GODEFROIDT AMELIE - THESIS - The structure and impact of citizenship

UNIVERSITY OF LEUVEN

FACULTY OF SOCIAL SCIENCES

MASTER OF SCIENCE IN COMPARATIVE AND

INTERNATIONAL POLITICS

Analysing the structure and impact of citizenship in an

African context

An empirical analysis of citizenship and political trust in Ghana

Supervisor: Prof. Dr. A. LANGER MASTER THESIS Reporter: Prof. Dr. B. MEULEMAN submitted to obtain the degree

of Master of Science in Comparative and International Politics by

Amélie GODEFROIDT

Academic year 2014-2015

UNIVERSITY OF LEUVEN

FACULTY OF SOCIAL SCIENCES

MASTER OF SCIENCE IN COMPARATIVE AND

INTERNATIONAL POLITICS

Analysing the structure and impact of citizenship in an

African context

An empirical analysis of citizenship and political trust in Ghana

Supervisor: Prof. Dr. A. LANGER MASTER THESIS Reporter: Prof. Dr. B. MEULEMAN submitted to obtain the degree

of Master of Science in Comparative and International Politics by

Amélie GODEFROIDT

Academic year 2014-2015

II

III

Summary

Who is a foreigner and who an autochthone according to the man-in-the-street? And do these ideas about citizenship matter for the development of political trust? Those questions are of utmost importance for many African countries struggling to increase their legitimacy and to foster a sense of national identity that overrides other ethnic, religious, social or political affiliations. The concept of citizenship helps us to answer those two questions. It revolves around membership and belonging to a group; and conventional wisdom assumes that it can be divided into two representations (i.e. civic and ethnic citizenship). However, no consensus has been reached yet neither on the precise conceptualisation of citizenship nor on the relative importance of a national identity for political trust when other variables are entered into the equation – especially not within an African context. Using the National Service System Survey and employing factor analyses as well as hierarchical regression techniques; this study makes at least two contributions to the western-dominated literature.

First, it contradicts these western-based citizenship theories by demonstrating that Ghanaian university students (N=3,264) experience a convergence of both civic and ethnic citizenship conceptions – rather than some students adhering to the civic form while others endorsing the ethnic notion. When assessing the validity of a two- or three-factor model, it is immediately clear that the two or three factors are highly correlated with each other. Thus, we conclude that citizenship ideas in Ghana complement, instead of contradict, each other resulting in a complex citizenship conception.

Second, it adds to studies of political trust by developing a multivariate model of political trust. This is especially useful for Ghanaian politicians challenged with low levels of political trust among their students. The results indicate that a stronger national identity indeed contributes to trust in political institutions and actors

IV

among Ghanaian adolescents. However, a positive evaluation of the government’s effectiveness in providing public goods is even more important for building political trust among Ghanaian students. In addition, there is no such thing as a ‘trusting personality’ since trust in government is barely influenced by socio-demographic factors or interpersonal trust. These results thus support the superiority of institutional performance theories while providing limited support for cultural explanations. Consequently, governments should always respond punctually and effectively to public priorities in order to raise trust.

Both conclusions urgently ask for further research on citizenship in African countries and for country-specific citizenship measurements as well as for more studies on trust-building in developing countries trying to establish working institutions.

KEYWORDS: Ghana; civic, ethnic and cultural citizenship; political trust; cultural theories; institutional performance theories

V

Table of contents

Summary ........................................................................................ III

Table of contents .............................................................................. V

List of tables ....................................................................................IX

List of figures ..................................................................................IX

List of maps ...................................................................................... X

List of abbreviations ........................................................................ X

Acknowledgments ...........................................................................XI

1. Introduction ............................................................................... 1

1.1. Research subject and objectives ........................................... 1

1.2. Significance of the study ...................................................... 5

1.3. Structure of the study ........................................................... 6

2. Theoretical and conceptual foundations ................................. 9

2.1. Structure of citizenship......................................................... 9

2.1.1. Civic, ethnic and cultural citizenship conceptions ........ 9

2.1.2. Dual identity and its implications for citizenship ........ 16

2.2. Impact of citizenship on political trust ............................... 18

2.2.1. Citizenship and political trust ...................................... 18

2.2.2. Cultural versus institutional performance theories ...... 22

2.3. The African context ............................................................ 26

2.3.1. Post-colonial Ghana .................................................... 27

2.3.2. Ethno-regional and religious composition of Ghana ... 28

3. Research objectives, questions and hypotheses .................... 33

3.1. Structure of citizenship....................................................... 33

VI

3.2. Impact of citizenship on political trust ............................... 35

3.2.1. Structure and level of political trust ............................ 36

3.2.2. Origins of political trust .............................................. 37

4. Methodological consideration ................................................ 41

4.1. Data source and collection ................................................. 42

4.1.1. Context and country .................................................... 42

4.1.2. N3S study: selection of schools and students .............. 43

4.2. Measurements of the variables in the N3S study ............... 45

4.2.1. Citizenship................................................................... 45

4.2.2. Dependent variable: political trust .............................. 46

4.2.3. Independent variables .................................................. 46

4.2.3.1. Interpersonal trust ................................................. 47

4.2.3.2. Sense of institutional effectiveness ...................... 48

4.2.3.3. General background characteristics ...................... 50

4.3. Data reduction: statistical analyses .................................... 51

4.3.1. Dimensionality of citizenship: factor analyses ............ 51

4.3.2. Origins of political trust: linear regression analyses ... 52

5. Results ...................................................................................... 53

5.1. What is the structure of citizenship? .................................. 53

5.1.1. RQ1: Exploratory Factor Analysis .............................. 54

5.1.2. RQ1-2: Confirmatory Factor Analysis ........................ 57

5.2. What is the impact of citizenship on political trust? .......... 64

5.2.1. RQ3: Exploratory Factor Analysis .............................. 64

5.2.2. RQ4: Frequency Analysis ........................................... 66

5.2.3. RQ5-9: Regression Analyses ...................................... 67

VII

6. Discussion ................................................................................. 73

6.1. Making a nation, creating a stranger: answering the research questions and hypotheses ............................................................. 73

6.1.1. Citizenship representations in Ghana .......................... 74

6.1.2. Cultural and institutional theories of political trust ..... 79

6.2. Limitations of the study and subsequent avenues for future research ......................................................................................... 82

7. Conclusion ................................................................................ 85

References ....................................................................................... 89

Appendix 1: survey ......................................................................... 99

Appendix 2: descriptives of key variables .................................. 109

VIII

IX

List of tables

Table 1: Historical, conceptual and theoretical features of the civic-versus-ethnic dichotomy. .......................................................... 11

Table 2: Diverse operationalisations of citizenship conceptions. .... 14 Table 3: Ethnic composition of Ghana’s administrative regions in

2010 (percentage). .................................................................... 30 Table 4: Religious composition of Ghana’s administrative regions in

2010 (percentage). .................................................................... 31 Table 5: Descriptives of interpersonal trust items and scale. ........... 47 Table 6: Descriptives of sense of institutional effectiveness items and

scale. ......................................................................................... 48 Table 7: Descriptives of NSS contribution items and scale. ............ 50 Table 8: Descriptives of citizenship items. ...................................... 54 Table 9: KMO and Bartlett’s Test. .................................................. 55 Table 10: Exploratory factor analysis of citizenship items. ............. 56 Table 11: KMO and Bartlett’s Test. ................................................ 64 Table 12: Exploratory factor analysis of political trust items. ......... 65 Table 13. Levels of trust in diverse nation-wide, legal-political

institutions and actors ............................................................... 66 Table 14: Determinants of political trust – hierarchical multiple

regression analyses. .................................................................. 70

List of figures

Figure 1: Ghanaian versus ethnic identity by ethnic groups. ........... 34 Figure 2: Comprehensive model of the origins of political trust. .... 40 Figure 3: Scree plot of Eigenvalues as a result of an EFA with

Promax rotation. ....................................................................... 55 Figure 4: Complex citizenship. ........................................................ 60 Figure 5: CFA one-factor model – Unstable. .................................. 61 Figure 6: CFA one-factor model – Stable. ....................................... 61 Figure 7: CFA two-Factor model – Unstable. ................................. 62

X

Figure 8: CFA two-factor model – Stable. ...................................... 62 Figure 9: CFA three-factor model – Unstabe. ................................. 63 Figure 10: CFA three-factor model – Stable. .................................. 63 Figure 11: Scree plot of Eigenvalues as a result of an EFA with

Promax rotation. ....................................................................... 65

List of maps

Map 1: Ghana's administrative regions and cities of universities. ... 43

List of abbreviations

CFA Confirmatory Factor Analysis CFI Comparative Fit Index EFA Exploratory Factor Analysis KMO Kaiser-Meyer-Olkin measure of sampling adequacy NSS National Service Scheme N3S National Service Scheme survey PCA Principle Component Analysis RMSEA Root Mean Square Error of Approximation TLI Tucker-Lewis Index

XI

Acknowledgments

This master thesis is the result of a year of hard, solitary work. However, it would not have been possible without the help of others. Therefore, I sincerely want to thank a few people without whom the result would not have been the same.

First and foremost, I want to thank my supervisor, Prof. Dr. Arnim Langer, who gave me the opportunity to work on this topic and to use his National Service System Survey. Through his different courses over the last year, I gained insight into the subject of identity and ethnicity as well as into research methods and statistics. Our discussions about the topic of this thesis, cooperation and mutual trust lifted this thesis to a higher level. Special thanks also go to Prof. Dr. Bart Meuleman for his statistical support and corrections. Last but not least, I want to thank all my friends and family; but especially my friends from the VIP-master for joining me in the library and making fun during the breaks, my roommates Tine and Valerie for the pleasant moments during the hectic days, my boyfriend Ludovic for his listening ear and everlasting patience, and my parents, brother and sister for their unconditional support and belief in me.

XII

1

1. Introduction

Democracy requires trust; trust requires a national identity.

1.1. Research subject and objectives There is a widespread agreement that the borders of many African states are arbitrarily drawn by the great colonial powers with little respect for social and linguistics alliances. As a result, it is assumed that African people identify little with their new nation within whose boundaries they have been boxed – particularly when these boundaries divided ethnic groups into separate states. This ethnically heterogeneous situation left the postcolonial authorities with the difficult task of developing a stable nation-state, maintaining national cohesion and cultivating political trust (Herbst, 1989; Miles & Rochefort, 1991). In order to succeed, most African countries have adopted policies and processes of nation-building to create a shared national identity superseding ethnic, religious or political affiliations and increasing political trust (Langer, Meuleman, Oshodi, & Schroyens, 2014). Some countries have largely succeeded in fostering national unity and promoting political trust. Ghana, for instance, has managed to peacefully unite inter-ethnic and -religious communities – with the notable exception of some severe communal tensions in its northern regions (Langer & Ukiwo, 2009). On the other hand, exclusive beliefs about citizenship can also lie at the heart of conflicts. The Rwanda genocide or the unrests in Côte d’Ivoire at the beginning of the 21th century revolved mainly around the redefinition of citizenship and discourse of autochthony (Marshall-Fratani, 2006). All those issues revolve around the questions: who can be considered an autochthone and who a foreigner within a particular nation? And what is the impact of such a national identity on the development of political trust?

2

This thesis tries to answer those two questions by focussing on the concept of citizenship – one of the most fertile areas in research on ethnic relations, national identity and nationalism (e.g. Brubaker. 1990; Connor. 1978; Kymlicka. 2001; Liebich. 2006; Shulman. 2002). It formulates beliefs about the organisation of a society and about who can be regarded as a national in-group member; which generates privileges, rights and obligations (Bereketeab, 2011). Moreover, a strong national identity, binding people together within a distinct geographical territory, is seen as a significant form of nation-state cohesion and a precondition for political trust. It makes people cooperate with each other more willingly than with other people, unite themselves under the same government and willing to be governed exclusively by a representative delegation of themselves (Berg & Hjerm, 2010; Mill, 1861).

Conventional wisdom traditionally assumes that citizenship models can be divided into two broad but still distinct categories: civic and ethnic citizenship. Civic citizenship traditionally stems from the territoriality principle (ius soli) and can, therefore, be regarded as open based on political loyalty; whereas ethnic citizenship is derived from the law of blood (ius sanguinis) and is, thus, considered as rather closed based on ancestry (Brubaker, 1992; Reeskens & Hooghe, 2010; Shulman, 2002, 2004; Smith, 2010). More recent, a cultural notion of citizenship has been proposed holding the middle ground between civic and ethnic citizenship. It is not entirely open or voluntary nor completely closed or inherited. Cultural citizenship is based on adhering to and preserving the nation’s culture, including its language and religion (Kymlicka, 1999, 2001; Nielsen, 1999; Reijerse, Acker, Vanbeselaere, Phalet, & Duriez, 2013). Although the general idea of the relationship between national identity and political trust is clear (i.e. a stronger national identity leads to a higher level of political trust), it stays unclear how the different forms of citizenship impact political trust differently.

3

There are, however, still considerable shortcomings both in the citizenship and political trust literature. First considering citizenship, even 100 years after the first references to the civic/ethnic distinction, no consensus has been reached yet on the structure of citizenship. Citizenship scholars argue about two or three competitive citizenship conceptualisations, while literature on national identity assumes that citizens can identify with both national (i.e. civic) and ethnic elements. More importantly, the ideas on citizenship representations and national identity are mainly based on European countries (e.g., Lalonde, 2002; Moreno & Arriba, 1996; Moreno, 2006; Reeskens & Hooghe, 2010; Reijerse, 2012). As a result, information about the empirical validity of citizenship in a non-western context is lacking. It is highly questionable whether citizenship can also be conceptualised into a civic, ethnic and/or cultural conception within the African context. This is a shortcoming shared with the literature on political trust. King et al. are right on the mark when they question the cross-cultural comparability and the validity of the political trust studies outside a western context (King, Murray, Salomon, & Tandon, 2003). Moreover, research on the impact of national identity on political trust (Berg & Hjerm, 2010) tend to neglect other possible determinants of political trust such as socio-demographic variables, interpersonal trust or evaluations of institutional performances.

As a result, profound knowledge on the dimensionality of citizenship and diverse individual-level origins of trust in developing countries is quite absent. This lacuna is intriguing because insights on citizenship and political trust provide more precise means of explaining the relationship between identity, authority and legitimacy in African, developing countries (Ndegwa, 1997). Hence, the purpose of this research is to attribute to the pool of knowledge on citizenship and political trust. Because it seems crucial to gain more in-depth understanding of the conceptualisations of citizenship and the origins of political trust within the African context, we ask: what is the structure and impact of citizenship within an African context?

4

To answer this question, we will conduct a quantitative analysis based on public opinion data from the ‘National Service Scheme Survey’ (N3S) gathered in Ghanaian universities in 2013 (N = 3,264). This N3S study aims to analyse inter-group relations and national identities among Ghanaian students1 and contains valuable information on citizenship notions and political trust as well. This data allows us to study the underlying dimensions of citizenship first, and to what extent these dimensions coincide with the civic/ethnic/ cultural division (RQ1). Second, this research addresses the question whether those civic, ethnic and cultural conceptions of nationhood are mutually exclusive or complementary (RQ2). Third, the structure (RQ3) and level (RQ4) of our dependent variable is examined before analysing the impact of citizenship on political trust. Next, in order to fully capture the determinants of political trust, the effects of socio-demographic variables are scrutinised first (RQ5) with the purpose of including the significant variables as control variables in the analyses of the effects of citizenship (RQ6), interpersonal trust (RQ7) and perceived institutional effectiveness (RQ8). Last, these divers explanations of political trust are combined to test an empirical model of political trust (RQ9). In doing so, this study will be one of the first to answer these questions within an African context thereby largely contributing to the citizenship and political trust framework.

1 It is worth noting that this study is based on only one country (i.e. Ghana) and one specific subsample of that country (i.e. student sample). Ghana is chosen because (1) it gives us the opportunity to explore citizenship and political trust in a completely different culture and because (2) it has been identified as one of the least-trust societies. This study is, however, not representative for the African region. Students are considered an interesting subject of study because today’s students are tomorrow’s societal and political elites. Hence, these students are more likely to impact and shape citizenship perceptions of others such as their peers and family. For more information on the sample, see 4.1.2.

5

1.2. Significance of the study Although citizenship is a crucial and complex topic with far reaching social and political consequences, little research has been done considering the way in which identity and citizenship is negotiated within African state borders and its implication for political trust. From a societal and political perspective, it is of utmost importance to obtain a more detailed understanding of the operationalisation of citizenship in Africa because most African governments are still exploring who should be a part of their nation-state. These multi-ethnic societies with their artificially drawn borders are challenged to stimulate a sense of national identity which over-rides ethnic affiliations (Herbst, 1989; Langer et al., 2014; Miles & Rochefort, 1991). More insights on the criteria considered important to let someone into the national in-group are therefore crucial.

The second part of our research (i.e. the determinants of political trust) also contributes to both politics and academics. By establishing a model of political trust, we can help Ghanaian politicians and policymakers with cultivating more legitimacy from their society. By examining the relationship between individual trust in political institutions and actors on the one hand and citizenship, social characteristics, interpersonal trust and evaluations of institutional effectiveness on the other hand, this study generated original insights from different theoretical and empirical research traditions. This comprehensive case study can be used as the base for more extensive, cross-national, comparative research.

The Ghanaian context is of particular interest for both of our research objectives. First and concerning the structure of citizenship, Ghana wanted to create a pluralistic nationality after its independence guaranteeing universal rights to all citizens, including people of foreign origin (i.e. civic citizenship). But the Ghanaian politicians struggled immensely to redefine the nationality status of the descants of the numerous labour migrants that came to Ghana during the

6

colonial era. The popular and ethnic conception (i.e. only indigenous people are ‘true’ citizens) led to protest against the constitution that recognised citizenship by birth in Ghana regardless of the origin of the parents. Ghana with its unique efforts of nation-building, therefore, provides an excellent case study for exploring the ambiguities and malleability of post-colonial citizenship (Kobo, 2010). The Ghanaian example also stresses the discrepancy between policy conceptions of citizenship and popular beliefs as well as between ethnic and civic notions of citizenship. We therefore opted to empirically assess the structure of citizenship based on public opinion data. Second and as to the origins of political trust, Ghana has been identified in previous research as one of the least-trust societies (Delhey, Newton, & Welzel, 2011) and, thus, benefits the most from an in-depth country study on political trust.

