-for-peace

programs as well as for policy. For quality of contact, the expected relationship with the willingness

to live together is captured in the following hypothesis.

Hypothesis 2: A better quality of intergroup contact in a sport-for-peace program will lead to more willingness to live together in a deeply divided society.

It should be noted that the survey data used for the analysis within this research conveys answers from adolescents, of whom a large part are minors (between 12 and 17 years of age). This data has been collected by the sport-for-peace organization in close collaboration with a professional data collection company in 2016. These surveys have been conducted with informed consent of each individual participant as well as from their parent(s) or guardian(s). Moreover, the data is completely anonymous, because no personal information of any of the participants was collected or included in the datafile.

2. Theory

2.1 Positioning the study

While this research is firmly situated within the field of Sport for Peace, it is important to reflect on the larger field of Sport for Development and Peace (SDP) from which it originates. SDP is broadly defined as programmes that use sport to build peace and social cohesion in underdeveloped, conflict or post-conflict areas (Lyras and Welty Peachey 2011). Due to the many interpretations of and (sub-)disciplines that are associated with all three key terms of SDP – sports, development and peace – the research within the field is concerned with a large variety of thematic areas, theoretical frameworks, and methods (Webb and Richelieu 2015). This paragraph does not serve the purpose of explaining all of these in detail, but rather paints a general picture of the field before concentrating on the field of Sport for Peace. Thematically, SDP research investigates the effect of sports (programs) on outcome variables related to disability, gender, health, livelihoods, peace, social cohesion, infrastructure or a combination of two or more of these (Svensson and Woods 2017). As a result, theoretical frameworks from a variety of disciplines have been used to study these outcome effects, notably management, sociology, health, public policy, gender studies, education, media and peace and conflict studies.

Despite the large differences in theme, theory and approach used in SDP research, there are also some commonalities. First, the larger part of SDP is grounded in social psychology and sociology (Lederach 1997; Kidd 2008; Webb and Richelieu 2015). This is because the functioning of SDP programs is often investigated through the evaluation of contact, interaction, and/or relationships.

Second, Schulenkorf et al. (2016) found that positive youth development and social capital

frameworks are most frequently used within SDP research. Both of these frameworks are

borrowed from the fields of sociology and social psychology. Positive youth development theory investigates how participation in SDP programs could contribute to the healthy development of young people, especially of those in disadvantaged situations, whereas social capital theory considers how involvement in SDP programs affect the social engagement and mobility of participants (Darnell et al. 2018). Thirdly, most programs and research within SDP look at youth populations in post-conflict, post-disaster, or poverty-affected areas and use football (Darnell, Whitley, and Massey 2016; Schulenkorf, Sherry, and Rowe 2016). Youth are often one of the most vulnerable population in these contexts and development through sport might lead to empowerment of youth in the face of marginalization (Coakley 2015; Darnell et al. 2018). Football is one of the most popular sports on the planet and therefore also very commonly used in SDP programs. Fourthly, most research within SDP has a qualitative research approach (Schulenkorf, Sherry, and Rowe 2016) as is common to most of its parental disciplines. Lastly, SDP research generally investigates one of the following broad topics: (1) the positive effects of SDP programs in macro-sociological studies; (2) field and case studies; (3) the management and evaluation of SDP programs; and (4) the existing literature of SDP in large-scale reviews (Gadais 2020).

A clear distinction between ‘development’ and ‘peace’ is not often made when researching the effect of sports programs on particular developmental or peace outcomes. As a result, a research focus on either Sport for Development or Sport for Peace is often not indicated. What could be observed, however, is that up until this point the predominant focus has been on Sport for Development while Sport for Peace remains understudied (Cardenas 2013). To illustrate, 92% of SDP research before 2016 focused on development-related outcomes while only 8% looked at peace-related outcomes (Schulenkorf, Sherry, and Rowe 2016). In a recent scoping review, Clarke et al (2021) mapped all research between 1990 and 2021 that zoomed in on sports programs’

contribution to peacebuilding outcomes in post-conflict situations, which resulted in 28 articles and 2 books. This research was all (1) on communities that have moved into a peacebuilding stage after experiencing a recognized conflict; (2) submitted after the end of the cold-war, and; (3) on sport and physical exercise programs at any level and in any community demographic that had a clear peacebuilding objective (Clarke, Jones, and Smith 2021).

Two observations could be made that are related to the results of the scoping review by Clarke et

al (2021). First, the academic community has given far less attention to sport-for-peace programs

than to sport-for-development programs. While Clarke et al (2021) have estimated a total of 30

peer-reviewed publications have been dedicated to sport-for-peace programs, Schulenkorf et al

(2016) found five years earlier that a total of 437 publications have been dedicated to sport-for- development programs. Considering research on SPD has continued to grow exponentially, this number is even higher by now. The second observation is that most of the sport-for-peace publications do not make use of peace or conflict (resolution) theories and involve qualitative methods. Of the 30 publications, there were two that had authors with a background in Peace and Conflict Studies while the others were from the fields of Sociology, Social Sciences, (Public) Health, Management or Sports Psychology. Five of the articles were published in a journal that specifically relates to either ‘peace’ or ‘peacebuilding’. Moreover, except for the five publications in which programs in Northern Ireland feature, all research is on SDP programs within OECD countries. Lastly, while there are differences in type, all publications made use of a qualitative approach to their research (Clarke, Jones, and Smith 2021).