1.3. Structure of the study This study is divided as follows. The next section will review prior research that has been conducted on the competing representations of citizenship as well as on the concept of dual identity that incorporates both civic and ethnic notions. Moreover, the theoretical framework will evaluate both cultural and institutional performance perspectives on political trust. The last part of the literature review will revise the African and Ghanaian contexts in more detail. In the third section, the two-folded research objective will be expounded with the research questions and hypotheses that have guided our research. Then, all methodological considerations will be discussed: the data source and collection, operationalisation and measurement of the variables and used statistical analyses. Further, chapter five will display all analyses and associated results needed to answer our research questions and test the hypotheses. A profound discussion of our Ghanaian study will then be possible. In addition to answering

7

the research questions, we will point to some limitations of our study and make some suggestions for future research. The last section will draw some conclusions regarding the relationship between citizenship and trust in the case of Ghana.

Because this thesis has a two-folded research objective, most parts of the study are also divided into two main parts: an analysis of the structure of citizenship and an examination of the impact of citizenship on political trust. In addition to the impact of citizenship, the effects of other cultural and institutional performance variables are examined as well to paint a complete picture.

8

9

2. Theoretical and conceptual foundations

This study will be one of the first to unravel the structure of citizenship and its impact on political trust within an African context. In order to do so properly, this literature review consists of three parts. First, we will give some more background information on the competing visions on citizenship. Literature on citizenship tends to focus on the civic, ethnic and cultural division while overlooking the contributions on national identity made by Luis Moreno. We will try to integrate both strands of research into one comprehensive vision. In the second part, we will give an overview of what is known on the determinants of political trust. Besides a strong national identity, other cultural and institutional variables are important as well. In the last part, we will give a short impression of the background of Africa and Ghana with regards to the post-colonial efforts of nation-building as well as to the ethnic and religious composition.

2.1. Structure of citizenship

2.1.1. Civic, ethnic and cultural citizenship conceptions

What criteria do people need to satisfy in order to be considered as a full citizen of a particular nation? The concept of citizenship – one of the most fertile areas in research on ethnic relations, national identity and nationalism (e.g. Brubaker. 1990; Connor. 1978; Kymlicka. 2001; Liebich. 2006; Shulman. 2002) – helps to answer this question. Citizenship revolves around membership and belonging to a group; which generates privileges, rights and obligations (Bereketeab, 2011). Since the pioneering works of Meinecke (1907) and Kohn (1945), conventional wisdom traditionally assumes that citizenship models can be divided into two broad but distinct categories: civic and ethnic citizenship. Both theoretical forms of citizenship have in common that they formulate beliefs about the organisation of a

10

society and about who can be regarded as a national in-group member. Civic citizenship traditionally stems from the territoriality principle (ius soli) and can, therefore, be regarded as open based on political loyalty; whereas ethnic citizenship is derived from the law of blood (ius sanguinis) and is, thus, considered as rather closed based on ancestry. Further it is argued that western nationalism is predominantly political and based on the idea of citizenship, while eastern nationalism consolidates around the common heritage of people and the idea of the folk. Civic notion of nationhood are thus more prevalent in Western Europe and America whereas ethnic nationalism became the dominant ideology in Eastern Europe and the East (Kohn, 1945; Shulman, 2004). This conceptual distinction has been persistent among political scientists, politicians and journalists for decades (Brubaker, 1992; Reeskens & Hooghe, 2010; Shulman, 2002, 2004; Smith, 2010).

Since the nineteen-nineties, this distinction between civic and ethnic citizenship has been widely used in political and sociological sciences, politics and journalism. Besides the West/East division, the dichotomy quickly went hand in hand with several other sets of dichotomous concepts (e.g. liberal/illiberal, individualist/collectivist, instrumentalist/essentialist) and normative labels (e.g. open/closed, rational/irrational, inclusive/exclusionist). Table 1 presents the conceptual, historical and theoretical features of these dichotomies used in the discourse of citizenship. It goes without saying that those distinctions are not without political consequences because the endowment of normative virtues leads to interpret ethnic as ‘bad’ and civic as ‘good’ (Bereketeab, 2011). Moreover, because of recent developments (e.g. resistance against the Maastricht treaty, separatist referendums in northern Ireland and Catalonia, increasing ethno-political tensions in Belgium), it becomes impossible to maintain the spatial distinction with an uncritical view of the essentially ‘civic’ quality of West European nationalism (Brubaker, 2004).

11

Table 1: Historical, conceptual and theoretical features of the civic-versus-ethnic dichotomy.

Conceptual dichotomies

Civic Modern, constructivist Instrumentalist Individualist

Ethnic Primordial Essentialist Collectivist

Historical foundation (Meinecke, 1907)

Staatsnation Kulturnation

Nation types and basic concept (Kohn, 1945)

State precedes nation Political community Supra-ethnic, national Citizenship

Nation precedes state Cultural community Ethnic, sub-national Folk

Citizenship principle

Ius soli (law of territory) = Legalistic, formalistic

Ius sanguinis (law of blood) = Sociologic, non-formalistic

Citizenship modalities

Obeying the law Accepting values and institutions Participation = Acquired

Common descent Common (cultural) heritage = Inherited

Normative values

Rational Liberal Universalistic Egalitarian Democratic Voluntaristic Descriptive Open Inclusive (towards migrants)

Irrational Illiberal Particularistic Essentialistic Undemocratic Deterministic Ascriptive Closed Exclusionist (towards migrants)

Examples France, England, the Netherlands, Switzerland and the United States

Germany, Japan, Eastern and Central Europe

12

Moreover, Kymlicka (1999, 2001), Nieguth (1999) and Nielsen (1999) have more recently argued that citizenship predominantly based on culture is often mislabelled as ethnic citizenship in literature. They regard membership of a nation not just as a question of law (i.e. civic) of ethnic heritage (i.e. ethnic), but also of culture (Kymlicka, 2001). Hence, they have pleaded for the decomposition of ethnic citizenship into a cultural citizenship where citizenship is based on the adherence to the dominant culture (with a special focus on language and religion). Cultural citizenship is based on adhering to and preserving the nation’s culture, including its language and religion (Kymlicka, 1999, 2001; Nielsen, 1999; Reijerse et al., 2013). Consequently, anyone who adopts, cherishes and preserves the national culture may be regarded as a citizen (Reijerse et al.. 2013). Cultural citizenship holds the middle ground between civic and ethnic citizenship. Membership is neither entirely open or voluntary (i.e. you cannot instantly acquire a culture), nor completely closed or inherited (i.e. children of foreigner can be regarded as a member when they adhere the culture) (Nielsen, 1999).

Nonetheless, whether a distinct cultural representation of citizenship is manifested at the individual level remains a highly contested issue (Reijerse et al., 2013). Some studies that explicitly analysed the prevalence of a cultural citizenship representation failed to find statistical evidence to suggest that their respondents manifested a distinct cultural citizenship representation – besides the ethnic and cultural representations. For instance, using an exploratory (EFA) and confirmatory factor analysis (CFA). Reeskens and Hooghe (2010) found only a two-factor structure of citizenship in their measure of citizenship representations among people across 33, mainly western, countries (i.e. a civic and ethnic factor, both including a cultural element). In other words, some research suggests that it is more appropriate to incorporate cultural items in the ethnic and civic scales than to construct a separate, cultural scale (Reijerse et al., 2013). More recently, however, some evidence for a third

13

citizenship representation has emerged. For example, also using factor analyses, Janmaat (2006) and Reijerse et al. (2013) found a clear three-factor solution for their citizenship measure (i.e. a civic, ethnic and cultural factor). This finding is remarkable, especially because Reijerse et al. (2013) used similar data from seven countries, all also included in Reeskens and Hooghe (2010).

It is surprising that, even 100 years after the first references to this distinction, some major questions have remained unanswered. In addition to the lack of research on African countries (see 2.3), at least two problems can be detected. First, different scholars have attached different features to each component of citizenship resulting in a tremendous lack of conceptual clarity. Consequently, there is no consensus about the specific items reflecting a specific type of citizenship representation nor about the questions that should be asked to the respondents (Brubaker, 2004; Reijerse et al., 2013). Shulman (2002), for instance, circumvents this problem by a priori assigning items to the concepts of civic, ethnic and cultural citizenship. Reeskens and Hooghe (2010) are right on the mark when stating that this method is ‘hardly in line with accepted practices in survey research or in data reduction’. They, just like other scholars (e.g. Janmaat. 2006; Reijerse et al.. 2013), opted to empirically test the underlying structure of citizenship by means of factor analyses.

In Table 2, we summarise diverse operationalisations of the three citizenship representations. It is immediately clear that some criteria have been added to different types of citizenship. Speaking the national language, for instance, has been associated with civic (Reeskens & Hooghe, 2010; Wimmer, 2008), ethnic (Bereketeab, 2011) and cultural (Janmaat, 2006) citizenship in previous studies. Consequently, the operationalisation of the concepts still needs to be carefully ascertained what raises concerns about the empirical validity of previously used citizenship representations. It is the aim of this study to carefully and empirically assess the operationalisation of citizenship based on public opinion data gathered in Ghana (RQ1).

14

Table 2: Diverse operationalisations of citizenship conceptions.

Civic citizenship

National language (Reeskens & Hooghe, 2010a; Wimmer, 2008a) National culture (Wimmer, 2008a) Institutions/legal system (Bereketeab, 2011b; Janmaat, 2006a; Reeskens & Hooghe, 2010a; Shulman, 2002b) Rights and duties (Janmaat, 2006a; Reijerse et al., 2013a; Shulman, 2002b) Social security and economy (Janmaat, 2006a) Territory (Shulman, 2002b) Legal citizenship status (Shulman, 2002b) Adhering to basic state ideology (Shulman, 2002 b) Participation/joining political community, accepting participation of members of all cultural groups (Reijerse et al., 2013a; Shulman, 2002b) Origin/cultural background no reason to deny citizenship (Reijerse et al., 2013a)

Ethnic citizenship

Common descent/ancestry (Bereketeab, 2011b; Janmaat, 2006a; Reeskens & Hooghe, 2010a; Reijerse et al., 2013a; Shulman, 2002b) Common history/heritage (Janmaat, 2006a) Being born in [country] (Reeskens & Hooghe, 2010a; Reijerse et al., 2013a) Living in [country] for most of one’s life (Bereketeab, 2011b; Reeskens & Hooghe, 2010a) Growing up in [country] family from an early age (Reijerse et al., 2013a) National language (Bereketeab, 2011b) Religion (Reeskens & Hooghe, 2010a) Cultural traits related to ethnic commonality (e,g, values, norms, dress, food) (Bereketeab, 2011b) Belonging to the dominant ethnic/racial group (Shulman, 2002 b)

Cultural citizenship

National language (Janmaat, 2006a) National culture (Janmaat, 2006a) National religion (Shulman, 2002b) Traditional [country] lifestyle/traditions (Reijerse et al.,

15

2013a; Shulman, 2002b) Respecting Christian origin of [country] culture (Reijerse et al., 2013a) Respecting [country] symbols (Reijerse et al., 2013a) Protecting [country] culture (rapid changes) (Reijerse et al., 2013a) Passing on [country] culture (next generations) (Reijerse et al., 2013a) Preserving [country] culture (Reijerse et al., 2013a)

a assignment based on factor analysis b a priori assignment

Second and in the same vein, the interrelatedness of the notions

has not been clarified yet (RQ2). In theory, ethnic and civic conceptions should be negatively related (i.e. be competitive). Concerning cultural citizenship, it is acknowledged that both civic and cultural citizenship are open to outsiders, whereas ethnic citizenship is not. Thus, cultural and civic citizenship representations should be positively related to each other but negatively to ethnic citizenship representation (Reijerse et al., 2013). The empirical evidence, however, is not so straightforward. While some research did find a negative relationship between ethnic and civic scales, other research tended to detect a positive relation. Reeskens and Hooghe (2010), for instance, concluded that civic and ethnic citizenship conceptions are not mutually exclusive as they found an extremely strong correlation of 0.80. This finding raises serious questions about the validity of two separate concepts and adds to the debate about whether citizenship representations should be regarded as competing or complementing concepts. It has been suggested in literature that popular notions of nationhood cannot be conceived as a dichotomy (Janmaat, 2006) or that individuals posses multiple identities by simultaneously adhering to diverse (or to no) citizenship criteria (Kymlicka, 1998). This observation is in line with the concept of dual identity captured by the ‘Moreno Question’.

16

2.1.2. Dual identity and its implications for citizenship

Literature on citizenship conceptions tends to neglect the contributions made by Luis Moreno by the turn of the century (Moreno, Arriba, & Serrano, 1998; Moreno & Arriba, 1996; Moreno, 2006). His survey research on dual identities in Scotland and Catalonia greatly resemble to the conceptualisation of citizenship2. He also argues that people tend to internalise common symbols to form a collective identity. An ethnic identity fosters the development and maintenance of group boundaries, separating ‘us’ from ‘them’. Members of ethnic groups emphasise origin, collective solidarity, cultural uniqueness, unity or territorial integrity to varying degrees. The ‘ethnic ingredients’ Moreno puts forward in his essays (i.e. a common culture and distinguishable traits, kinship, history, territory or religion) significantly overlap with the items measuring an ethnic citizenship conception (e.g. cultural traits, descent/ancestry, history/ heritage, living in the country, religion) (Moreno & Arriba, 1996).

Besides this local/ethno-territorial self-ascribed identity closely associated with sub-state communities and ethnic groups, Moreno (1998; 1996; 2006) also identifies a state/national identity. This identity representation is similar to the notion of civic citizenship as Moreno (1998; 1996; 2006) regards political integration in the state-building process as the crucial element to form a national identity. In this way, national identity can also be operationalised as a bipolar variable in multi-ethnic states: the citizens who exclusively identify with the national (e.g. Ghanaian) or the subnational group (e.g. their ethnic group). Connecting the citizenship literature with studies on national identity, citizens who identify with the national group (i.e. 2 In his 1986-Ph.D. thesis, Moreno introduced a self-identification scale to measure how citizens regard themselves. Scottish and later Catalan respondents had to choose between options ranging from feeling a national citizen to feeling a member of their ethno-regional group. This question – although slightly modified in subsequent surveys – became known as the ‘Moreno question’.

17

state/nation identity) will endorse more civic requirements in order to let an outsider into the national in-group. In contrast, citizens who identify more with the subnational group (i.e. local/ethno-territorial identity) will adhere more to ethnic criteria to accept someone else into his subnational in-group. This idea of national identities seems to reinforce the mutually exclusiveness of civic and ethnic citizenship conceptions.

Moreno (1998; 1996; 2006), however, does not see those concepts as mutually exclusive – as the citizenship literature does. On the contrary, he takes his analysis one step further by arguing that there are considerable degrees of duality in the way citizens identify themselves. Citizens can share loyalty to both the ethnic group and the state without any fracture between them resulting in a multiple national identity. He is right on the mark with his assessment that mutually exclusive concepts are not tenable because when citizens in a sub-state or multi-ethnic community identify themselves in an exclusive manner with their ethnic identity, the institutional outcome of such antagonism will also be exclusive (e.g. secessionism or forced assimilation). Constant ethno-territorial conflicts would be the result of a population strongly adhering to its ethnic identities (Moreno et al., 1998; Moreno & Arriba, 1996; Moreno, 2006). His solution, therefore, is the construct of a dual identity concerning the way citizens identify themselves in communities within pluri-ethnic polities. This concept incorporates – in variable proportions, individually asserted – the local/ethno-territorial identity (i.e. ethnic) and the state/national identity (e.g. civic) with no apparent exclusion. It is thus perfectly possible that citizens simultaneously adhere to both ethnic and nationalistic items – a fact mainly neglected by the citizenship literature. His data of ethnic minorities in the United Kingdom and Spain confirms this, concluding that Scottish and Catalan people feel part of their ethnic group as well as of the country as a whole (Moreno, 2006).

18

To wrap up, citizenship literature puts, at most, three distinct citizenship conceptions forward. The empirical validity of the three notions as well as their interrelatedness, however, stays unclear. Furthermore, the ‘Moreno question’ – although not measuring exactly the same concepts – suggests that group identities are not as exclusive as suggested in the citizenship literature. Therefore, Moreno (1996) proposes a dual identity encompassing elements that resemble the ethnic and the civic citizenship items. Consequently, we examine whether our data reveals separate civic, ethnic and cultural citizenship representations or a combined representation – which we will call complex citizenship in this study – instead of a priori suggesting a two- or three-folded solution. After assessing the structure of citizenship, its impact on political trust will be examined.

2.2. Impact of citizenship on political trust

2.2.1. Citizenship and political trust

Democracy requires trust (Mill, 1861). Trust between individuals as well as trust between citizens and political actors is supposed to be vital for a liberal democracy (Berg & Hjerm, 2010). Putnam (1993; 1995), for example, identified trust as one of the main features – besides norms and networks – of the social capital that ‘makes democracy work’. In particular, democracy requires political trust since that is what links citizens to the institutions intended to represent them, thereby enhancing both legitimacy and effectiveness of democratic regimes (Mishler & Rose, 2001). Political trust has been defined in previous studies as ‘the trust individuals have in their state-wide legal-political institutions and actors’ (Berg & Hjerm, 2010, p. 391). Although this definition3 suggests that political trust is

3 Specific information on the conceptualisation, language and terminology of most variables can be found in the methodology section (see 4.2).

19

two-dimensional representing institutional trust on the one hand and political trust on the other hand, pubic opinion data reveals one comprehensive political trust-scale. We will first examine whether this is the case in Ghana as well (RQ3). Regardless of the precise operationalisation of political trust, without political trust democracy risks losing legitimacy and thus ability to function satisfactorily. During a recession, for instance, democratic stability requires sufficient political trust from citizens to accept temporary economic constraints in return for better conditions in the future. This is especially important for democratic governments since they cannot rely on coercion to the same extent as other regimes (Catterberg & Moreno, 2006; Mishler & Rose, 1997, 2001). Hence, political trust serves as ‘the creator of collective power’ (Gamson, 1968, p. 42), or as the essential element ‘to the proper functioning of democracy’ (Hetherington, 1998, p. 792). Nonetheless, Gamson (1968) illustrates that political trust is double edged. Although insufficient trust signals the disintegration of civil society, excessive trust also fosters political apathy and diminished citizens’ control of government.