There is also some research on sport-for-peace programs that are conducted in active conflicts.

Most scholarship tends to focus on programs that are operated in Israel/Palestine, but some also focus on local conflicts (for an example on the role of sport in a pastoralist conflict, refer to Aronson-Ensign 2018) or other geographical locations (for an example on a soccer program in Iraq, refer to Mousa 2020). What comes forward in most research is the ability of sport-for-peace programs in conflict areas to ameliorate intergroup attitudes (Galily, J. Leitner, and Shimion 2013;

Schulenkorf and Sugden 2011; M. J. Leitner, Galily, and Shimon 2014; Mousa 2020) or to improve willingness for coexistence (M. Leitner, Galily, and Shimon 2013; Sugden 2006; 2008). Sugden (2006) does emphasize the need for sport-for-peace programs in conflict areas to be locally grounded, carefully designed, and professionally managed, because only then they could make a modest contribution to the wider conflict resolution process. However, there are also critics of the role of sport-for-peace organizations in conflict areas. A recent assessment of thirteen sport- for-peace organisations in Israel/Palestine argues that sport might not be the right tool for reconciliation between the Israeli and Palestinian communities as it is deemed only appropriate in post-conflict situations where peace has already been established (Dart 2019). Future research has to determine the exact role of sport-for-peace programs in conflict areas.

2.2 Theoretical framework

The theoretical framework for this study finds its basis in relational peace theory. Relational peace

theory analyzes peace in terms of relationships between actors or groups. These relationships, or

dyads, form the unit of analysis and together constitute a larger web of relationships that make up

societies. If a cluster of these relationships within one such context is studied, then relational peace

could be properly analyzed and understood (Jarstad et al. 2019).

According to Söderström et al. (2020) relational peace is made up by three components: (1) behavior such as non-domination, deliberation, and cooperation); (2) attitudes (mutual recognition and trust); and, (3) perception of the relationship (fellows or friends). Non-domination simply means that no individual (or group) dominates over another. Deliberation implies that actors in the dyad feel free to exchange their views and to provide reasons for it (Söderström, Åkebo, and Jarstad 2020). This is considered particularly important in post-war settings because it allows for differences to be communicated, considered, and accepted (Jarstad et al. 2019). Cooperation means that actors in the dyad choose to work together rather than to compete on shared issues. The attitudes of each actor towards the other in (all of) the dyad(s) should also be considered. Mutual recognition means that there is respect between the actors. At the most basic level this should mean that each actor accepts the existence of the other. Trust is often seen as the foundation for (relational) peace while it also stimulates cooperation between the actors (Jarstad et al. 2019).

Lastly, actors’ own view of the relationship is important. For relational peace to be considered present actors have to see the other as a fellow, a legitimate and equal partner to interact with, or as a friend (Söderström, Åkebo, and Jarstad 2020).

Relational peace theory is very well equipped to analyze sport-for-peace programs because these programs evolve around contact and relationship-building. Sport-for-peace programs make use of sports to bring together people from different communities in (mostly) post-war societies. Sports almost always involves contact. In addition, the repetition of these contact moments allows for (positive and negative) relationships to be formed over time. By analyzing both the contact setting and the relationships through a relational peace theory framework, it could be tested not only whether these relationships in itself are peaceful (or not), but also how they connect to the peace- related ambitions and/or outcomes of sport-for-peace programs. The latter has not been investigated yet in existing research.

This relational approach to peace is also found in the sociology-orientated conflict transformation

literature (Jarstad et al. 2019). Contact theory provides one well-known example. Contact theory

was developed by social-psychologist Gordon Allport (1958) and he posits in his book “Nature of

Prejudice” that positive effects only flow from intergroup contact in situations where the following

key conditions are present: (1) equal group status within the situation; (2) common goals; (3)

intergroup cooperation; and, (4) support of authorities. In their influential meta-analysis of

intergroup contact theory, Pettigrew and Tropp (2006) find that contact under Allport’s contact

conditions lead to a greater reduction in prejudice compared to situations where none of these

conditions were present (see also Hewstone and Swart 2011). Lemmer and Wagner (2015) have come to the same conclusion while accounting solely for real-world contact-based interventions.

More recently, Paluck, Green, and Green (2019) have also evaluated the contact hypothesis focusing on studies that feature random assignment and delayed outcome measures and came to a similar conclusion: intergroup contact generally reduces the level of prejudice between groups.

Even though contact theory was originally conceptualized to investigate prejudice between different ethnic groups, contact theory has also been deemed applicable to investigate peace- related outcomes (Gibson and Claassen 2010; Hewstone et al. 2014; Tropp et al. 2017; Grady 2020;

Mousa 2020; Ioannou, Al Ramiah, and Hewstone 2018; Hewstone et al. 2008). One reason for this is that in deeply-divided and segregated societies contact between different communities is considered as a very important step in the bigger process of reconciliation or improved coexistence. Nonetheless, little is known about the effect of contact on concrete measures for peaceful coexistence such as future contact intentions (for an example of this, see (McKeown and Psaltis 2017). This research grounds contact theory within a relational peace theoretical framework to address that gap. Specifically, this research investigates the effect of repeated contact in a sport- for-peace program on the willingness to live together. The steps within this theoretical framework are explained below.