In the post-colonial regimes of the African region, excessive trust is not a concern. On the contrary, these developing countries are often confronted with a low initial level of political trust, partly as a result of the suppression under authoritarian regimes. This low level of trust in government signifies the low degree of legitimacy ascribed to the state by the individual. This situation is problematic because governments cannot govern effectively which is further undermining political trust and a vicious cycle is created (Miller, 1974). The immediate issue for these developing regimes is, therefore, to override this abiding distrust in order to establish legitimate and long-lasting institutions. Raising trust among the population raises the credibility of the government and can facilitate peace in previously fractured and divided societies. Although an understanding of the origins of trust is vital for consolidating stabile democratic regimes, empirical evidence about the underlying causes

20

of trust in African societies is lacking. Conventional wisdom agrees that the African region urgently needs more insights on building regime legitimacy and raising political trust (Hutchison & Johnson, 2011; Mishler & Rose, 1997). This is especially vital in the case of Ghana since it has been identified as one of the least-trust societies (Delhey et al., 2011). We first test whether this is also the true for our subsample (RQ4). Subsequently, it is highly relevant to ask: how can the Ghanaian government develop more trust with their citizens?

Since J.S. Mill, it has been taken for granted that trust comes from a common nationality (Mill, 1861). Decades ago. Easton also claimed that ‘there must be some cohesive cement – a sense of feeling of community amongst the members. Unless such identity emerges, the political system (…) may not survive’ (Easton, 1975, p. 176). The development of a nation is considered as an essential precondition for the establishment of stable democratic states because it fosters a national identity. When people feel connected within a distinct geographical territory, a significant form of nation-state cohesion arises. This national in-group identification or a territorially restricted ‘we-feeling’ serves in his turn as a facilitator for political trust by making people more willingly to cooperate with each other, to unite themselves under the same government and to be governed exclusively by a representative portion of themselves. The rationale behind this is that a common national identity overrides sub-group interests (Berg & Hjerm, 2010; Hjerm, 1998; Mill, 1861). In understanding cohesion as a possible basis for democratic political institutions, it appears that we have to take national identity and its relationship to trust into consideration (Löden, 2014).

The general idea of the relationship between national identity and political trust is well-founded in theory. National identity binds people in a geographical distinct nation together, which is ought to be necessary for gaining trust in common democratic institutions. Thus, the stronger one’s national identity, the higher the levels of trust in the national political institutions should be. It is, however, suggested

21

that it is important to take the content or form of this national identity into account as well (Berg & Hjerm, 2010). As explained in the previous sections, conventional wisdom traditionally assumes that citizenship can be divided into two broad but distinct categories: civic and ethnic citizenship. Both theoretical forms of citizenship have in common that they formulate beliefs about the organisation of a society and about who can be regarded as a national in-group member. Civic citizenship traditionally stems from the territoriality principle (ius soli) and can, therefore, be regarded as open based on political loyalty; whereas ethnic citizenship is derived from the law of blood (ius sanguinis) and is, thus, considered as rather closed based on ancestry (Brubaker, 1992; Kohn, 1945; Reeskens & Hooghe, 2010; Shulman, 2002, 2004; Smith, 2010). Given these different connotations, it has been suggested that they impact individuals’ political trust in different ways. Two opposite hypotheses can be found in theory. On the one hand, more traditional political theorists have assumed that a culturally homogenous community is more likely to produce legitimate, functional political institutions. Ethnic heterogeneity was seen as detrimental for trust and thus for the existence of democratic institutions (Mill, 1975). Other scholars, on the other hand, argue that a civic national identity is equally likely to produce legitimate institutions because of its inherent connection to common procedures, laws and institutions (Berg & Hjerm, 2010).

In sum, we expect that a collective national identity matters for political trust. The firmer this national cohesion, the more confidence citizens will have in their nation-wide politicians and legal-political institutions. Concerning the specific form of this national identity, two opposite hypotheses can be deduced. It is, moreover, not certain that we will find two or three different forms of citizenship within our African subsample. We will therefore focus on the effect of the strength of national identity upon individuals’ trust in political institutions based on individual-level survey data gathered in Ghana.

22

2.2.2. Cultural versus institutional performance theories

Although some form of social cohesion or shared idea about citizenship amongst members of society seems needed in order for political institutions to exist over time (Cederman, 2001), scholars focussing on the relationship between national identity and political trust (e.g. Berg & Hjerm, 2010; Löden, 2014) tend to forget other underlying causes of trust. Two general perspectives predominate the discussion about the sources of political trust and, thus, provide different means to generate trust for newly established institutions: the cultural and institutional performance approaches.

Cultural theories, including both sociological and social-psychological explanations, propose numerous affective explanations of political trust – such as group identification but also socialisation patterns and interpersonal trust. Trust is a developing feature linked to basic forms of social relations and communicated through early-life socialisation. Because of different socialisation patterns, cultures can differ greatly in their levels of trust or distrust of others. Besides this macro-level idea, micro-level cultural theories emphasise substantial within-country differences in values linked to social, economic and political differences. Political beliefs are learned through early-life experiences associated with individuals’ position in society as reflected in their education, gender, age and socio-economic status (Mishler & Rose, 1997). Girls, for instance, seem on average mere trusting than boys (Putnam, 2000).

Besides national in-group identification and socialisation patterns, cultural and social capital scholars consider interpersonal trust as equally important to generate political trust. Political trust is seen as an extension of interpersonal trust projected onto the political institutions, thereby creating a civic culture. People who have more trust in each other are also more likely to support both formal and informal institutions (Mishler & Rose, 2001; Putnam et al., 1993; Putnam, 1995).

23

Thus, cultural theories assume that political trust is exogenous to or originated outside the political sphere in the cultural values of the people. In general, these theories offer three basic hypotheses about the origins of political trust: national identity or citizenship, socialisation and interpersonal trust (Berg & Hjerm, 2010). Because cultural theories consider political trust as culturally determined and rooted in deep-seated societal norms or socialisation pattern, they predict that it takes decades or generations to cultivate sustainable trust in institutions and political actors (Mishler & Rose, 2001). In this article we are interested how important national cohesion and citizenship conceptions are for cultivating political trust among Ghanaian students compared to interpersonal trust and socialisation characteristics.

Besides these cultural explanations, another main strand of research has strongly suggested that ‘political trust seems to be a reflection of the external or objective conditions’ (Newton, 2001, p. 205). Institutional performance theories (e.g. Citrin, 1974; Hutchison & Johnson, 2011; Miller, 1974; Mishler & Rose, 1997, 2001; Williams, 1985) are founded in the rational choice perspective. This research tradition argues that trust in legal-political institutions and in actors is rational, based on evaluations of the political world and its capabilities. Trust arises from reasonable responses to institutional performances. Political trust increases as a result of the successful delivering of public goods by the government and decreases when government fails. Politicians prove themself worthy of trust by, for instance, building roads, improving the economy or providing jobs. The deliberate choice of institutional designs and policy options has tangible consequences for government performance and, hence, for public trust in politicians as well. Institutions performance theories assume, therefore, that trust in political institutions and actors is endogenous or originated inside the political sphere (Mishler & Rose, 2001). Moreover, individual evaluations of institutional performance seem important regardless of the outcome of policies (Huseby, 2000).

24

It is worth noting that one has to clearly differentiate between political trust and evaluations of government performances because these concepts are not interchangeable notions (Catterberg & Moreno, 2006). In this paper, as mentioned above, political trust is operationalised as the trust people have in diverse actors and institutions whereas institutional performances is measured by the perceived effectiveness of the government in taking diverse responsibilities. We propose that perceived government capability, in this case effectiveness, is an important factor in shaping individuals’ ascription of trust to political actors and legal-political institutions.

Although political trust is always seen as endogenous by adherents of this approach, trust can be cultivated differently in established and developing democracies. Because the structure and character of political institutions are already established in consolidated democracies, the policy performances are more important. More specifically, institutions are trusted or distrusted in established regimes to the extent that they produce the preferred policy outcomes, mainly of economic nature. However, in new democracies – such as most post-colonial African countries – the political character matters as much as the policy outputs and political outputs matter as much as economic and social ones. Consequently, new democracies have numerous opportunities to increase trust by ameliorating several of these aspects; from system and party reforms to policy initiatives increasing human well-being. By taking those measures that may take months or even years but not need to take decades or generations, new democratic institutions can generate increased trust rather quick (Miller, 1974; Mishler & Rose, 2001).

Relative few studies have investigated the cultural and institutional determinants of political trust and the empirical evidence is not unambiguous. Regarding the cultural approach, previous studies have firstly found positive effects of social background, gender, age, education and occupational status on political trust (Christensen & Lægreid, 2005; Schoon & Cheng, 2011), whereas

25

others have found negative (Döring, 1992) or non-significant associations (Newton & Norris, 2000; Newton, 2001). The overall picture suggests that socio-demographic variables are not very important in explaining levels of political trust. Moreover, it is a commonly held belief that those variables are correlated with many others and thus should be included as control variables. As a result, this study will examine the impact of socio-demographic variables first (RQ5) before including the significant variables as control variables in further analyses. Secondly, focussing on national identity. Berg and Hjerm (2010) have concluded that a strong civic national identity has a positive impact on political trust whereas a strong ethnic national identity has a negative impact on political trust. As we are uncertain of the dimensionality of citizenship within Ghana, we focus on the impact of the strength of citizenship on political trust while controlling for the significant socio-demographic variables (RQ6). Thirdly, although Putnam (1993; 1995) convincingly claims that interpersonal trust increases political trust (RQ7), the linkage and direction between interpersonal and political trust has also been increasingly criticised in empirical studies (Kaase, 1999; Mishler & Rose, 2001). Finally, in support of the institutional performance perspective (RQ8), Catterberg and Moreno (2006) and Hutchison and Johnson (2011) have demonstrated that political trust is closely tied to government performance and the system’s capability to increase or maintain well-being.

Some studies have compared both traditions but this has not resulted in a substantive clarification (RQ9). Based on macro- and micro-level data, Mishler and Rose (2001) have argued that governments in post-Communist countries can generate trust the ‘old-fashioned way: they can earn it by responding promptly and effectively to public priorities’. Their results strongly supported the superiority of the institutional approach while disregarding the cultural one – a conclusion shared by Wong et al. (2011). Nevertheless, other scholars have pointed out that both socialisation

26

or cultural experiences and evaluations of institutional performances influence levels of political trust (Mishler & Rose, 1997). In addition to this empirical ambiguity, King et al. (2003) are also right on the mark with their concerns about cross-cultural comparability of measurements of complicated concepts in survey research. It is highly questionable whether one of these conclusions will be the same within a completely different context such as the Ghanaian one.

2.3. The African context It is extremely notable that most of previous studies concerning citizenship, national identities and political trust have been conducted on western policies and people (Berg & Hjerm, 2010; Christensen & Lægreid, 2005; e.g., Kohn, 1945; Meinecke, 1907; Mishler & Rose, 1997, 2001; Moreno et al., 1998; Moreno & Arriba, 1996; Moreno, 2006; Reeskens & Hooghe, 2010; Reijerse et al., 2013; Reijerse, 2012; Shulman, 2002, 2004). The lack of research about the ways in which nationalism, citizenship and political trust is negotiated within African state borders is surprising considering the far-reaching social and political consequences of the topic. The African continent provides a unique history and accompanying narratives of citizenship claims. Fostering a national identity overriding ethnic affiliations is still a fundamental challenge in multi-ethnic societies in general and in post-colonial African countries in particular (Langer et al., 2014). The concept of citizenship is, therefore, extremely helpful in understanding ethnic politics and claims for self-determination within emergent democracies in the developing world because it introduces a discourse of rights and obligations as well as opportunities and constraints in a political community (Ndegwa, 1997).

Ethnic tribes and group are preeminent in African countries occasionally causing political and social tensions. Kaplan (1994) noted that in much of the developing world (e.g. most of the African continent), ethnic groups are likely to be involved. Politicians in the

27

developing world also frequently draw upon ethnic blocs of voters as illustrated, for instance, by the ‘baoulisation’ of Bédié in Côte d’Ivoire (Langer, 2005). Other scholars, however, argue that conflict remains a rare event even when there is a high level of ethnic diversity. In order to clear this ambiguity, it seems important to determine the conditions under which ethnic politics turns violent (Bates, 2012). Of course, this is an extremely complex question involving many different aspects such as political exclusion and horizontal inequalities among others, but first and foremost it seems important that people adhere to their ethnic group and exclude others from them based on ethnic criteria. In other words, an adherence to a civic, ethnic, cultural or multiple citizenship conception can make an important difference.

Furthermore, little is known about the specific means needed by African politicians and institutions to increase their legitimacy with the population. It is, however, important to explore the determinants of trust within the African context because most of these post-colonial countries – such as Ghana – suffer from extremely low level of initial political trust (Delhey et al., 2011; Hutchison & Johnson, 2011).

2.3.1. Post-colonial Ghana

In this vein, the Ghanaian context is of particular interest considering its post-colonial efforts of nation-building. After its independence, Ghana wanted to create a pluralistic nationality guaranteeing universal rights to all citizens, including people of foreign origin. Ever since, Ghanaian political elites have aimed at formally and informally reducing ethnic electoral politics and ensuring that political parties are national in character. Furthermore, they also tried to maintain the ethno-regional balances in political sphere as well as the cultural and religious ‘neutrality’ of the state. At last, the symbolic impact of the continuous efforts to reduce the

28

developmental gap between ethnic groups was also influential in maintaining national integration and reducing ethnic identities (Langer, 2007). Some indigenous Ghanaians, however, declared the non-autochthonous population as ‘aliens’ who should be excluded from the economical, political and societal benefits of citizenship. Additionally, Ghana had been characterised by an extremely low level of trust on both the interpersonal and political level (Delhey et al., 2011). Consequently, Ghanaian politicians struggled immensely to redefine the nationality status of those migrant-descants and to consolidate their post-colonial nation-state. Popular conception (i.e. only autochthonous are ‘true’ citizens) led to protest against the constitution that recognised citizenship by birth in Ghana regardless of the origin of the parents (Kobo, 2010). However, despite these tensions and the low level of trust, Ghana has been characterised by a notable low level of national ethnic tensions (Langer, 2007).

2.3.2. Ethno-regional and religious composition of Ghana

In order to come to grips with our country-study, some background information on Ghana in general and specifically as to its ethnic and religious composition is needed. The Republic of Ghana is an independent country since 1957 after more than 50 years of British colonisation. It is located in West Africa, bordered by Burkina Faso in the North, the Gulf of Guinea in the South and Ivory Coast in the West. The capital of Ghana, Accra, is situated in the utter South of the country in the region of Greater Accra.

Since its interdependence, Ghana’s population increased from over the six million people to approximately 25 million people – an increase of three-and-half-times within fifty years. As a result, the population density more than tripled between 1960 and 2010, which presents a challenge for the political management of Ghana. Ten administrative regions divide the Ghanaian population (see Map 1 on page 43). According to the 2010 census, the Akan are by far the

29

largest ethnic group in the country (47.3%), followed by Mole-Dagbani (16.6%), Ewe (13.9%), Ga-Dangme (7.4%), Gurma (5.7%) and Guan (3.7%). Every ethnic group in turn consists of smaller ethnic groups, as there are some 60 to 75 different groups present in Ghana in total (Ghana Embassy, s.d.; Langer & Ukiwo, 2007). The majority group, the Akan, are concentrated mainly in the Western, Central, Eastern, Ashanti, Greater Accra and Bron Ahafo regions. The great majority of the population of Volta is Ewe and for the Northern, Upper East and Upper West regions the predominant group was the Mole-Dagbani. Table 3 on the next page shows the ethno-regional composition of Ghana.

Considering the religious composition of Ghana, more than 70 percent of the population reports to be Christians in 2010 (i.e. Catholic, Protestant, Pentecostal/Charismatic and other Christian), followed by Islam (17.6%) and Traditionalists (5.2%). Pentecostal/ Charismatic (i.e. Christian) is the dominant religious group in seven out of the ten administrative regions: Western, Central, Greater Accra, Volta, Eastern, Ashanti and Brong Ahafo. The Northern regions are the notable exceptions: three out of five residents in the Northern region are affiliated to Islam (60.0%). the people in the Upper West are mainly Catholics (35.7%) and the ones of the Upper East report to be mainly Protestants (24.8). Table 4 on page 43 shows the distribution of the religious affiliations by region.

In addition, young people dominate the age structure of the Ghanaian population with about 40 percent under the age of 15. The elder people (+65) only account for 4.7 percent of the total population – making our student sample more relevant. It is also worth noting in the light of our research that the proportion of the population recorded as Ghanaian by birth has increased while the proportion of Ghanaians by naturalization and of the non-Ghanaian population has consistently declined since 1960.

30

Tabl

e 3:

Eth

nic

com

posit

ion

of G

hana

’s a

dmin

istra

tive

regi

ons i

n 20

10 (p

erce

ntag

e).

Sour

ce: G

hana

Sta

tistic

al S

ervi

ce, 2

010,

p. 6

2.

SO

UTH

ERN

REG

ION

S

NO

RTH

ERN

REG

ION

S

Gha

na

W

este

rn

Cen

tral

Gre

ater

A

ccra

V

olta

Ea

ster

n A

shan

ti B

rong

A

hafo

Nor

ther

n U

pper

Ea

st

Upp

er

Wes

t

Aka

n 78

.2

81.7

39

.7

2.8

51.1

54

.2

58.9

3.1

47

.3

Ga-

Dan

gme

3.1

2.5

27.4

1.

5 17

.9

1.2

1.3

0.

3 0.

1 0.

1

7.4

Ewe

6.2

6.2

20.1

73

.8

18.9

3.

8 3.

7

1.7

0.3

0.4

13

.9

Gua

n 0.

8 5.

3 1.

9 8.

1 5.

3 1.

5 4.

1

8.6

0.3

0.8

3.

7 G

urm

a 0.

9 0.

9 1.

6 11

.3

1.6

2.8

6.9

27

.3

4.7

1.2

5.

7 M

ole-

Dag

bani

8.

6 1.

7 5.

2 0.

5 3.

2 11

.3

18.2

52.7

74

.7

73.0

16.6

Gru

si

0.8

0.5

1.3

0.1

0.8

2.0

3.9

3.

7 8.

6 20

.6

2.

5 M

ande

0.

8 0.

4 0.

7 0.

1 0.

3 2.

0 1.

8

0.5

5.6

0.3

1.