Step 1: Intergroup contact

In this research, intergroup contact is defined as face-to-face interaction between members of clearly defined groups (Thomas F. Pettigrew and Tropp 2006). Face-to-face interaction connotes direct contact. While outside the scope of this particular research, the contact theory literature has also thoroughly investigated the effects of indirect intergroup contact (see for example Dovidio, Eller, and Hewstone 2011). The groups under study in this research are defined by their ethnicity.

Moreover, it is important to reflect on both the quantity and quality of contact within a contact situation. Quantity of contact simply relates to the number of contact moments whereas the quality of contact looks at the content of the contact (for more information on the latter, refer to the next paragraph). Moreover, Jarstad et al. (2019) argue that repeated interaction over time will lead to either stronger or weaker relationships in terms of peace and therefore constitutes a more reliable indicator.

As stated before, an optimal contact situation in which a high quality of contact is guaranteed, is

defined by the following conditions: (1); equality within the situation, (2); common goals, (3);

intergroup cooperation, and (4); authority support (Allport 1958). While Allport (1958) defined equality as a similarity between actors in socioeconomic background, wealth, skills and experiences, this research follows a more minimal definition following Söderström et al. (2020) and define equality as the non-domination of one (member of a) group over another. Moreover, an optimal contact situation is characterized by having both (individuals of) groups work together to reach a shared objective. Söderström et al. (2020) argue that deliberation, that is to exchange (different) views and reasoning behind it, should be an integral part of the cooperation to obtain the common goal. It is also important that the contact situation enjoys the support of an authority, one that is also recognized and accepted by the actors in the contact situation (Allport 1958). The authority could be a local NGO to a supranational organization such as the European Union and everything in between. Lastly, in order for (members of) clearly defined group to consider the other group as equal and to cooperate together there needs to be mutual recognition as well as trust between them (Söderström, Åkebo, and Jarstad 2020). The presence of all these conditions creates an environment in which positive relationships and/or friendships could be formed.

Step 2: positive relationships and/or friendships

Relationships between the actors are formed over time and are a result of repeated contact. These relationships are more likely to be positive if the conditions outlined above are present (Hewstone and Swart 2011). The minimum definition of a positive relationship is that both actors in a dyad recognizes the other as a fellow human being and would be willing to interact with one another.

This is what Söderström et al. (2020) specify as a relationship between fellows. The maximum definition of a positive relationship would be friendship. Cross-group friendships are regarded as the most effective form of intergroup contact (Hamberger and Hewstone 1997; Thomas F.

Pettigrew 1997; Hewstone and Swart 2011). While Allport formulated the conditions to explain

‘when’ intergroup contact leads to positive effects, Pettigrew (1997; 1998) investigated ‘how’

intergroup contact translates to positive effects and argues that the level of friendship potential in a situation of intergroup contact explains whether and to what extend these positive effects will occur. His theoretical argument could be shortly summarized in four succeeding steps:

1. Intergroup contact situations offer the possibility to learn about the outgroup;

2. The behavior between members of different groups changes;

3. Affective ties between members of these groups are established;

4. Each group reappraises the other (Thomas F. Pettigrew 1997; T. F. Pettigrew 1998).

According to Söderström et al. (2020), peaceful relationships are either those between fellows or friends. A fellowship relates to what is called a ‘positive relationship’ above and is formed when actors have come to accept that they share a common space or community with one another. A friendship is different from a fellowship in that it involves affection and some form of altruism (Söderström, Åkebo, and Jarstad 2020). Peaceful relationships, that is positive relationships and friendships, are thought to have a positive effect on peacebuilding (Lederach 1997).

Step 3: Willingness to live together

The peace-related outcome considered in this research is the willingness to live together. This means whether actors in the contact situation are open to the idea of living together in the same space, for example a neighborhood, with actors from the other group. Regular intergroup contact gives actors an idea interacting with one another on a regular basis and as a result might help to establish whether that would be a long-term solution. In the context of a post-war society, this is extremely relevant as one of the major challenges in divided societies is to overcome geographical separation of different population groups (Yucel and Psaltis 2020). A post-war society relates to a society in which the issues underlying the conflict remain unresolved, although (large-scale) (Söderström, Åkebo, and Jarstad 2020). While intergroup contact has been directly linked to improved social cohesion in deeply divided societies (Cox and Sisk 2017), there is very little known about the consequences for future contact intentions. One of the few studies that has investigated the effect of contact on willingness to live together is one by Yucel & Psaltis (2019). These authors analyzed the (in)direct effect of intergroup contact through prejudice and trust on the willingness for cohabitation in Cyprus, thereby specifically looking at variation for different age groups. They found support for their hypothesis that intergroup contact (in)directly positively affects the willingness to live together. The study in this paper also investigates the effect of intergroup (quantity and quality) of contact, but argues that positive relationships or friendships acts a mediating variable between contact and willingness to live together. The theoretical argument is summarized in Figure 1.