1 O

ther

s 0.

6 0.

8 2.

0 1.

8 0.

8 1.

1 1.

3

2.1

3.4

2.1

1.

4

30

31

Tabl

e 4:

Rel

igio

us c

ompo

sitio

n of

Gha

na’s

adm

inist

rativ

e re

gion

s in

2010

(per

cent

age)

.

Sour

ce: S

ourc

e: G

hana

Sta

tistic

al S

ervi

ce, 2

010,

p. 6

3.

SO

UTH

ERN

REG

ION

S

NO

RTH

ERN

REG

ION

S

Gha

na

W

este

rn

Cen

tral

Gre

ater

A

ccra

V

olta

Ea

ster

n A

shan

ti B

rong

A

hafo

Nor

ther

n U

pper

Ea

st

Upp

er

Wes

t

No

relig

ion

6.7

6.6

3.4

6.6

6.5

5.4

7.3

2.

7 2.

8 3.

5

5.3

Cat

holic

16

.2

11.1

7.

5 17

.6

7.9

12.7

20

.1

7.

6 19

.9

35.7

13.1

Pr

otes

tant

21

.1

21.0

22

.3

21.5

24

.8

19.7

17

.7

5.

0 7.

1 3.

0

18.4

Pe

ntec

osta

l/ C

haris

mat

ic

29.5

29

.8

44.6

26

.6

36.3

30

.1

24.5

6.3

11.8

4.

3

28.3

Oth

er

Chr

istia

n 15

.2

21.4

8.

9 7.

1 15

.5

15.3

9.

9

2.1

2.9

1.3

11

.4

Isla

m

9.4

8.7

11.9

5.

7 6.

7 15

.2

17.0

60.0

27

.1

38.1

17.6

Tr

aditi

onal

ist

0.8

0.6

0.5

14.1

1.

4 0.

7 2.

7

16.0

27

.9

13.9

5.2

Oth

ers

1.0

0.9

1.0

0.8

0.9

0.8

0.7

0.

4 0.

6 0.

3

0.8

31

32

33

3. Research objectives, questions and hypotheses

This research sheds light on the structure of citizenship on the one hand and on its impact on political trust on the other hand. Based on the above literature review, eight broad research questions and subsequent specific hypotheses are generated in order to examine our two-folded research objective.

3.1. Structure of citizenship From the late nineteenth-century, the study of citizenship has been marked by intractable ambiguities and clear black spots. Both two- and three-factor solutions are suggested but the ‘Moreno Question’ – stating that people can adhere to both ethnic and civic criteria at the same time – has clearly contradicted this civic/ethnic dichotomy. Consequently, it is possible to find a one-, two-, or three-factor solution. Studies investigating the underlying structure of citizenship conceptions and their interrelatedness have not resulted in a substantive clarification (Wright, Citrin, & Wand, 2012). On the contrary, the scare information about their interrelatedness seems to suggest that those factors are not that different from each other at all. Nevertheless, they are treated as opposites by academics and politicians. Finally, information on the conceptualisation of citizenship and the relation between the different representations within an African context is non-existent. This study will therefore first empirically analyse the structure of citizenship in Ghana. In order to do so, two specific research questions with accompanying hypothesis are addressed.

First, we explore whether there are underlying citizenship dimensions in students’ opinion, and if so, to what extent these dimensions coincide with the civic/ethnic/cultural division (RQ1). A preliminary analysis of our data does suggest that a very high proportion of Ghanaian students claim some form of dual identity

34

regardless of their ethnicity (see Figure 1). More than half of the sample claims to feel equally Ghanaian and a member of his ethnic group. Thus, it seems perfectly possible that Ghanaian students will incorporate both ethnic and civic requirements in order to form their Ghanaian in-group. The logic is straightforward: if students feel equally Ghanaian and a member of their ethnic group, they will appreciate both civic and ethnic requirements for being truly Ghanaian. We therefore predict that Ghanaian students will adhere to numerous (or no) citizenship requirements simultaneously – we call this complex citizenship (H1).

Ghanaian versus ethnic identity

Figure 1: Ghanaian versus ethnic identity by ethnic groups.

35

Second, this article addresses the questions whether civic, ethnic and cultural conceptions of nationhood are mutually exclusive or complementary (RQ2). In much of the literature, there is an implicit understanding that ethnic and civic citizenship conceptions exclude each other. The fact that Ghanaian students feel equally Ghanaian and a part of their ethnic group, however, also suggests that civic, ethnic and cultural notions of nationhood will be highly complementary instead of competing or exclusive (H2). Hence, this study addresses following questions:

RQ1: What is the dimensionality of citizenship notion among Ghanaian university students?

H1: Based on the ambiguity about the interrelatedness of citizenship conceptions and on the ‘Moreno Question’, a one-factor solution of citizenship will be found indicating a complex citizenship conception.

RQ2: Are civic, ethnic and cultural citizenship notions mutually exclusive or complementary?

H2: Civic, ethnic and cultural citizenship conceptions will be highly correlated.

3.2. Impact of citizenship on political trust The second objective of this master thesis is to examine the impact of citizenship on political trust. In order to do so properly, we have to dissect the structure and level of our dependent variable first. Second, the determinants of political trust are assessed. In addition to analysing the impact of citizenship, the effects of interpersonal trust and perceived institutional effectiveness are explored as well while always controlling for significant socio-demographic variables. The research questions and hypotheses will allow us to develop a comprehensive model of political trust in the end.

36

3.2.1. Structure and level of political trust

Trust between citizens and the political institutions representing them is assumed to be fundamental for democracies. This is especially important in the context of developing countries trying to establish legitimate and effective institutions. Because Ghana is identified as one of the least trusting countries (Delhey et al., 2011), we try to develop a comprehensive model of trust incorporating diverse theoretical arguments. Before assessing the impact of divers variables on political trust, it is important to empirically assess the dimensionality and level of political trust (RQ3, RQ4).

Political trust is defined in this study as the trust citizens have in the ‘state-wide legal-political institutions and actors’ (Berg & Hjerm, 2010, p. 391) and is differentiated from generalised trust. A work that should be taken into account when talking about political trust is the one of David Easton (1957, 1965) who developed a two-dimensional framework. His first dimension contains support for the political authorities, community and regime, while the other ranges from specific to diffuse political support. Moreover, Norris (1999) also stresses the multi-dimensionality of political support and develops a five-folded conceptualisation drawing a line between different levels or objects of political support (i.e. the political community, regime principles, regime performance, regime institutions and political actors). Thus, many dimensions can be distinguished theoretically; but major problems arise when attempting to measure these dimensions within the public opinion. Most empirical analyses find only one factor including trust in political institutions and actors (H3)

Additionally, Ghana has been identified as a relatively low trusting society in previous research (Delhey et al., 2011). We therefore suspect that our Ghanaian subsample will also display a relatively low level of political trust (H4). Thus, the following research question and hypotheses about the dimensionality and level of political trust are formulated:

37

RQ3: What is the dimensionality of political trust according to Ghanaian university students?

H3: Based on previous empirical studies, a one-factor solution of political trust will be found indicating an all-embracing political trust.

RQ4: What is the level of political trust according to Ghanaian university students?

H4: Political trust among Ghanaian university students will be considerable low.

3.2.2. Origins of political trust

First and foremost, we aim to assess the impact of citizenship on political trust. However, in order to do so properly, we have to control for important socio-demographic variables correlating with both independent and dependent variables. Furthermore, the cultural perspectives (e.g. Mishler & Rose, 2001) argue that gender, age, socio-economic status and many other socio-demographic variables also independently influence political trust (RQ5). Nevertheless, the majority of the researchers agree that there is no such thing as a ‘trusting personality’ influenced by socialisation and demographic factors (H5) – although the evidence remains highly inconclusive4. Thus, we will first assess the independent effects of socialisation variables before controlling for the significant variables in further analyses.

Second, our main assumption is that political trust is based on a common nationality. Based on previous theoretical and empirical studies (e.g. Berg & Hjerm, 2010; Löden, 2014; Mill, 1861), we believe that political trust partly reflects the strength of citizenship. 4 Previous studies have found positive effects of social background, gender, age and education on political trust (Christensen & Lægreid, 2005; Schoon & Cheng, 2011), whereas others have found negative (Döring, 1992) or non-significant associations (Newton & Norris, 2000; Newton, 2001).

38

More strict beliefs about who can be regarded as a national in-group member indicates stronger Ghanaian in-group identifications or a more solid sense of Ghanaian ‘we-feeling’. Such a stronger identification with the Ghanaian in-group should, in turn, foster greater trust in political institutions and actors (RQ6, H6).

Third, social capital theories (e.g. Putnam et al., 1993; Putnam, 1995) also claim that interpersonal trust has an important influences on political trust because it ‘spills over onto political institutions and actors’ (RQ7). Therefore, we hypothesise that individuals with higher levels of interpersonal trust are more likely to trust their government compared to individuals with lower levels of interpersonal trust (H7).

In addition to these cultural theories, institutional performance studies (e.g. Hutchison & Johnson, 2011; Newton, 2001) offer equally viable ways to cultivate political trust. Those theories assume that political trust is dependent of individuals’ evaluations of institutional effectiveness (RQ8). We believe that the observed effectiveness of the government to implement policies should be positively related to levels of popular trusts in government. If the leadership is effective, competent and efficient in delivering the desired public goods, political trust will increase (H8). In conclusion, four research questions with accompanying hypotheses will be tested in order to carefully evaluate the impact of different variables, including citizenship, on political trust:

RQ5: What is the impact of socio-demographic variables on

political trust? H5: The impact of divers socio-demographic variables will be

non-substantial to insignificant. RQ6: What is the impact of citizenship on political trust, while

controlling for significant socio-demographic variables? H6: Students with stronger citizenship beliefs will be more likely

to show a higher level of trust in political institutions and actors (cultural theory: national identity).

39

RQ7: What is the impact of interpersonal trust on political trust, while controlling for socio-demographic variables?

H7: Students with more interpersonal trust will be more likely to show a higher level of trust in political institutions and actors (cultural theories: interpersonal trust).

RQ8: What is the impact of the sense of institutional effectiveness on political trust, while controlling for significant socio-demographic variables?

H8: Students with more positive evaluations of their governments’ performance will be more likely to show a higher level of trust in the same political institutions and actors (institutional theory: effectiveness).

Although both cultural and institutional theories have often been characterised as incompatible (Schoon & Cheng, 2011), one should not overestimate the opposition between cultural and institutional theories. Arguably, the key difference between both perspectives is their time frames. The cultural theories emphasise the primacy of early-life socialisation and group formation while the institutional performance theories focus more on recent and contemporary experiences with government activities. Both experience-based theories could therefore be integrated into a ‘developmental or lifetime learning model’ in which political trust starts to develop in early life and evolves continuously thereafter through subsequent experiences (Mishler & Rose, 1997, 2001).

Hence, our research questions and hypotheses should not be considered as mutually exclusive. On the contrary, individual cultural characteristics and performance evaluations can reinforce each other in influencing political trust. Consequently, this article tries to supplement the trust-building force of a national identity with other mechanisms in order to create one comprehensive model of political trust in Ghana. The comprehensive model – based on both cultural and institutional theories – that will be tested in this study can be

40

found in Figure 2. Such a comprehensive model is indicative for the Ghanaian government trying to establish more trust with its community. This model can be used as basis for other single-country or comparative studies on trust-building in Africa. It is, however, possible that one theory (i.e. the cultural or institutional performance theory) will be more powerful in explaining political trust compared to the other (RQ9). As explained in the literature review, most studies tend to slightly prefer the institutional performance perspective above the cultural one (H9). We therefore ask: RQ9: What is the relative importance of cultural and institutional

performance theories when integrating all variables into a comprehensive model?

H9: The institutional performance variables are more powerful in explaining political trust compared to the cultural variables.

Figure 2: Comprehensive model of the origins of political trust.

41

4. Methodological consideration

As explained in the introductory chapter, this master thesis aims at a deductive, quantitative analysis technique. This means that we have deducted hypotheses based on theoretical considerations that must now be subjected to empirical scrutiny (Bryman, 2012). To test our ideas accurately, we conducted a secondary analysis relying on a wealth of primary data collected by the 2013-‘National Service Scheme Survey’ (N3S) researchers in Ghana. The purpose of this longitudinal survey study is to investigate inter-group relations and national identities (Langer et al., 2014). The secondary analysis offered us the prospect of having access to high-quality data, which saved cost and time to focus on extensive data-analysis. Moreover, this public opinion data made sure that we could draw from a representative large sample (N = 3,264) to make inferences regarding citizenship and its effects. It is, however, equally true that this quantitative method left no room for more comments or ideas of the participants (Bryman, 2012). Future research could, for instance, complement our findings with more in-depth and qualitative insights (for more suggestions, see 6.2).

The following paragraphs continue specifying our cross-sectional research design. We will first shed light on the data source and collection. It is of equal importance to assess the rationales behind the continent and country as well as behind the selected universities and students of the N3S study. Second, the operationalisation and measurement of the variables will be explained with attention for the used language, conceptualisation and terminology. Last, the used statistical analyses are briefly evaluated. Overall, it is noteworthy that the external validity of our county study, that is the degree of generalisation across social settings (Bryman, 2012), remains questionable because we concentrated on a specific subsample (i.e. university students) within only one African country (i.e. Ghana).

42

4.1. Data source and collection

4.1.1. Context and country

As discussed above (see 1.2 or 2.3), information on the structure and impact of citizenship in a non-western context is lacking – especially within the context of recently established, post-colonial, African countries. Besides some theoretical arguments, little is known about the structure of citizenship based on African public opinion data or about the means needed by African politicians and institutions to increase their legitimacy with the population. Consequently, one of the biggest innovations of this study is the fact that the analysis is based on individual-level data gathered in an African country. Africa was an ideal region for this study because it offered the opportunity to study citizenship and political trust in a completely other culture with different levels of citizenship, trust or government effectiveness.

Ghana was chosen first because of pragmatic reasons as the N3S study gave us unique, reliable and recent data on citizenship, political and interpersonal trust and institutional effectiveness evaluations. Moreover, Ghana was a good example of a post-colonial African country because it of its specific post-colonial nation-building policies (Langer et al., 2014) and because it has been identified as one of the least-trust societies (Delhey et al., 2011). Hence, Ghana will benefit the most of more knowledge about the underlying causes of trust. As mentioned before, this study is not representative for the African region, but it can nevertheless be treated as an exemplifying case for the West-African region (Bryman, 2012) and it nonetheless provides profound and state-of-the-art insights on the business of citizenship. These insights could be used in future studies on nation- and trust-building in other African countries.

43

4.1.2. N3S study: selection of schools and students

The N3S scholars carefully selected three universities (see red circles on Map 1) in a stratified way – University of Ghana (UG) in Legon, Kwame Nkrumah University of Science and Technology (KNUST) in Kumasi and University for Development Studies (UDS) in Tamale and surrounding areas5 – for three reasons. First, the universities are geographically spread in terms of location, which ensured that the survey had a proper ethno-regional and religious mix of respondents. Second, this school selection was in keeping with the available resource mix of scholars, funds and time. Third, the universities are among the biggest public universities in Ghana. Because the vast majority of Ghanaian students are enrolled in one of their country’s nine public universities, we were confident that by selecting these three universities we would obtain a representative sample of the Ghanaian student population (Langer et al., 2014).

5 UDS consists of several campuses, namely Wa (Upper West Region), Navrongo (Upper East Region), Nyampkala and Tamale (Northern Region).

Map 1: Ghana's administrative regions and cities of universities.

44

Having selected three public universities, the next step was to recruit a sample of students within the populations of these universities. This study relies on a subsample interviewed at a second stage and based on the multi-stage cluster sample (Bryman, 2012) of the first in-class survey stage. 5368 students were questioned by means of a web self-administered questionnaire (WSAQ) and 3264 students at least partially completed the survey. It is worth noting that 60.3 percent is a notably high response rate, and this can be partially explained by the extensive efforts made by the scholars to maximise the response rate (i.e. by implementing principles of the Tailored Design Method. diverse conditional and unconditional incentives and reminders via different means of communication). Additionally, the ratio between the number of completed surveys and the total number of started surveys equalled 90.3% (i.e. the completion ratio). More information on the methodology that was developed for the N3S study can be found in Langer et al. (2014).

Some important remarks have to be made as to the composition of our sample. The students ranged in age from 16 to 53 with a mean age of 23. An abundance of research (e.g. Galston, 2001; Sapiro, 2004) has shown that this age (i.e. adolescence) is crucial in the development of young people’s political thinking and socialisation. It is thus presumable that students will form their notions about the national in-group and about the trustworthiness of their government mainly at this age. It is also worth noting that only one-third of our respondents was female. This corresponds with Atuahene and Owusu-Ansah (2013) finding that only 37 percent of university students in Ghana were female in 2011. Moreover, almost one third of the students came from the Greater Accra region and a quarter belonged to the Assante ethnic group. A great majority of the students was Christian whereas only a small 10 percent was Muslim – which also corresponds with the religious composition of Ghana (GSS, 2013). Moreover, half of the respondents clearly lacked political affiliation, as 50.3% of the students did not feel close to any

45

Ghanaian political party. This could be a first indicator of a relatively low level of political trust. If they did feel close to a political party, they mostly adhere to the NPP (33.7%) or NDC (12.6%). In addition, parental education (i.e. a proxy for socio-economic status) was also examined. On average parents completed secondary school or a post-secondary qualification (other than university). More information on the descriptives of the key variables can be found in Appendix 2 on page 109 and on the socio-demographic variables in section 4.2.3.3. It is true that this sample contains a particular subgroup of the population in terms of educational level, literacy, financial resources, mobility and access to new technologies. Nevertheless, we argue that they form an extremely interesting subject of study because today’s students are tomorrow’s societal elites. Hence, these students are more likely to impact and shape citizenship perceptions of others such as their peers and family. It is also especially important to raise political trust among this subsample because the students are less likely to engage in politics later on if they lack political trust. Finally, an online survey strategy appears to be very useful to reach this subgroup indicated by the 60% response rate.

4.2. Measurements of the variables in the N3S study

4.2.1. Citizenship

In the survey, the respondents were questioned about the criteria they considered as important for being truly Ghanaian in order to answer the first two research questions. More specifically, they could rate eight different criteria on a 10-point Likert scale ranging from ‘not important at all’ to ‘very important’. The citizenship items were: being born in Ghana, respecting Ghana’s political institutions and laws, having lived in Ghana for most of one’s life, speaking English, feeling Ghanaian, having Ghanaian ancestry, speaking the a locale language and being married to a Ghanaian.

46

It is clear that these items are designed to test citizenship beliefs in Ghana; but in the literature thus far we do not find consensus about the validity of the three separate conceptions.

4.2.2. Dependent variable: political trust

Political trust was defined in this study as the trust citizens have in the ‘state-wide legal-political institutions and actors’ (Berg & Hjerm, 2010, p. 391) Although multiple dimensions can be distinguished theoretically; major problems arise when attempting to measure these dimensions within the public opinion. This is why we first assess the dimensionality of political trust within the Ghanaian public opinion.

In accordance with the bulk of survey-based research on the subject, political trust was operationalised by capturing trust in the most important institutions of a political system. The question was phrased: ‘On a 0 to 10 scale, how much do you personally trust each of the following institutions’. The institutions included: parliament, legal system, traditional leaders, police, politicians, NSS administration and the president. Respondents could rate those seven items on an 11-point Likert scale ranging from ‘not at all’ (= 0) to ‘very much’ (= 10).

4.2.3. Independent variables

Although citizenship and political trust were the main variables in this research, other variables were used as well in order to accurately estimate the effect of citizenship on political trust. This section outlines those other variables. More information on the variables and their descriptives can be found in the survey (Appendix 1) on page 99 and in the descriptives of the key variables (Appendix 2) on page 109.

47

4.2.3.1. Interpersonal trust

Interpersonal trust is defined as the trust between people (Campbell, 2006) and is seen as crucial for a civil society as it helps solving the collective action problems (Putnam, 2000). In this study, it was operationalised by measuring trust in people with diverse characteristics. The question was phrased: ‘How much trust do you have in people with the following characteristics?’. The options included: people from your ethnic group, from different ethnic groups, of your religion, of a different religion, from your region of origin, of different regions. Respondents could rate those six items on a four-point Likert scale ranging from ‘no trust at all’ (= 1) to ‘a lot of trust’ (= 4). This has been recoded into an 11-point Likert scale in order to enhance the comparability with the other variables. A principal component analysis (PCA) using Varimax rotation indicated that these six items load strongly on a singly dimension, explaining 66.52 percent of the total variance with an Eigenvalue of 3.99 and Chronbach’s α of .90. A composite index was then computed by averaging the scores on the six items (M = 6.53, SD = 1.49). Higher scores indicate higher levels of interpersonal trust.

Table 5: Descriptives of interpersonal trust items and scale.

n Mean SD Rank Q83_1. Your ethnic group. 2611 2.74 .73 2 Q83_2. Different ethnic groups. 2584 2.49 .69 4 Q83_3. Your religion. 2669 2.93 .77 1 Q83_4. Different religion. 2586 2.44 .74 6 Q83_5. Your region of origin. 2591 2.62 .74 3 Q83_6. Different regions. 2547 2.44 .70 5 Overall interpersonal trust 2731 6.53 1.49 Note. Entries represent average score. 0 = no trust at all. 4 = a lot of trust. Each item is preceded by the phrase: “How much trust do you have in people with the following characteristics? People from…”.

48

4.2.3.2. Sense of institutional effectiveness

To start with, it is valuable to look into the conceptualisation of perceived institutional effectiveness. Institutional performance is closely linked with the external dimension of the sense of political efficacy looking at the responsiveness of the political system (Claes, 2010). Our Ghanaian students were also questioned about their perceptions of the government’s effectiveness. More specifically, they could rate five different criteria on an 11-point Likert scale ranging from ‘not effective at all’ (= 0) to ‘very effective (= 10). These items were: combating poverty, creating employment opportunities, providing education, supporting poor and deprived people, promoting national unity. A PCA using Varimax rotation indicated that these seven items load strongly on a singly dimension, explaining 75.78 percent of the total variance with an Eigenvalue of 3.79 and Chronbach’s α of .92. A composite index (i.e. ‘sense of institutional effectiveness’) was then computed by averaging the scores on the seven items (M = 3.28, SD = 2.03). Higher scores indicate better evaluations of the government’s performances. Table 6: Descriptives of sense of institutional effectiveness items

and scale.

n Mean SD Rank Q42_1. Combatting poverty. 3012 2.72 2.15 4 Q42_2. Creating employment opportunities. 3013 2.50 2.11 5 Q42_3. Providing education. 3009 4.13 2.44 2 Q42_4. Supporting poor people. 3007 2.92 2.26 3 Q42_5. Promoting national unity. 3003 4.14 2.69 1 Overall sense of institutional effectiveness 3019 3.28 2.03 Note. Entries represent average score. 0 = not effective at all. 4 = very effective. Each item is preceded by the phrase: “The Ghanaian government takes on certain responsibilities. Can you indicate for each of the following items to what extent you think the government is effective or not? In …”

49

The N3S Survey also gives us the unique opportunity to evaluate the extent to which evaluations of the contributions of the National Service Scheme (NSS) influence political trust. Most African countries embarked upon a process of nation-building after gaining independence in the 1960s (Berman, 2010). In this respect, African countries have adopted a range of policies in order to foster stronger national identities; including establishing national youth service programs (Bandyopadhyay & Green, 2013). The national service system is an important policy tool pervading the lives of adolescence and it is, therefore, interesting to analyse the effect of their views on the contributions of this program. In the survey, the students were asked about their perceptions of the contribution of the NSS program. The question was phrased: ‘The NSS has several objectives, indicate for each of them to what extent you think the NSS program makes a substantial contribution’. The options included: improving national unity, diminishing poverty, diminishing illiteracy, diminishing unemployment, delivering essential amenities to the rural areas in Ghana. Respondents could rate those seven items on an 11-point Likert scale ranging from ‘not at all’ (= 0) to ‘very much’ (= 10). A PCA using Varimax rotation indicated that these five items load strongly on a singly dimension, explaining 67.34 percent of the total variance with an Eigenvalue of 3.37 and Chronbach’s α of .88. A composite index (‘sense of NSS contribution’) was computed by averaging the scores on the items (M = 3.76, SD = 2.21). Higher scores indicate better evaluations of the NSS contribution.

50

Table 7: Descriptives perceived NSS contribution items and scale.

n Mean SD Rank Q28_1. Improving national unity. 2901 4.54 2.92 1 Q28_2. Diminishing poverty in Ghana. 2895 3.05 2.35 5 Q28_3. Diminishing illiteracy in Ghana. 2889 3.84 2.69 2 Q28_4. Diminishing unemployment. 2890 3.79 2.60 3 Q28_5. Delivering of essential amenities to the rural areas in Ghana.

2891 3.67 2.78 4

Overall sense of NSS-contribution 2928 3.76 2.21 Note. Entries represent average score. 0 = makes no contribution at all. 10 = a lot of contribution. Each item is preceded by the phrase: “the NSS has several objectives, indicate for each of them to what extent you think the NSS program makes a substantial contribution.”

4.2.3.3. General background characteristics

Finally, one has to consider socio-demographic characteristics when looking into the structure and impact of citizenship. These variables provide general information about a person’s life and are considered as common controls in many social sciences research. Moreover, literature on political trust indicates that these background variables independently influence the levels of political trust as they represent different political socialisation experiences. Consequently, we also included many variables linked to different socialisation experiences. The individual impact of those socio-demographics was assessed first because they could influence other effects. The significant variables were then included as control variables in all other analyses. It is argued that gender, age, town size, socio-economic status (measured by parental education), ethnicity, religion and extent of religiousness, and political affiliation can be, to varying degrees, related to levels of trust. In accordance with previous research (Mishler & Rose, 2001), these variable are included as proxies for political socialisation. For information on the descriptives of these variables, see Appendix 2.

51

4.3. Data reduction: statistical analyses For the purpose of this study, all statistical analyses were performed with SPSS Statistics (version 22.0) or AMOS (version 23.0). As we had a two-folded research objective, we recurrently relied on two statistical tests: factor analyses and linear regression analyses.

4.3.1. Dimensionality of citizenship: factor analyses

Several factor analyses were conducted to examine the underlying structure of citizenship (RQ1, H1) and the interrelatedness (RQ2, H2) of the citizenship conceptions. Factor analysis makes it possible to reduce the citizenship criteria to one or more factors. Because we did not want to make an a priori assumption on the number of factors, we used exploratory factor analysis (EFA) on the eight items of the citizenship battery. We were interested in the number of factors that account for significant amount of common variance; and in whether or not the content of these founded factors corresponded to the two- or three-factor structure suggested in previous studies (Reeskens & Hooghe, 2010; Reijerse et al., 2013). Because we assumed that the citizenship conception factors are not independent from each other and thus correlated, we applied principle component analysis (PCA) with an oblique rotation (Promax) (Field, 2009). Second, next to this exploratory technique, confirmatory factor analyses (CFA) are used in order to really check the stability of the extracted solution. Based on the EFA-results and theoretical considerations, we constructed a one-, two- as well as a three-factor model to test in the CFA. This frequently used method gave us the opportunity to fully scrutinise the interrelatedness of the factors. It is worth noting that EFA is also used to analyse the structure of political trust (RQ3, H3) before using simple frequency techniques to explore the level of political trust (RQ4, H4).

52

4.3.2. Origins of political trust: linear regression analyses

Micro-level theories assume that political trust varies within a country as a result of dissimilar views on citizenship (RQ6, H6) as well as because of different socialisation experiences linked to gender, age, SES, town size or other social-structural influences (RQ5, H5), the level of interpersonal trust (RQ7, H7) and/or dissimilar evaluations of political and economical policy efficiency (RQ8, H8). Moreover, this study examined the relative importance on political trust of all suggested variables (RQ9, H9).

In order to predict values of the dependent variable (i.e. political trust) from the several independent variables, divers hierarchical linear regression analyses were conducted (Field, 2009). First, the effect of the socio-demographic characteristics (i.e. gender, age, SES, town size, religion, religiousness, ethnicity and political affiliation) were assessed (Model 0) because these variables could distort further results (Christensen & Lægreid, 2005). Next, the individual effects of national identity (Model 1), interpersonal trust (Model 2) and institutional effectiveness (Model 3) were analysed while controlling for the significant socio-demographic variables identified in the previous analysis. Last, the whole comprehensive model of political trust was examined to test the robustness of all variables included (Model 4).

It is important to briefly assess the issue of causality inherent to these analyses before expounding the results. The cross-sectional nature of our research design only made it possible to examine relationships between variables. Firm conclusions about causality require longitudinal data. There was, however, no time ordering to our variables because the all the data was collected more or less simultaneously. This created the problem that Bryman (2012) calls the ‘ambiguity about the direction of causality’. At the moment, The N3S researchers are up to collecting longitudinal data, which enables us to make firmer conclusions about causality in the future (see 6.2).

53

5. Results

This section displays all results needed to answer our nine research questions and validate the accompanying hypotheses. Firstly, an overview of the importance attached to the citizenship criteria is given before answering the questions about the structure of citizenship notions (RQ1, H1) and their interrelatedness (RQ2, H2). Secondly, the dimensionality (RQ3, H3) and level (RQ4, H4) of political trust is presented before answering the questions about the origins of political trust. After investigating the independent effect of socio-demographic variables on political trust (RQ5, H5), the impact of citizenship (RQ6, H6), interpersonal trust (RQ7, H7) and sense of institutional effectiveness (RQ8, H8) is assessed as well while controlling for the significant socio-demographic variables. Lastly, all determinants of political trust are combined into a comprehensive model in order to assess the relative importance of all variables included (RQ9, H9).

5.1. What is the structure of citizenship? Before answering research question one and two on the structure of citizenship in Ghana, the mean scores of the eight citizenship criteria are presented in Table 8. These descriptives made clear that various respondents considered most of these criteria as neither important nor unimportant. Respecting Ghana’s political institutions and laws, speaking a local language and feeling Ghanaian obtained higher ratings while being married to a Ghanaian received a substantially lower score. It is also noteworthy that all standard deviations were rather large meaning that the data spread far from the mean and, thus, that there is a lot of variability in importance attached to the criteria (Field, 2009). Because these descriptives did not reveal which items empirically belong together, we still had to assess the latent structure of citizenship using exploratory and confirmatory factor analyses.

54

Table 8: Descriptives of citizenship items.

Citizenship items n Mean SD Q59_1. To have been born in Ghana 2995 5.99 3.14 Q59_2. To respect Ghana’s political institutions and laws.

2992 6.41 2.99

Q59_3. To have lived in Ghana for most of one’s life.

2987 5.02 3.00

Q59_4. To be able to speak English 2993 5.02 3.58 Q59_5. To feel Ghanaian 2986 6.35 3.23 Q59_6. To have Ghanaian ancestry 2986 5.93 3.18 Q59_7. To speak a local language 2988 6.39 3.21 Q59_8. To be married to a Ghanaian 2983 4.56 3.26 Note: Entries represent row percentages. Each item is preceded by the phrase: “Some people say that the following things are important for being truly Ghanaian. Others say they are not important. How important do you think each of the following things are?”.

5.1.1. RQ1: Exploratory Factor Analysis

To investigate the latent structure of citizenship perceptions among Ghanaian students and to establish the empirical validity of the civic/ ethnic dichotomy (RQ 1, H1), we conducted an exploratory factor analysis (ECA) – more specifically, a principle component analysis (PCA) with oblique rotation (Promax) – on the eight items of the citizenship battery. The Kaiser-Meyer-Olkin measure of sampling adequacy (KMO) and Bartlett’s test of Sphericity both confirmed that conducting a factor analysis was an appropriate technique for this study. The KMO test statistics obtained a value of .910 which is generally considered as ‘superb’ (Hutcheson & Sofroniou, 1999). Thus, we could be confident that the sample size is adequate for factor analysis. Bartlett’s test was also significant (χ2 = 12 670.04, df = 28, p < .001), indicating that the correlations between the items were sufficiently large for an EFA (Field, 2009).

55

Table 9: KMO and Bartlett’s Test.

Kaiser-Meyer-Olkin Measure of Sampling Adequacy

.91

Bartlett’s Test of Sphericity Approx. Chi-Square 12670.04 df 28 Sig. .000 The following and main step in the analysis was to determine the number of underlying dimensions that can be well measured by the items provided our survey. For that purpose, the Cattell’s scree test (Cattell, 1966) and Kaiser-Guttmann rule were employed as some of the commonly accepted factor selection procedures based on the Eigenvalues (Brown, 2006). The Eigenvalues of the factors have been plotted in the scree plot in Figure 3. In line with the expectations, the scree plot clearly suggested a one-factor solution – instead of a two- or three-factor solution. Only one factor had an eigenvalue greater than 1.0 (4.75), explaining almost 60 percent of the variance. Consequently, the two- or –three-factor solution as commonly suggested by researchers was not presented in our data based on Ghanaian students’ opinion. Furthermore, all citizenship items loaded higher than .67 on this single factor indicating a strong loading; as can be seen in the EFA-results displayed in Table 10.

Figure 3: Scree plot of Eigenvalues as a result of an EFA with Promax rotation.

56

Table 10: Exploratory factor analysis of citizenship items.

Citizenship items Factor(s) Q59_1. To have been born in Ghana .78 Q59_2. To respect Ghana’s political institutions and laws. .73 Q59_3. To have lived in Ghana for most of one’s life. .79 Q59_4. To be able to speak English .67 Q59_5. To feel Ghanaian .83 Q59_6. To have Ghanaian ancestry .81 Q59_7. To speak a local language .84 Q59_8. To be married to a Ghanaian .70 Note: Entries represent factor loadings obtained with an exploratory factor analysis for categorical data (in SPSS 22) with Promax as rotation method and PCA as estimator. The figures in bold represent membership of citizenship items to one latent factor. All factor loadings are significant at 5% level.

Our EFA-results clearly contradicted the widespread belief of a distinct civic, ethnic and/or cultural citizenship representation commonly found in the western-dominated literature on citizenship (Reeskens & Hooghe, 2010; Reijerse et al., 2013; Wright et al., 2012). Citizenship was not categorised into two or three exhaustive and mutually exclusive groups in this specific casus as we could only distinguish one, overarching citizenship factor – labelled complex citizenship in this study. So, the first research question has been answered and the first hypothesis confirmed. In the next step, we used a confirmatory factor analysis (CFA) to assess the stability of this one-factor solution. In addition, we conducted a CFA of a two- and three-factor model based on theoretical grounds to obtain more insight on the reason why we found only one factor and to examine their correlations. These CFA’s enabled us to answer the second research question and validate the second hypothesis.

57

5.1.2. RQ1-2: Confirmatory Factor Analysis

In three confirmatory factor analyses (CFA), we formally assessed the stability of the one-, two- and three-factor configurations found in the EFA-results and in theory. These analyses allowed us to strengthen our one-factor solution (RQ1, H1) as well as to scrutinise the theoretically suggested two- or three-factor solutions and their interrelatedness (RQ2, H2). In order to determine the model fit, we used different fit indices. The first test statistic one should consider is the chi-square test. However, because of the sensitivity to sample size of the chi-square test statistic and because our sample size was considerable large (N = 3,264), this test was not sufficiently sensitive as an indicator for the model fit in this casus. We, therefore, focussed on two other types of fit indices: the root mean square error of approximation (RMSEA) as a measure of an absolute fit on the one hand; and on the other hand the Tucker-Lewis index (TLI) and Comparative fit index (CFI) to assess a comparative fit of the model. It is suggested to reject models with RMSEA values greater than 0.8 with an ideal RMSEA value less than 0.5. The CFI and TLI values greater than 0.95 generally demonstrate adequate model fit (Browne & Cudeck, 1992; Hu & Bentler, 1998).

First, we tested the one-factor solution found in the EFA by means of a CFA. Accordingly, a theoretical model was specified that included only one factor measured by the eight indicators addressing a complex citizenship conception. The initial results of the confirmatory factor analysis using a one-factor solution were not sufficiently stable: a RMSEA of 0.121, a CFI of 0.929 and a TLI of 0.901 cannot be considered acceptable. To improve the model fit, I introduced three modifications to the original model. First, I removed item Q59_8 as suggested by the standardised Residual Covariances. Second, I allowed the error terms of Q59_1 and Q59_3, and the ones of Q59_4 and Q59_7 to covary as suggested by the Modification Indices for Covariances and based on theoretical grounds. Being born

58

in Ghana is intrinsically connected to living there for most of one’s life. Speaking English and speaking a local language are also two sides of the same ‘language’-coin. Last, as the covariances of Q59_1 and Q59_3 were also considerably high, we allowed them to covary as well. The adjusted model confirmed the validity of a one factor-solution (RMSEA of .054, a CFI of .991 and a TLI of .983).