Figure 1: Theoretical argument

Willingness to live together Relationship

(between friends or fellows) Intergroup

contact

3. Research Design

This chapter elaborates on the elements that together provide the framework for the research conducted in this study. The chapter commences by introducing the selected program as well as relating it to the scope conditions that were set out for eligible programs. The second sub-chapter provides the reader with some important background to the case. In the third sub-chapter, the quantitative and qualitative data and methods are introduced. This research will conduct a quantitative analysis of a survey dataset and a qualitative analysis of semi-structured interviews in order to investigate the case. The fourth sub-chapter gives the ethical considerations regarding the survey dataset and the semi-structured interviews. The operationalizations for the dependent variable, both the independent variables, and the control variables are discussed in fifth sub- chapter. Lastly, the sixth sub-chapter discusses the validity and reliability of all of these variables.

3.1 Program selection

This study investigates the role of contact in sport-for-peace initiatives on reconciliation efforts, specifically the willingness to live together. As a result, the case considered for this research has to meet certain conditions. First, as the objective is to build peace, the sport-for-peace organization has to be located in a country or region that is divided by conflict. Thus, the organization has to operate its program in a post-conflict or post-war environment in which societal groups are divided by previous conflict.

Second, the organization’s programs need to have a clear focus on peace-related outcomes. This means that there should be more to a sport-for-peace program than just sport, because otherwise it does not differ from a regular sports program. For example, besides the activities related to sports, sport-for-peace programs offer classroom sessions on peace-related topics or organize additional social activities that aim to bring together different population groups. Moreover, the focus on peace-related outcomes needs to be distinguished from development-related outcomes.

Sport-for-peace programs aim to improve the ‘life skills’ of individual participants while sport-for- peace programs aims to improve these and the relationships between the individual participants from different population groups for purposes such as reconciliation or cohabitation.

The third condition is that the organization needs to make use of team-related sport as a medium

to establish contact between two or more population groups. This also means that the participants

at the organization should be from at least two different population groups that have encountered

some degree of division due to conflict. Despite some examples of individual sports used to obtain

peace objectives (see Hayhurst et al. 2014; Thorpe 2016; Collison and Marchesseault 2018), team- related sports are predominant within sport-for-peace initiatives. A first reason for this is that team-related sports simply require participants to communicate and come into contact with one another. Moreover, research shows that when participants play together in team sports they are able to take distance from their own as well their family’s prejudices (Gilbert and Bennett 2012).

Lastly, the presence of Allport’s conditions lead to more positive outcomes according to contact theory. The conditions were (1) equal group status within the situation; (2) common goals; (3) intergroup cooperation; and, (4) the support of authorities. For this research that means that these conditions need to be present within the sport-for-peace program in order to arrive at the desired peace outcome. Due to the fact that Allport’s contact hypothesis has received both academic support and critique, the academic community has asked for research that investigates whether these conditions are really necessary for positive effects. Therefore, this research will consider the conditions of Allport as scope conditions and apply them to the case at hand, even though the investigation into the Allport’s conditions is not the main objective of this research.

Based on these conditions, the sport-for-peace program of PeacePlayers Cyprus was identified as an eligible program. The program uses the game of basketball to bring together youth from the Greek Cypriot (GC) and Turkish Cypriot (TC) communities on the island. These communities have been divided by a physical border since 1974. PPCY is the only year-round bi-communal youth sports organization on the island with mono-communal trainings every week and bi- communal events once per month. Bi-communal events consist of tournaments, summer camps, retreats and shared practices between GC and TC teams that are called ‘twinnings’. Moreover, there are additional bi-communal events and classroom sessions for PPCY participants that are in the Leadership Development Program (LDP), a program for those that want to learn more about conflict resolution and the Cyprus conflict. Every year approximately 250 Cypriot children participate in the events of PPCY. The overall goal of PPCY is to unite the divided GC and TC communities through basketball with the underlying motto that “children that play together can learn to live together”.

All the scope conditions could be found within the PPCY program. The program uses the game

of basketball (scope condition 3) to bring together GC and TC youth that are currently living in a

divided post-war environment (scope condition 1) and has the underlying intention to make

participants more willing to live together with one another (scope condition 2). The presence (or

absence) of Allport’s contact conditions was determined through the use of semi-structured interviews with alumni, staff members and (assistant) coaches at PPCY in which specific questions regarding these conditions were asked (the qualitative methods are discussed on page 20). First, while there is some inequality between GCs and TCs due to reasons outside of the contact environment, for example as a result of checkpoint regulations or differences in international opportunities, interview participants univocally stated that there is equality within the specific contact situation of PPCY. Second, three common goals were mentioned: to play basketball, to have fun, and to interact with the other community. Third, cooperation was deemed an essential part of PPCY programs, both on the basketball court and off the court where classroom sessions also involve a focus on collaboration. Lastly, interview participants have indicated that although most authorities support the contact situation at PPCY, the contact situation is also hindered by checkpoint regulations and biased school curriculum on both sides. Therefore, authority support is considered to be moderately present within the PPCY program. Table 1 gives a summary of the scope conditions and how they apply to the PPCY program.

Table 1: Overview of the presence/absence of scope conditions in the PPCY program

# Scope condition Present or absent in

PPCY program

1 Post-conflict or post-war environment Present

2 Focus on peace-related outcome(s) Present

3 Use of a team-related sport Present

4a Equal group status within the contact situation Present

4b Common goals within the contact situation Present

4c Cooperation within the contact situation Present

4d Authority support for the contact situation Moderately present

3.2 Historical context of the program

This section provides some relevant background information to the Cypriot conflict. This is relevant as the current situation in Cyprus is the direct result of a series of events that took place in the past fifty years.

Cyprus gained independence from the British Empire in 1960, following the Zürich and London Agreement in 1959 between the United Kingdom, Greece, Turkey and the Cypriot community leaders. Representatives from Greece and Turkey were present as there were large Greek and Turkish populations on Cyprus because both Greece and the Ottoman empire had occupied Cyprus in earlier history. The population at the time consisted for 77.1% of GCs, 18% of TCs and 4.9% of other nationalities (Lyras 2012) and the provisions of the

1

The underlying tensions between the Greek Cypriot and Turkish Cypriot

agreements ensured political rights for both population groups: (1) the president was to be a GC and the vice-president a TC; (2) government posts and public offices were allocated by ethnic quotas (30% would be TC); and, (3) the vice-president had the power to veto any decision (Conference on Cyprus, 1959). Both sides immediately showed discontent with these power- sharing provisions of the agreement. Instead, the GC and TC populations rather wanted to unite with what they each considered their homeland, Greece and Turkey, respectively. Tensions between the population groups rose and eventually came to a boiling point in 1963 when interethnic violence erupted between them. The United Nations intervened and set up the United Nations Peacekeeping Force in Cyprus (UNFICYP) to prevent further fighting between the GC and TC communities.

However, when the GC community ousted president Makarios III in a Greek-led coup and replaced him with someone that was much more in favour of unification with Greece, Turkey saw this as a breach of the Zürich agreement and responded with a military intervention in Cyprus.

Within the next couple of days, tens of thousands of Turkish troops arrived on Cyprus and interethnic violence between GCs (and Greece) and TCs (and Turkey) intensified. Eventually, the UNFICYP brokered a ceasefire on 16 August 1974 and established a UN Buffer zone between the TCs and Turks in the north (36% of the island) and the GCs and Greeks in the south (64% of the island). It is estimated that hundreds of GCs and TCs lost their life, 160.000 GCs and 50.000 TCs became displaced and more than 2000 people went missing during the 1974 war (Ker-Lindsay 2011).

North of the UN Buffer Zone, the TCs established the Turkish Republic of Northern Cyprus (TRNC), which is only recognized as an independent country by Turkey. The GCs established the Republic of Cyprus in the South, which is internationally recognized as an independent country and is a member state of the European Union since 2004. Ever since the establishment of the UN Buffer Zone by UNFICYP in 1974, the GC and TC communities have lived segregated from one another. The border was closed for Cypriot citizens until 2003, when it was partially opened again and people were allowed to cross through a number of checkpoints along the Buffer Zone.

Although the re-opening of the border allows for more intergroup contact between the two

communities, the contact has been limited as a result of each community’s selective history and

education, the negative memories passed on by older generations, and the structures of the divided

nations, among other things (Lyras 2012). Especially among the youth population there is very

little intergroup contact observed (Ioannou, Al Ramiah, and Hewstone 2018). The PPCY program

brings together youth from both communities to play sport and get to know each other. This gives youth a chance to form their own opinion about people from the other community and whether they would prefer coexistence in the future.

3.3 Data & Methods

This research uses a mixed methods design. The quantitative analysis of survey data investigates the causal relationship between (1) the independent variable ‘quantity of contact’ and the dependent variable ‘willingness to live together’ and (2) the independent variable ‘quality of contact’ and willingness to live together. The causal mechanism for these relationships is listed in Figure 2. The qualitative analysis supports the quantitative analysis with information gathered from a set of semi-structured interviews with PPCY staff, coaches and alumni. This is necessary considering that the survey dataset consists of relatively few observations (N = 140) and does not have data on all theoretical variables outlined in the causal mechanism (e.g., on positive relationships). Semi-structured interviews provide a tool to gather for more information on key variables and the relationships between them that are of interest for this study. The interviews also give an opportunity to learn more about the data collection for this dataset as well as the participants. Thus, the logistic regression in the quantitative analysis will form the basis for the results about the relationship between each of the independent variables with the dependent variable. The qualitative evidence will be used to further substantiate these claims as well as to provide additional insights. The next two sections discuss the quantitative and qualitative methods in more detail.

Figure 2: Causal mechanism

3.3.1 Quantitative methods

This study has access to a unique dataset containing survey answers from PPCY participants are between the age of 12 and 19. This survey was conducted by staff members of PPCY and in collaboration with a Cypriot data collection organization. It was conducted at various PPCY locations on both sides of the border, notably in Nicosia, Dali, Iskele, Kiti, and Lapta. Of the 272 participants that were in PPCY in 2016, 142 (52%) people filled out the survey. After accounting for missing values on the independent variables and the dependent variable 140 survey

More willingness to live together More positive

relationships More (quantity

and quality) of

intergroup contact

respondents remained. Of these 140 participants, 72 identified as female and 68 as male. Moreover, 66 of the survey respondents are TCs and 74 are GCs. This indicates that the sample size is a fair representation of the population of PPCY participants.