Second, since prior research clearly revealed that items belong to either civic or ethnic scales, a dichotomous factor solution was tested as well. Based on prior empirical studies (Janmaat, 2006; Reeskens & Hooghe, 2010), we specified a theoretical model that included two factors (i.e. civic versus ethnic citizenship conceptions) measured by indicators addressing a variety of citizenship attitudes. In line with Reeskens and Hooghe (2010), we ascribed speaking the national language (i.e. English) (Q59_4), respecting political in the political institutions/laws (Q59_2) and feeling Ghanaian (Q59_5) to a civic factor while we ascribed being born in Ghana (Q59_1), having lived in Ghana most of one’s life (Q59_3) and having Ghanaian ancestry (Q59_6) to the ethnic factor. Moreover speaking a local language (Q59_7) was also connected to the ethnic factor because learning a local language is difficult making it a more ascriptive, inherited aspect. Being married to a Ghanaian (Q59_8) was also linked to the ethnic factor as ethnic membership is not only conferred by birth or adoption but also by marriage (Balaton-Chrimes, 2015).

The results of the CFA using a two-factor solution proved to be insufficiently stable: a RMSEA of .119, a CFI of .935 and a TLI of .904. To improve the model fit, I introduced two modifications to the original model. First, I removed items Q59_4 and Q59_8 as suggested by the standardised Residual Covariances. Second, I allowed the error terms of Q59_1 and Q59_3 to covary as suggested by the Modification Indices for Covariances. Being born in Ghana is also intrinsically connected to living there for most of one’s life. The adjusted model confirmed the validity of a two factor-solution (RMSEA of .058, a CFI of .993 and a TLI of .984).

59

Although this two-factor model was acceptable, it was also clear that the two factors were extremely correlated to each other (.94). This implies that 94 percent of the respondents who are considering civic citizenship requirements as important also tend to adhere to an ethnic logic. So, it is preferential to speak about one comprehensive citizenship representation (i.e. complex citizenship).

Third, since more recent studies revealed that items could load on three factors (i.e. civic, ethnic and cultural citizenship), the validity of a three-factor solution was also tested. Based on prior empirical studies (Reijerse et al., 2013), we specified a theoretical model that included three factors (i.e. civic vs. ethnic vs. cultural citizenship conceptions) measured by indicators addressing a variety of citizenship attitudes. In line with Reijers et al. (2013) and Janmaat (2006), we ascribed respecting political in the political institutions/ laws, and feeling Ghanaian to a civic factor. We ascribed being born in Ghana, having lived in Ghana most of one’s life, having Ghanaian ancestry to the ethnic factor. The two language items as well as the marriage item reflect a cultural sense of national identity as these items represent traditions (Janmaat, 2006).

The results of the CFA using a three-factor solution proved to be insufficiently stable: a RMSEA of .107, a CFI of .923 and a TLI of .954. To improve the model fit, I introduced two modifications to the original model. First, I removed item Q59_8 as suggested by the standardised Residual Covariances. Note that this item has not been included in previous empirical studies and that it was also removed in the one- and two-factor solutions. Hence, we carefully conclude that this item is not appropriate to measure citizenship notions. Second, I allowed the error terms of Q59_1 and Q59_3 to covary as suggested by the Modification Indices for Covariances – just as with the one- and two-factor model. Again, this is not improbable because being born in Ghana is intrinsically connected to living there for most of one’s life. The adjusted model confirmed the validity of a three factor-solution (RMSEA of .054, a CFI of .992 and a TLI of .983).

60

Although this three-factor model was acceptable, it was also clear that the three factors were extremely correlated to each other (between .84 and .91). This implies that more than 80 percent of the respondents who consider civic citizenship requirements as important also tend to adhere to an ethnic logic and to a cultural notion. Just as concluded based on the two-factor model, it is seems preferential to speak about one comprehensive citizenship representation (i.e. complex citizenship) instead of two or three representations. The CFA results, thus, further explain the one-factor solution found by the EFA. The six discussed models are graphically displayed in Figures 5 to 10 on page 61 to 63.

In conclusion, the markers of national identity clearly clustered into only one factor: complex citizenship (RQ1, H1). When assessing the validity of two or three distinct dimensions, it was clear that these dimensions were highly correlated which indicates complementary rather than mutually excluding notions of nationhood (RQ2, H2). Consequently, we argue to reassess the crude ethnic-civic dichotomy in favour of a multi-dimensional model – at least within an African, highly educated subsample. The citizenship framework be not be conceived as a civic-ethnic continuum where moving to one end implicates moving further away from the other end. If we are at all to visualise the conceptual nature, it seems more appropriate to construct a continuum ranging from indifferent citizenship to rigid citizenship whereby citizenship includes divers criteria (Figure 4).

Figure 4: Complex citizenship.

61

χ2= 874.032; df = 20; p = 0.000 CFI = .929; TLI = .901; RMSEA = .1216

χ2= 103.528; df = 11; p = 0.000 CFI = .991; TLI = .983; RMSEA = .0546

6 Note: Confirmatory factor analysis for continuous data (N = 2901) with maximum likelihood. CFI – comparative fit index; TLI – Tucker-Lewis index; RMSEA – root mean square error of approximation.

Figure 5: CFA one-factor model – Unstable.

Figure 6: CFA one-factor model – Stable.

62

χ2= 802.916; df = 19; p = 0.000 CFI = .935;TLI = .904; RMSEA = .1196

χ2= 74.192; df = 7; p = 0.000

CFI = .993;TLI = .984; RMSEA = .0586

Figure 7: CFA two-Factor model – Unstable.

Figure 8: CFA two-factor model – Stable.

63

χ2= 576.079; df = 17; p = 0.000 CFI = .923; TLI = .954; RMSEA = .1076

χ2= 93.647; df = 10; p = 0.0006

CFI = .992; TLI = .983; RMSEA = .054

Figure 9: CFA three-factor model – Unstabe.

Figure 10: CFA three-factor model - Stable.

64

5.2. What is the impact of citizenship on political trust?

5.2.1. RQ3: Exploratory Factor Analysis

It was important to firstly assess the dimensionality of political trust before examining its determinants. To investigate the latent structure of political trust among Ghanaian university students (RQ3, H3), we conducted an EFA – namely a PCA with oblique rotation (Promax) – on the seven items of the political trust battery. Just as with the factor analysis for citizenship, the KMO and Bartlett’s test of Sphericity both confirmed that conducting a factor analysis was an appropriate technique for this study. The KMO test statistics obtained a value of .902 which is generally considered as ‘superb’ (Hutcheson & Sofroniou, 1999). Thus, we were confident that the sample size was adequate for factor analysis. Bartlett’s test was also significant (χ2 = 11 182.32, df = 21, p < .001), indicating that the correlations between the items were sufficiently large for an EFA (Field, 2009).

Table 11: KMO and Bartlett’s Test.

Kaiser-Meyer-Olkin Measure of Sampling Adequacy

.90

Bartlett’s Test of Sphericity Approx. Chi-Square 11182.32 df 21 Sig. .000 The following step was to determine the underlying dimension of political trust. The same procedure as in section 5.1.1 was used. The Eigenvalues of the factors have been plotted in the scree plot in Figure 11. In keeping with the expectations, the scree plot clearly suggested a one-factor solution because only one factor had an eigenvalue greater than 1.0 (4.38), explaining over 60 percent of the variance. All political trust items loaded higher than .76 on this single factor indicating very strong loadings; as can be seen in the EFA-results displayed in Table 12.

65

Table 12: Exploratory factor analysis of political trust items.

Political trust items Factor(s) Q40_1. Trust in parliament. .83 Q40_2. Trust in the legal system. .83 Q40_3. Trust in traditional leaders. .76 Q40_4. Trust in the police. .80 Q40_5. Trust in politicians. .75 Q40_6. Trust in the NSS administration. .79 Q40_7. Trust in the president. .78 Note: Entries represent factor loadings obtained with an exploratory factor analysis for categorical data (in SPSS 22) with Promax as rotation method and PCA as estimator. The figures in bold represent membership of citizenship items to one latent factor. All factor loadings are significant at 5% level.

Figure 11: Scree plot of Eigenvalues as a result of an EFA with Promax rotation.

66

5.2.2. RQ4: Frequency Analysis

Besides the dimensionality, the level of political trust was examined as well before scrutinising the underlying causes of trust (RQ4, H4). Table 13 lists the mean values and standard deviations of the trust that the Ghanaian students have in each of the seven institutions as well as the mean value of the overall political trust index. The overall mean for political trust was 3.09 on a 10-point scale indicating that the students do not have much trust in the Ghanaian legal-political institutions and actors. Moreover, it is worth noting that all means were below the scale midpoint of 5. Hence, students tend not to segregate their (dis)trust very much – although there was one notable differences among the actors and institutions. Ghanaian students had least confidence in the politicians (M = 1.43; SD = 1.86) and president (M = 2.64; SD = 2.63) whereas the trust the legal system (M = 4.09; SD = 2.59) was the highest. This indicates a high level of distrust among Ghanaian students – not so much related to the system as such but more to the central actors who are active within it.

Table 13. Levels of trust in diverse nation-wide, legal-political institutions and actors

n Mean SD Rank Q40_1. Trust in parliament. 3012 3.12 2.40 5 Q40_2. Trust in the legal system. 3011 4.09 2.59 1 Q40_3. Trust in traditional leaders. 3011 3.61 2.52 3 Q40_4. Trust in the police. 3009 3.13 2.47 4 Q40_5. Trust in politicians. 3011 1.43 1.86 7 Q40_6. Trust in the NSS administration. 3013 3.61 2.43 2 Q40_7. Trust in the president. 3014 2.64 2.63 6 Overall political trust 3022 3.09 1.92 Note: Entries represent average score. 0 = no trust at all. 10 = very high level of trust. Each item is preceded by the phrase: “On a scale from 0-10. how much do you personally trust each of the following institutions?”.

67

5.2.3. RQ5-9: Regression Analyses

To investigate the effect of diverse variables on political trust (RQ5-9, H5-9), we conducted a series of regression analyses. Table 14 displays the results of the five multiple linear regression analyses. The table reports unstandardised coefficients with accompanying standard error and significance level, while the standardised coefficients and test-statistics are reported in the text below. Several important findings can be observed in the results.

First, concerning the impact of the socio-demographic variables (RQ5, H5), only gender (ß = -.125), the strength of religiousness (ß = .228) and political affiliation (ß = .205; ß = .062) were significant predictors of political trust. Men, more religious students and students who adhere the NPP were significantly more likely to trust political institutions and actors compared to women, less religious students and students who vote for the NDC or CPP. We controlled for these significant variables in further analysis to be more accurate. Further, neither age (ß = .040), SES (ß = -.001), town size (ß =-.021), religion (ß = -.026) nor ethnicity were significant predictors of political trust and were therefore excluded from further analyses. Together, the eight socialisation variables accounted for 14.7% of the total variance in political trust when they are not competing with any other bloc of variables (R2 = 0.147, F(15; 1278) = 14.636, p < 0.001).

Second, concerning the impact of citizenship on political trust (RQ6, H6), people with stronger citizenship beliefs (ß = .430) were more likely to show a higher level of trust in political institutions and actors. When controlled for significant socio-demographic variables, the complex citizenship variable accounted for an additional 16.2% of the total variance in political trust (R2 = 0.162, F(7; 1373) = 90.639, p < 0.001).

Third, concerning the impact of interpersonal trust on political trust (RQ7, H7), students who have more trust in each other were also more likely to have confidence in government (ß = .129).

68

Nonetheless, the contribution of interpersonal trust was not substantial. When controlled for the significant socio-demographic variables, the interpersonal trust variable accounted for an additional 1,6% of the total variance in political trust – a miniscule level (R2 = 0.016, F(5; 1315) = 48.897, p < 0.001). The new model (including interpersonal trust) was not better in predicting the outcome variable compared to our initial model (because the F-ratio was less significant). This result clearly rejects Putnam’s cultural hypothesis that interpersonal trust ‘spills up’ to create political trust (Putnam, 1993, 1995).

Fourth, concerning the impact of perceived polity effectiveness on political trust (RQ8, H8), Ghanaian students who have more favourable evaluations of policy efficiency tended to express a higher level of political trust. Coefficients of both the indices of the overall sense of institutional effectiveness (ß = .636) and of the sense of the NSS administration’s contributions (ß = .149) were significantly positive, as predicted. When controlled for important socio-demographic variables, the combination of performance measures accounted for an additional 45.1% of the variance in political trust (R2 = 0.451, F(6; 1298) = 302.229, p < 0.001).

Last but not least, the whole model was evaluated in order to assess the relative importance of the cultural and institutional theories (RQ9, H9). The comprehensive model was able to account for 60% of the total variance in political trust (R2 = 0.600, F(8; 1235) = 231.75, p < 0.001). Some changes are noteworthy. First, the effects of the socio-demographic factors were even weaker when we accounted for other cultural and institutional variables. This was especially true for political affiliation, as it had no significant effect any more on Ghanaian students’ political trust. Second, both citizenship (ß = .124) and interpersonal trust (ß = .064) were substantially weaker predictors compared to evaluations of institutional performance (ß = .595). In other words and in accordance with previous studies (Catterberg & Moreno, 2006;

69

Mishler & Rose, 2001; Wong et al., 2011), institutional performance appears to be an extremely important determinant of political trust, even in an African country, and such performance is not limited to economic aspects.

All research questions have been answered and the hypotheses concerning the origins of political trust are all positively confirmed. Taken as a whole, male students who are more religious, tend to strongly adhere to citizenship criteria, have more trust in other people and perceive the government as effective will also have more faith in political institutions and actors. Furthermore, we can conclude that the institutional approach is undoubtedly more powerful than the cultural approach in explaining political trust in our Ghanaian casus. This is consistent with the findings of Mishler and Rose (2001) considering 10 post-communist countries in Eastern and Central Europe and the ones of Wong et al. (2011) concerning six Asian societies. National identity also remains a strong predictor of trust in political institutions and actors, which resembles Berg and Hjerm’s (2010) conclusion. Socio-demographic factors and interpersonal trust, however, cannot be regarded as important causes of political trust among Ghanaian students – leading us to the conclusion that there is no such thing as a ‘trusting personality’.

70

Table 14: Determinants of political trust – hierarchical multiple regression analyses.

Model 0 Model 1 Model 2 Model 3 Model 4 Variable B(SE) B(SE) B(SE) B(SE) B(SE) (constant) .987(.566)ns .362(.168)* .231(.293)ns .232(.132)ns -.551(.210)** Socialisation

Gender (female=1) -.539(.117)*** -.373(.099)*** -.503(.112)*** -.296(.078)*** -.255(.080)***

Age .028(.019)ns SES -.002(.047)ns Town size -.061(.080)ns

Religion (Muslim=1) -.179(.262)ns

Religiousness .202(.023)*** .084(.021)*** .217(.023)*** .070(017)*** .047(.018)**

Akan vs. Da-dangme .380(.196)ns

Akan vs. Ewe .283(.188)ns Akan vs. Guan .108(.326)ns Akan vs. Gurma .480(.302)ns

Akan vs. Mole Dagbani .318(.245)ns

Akan vs. Grusi -.224(.449)ns Akan vs. Mande -.165(.579)ns

70

71

NPP vs. NDC .910(.143)*** .978(.100)*** 1.043(.112)*** .070(.085)ns .102(.086)ns NPP vs. CPP .675(.299)*** .673(.253)*** .782(.286)** .224(.198)ns .257(.200)ns Complex citizenship .342(.019)*** Interpersonal trust .183(.036)*** .090(.026)*** Institutional effectiveness

Institutional performance .602(.022)*** .562(.023)***

NSS contribution .132(.019)*** .104(.020)*** R2 Change 14.7% 16.2% 1.6% 45.1% 46.3% R2 14.7% 29.7% 15.7% 58.3% 60.0% Note. ns non-significant; * p < .05; ** p < .01; *** p < .001. Source: National Service System Survey, Ghana, 2013 (N = 3,264).

71

72

73

6. Discussion

In this part, the findings are discussed in relation to the theoretical background, research questions and hypotheses. Additionally, attention is paid to the limitations of this study as well as to possible avenues for future research.

6.1. Making a nation, creating a stranger: answering the research questions and hypotheses

This quantitative, cross-sectional study had a two-folded research objective. Firstly, we uncovered the dimensionality of citizenship and assessed the relation between the different dimensions. Secondly, the impact of citizenship on political trust was examined. We considered it important to explore the effect of divers variables in order to fully capture the contribution of citizenship to political trust. To achieve those two research goals, nine research questions with as much corresponding hypotheses were formulated based on the literature review. We opted to focus our research on an African country (i.e. Ghana) because of the noticeable lack of research on citizenship and political trust in Africa. Subsequently, this study relied on the public opinion data (N = 3,264) gathered in Ghanaian universities by the N3S researchers for answering the questions. The purpose of the longitudinal N3S study is to examine the NSS program, inter-group relations and national identities (Langer et al., 2014). It also contains valuable primary data on citizenship and political trust. In the following two parts, we revise to what extent our findings correspond with the nine questions and hypotheses that have guided our research.

74

6.1.1. Citizenship representations in Ghana

Who is a foreigner and who an autochthone according to the man-in-the-street? This question is of utmost importance for many African countries struggling to foster a sense of national identity that overrides other ethnic, religious, social or political affiliations. The concept of citizenship helps to answer this question. The civic/ethnic dichotomy has dominated the research on citizenship for over a century whereby civic citizenship is more open based on respecting laws and institutions, whereas ethnic citizenship is rather closed based on ancestry. Recently, a cultural factor has been introduced with an emphasis on cultural aspects including the national language and religion. Most of the literature assumes a priori a two- or a three-factor model in which respondents are classified as either adhering to a civic or ethnic citizenship conception (Reeskens & Hooghe, 2010) or to a civic, ethnic or cultural citizenship representation (Reijerse et al., 2013). The literature on national identity, however, suggest that people can incorporate both civic and ethnic items of citizenship creating a ‘dual identity’ (Moreno & Arriba, 1996).

In addition to the fact that the citizenship and national identity literature has not reached a consensus yet, it is also notable that most of those studies had been conducted on western policies and people. Despite the substantial social and political consequences of citizenship ideas and the pervasiveness of ethnic notions within the political, economic and social scene in Africa, little empirical research has been conducted on citizenship representations in Africa. Therefore, the first main purpose of this article was to examine the underlying dimensionality of citizenship representations based on public opinion data gathered from Ghanaian university students. Our study has produced some interesting results that clearly contradicted the findings of previous studies on citizenship. In general, we argue that indicators of citizenship are numerous and diverse, and that they differ from country to country in structure and interrelatedness.

75

First and foremost, we conducted an EFA on the eight items of the citizenship battery to answer our first research question about the structure of citizenship in Ghana. The results revealed a clear one-factor solution instead of a two- or three-factor solution. Thus, our first hypothesis predicting such a one-factor solution was confirmed. This result was more in line with the literature on national identity and with the concept of an ‘dual identity’ than with the literature on citizenship. Moreover, it revealed that Ghanaian university students attach importance to all or no citizenship criteria at the same time. The descriptives, however, suggested that the average participant in our sample considered citizenship criteria as neither unimportant nor important to be regarded as a true Ghanaian citizen – although previous studies indicate that students’ opinion on average deviates from the midpoint of the Likert scales.

Second, to examine the second research questions about the interrelatedness of two or three factors in more detail, we assessed the validity of a one-, two- and three-factor solution in a CFA based on both EFA-results and theoretical grounds. In keeping with our second hypothesis, it was clear that the two or three factors (i.e. civic, ethnic and cultural citizenship) were correlated for more than 84 percent. This is a fundamental finding. Many authors assume civic and ethnic visions of citizenship as mutually excluding concepts. Although the civic/ethnic dichotomy could be a useful tool for academics, politicians and journalists, it evidently does not apply for Ghanaian popular notions of citizenship.

Consequently, we conclude that we cannot find two or three competitive citizenship representations within our substantive Ghanaian student sample. On the contrary, Ghanaian adolescents experience a convergence of both civic and ethnic citizenship conceptions – rather than some students adhering the civic and others to the ethnic form. We therefore argue to reassess the civic-versus-ethnic dichotomy or the civic/ethnic/cultural classification and opt for a more proportional, multi-dimensional model mixing civic, ethnic

76

and cultural practices. Another alternative is to collapse the civic/ethnic scale into a unipolar scale measuring the intensity of attachment to the national in-group as such.

This stark difference could be explained by a lack of cross-nation equivalence of the citizenship measurement. In other words, the same concept (i.e. citizenship) is measured in the different ways across various subgroups of respondents and, thus, the measured attributes cannot be compared easily across different populations. Both item and construct bias could have affected the non-equivalence between the western measurements and ours (Davidov, Meuleman, Cieciuch, Schmidt, & Billiet, 2014). First, the items may measure different conceptions in different countries. Based on a cross-national scalar equivalence test, Reeskens and Hooghe (2010) also concluded that scores on the civic or ethnic citizenship representation could not be compared easily across the various countries they investigated. In a similar vein, Svallfors (1996) indicated that attitudes are context- and culture-dependent, which means that certain questions will be understood differently in different countries. There is, therefore, a risk that the questions may become trivial to have universal applicability. The framing or wording of the questions could also be interpreted differently from country to country (Hjerm, 1998; Svallfors, 1996). Second, and more problematic, is the fact that the concept (i.e. citizenship) itself could have a completely different meaning across groups. Such ‘emic concepts’ have meanings that are strongly nation-or culture-specific (Davidov et al., 2014). In this respect, it is this arguable that the African context deviates substantially from the normally investigated western contexts. The fact that Ghanaian students seem indifferent about citizenship requirements could suggest that the questions asked were not appropriate for an African country with its specific culture and habits. Therefore, we urgently ask for more culture-specific measurements of citizenship in general and specifically for more Africa-specific measurements.

77

Another explanation for the one-factor solution can be found in the logic that Hjerm (1998) is pursuing. He suggests that – depending on the political community – the civic identity may subsume the ethnic one. In his 1998 article, he introduced the ‘multiple national identity’ type of citizenship in which the feeling of the national identity is based on both civic and ethnic components at the same time. His ‘pluralist’ type, in contrast, is characterised by a low identification with both the civic and ethnic criteria of citizenship (Hjerm, 1998). It seems that our Ghanaian students also pursue a multiple versus pluralist dichotomy instead of a civic versus ethnic dichotomy (see Figure 4 on page 72).

In a similar vein, Hjerm also offers a third explanation for both the one-factor solution. He assumes that it is ‘primarily citizenship and immigration policy regimes that will affect the constitution of a national identity’ (Hjerm, 1998, p. 452). It is possible that citizenship ideas change over time due to policy efforts. This is in keeping with the ideas of Wimmer and Kuzio. Wimmer (2008) evaluated different forms of ethnicity and argued that boundary-making is a ‘process of constituting and re-configuring groups by defining the boundaries between them’. In other words, one should not regard citizenship as fixed but rather as continually evolving (Wimmer, 2008, p. 1027 - original emphasis). Likewise, Kuzio (2002) outlined that nationality is a ‘process of change that incorporates tensions between civic universalism and ethnic pluralism’. Hence, he concluded that ‘pure civic or ethnic states only exist in theory’ because each nation contains elements of both types of nationalism (Kuzio, 2002).

After their independence, Ghanaian politicians have underscored a pluralistic nationality guaranteeing universal rights to all citizens, including people of foreign origin, based on birth (Kobo, 2010). Furthermore, the political elite has aimed – in both formal and informal ways – at reducing ethnic electoral politics and ensuring that political parties are national in character. They also tried to maintain the ethno-regional balances in political sphere as well as the cultural

78

and religious ‘neutrality’ of the state. At last, the symbolic impact of the continuous efforts to reduce the developmental gap between ethnic groups was also influential in maintaining national integration and reducing ethnic identities (Langer, 2007). It is therefore possible that the civic logic has gradually subsumed the ethnic one as a result of decades of policy efforts. In other words, the Ghanaian population could have incorporated more civic elements into their rather ethnic citizenship conception over time. Thus, ethnic markers appear less much salience in the social sphere. As a result, it is plausible that older people still endorse a more ethnic citizenship notion, but the N3S data did not allow us to test this assumption

To wrap up, we departed from the theoretical ethnic, civic and cultural conceptualisations of citizenship that are identified and developed in North American en Western Europe. The same western literature tends to offer some empirical evidence of this two- or tree-folded vision. This literature tradition, however, tends to neglect the ideas on national identity claiming that people can incorporate both civic and ethnic ideas into a dual identity. It was clear from our quantitative research that the Ghanaian context differs from the western one. We failed to find empirically evidence for the two- or tree-folded vision on citizenship represented in previous studies. Ghanaian students seem to adhere to or to neglect diverse citizenship criteria at the same time. On average, however, they seem quite indifferent about the citizenship criteria contrasting the strong civic citizenship attitudes found in previous, western, research. This conclusion urgently asks for future research on citizenship in other African countries. It would be extremely interesting to find out whether or not public opinion data derived from other African populations do represent a two- or tree-folded citizenship representation. If researchers also fail to find civic, ethnic and/or cultural representations in other African countries, serious questions can be raised about the validity of the citizenship dichotomy within the African context.

79

6.1.2. Cultural and institutional theories of political trust

The logical question that follows from our first research objective is: citizenship, what does it matter? Do ideas concerning citizenship impact people’s level of political trust? Trust in political institutions and actors is crucial to democracy. A certain degree of political trust is a prerequisite for effective, legitimate and efficient governance. In numerous post-colonial, African countries, however, scepticism and distrust in the institutions and political actors are widespread. Cultural and institutions theories offer competing explanations for the formation of political trust and, thus, suggest different ways to develop more trust. Cultural theories offer affective explanations grounded in early-life experiences and socialisation while the institutional approach emphasises experiences with the government on a later stage. Accordingly, both theories are not mutually exclusive and can reinforce each other in developing more political trust. However, empirical, individual-level information on the power of each perspective in explaining popular trust in political institutions and actors within an African context remains largely absent.

As a result, the second aim of this research was to develop a comprehensive model of the determinants of political trust, including citizenship, based on the unique N3S data. Before we could develop such a model of political trust, we examined the dimensionality of our dependent variable. In line with the third hypothesis, the public opinion revealed a one-dimensional, all-embracing trust in political institutions and actors – although the theory tends to suggest multi-dimensional definitions of political trust. Furthermore, the fourth research question explored the levels of political trust in Ghana as previous research has pointed Ghana as one of the least-trusting countries (Delhey et al., 2011). Ghanaian university students are the political leaders of tomorrow but it appeared that they have lost faith in politics as shown by the extremely low levels of political trust. Although variations in trust among the different institutions and

80

actors were relatively small, it seemed that Ghanaian students were mostly disappointed in their politicians and president while trust was highest in the legal system. This adds to the relevance of this study; by establishing a model of political trust, we can assist Ghanaian policymakers with cultivating more legitimacy within their society. This study also contributes to the academic world by generating original insights from different theoretical and empirical research traditions and by laying the foundations of a model of trust that can be used in future cross-national, comparative research in Africa.

Before integrating the variables into a comprehensive model of political trust, the individual strength of each of the predictors was assessed. More specifically, we examined the possible effects on political trust of various cultural (i.e. socio-demographics, national identity and interpersonal trust) and institutional factors (i.e. sense of institutional effectiveness and the NSS contribution), with a view of finding out which approach is more powerful in explaining political trust. Our results showed that trust in political institutions and actors was both exogenous and endogenous, determined by both social characteristics and evaluations of policy effectiveness. However, evidence on the cultural explanations was mixed. While citizenship did have a significant and substantial impact on political trust, the contributions of socio-demographics and interpersonal trust were negligible or insignificant. By contrast and in keeping with our eighth hypothesis, the individual evaluations of policy performances and the NSS contribution were significant and substantial predictors of political trust.

When examining the relative importance of all variables in one comprehensive model, it was immediately clear that individual evaluations of political and economic performances were substantially more important compared to either socialisation experiences, the strength of the national identity or interpersonal trust. The effects of the socialisation variables and interpersonal trust were especially weak. This supported our conclusion that there is no

81

such thing as a ‘trusting personality’. When all variables were included in the model, only two of the eight variables were statically significant – despite the sample size of over 3,000 Ghanaian students. Male students exhibited slightly but significant higher levels of trust in institutions than their female counterparts and more religious students were more trusting as well. These findings are at odds with previous studies claiming that women have somewhat higher political trust (Berg & Hjerm, 2010; Christensen & Lægreid, 2005; Schoon & Cheng, 2011). At first glance, students supporting the NPP also tend to support the political system more but this effect disappeared when all variables were included. Although education, age and town size are often acknowledges as predictors of political trust in established countries, they were not significant within our Ghanaian university context and neither was socio-economic status, religion or ethnicity. These non-significant results are contradicting the significant results of those variables in previous studies (Berg & Hjerm, 2010; Catterberg & Moreno, 2006; Christensen & Lægreid, 2005; Mishler & Rose, 2001). This could suggest that the Ghanaian context is, indeed, not comparable to western (Christensen & Lægreid, 2005) or post-communist societies (Mishler & Rose, 2001).

In sum, we departed from the literature on citizenship and political trust claiming that a stronger national identity leads to more political trust. In order to fully explore the impact of citizenship ideas on political trust, we incorporated the effects of divers cultural and institutional variables into our research. In line with previous studies (Mishler & Rose, 2001; Wong et al., 2011), we can conclude that the institutional approach appears to be much more powerful than the cultural approach in explaining political trust among Ghanaian students. Short-term specific experiences with good governance seem more important than long-term general identities. This finding should be seen in the context of the Ghanaian public sector, which has a relatively low level of performance and it affirms the importance of policy performance in strengthening political trust in societies.

82

Consequently, governments should always respond punctually and effectively to public priorities in order to raise trust. By combatting poverty, creating employment opportunities, providing education, supporting poor and deprived people, politicians could gain more confidence from their constituents. Thus, expanding the economic and administrative apparatus of Ghana and promoting government involvement in the daily lives of the population can help politicians in culminating political development. For instance, the regulation of primary commodity exports and taxes together with the strengthening of the accompanying administrative structure could positively contribute to state-society relations. Nonetheless, one cannot easily conclude that cultural factors are irrelevant. Especially promoting a national identity and national unity seems important to raise political trust. However, the other side of the coin is that a stronger national identity or a stronger in-group identification also tends to foster out-group derogation (Meeus, Duriez, Vanbeselaere, & Boen, 2010). Future research can build on these results to extend our model of political trust and to test it in other, more representative, African settings. In the last two paragraphs, we further outline the limitations we faced in this study and the avenues for future studies.

6.2. Limitations of the study and subsequent avenues for future research

Obviously, this study had some noteworthy limitations. Our cross-sectional country study based on the opinion of university students contained at least five, mainly methodological, limitations. First, this cross-country study was focussed solely on Ghana, which makes generalisations to the larger African public difficult. However, one should also not forget that the determinants of political trust are extremely context-specific and can only be fully unfolded and understood in relation to the specific institutions and contexts of a

83

particular society. In other words, social science research always needs case and country-specific studies in order to achieve a full understanding of political trust in particular and of socio-political phenomena in general (Wong et al., 2011).

This country-specific study suggested first that indicators of citizenship differ from country to country in structure and interrelatedness. Therefore, we urgently ask for more culture-specific measurements of citizenship in general and specifically for more Africa-specific measurements. It would be extremely interesting to supplement our quantitative study with more qualitative ones. Because we failed to find two or three statistically valid factors and because our Ghanaian students seemed rather indifferent about citizenship, it could be more useful to ask open-ended descriptions of their citizenship attitudes or discuss citizenship through in-depth interviews. By looking for patterns in those descriptions, alternative ways of representing citizenship in the African context could be revealed. In this way, future studies can greatly contribute to the citizenship literature by constructing a continent- or culture-sensitive measurement scale.

Second and in the same vein, this study was based on a non-representative sample, making generalisation to the whole Ghanaian population – let alone to the African continent – difficult. Our young, chiefly male participants formed a particular subgroup of the population in terms of educational level, literacy, and economic and financial resources. This constraint could have distorted the effect of the socio-demographic variables on political trust. For instance, we have hypothesised that older people will still endorse the ethnic criteria more strongly. Future research should compare students’ citizenship notions with the ones of older people to test this assumption. Nevertheless, we argue that this limitation can be abated by the fact that these students are considered the political and societal leaders of tomorrow. Hence, these students are more likely to impact and shape citizenship perceptions of others such as their peers and

84

family. It is also especially important to raise political trust among this subsample because adolescence is a critical age for the development of attitudinal dispositions and because the students will be less likely to engage in politics later on if they lack political trust.

Third, the cross-sectional nature of this study makes firm causal inferences impossible. Clearly, scholars need to conduct more comparative, representative and longitudinal studies beyond the established democracies in order to examine causal mechanisms of cultural and institutional factors on political trust in Africa. Currently, the N3S researchers are collecting longitudinal data on the NSS program, national identities and cohesion, and inter-group relations in Ghana. These additional data will enable us to make more firm claims about citizenship and political trust in the future. Longitudinal research can also detect evolvements in the Ghanaian public opinion vis-à-vis citizenship criteria or political trust.

Fourth, as we did not develop the survey ourselves and because the focus of the N3S scholars was more on inter-group relations and national cohesion, some important questions and variables were lacking. We should, for instance, acknowledge that the questionnaire did not contain enough questions to investigate a three-factor solution properly. We had only two items on the civic representation, while three items are generally considered as the bare minimum for a factor analysis (Claes, 2010). Having legal citizenship and a religion-item were criteria that are often included in research but were absent in this survey.

Last but not least, we opted to focus solely on the structure of citizenship and its impact on only one variable (i.e. political trust). Citizenship notions are, however, important for a wide range of other attitudinal and behavioural responses. In addition to other outcomes, it is also extremely interesting to detect the social, psychological and contextual factors influencing the development of someone’s citizenship notions.

85

7. Conclusion

Who is a foreigner and who an autochthone according to the ordinary man? And what is the impact of such national identity ideas on the development of political trust? These two questions are of utmost importance for many African countries struggling to foster a sense of national identity that overrides other ethnic, religious, social or political affiliations. The concept of citizenship helps to answer both questions. Citizenship revolves around membership and belonging to a group; and conventional wisdom assumes that it can be divided into two conceptualisations (i.e. civic and ethnic citizenship). Both forms of citizenship have in common that they formulate beliefs about the organisation of a society and about who can be regarded as a national in-group member. Civic citizenship traditionally stems from the territoriality principle (ius soli) and can, therefore, be regarded as open based on political loyalty; whereas ethnic citizenship is derived from the law of blood (ius sanguinis) and is, thus, considered as rather closed based on ancestry. More recently, a third notion of citizenship is added to the debate – namely a cultural citizenship representation, based on adhering to and preserving the nation’s culture. However, no consensus has been reached yet on the empirical validity of this dichotomy or three-folded classification.

Regardless of the precise conceptualisation of citizenship, it is traditionally assumed that some form of national identity is required to foster political trust. Democracy requires such trust. Popular distrust in political institutions and actors impairs the state’s legitimacy, sustainability and ability to intervene adequately. Cultural theorists take it for granted that a firm national identity increases political trust. Other theoretical approaches – such as the institutional performance explanations – are, however, equally plausible. While some studies have empirically tested particular relations with political trust, most of this work fails to integrate a diverse range of underlying causes or is concentrated on macro-level data.

86

The main lacuna in research on citizenship and political trust is the absent of African-based studies. This void is striking considering both the continued prominence of ethnic and national ideas within many Africa countries and the importance of developing political trust for the consolidation of these recently established, post-colonial democracies. This thesis was one of the first to fill this vacuum by examining the structure and impact of citizenship in Ghana. Specifically, we opted to focus on its impact on political trust while controlling for other important variables. Using the 2013 National Service System Survey in Ghana and employing both factor analyses and linear regression techniques; we explored the structure of citizenship and of political trust. After assessing the level of political trust in Ghana, the independent effects of diverse variables were determined. Finally, a comprehensive model of the determinants of political trust was build. This quantitative survey-analysis yielded some interesting findings.