Due to the nature of survey data (e.g., data is collected through the individual) the unit of analysis is the individual. The time period on which will be focused in this research is from 2006 to 2016.

The reason for this is that 2006 was indicated by survey respondents as the earliest year and 2016 as the last year in which they first participated in PPCY programming. Qualitative analysis from the semi-structured interviews will ascertain whether programming has changed significantly during this time period.

The choice of statistical method is dependent on the values of the dependent variable. In this research, the dependent variable ‘willingness to live together’ is a binary variable consisting of either the value 0 for ‘not willing to live together’ or 1 for ‘willing to live together’. A logistic regression requires the dependent variable to be binary and therefore has been chosen as method.

In fact, a dichotomous dependent variable is the main assumption of the logistic regression (Peng, Lee, and Ingersoll 2002)). Another assumption is that the desired outcome should be coded ‘1’.

The desired outcome in this research is ‘willing to live together’ and has already been assigned the value ‘1’. The model used in this research does not include any meaningless variables and all variables are independent from one another, which are a third and fourth assumption, respectively.

Lastly, logistic regression requires a large sample size. While the sample size is relatively small, it does fit the minimum requirement of 100 observations (Hosmer, Lemeshow, and Sturdivant 2013). Thus, logistic regression provides an adequate method for testing the hypotheses of this research.

3.3.2 Qualitative methods

The qualitative analysis serves to support the findings of the quantitative analysis. The qualitative information is mainly gathered through conducting semi-structured interviews with staff, coaches and alumni and from secondary sources such as PPCY reports, documents, and assessments. This qualitative data collection aims to collect primary data that is otherwise unavailable and contributes to the analysis by allowing a more holistic understanding of the cases (Höglund 2011).

In total, fourteen possible candidates were contacted for an interview. They were sent an interview

letter which informed them about the research and their rights. These particular candidates are

selected because they are or recently have been involved in the programs of PPCY as a member of the office, coach or alumnus. The active community of PPCY consists of staff members, (assistant) coaches, and alumni and therefore this reflects the selection of interview participants.

All participants in the interviews are adults, meaning they are 18 years or older. At the start of each interview, all were asked to give oral consent for the researcher to conduct and record these interviews. Of the fourteen candidates that were approached, twelve agreed to an interview while two declined due to limited availability and/or unwillingness to participate. Some general information of the interview participants is given in Table 2 and Figure 3 below. How interview participants are referenced in this study is indicated in the last column of Table 2, and it should be noted that some participants have experiences in several affiliations. Eleven interviews with twelve participants – two of the candidates preferred to have the interview together – were conducted through the video communication platform Zoom. The template for the semi-structured interviews and the interview letter are listed in the appendix. The interviews lasted for 53 minutes on average, the shortest taking 40 minutes and the longest 1 hour and 18 minutes.

Figure 3: General information of interview participants

Table 2: General information of interview participants

# Affiliation Date Reference within study

Participant 1 Office 08-04-2021 P1, Office

Participant 2 Office 08-04-2021 P2, Office

Participant 3 Office 09-04.2021 P3, Office

Participant 4 Alumni, Coach 09-04-2021 P4, Alumni/Coach

Participant 5 Alumni, Coach 09-04-2021 P5, Alumni/Coach

Participant 6 Alumni, Coach 09-04-2021 P6, Alumni/Coach

Participant 7 Coach 15-04-2021 P7, Coach

Participant 8 Office 15-04-2021 P8, Office

Participant 9 Alumni, Coach 17-04-2021 P9, Alumni/Coach

Participant 10 Office 19-04-2021 P10, Office

Participant 11 Alumni, Coach 19-04-2021 P11, Alumni/Coach

Participant 12 Alumni, Coach 20-04-2021 P12, Alumni/Coach

3.4 Ethical considerations

There are some ethical considerations with regards to the quantitative and qualitative data sources.

First, the dataset used for the quantitative analysis of this study involves minors - 134 of the 140 survey respondents (96%) are younger than 18 years old. The data collection was carried out by PPCY and several measures were taken to ensure that it was ethically sound. A first measure was that PPCY staff members were supported by a professional external organization to assure that ethical guidelines were followed. Moreover, surveys were only conducted if the individual survey respondent as well as their parent(s) or guardian(s) gave consent to participating in the survey.

Lastly, no personal information of any kind was collected in the surveys or during the collection period, thereby guaranteeing full anonymity to the participants. There was no mention of complications or problems during the data collection period by participants, PPCY staff or employers of the external organization.

Second, this study also adhered to ethical guidelines for the semi-structured interviews. Each prospective interview participant received an interview letter containing a description of the project and their rights as a participant. Moreover, at the beginning of each interview participants were asked whether they gave consent to the interview being conducted and the interview being recorded. No personal information was collected and stored, with the exception of the video recording of the interview. In order to safeguard the anonymity of the participants, the videorecording together with all other files and material related to each interview was stored on a password protected file located on a separate storage device (USB). All the files related to a certain interview will only contain the date of the interview. These files will remain on the USB until the completion of the thesis, after which they will be permanently deleted.

Third, the Cypriot situation as well as the sensitivity of certain topics needs to be carefully

monitored at all times. The situation in Cyprus has been relatively stable, that is without interethnic

violence for a long period of time. However, considering that the civil war took place quite recently

in the 1970s, the negative attitudes towards the other group are still deeply rooted into GC and

TC societies (Lyras 2012). This needs to be constantly considered while designing and

implementing the semi-structured interviews. The ethical risks most pertinent to this research

design are the sensitivity of the research questions and potential research fatigue from long-term

(inter)national involvement in the country (Höglund 2011). For this reason, the interview questions

have been designed with utmost care and they are not touching upon sensitive topics.

3.5 Operationalization of key variables

This section explains how the independent variables, dependent variable and control variables are operationalized within this research. In other words, how these concepts will be exactly measured.

The operationalizations for all variables are retrieved from the PPCY survey dataset. The first section will discuss how the dependent variable ‘willingness to live together’ is operationalized.

Second, the measurements for each independent variable – quantity and quality of contact – are outlined. Third, the measurements for the control variables are briefly discussed. The final chapter of this section will elaborate on the validity and reliability of the measurements for all of these variables.

3.5.1 The dependent variable ‘willingness to live together’

The dependent variable, willingness to live together, is measured by the answers to two of the survey questions, namely (1) “I am willing to live in the same neighborhood with GCs/TCs” and (2) “I am willing to live in a GC/TC neighborhood”. The survey respondents were asked to indicate to what extend they agreed with both of the statements on a scale from 1 (“Strongly Disagree”) to 5 (“Strongly Agree)". These questions refer to different situations of living together.

While the former question refers to the willingness to live in a mixed neighborhood the latter question asks about the willingness to live in a neighborhood inhabited for the majority by the other population group. The distribution of participants’ answers to both of these questions are summarized in Figure 4. The figure shows that survey respondents answer the first question more positively than the second.

Figure 4: The distribution of answers to the survey questions for the dependent variables

The scale for the DV “willingness to live together” was created by averaging the scores on the two questions. These values reflected a 0,5-point value on a scale between 1 (“Strongly Disagree”) and 5 (“Strongly Agree”). These values were then transformed a binary variable for the logistic regression. All values between 0 and 3.0 were scored a 0 (= ‘not willing to live together’) whereas all values between 3.0 and 5.0 were scored a 1 (= ‘willing to live together’). If the value was exactly 3.0, originally meaning ‘no opinion or neutral’, it was interpreted as ‘not willing to live together’

and assigned value of 0 in the original model. Chapter 4.2 on page 31 discusses the implications of models with alternative values for the dependent variable. Eventually, 83 survey respondents are willing to live together with the other community and 57 respondents are not.

3.5.2 The independent variable ‘quantity of contact’

The independent variable ‘quantity of contact’ is operationalized as the number of years that participants are in the PPCY program. The survey data did not contain any specific information on individual participation of each PPCY participant. In order to obtain such information through other means, interview participants were asked to indicate the number of bi-communal contact moments per month and the last nine annual program reports were analyzed for information on bi-communal contact moments per month. From this it was estimated that there is approximately one bi-communal contact moment per month for each PPCY participant within the program.

According to the interview participants, the number of bi-communal contact moments per month has not changed since 2006 and the content of these moments has also stayed relatively the same.

This is important as no changes to number and content of bi-communal contact moments per month between 2006 and 2016 – the time period of this study – need to be taken into account.

However, the PPCY dataset does only capture the years in the program for each individual survey

respondent. One year constitutes twelve bi-communal contact moments. As both the years in the

program and the corresponding bi-communal contact moments increase at the same rate

(multiplied by two), both are considered adequate measures for the independent variable ‘quantity

of contact’. This study has chosen for years in the program. In the survey, years in the program

was measured by asking participants what year they joined PPCY. The answers are interpreted as

follows: 2016 = 1 year in the program, 2015 = 2 years in the program, and so on. The distribution

of the answers is given in Figure 5. The answers range between 1 (2016) and 11 years (2006) in the

program. None of the respondents indicated that they are 8 or 10 years in the program. As only a

few respondents were already in the program from more than five years, all are grouped as ‘6 or

more years in the program’.

Figure 5: The distribution of answers to years in the PPCY program

3.5.3 The independent variable ‘quality of contact’

The quality of contact between participants from the GC and TC communities is measured by the answers to five different survey questions. These questions ask to rate whether the interaction with participants from the other community is (1) superficial, (2) natural, (3) pleasant, (4) cooperative and (5) distant. Participants could rate their answers on a scale from 1 (“Strongly Disagree”) to 5 (“Strongly Agree”). The original survey questions are listed in table 3 below.

Table 3: Survey questions for intergroup interaction

Below you will see a series of statements. For each of these, think about your general feelings about interaction with Greek Cypriots/Turkish Cypriots. How much do you either agree or disagree with what is written?: Strongly Disagree (1); Disagree (2); No Opinion\ Neutral (3); Agree (4); Strongly Agree (5)

Interaction with GC/TC people would be superficial

Interaction with GC/TC people would be natural

Interaction with GC/TC people would be pleasant

Interaction with GC/TC people would be cooperative

Interaction with GC/TC people would be distant

Three of the questions ask about positive attributes of the perceived interaction (e.g., natural, pleasant, and cooperative) while two ask about negative attributes (e.g., superficial and distant) and as a result their scales are exactly opposite. For example, a score of 5 for ‘natural interaction’ would indicate a positive answer whereas it would indicate a negative answer for ‘superficial interaction’.

This problem was solved by transforming the two negative attributes to positive attributes of interaction and reversing the scores. Thus, ‘superficial’ was changed into ‘genuine’ and ‘distant’

into ‘sociable’ and scores of 1 and 2 were changed into 4 and 5, respectively. This is visualized in

Figure 6 below.

Figure 6: Formation of the quality components ‘genuine interaction’ and ‘sociable interaction’

Taken together, the answers to these five questions give a very clear idea of the perceived quality of the interaction. The answers to each component of the independent variable ‘quality of contact’ are summarized in the figures below. Based on the figures, the answers for natural, pleasant and cooperative interaction show a similar distribution whereas the answers to sociable and genuine interaction deviate. For each individual survey respondent, an overall score was calculated by adding the scores for each of these components and dividing the total by five. This overall score represents the independent variable ‘quality of contact’ used in this thesis and is shown in Figure 12. The figure shows that answers are strongly skewed towards a positive perception of quality of contact.

Superficial interaction

Strongly Agree 19

Agree 25

Neutral 34

Disagree 42

Strongly Disagree 19

Genuine interaction

Strongly Disagree 19

Disagree 42

Neutral 34

Agree 25

Strongly Agree 19

Distant interaction

Strongly Agree 39

Agree 37

Neutral 32

Disagree 20

Strongly Disagree 11

Sociable interaction

Strongly Disagree 11

Disagree 20

Neutral 32

Agree 37

Strongly Agree 39

Figure 7: Distribution of answers for genuine interaction

Figure 8: Distribution of answers for natural

interaction

5.3.4 The control variables

Moreover, several control variables have been selected for the quantitative analysis. Three of these control variables constitute common demographic controls: gender, age and nationality. Of these demographic control variables, nationality is of particular interest. Research by Yucel and Psaltis (2019) showed that on average GCs are more willing to live together than TCs. This research provides an opportunity to confirm or dispute this finding. The final control variable is called

‘border crossing frequency’ and provides information on how frequently each survey respondent crosses a checkpoint on a scale from 1 (“never”) to 5 (“very frequently”). It is assumed that more frequent crossing of the border between the two communities is positively correlated with the dependent variable ‘willingness to live together’. An overview of the control variables and their summary statistics is given in table 4 on the next page.

Figure 9: Distribution of answers for pleasant

interaction Figure 10: Distribution of answers for cooperative

interaction

Figure 12: Distribution of answers for the independent variable ‘quality of interaction’

Figure 11: Distribution of answers for sociable

interaction

Table 4: Summary statistics for the control variables Gender

(Female = 1, Male = 2) Age Nationality

(TC =1, GC = 2) Border crossing frequency (1 = never to 5 = very frequently)

Min. :1.000 Min. :12.0 Min. :1.000 Min. :1.000

1st Qu. :1.000 1st Qu.:12.0 1st Qu.:1.000 1st Qu.:1.000

Median :1.000 Median :13.0 Median :2.000 Median :2.000

Mean :1.486 Mean :13.7 Mean :1.529 Mean :2.507

3rd Qu. :2.000 3rd Qu.:15.0 3rd Qu.:2.000 3rd Qu.:4.000

Max. :2.000 Max. :19.0 Max. :2.000 Max. :5.000

3.6 Validity and reliability

While the previous sections have outlined how the variables of interest in this study are measured, this aims to reflect upon the validity and reliability of these operationalizations. Validity refers to the accuracy of the measurements and is concerned with whether what is measured is in line with what was intended to be measured (Kellstedt and Whitten 2013). Reliability relates to the consistency of measures, so whether repeated measurements would yield the same results every time (Kellstedt and Whitten 2013). First, the measurement of the respondent’s willingness to live together with people from the other community is directly related to the dependent variable and has a strong validity. However, validity could be affected by the equal weighting of both components in measuring the dependent variable, especially because the living situation referred to in question 1 is less extreme than the one in question 2. Considering the difficulties related to establishing a valid unequal ratio between these two questions, this research has opted for equal weighting.

The reliability of the measurement for the dependent variable ‘willingness to live together’ is mostly affected by the nature of survey research. Survey data by definition consists of self-reported answers and therefore could potentially be prone to self-report bias, which occurs when a self- reported answer does not correspond with the actual true answer (Dixon, Durrheim, and Tredoux 2005). It also does not allow participants to fully express their own feelings regarding intergroup contact as they often have to choose between fixed answers. As a result, it is argued that repeating this particular survey might yield different results and affects the reliability of the measurements.

However, the reliability of the two components of the dependent variable could be established by

measuring Cronbach’s alpha. Cronbach’s alpha is a test used to measure whether scores for

questions on the same topic are reliable. It gives a score between 0 and 1 and as a rule of thumb a

scores above 0.70 indicate good reliability. The Cronbach’s alpha for the scores of the two survey

questions of the dependent variable are 0.85, indicating high reliability. Nonetheless, the concerns

related to the reliability are taken into consideration in this study.