First, the present study adds to the current literature by demonstrating that Ghanaian university students (N = 3,264) experience a convergence of both civic and ethnic citizenship conceptions, rather than some students adhering the civic form while others endorsing the ethnic notion. We, therefore, advocate for an overarching notion of citizenship incorporating various criteria instead of the civic/ethnic dichotomy. Because of the sharp difference with the western-based studies suggesting two or three competitive dimensions, we urgently ask for further research on citizenship in African countries and for country-specific citizenship measurements.

Second, we developed a multivariate model of political trust. The results indicated that national identity indeed contributes to trust in political institutions and actors. The government’s effectiveness in providing public goods, however, was seen as even more important for building political trust; according to the Ghanaian students. Next, trust in government was barely influenced by socio-demographic factors or interpersonal trust. Hence, we concluded that there exists

87

no such thing as ‘trusting personality’ in our casus. These results support the superiority of institutional performance theories while providing limited support for cultural explanations. This asks for future studies on trust-building in developing countries trying to establish working institutions.

In conclusion, the more importance Ghanaian students attach to the numerous criteria for considering someone as a true Ghanaian, the more trust they will have in their nation-wide, legal-political institutions and actors. Thus, by fostering a stronger Ghanaian identity, the government can also cultivate more political trust. This is, however, not the main strategy the government can apply. The perceived effectiveness of a wide range of policy implementations is the most significant contributor to political trust. Politicians trying to foster more trust with their society can use our model of political trust. The model is also useful for other scholars as they can expand and improve it. Although the nature of this cross-sectional study did not allow for wide generalisations; it forms, nonetheless, an interesting point of departure of future research.

88

89

References

Atuahene, F., & Owusu-Ansah, a. (2013). A Descriptive Assessment of Higher Education Access, Participation, Equity, and Disparity in Ghana. SAGE Open, 3(3), 1–16. doi:10.1177/21582 44013497725

Balaton-Chrimes, S. (2015). Ethnicity, Democracy and Citizenship in Africa: Political Marginalisation of Kenya’s Nubians. Farnham: Ashgate Publishing, Ltd.

Bandyopadhyay, S., & Green, E. (2013). Nation-Building and Conflict in Modern Africa. World Development, 45(), 108–118. doi:10.1016/j.worlddev.2012.09.012

Bates, R. H. (2012). Ethnicity. In K. G. Brown & A. Langer (Eds.), Elgar Handbook Of Civil War And Fragile States (pp. 16–20). Cheltenham: Edward Elgar.

Bereketeab, R. (2011). The Ethnic and Civic Foundations of Citizenship and Identity in the Horn of Africa. Studies in Ethnicity and Nationalism, 11(1), 63–81.

Berg, L., & Hjerm, M. (2010). National Identity and Political Trust. Perspectives on European Politics and Society, 11(4), 390–407. doi:10.1080/15705854.2010.524403

Berman, B. J. (2010). Ethnicity and Democracy in Africa (No. 22). Ethnic Diversity and Economic Instability in Africa (Vol. 22). doi:10.1353/arw.2006.0064

Brown, T. A. (2006). Confirmatory Factor Analysis for Applied Research. New York, NY: Guilford Press.

Browne, M. W., & Cudeck, R. (1992). Alternative Ways of Assessing Model Fit. Sociological Methods & Research, 21(2), 230–258. doi:10.1177/0049124192020004001

Brubaker, W. R. (1990). Citizenship and Nationhood in France and Germany. Columbia University.

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

90

Brubaker, W. R. (2004). Ethnicity Without Groups. Cambridge, MA: Harvard University Press.

Bryman, A. (2012). Social Research Methods. Oxford University Press.

Campbell, B. D. E. (2006). What is education’ s impact on civic and social engagement!? In R. Desjardins & T. Schuller (Eds.), Measuring the Effects of Education on Healt and Civic Engagement (pp. 25–126). Paris: Centre for Educational Research and Innovation, Organisation for Economic Cooperation and Development.

Cattell, R. (1966). The Scree Test for the Number of Factors. Multivariate Behavioral Research, 1, 245–276.

Catterberg, G., & Moreno, A. (2006). The Individual Bases of Political Trust: Trends in New and Established Democracies. International Journal of Public Opinion Research, 18(1), 31–48. doi:10.1093/ijpor/edh081

Cederman, L.-E. (2001). Nationalism and bounded integration: what it would take to construct a European demos. European Journal of International Relations, 7(2), 139–174.

Christensen, T., & Lægreid, P. (2005). Trust in Government: The Relative Importance of Service Satisfaction, Political Factors, and Demography. Public Performance & Management Review, 28(4), 487–511.

Citrin, J. (1974). Comment: The Political Relevance of Trust in Government. American Political Science Review, 68(3), 973–988. doi:10.2307/1959141

Claes, E. (2010). Schools and Citizenship Education. A Comparative Investigation of Socialization Effects of Citizenship Education on Adolescents. status: published. K.U.Leuven.

Connor, W. (1978). A nation is a nation, is a state, is an ethnic group is a... Ethnic and Racial Studies, 1(4), 379–388.

Davidov, E., Meuleman, B., Cieciuch, J., Schmidt, P., & Billiet, J. (2014). Measurement Equivalence in Cross-National Research.

91

Annual Review of Sociology, 40(1), 55–75. doi:10.1146/annurev -soc-071913-043137

Delhey, J., Newton, K., & Welzel, C. (2011). How General Is Trust in ‘Most People’? Solving the Radius of Trust Problem. American Sociological Review, 76(5), 786–807. doi:10.1177/0003122411420817

Döring, H. (1992). Higher education and confidence in institutions. West European Politics, 15, 126–146.

Easton, D. (1957). An Approach to the Analysis of Political Systems. World Politics, 9(03), 383–400. doi:10.2307/2008920

Easton, D. (1965). A Framework for Political Analysis. Englewood Cliffs, NJ: Prentice-Hall.

Easton, D. (1975). A Re-assessment of the Concept of Political Support. British Journal of Political Science, 5(04), 435–457. doi:10.1017/S0007123400008309

Field, A. (2009). Discovering Statistics using SPSS (3th ed.). London: Sage Publications.

Galston, W. a. (2001). Political Knowledge, Political Engagement and Civic Education. Annual Review of Political Science, 4, 217–234.

Gamson, W. A. (1968). Power and Discontent. Homewood, IL: Dorsey Press.

Ghanaian Statistical Service. (May 2013). 2010 Population & Housing Census. Accra.

Herbst, J. (1989). The creation and matintenance of national boundaries in Africa. International Organization, 43(4), 673–692. doi:10.1017/S0020818300034482

Hetherington, M. J. (1998). The Political Relevance of Political Trust. American Political Science Review, 92(4), 791–808.

Hjerm, M. (1998). National identity: A comparison of Sweden, Germany and Australia. Journal of Ethnic and Migration Studies, 24(3), 451–469. doi:10.1080/1369183X.1998.9976644

92

Hu, L., & Bentler, P. M. (1998). Fit indices in covariance structure modeling: Sensitivity to underparameterized model misspecification. Psychological Methods, 3(4), 424–453. doi:10.1037/1082-989X.3.4.424

Huseby, B. M. (2000). Government Performance and Political Support. A Study of How Evaluations of Economic Performance, Social Policy and Environmental Protection Influence the Popular Assessments of the Political System. Trondheim Norges teknisk-naturvitenskaplige universitet.

Hutcheson, G. D., & Sofroniou, N. (1999). The multivariate social scientist!: introductory statistics using generalized linear models - KU Leuven. London: London Sage.

Hutchison, M. L., & Johnson, K. (2011). Capacity to trust? Institutional capacity, conflict, and political trust in Africa, 2000-2005. Journal of Peace Research, 48(6), 737–752. doi:10.1177/0022343311417981

Janmaat, J. G. (2006). Popular conceptions of nationhood in old and new European member states: Partial support for the ethnic-civic framework. Ethnic and Racial Studies, 29(1), 50–78. doi:10.1080/01419870500352363

Kaase, M. (1999). Interpersonal trust, political trust and non‐institutionalised political participation in Western Europe. West European Politics, 22(3), 1–22.

Kaplan, R. D. (1994). The Coming Anarchy. Atlantic Monthly, 273(February 1994), 44–76. doi:10.1080/14662040408565567

King, G., Murray, C. J. L., Salomon, J. A., & Tandon, A. (2003). Enhancing the Validity and Cruss-Cultural Comparability of Measurement in Survey Research. The American Political Science Review, 97(4), 567–583.

Kobo, O. (2010). ‘We are citizens too’: the politics of citizenship in independent Ghana. The Journal of Modern African Studies, 48(01), 67–94. doi:10.1017/S0022278X0999022X

93

Kohn, H. (1945). The Idea Of Nationalism: A Study In Its Origins And Background. New York, NY: Macmillan.

Kuzio, T. (2002). The myth of the civic state: a critical survey of Hans Kohn’s framework for understanding nationalism. Ethnic and Racial Studies, 25(1), 20–39. doi:10.1080/01419870120112 049

Kymlicka, W. (1998). Finding Our Way: Rethinking Ethnocultural Relations in Canada. Oxford: Oxford University Press.

Kymlicka, W. (1999). Mislabelling Nationalism. In R. Beiner (Ed.), Theorizing Nationalism (p. 67). Albany: State University of New York Press.

Kymlicka, W. (2001). Politics in the vernacular: Nationalism, multiculturalism, and citizenship. Oxford: Oxford University Press.

Lalonde, R. N. (2002). Testing the social identity-intergroup differentiation hypothesis: ‘We’re not American eh!’. The British Journal of Social Psychology / the British Psychological Society, 41(4), 611–630. doi:10.1348/014466602321149902

Langer, A. (2005). Horizontal Inequalities and Violent Group Mobilization in Côte d’Ivoire. Oxford Development Studies, 33(1), 25–45. doi:10.1080/13600810500099634

Langer, A. (2007). The Peaceful Management of Horizontal Inequalities in Ghana. CRISE. Oxford.

Langer, A., Meuleman, B., Oshodi, T., & Schroyens, M. (2014). Can student populations in developing countries be reached by online surveys!? The case of the National Service Scheme Survey. Leuven.

Langer, A., & Ukiwo, U. (2007). Ethnicity , Religion and the State in Ghana and Nigeria!: Perceptions from the Street.

Langer, A., & Ukiwo, U. (2009). Subjective Realities: Perceptions of Identity and Conflict in Ghana and Nigeria. Journal of International Development, 21, 483–494. doi:10.1002/jid.1570

94

Liebich, A. (2006). Searching for the perfect nation: the itinerary of Hans Kohn (1891-1971). Nations and Nationalism, 12(4), 579–596. doi:10.1111/j.1469-8129.2006.00259.x

Löden, H. (2014). Citizenship education, national identity and political trust: The case of Sweden. Nordidactica, 2, 116–136.

Marshall-Fratani, R. (2006). The War of ‘Who Is Who’: Autochthony, Nationalism, and Citizenship in the Ivoirian Crisis. African Studies Review, 49(2), 9–43. doi:10.1353/arw.2006.0098

Meeus, J., Duriez, B., Vanbeselaere, N., & Boen, F. (2010). The role of national identity representation in the relation between in-group identification and out-group derogation: ethnic versus civic representation. British Journal of Social Psychology, 49(2), 305–320. doi:10.1348/014466609X451455

Meinecke, F. (1907). Weltbürgertum und Nationalstaat; Studien zur Genesis des deutschen Nationalstaates. Princeton, NJ: Princeton University Press.

Miles, W. F. S., & Rochefort, D. A. (1991). Nationalism Versus Ethnic Identity in Sub-Saharan Africa. The American Political Science Review, 85(2), 393–403.

Mill, J. S. (1861). Considerations on Representative Government. London: Parker, Son, and Bourn.

Mill, J. S. (1975). Three Essays. Oxford: Oxford University Press. Miller, A. H. (1974). Political Issues and Trust in Government:

1964–1970. American Political Science Review, 68(03), 951–972. doi:10.2307/1959140

Mishler, W., & Rose, R. (1997). Trust, Distrust and Skepticism: Popular Evaluations of Civil and Political Institutions in Post-Communist Societies. The Journal of Politics, 59(02), 418–451. doi:10.2307/2998171

Mishler, W., & Rose, R. (2001). What are the origins of political trust? Testing Institional and Cultural Theories in Post-

95

Communist Societies. Comparative Political Studies, 34(1), 30–62.

Moreno, L. (2006). Scotland, Catalonia, Europeanization and the ‘Moreno Question’. Scottish Affairs, (54), 1–21.

Moreno, L., & Arriba, A. (1996). Dual idenitity in autonomous Catalonia. Scottish Affairs, (17), 22–41.

Moreno, L., Arriba, A., & Serrano, A. (1998). Multiple identities in decentralized Spain: The case of Catalonia. Regional & Federal Studies, 8, 65–88. doi:10.1080/13597569808421060

Ndegwa, S. N. (1997). Citizenship and Ethnicity: An Examination of Two Transition Moments in Kenyan Politics. The American Political Science Review, 91(3), 599–616.

Newton, K. (2001). Trust, Social Capital, Civil Society, and Democracy. International Political Science Review, 22(2), 201–214.

Newton, K., & Norris, P. (2000). Confidence in public institutions: Faith, Culture or Performance? In S. Pharr & R. D. Putnam (Eds.), Disaffected democracies. What’s troubling the Trilateral Countries? (pp. 1–33). Princeton, NJ: Princeton University Press.

Nieguth, T. (1999). Beyond Dichotomy: Concepts of the Nation and the Distribution of Membership. Nations and Nationalism, 5(2), 155–173. doi:10.1111/j.1354-5078.1999.00155.x

Nielsen, K. (1999). Cultural Nationalism, Neither Ethnic nor Civic. In R. Beiner (Ed.), Theorizing Nationalism (pp. 67–90). Albany: State University of New York Press.

Norris, P. (1999). Critical Citizens!: Global Support for Democratic Government. Oxford: Oxford University Press.

Putnam, R. D. (1995). Bowling Alone: America’s Declining Social Capital. Journal of Democracy, 6, 65–78. doi:10.1016/S0362-3319(02)00190-8

Putnam, R. D. (2000). Bowling Alone: The Collapse and Revival of American Community. New York, NY: Simon & Schuster.

96

Putnam, R. D., Leonardi, R., & Nanetti, R. Y. (1993). Making Democracy Work: Civic Traditions in Modern Italy. Princeton, NJ: Princeton University Press.

Reeskens, T., & Hooghe, M. (2010). Beyond the civic-ethnic dichotomy: Investigating the structure of citizenship concepts across thirty-three countries. Nations and Nationalism, 16(4), 579–597.

Reijerse, A. (2012). What does it take for ‘them’ to become part of ‘us’? Acceptance of immigrants in EU countries. University of Leuven.

Reijerse, A., Acker, K. Van, Vanbeselaere, N., Phalet, K., & Duriez, B. (2013). Beyond the Ethnic‐Civic Dichotomy: Cultural Citizenship as a New Way of Excluding Immigrants. Political Psychology, 34(4), 611–630. doi:10.1111/j.1467-9221.2012. 00920.x

Sapiro, V. (2004). Not Your Parents’ Political Socialization: Introduction for a New Generation. Annual Review of Political Science, 7(1), 1–23. doi:10.1146/annurev.polisci.7.012003.1 04840

Schoon, I., & Cheng, H. (2011). Determinants of political trust: a lifetime learning model. Developmental Psychology, 47(3), 619–631. doi:10.1037/a0021817

Shulman, S. (2002). Challenging the Civic/Ethnic and West/East Dichotomies in the Study of Nationalism. Comparative Political Studies, 35(5), 554–585. doi:10.1177/001041400203 5005003

Shulman, S. (2004). The contours of civic and ethnic national identification in Ukraine. Europe-Asia Studies, 56(1), 35–56. doi:10.1080/0966813032000161437

Smith, A. D. (2010). Nationalism: Theory, Ideology, History (2nd ed.). Cambridge, MA: Cambridge Polity.

97

Svallfors, S. (1996). National differences in national identities? An introduction to the International Social Survey Programme. New Community, 22(1), 127–134.

Williams, J. T. (1985). Systemic Influences on Political Trust: The Importance of Perceived Institutional Performance. Political Methodology, 11(1/2), 125–142.

Wimmer, A. (2008). Elementary strategies of ethnic boundary making. Ethnic and Racial Studies, 31(6), 1025–1055. doi:10.1080/01419870801905612

Wong, T. K., Wan, P.-S., & Hsiao, H.-H. M. (2011). The bases of political trust in six Asian societies: Institutional and cultural explanations compared. International Political Science Review, 32(3), 263–281. doi:10.1177/0192512110378657

Wright, M., Citrin, J., & Wand, J. (2012). Alternative Measures of American National Identity: Implications for the Civic-Ethnic Distinction. Political Psychology, 33(4), 469–482.doi:10.1111/ j.1467-9221.2012.00885.x

98

99

Appendix 1: survey

This appendix only shows the variables used in this study. Codebook T1 Ghana University: University of respondent Value Label 1 UDS Tamale 2 UG Accra 3 KNUST Kumasi ParticipationT2: Respondent participated in T2 Value Label 0 Yes 1 No Finished T2: Respondent completed survey T2 Value Label 0 Partial completion 1 Finished survey 6 Not applicable V1: ‘What is your gender?’ Value Label 0 Male 1 Female 9 Missing V2: ‘How old are you?’ V3: ‘In which region were you born?’ Value Label 1 Greater Accra 2 Ashanti

100

3 BrongAhafo 4 Central 5 Eastern 6 Western 7 Volta 8 Northern 9 UpperWest 10 UpperEast 11 NotinGhana 99 Missing V4: ‘To what ethnic group do you belong?’ Value Label Value Label 1 Agona 27 Nkonya

2 Ahafo 28 Yefi, Achode, Nawuni, Nchumuru

3 Ahanta 29 Bimoba 4 Akuapem 30 Frafra/ Talensi 5 Akwamu 31 Kokomba 6 Akyem 32 Kyamba (Tchamba) 7 Aowin 33 Pilapila

8 Asante 34 Salfalba (Sabulaba) Wali, Dagaar

9 Asen Assin 35 Builsa (Kangyaga or Kanjaga)

10 Boron Brong 36 Dagarte (Dagaba), Lobi 11 Chokosi 37 Dagomba 12 Denkyira 38 Kusasi 13 Evalue 39 Mamprusi 14 Fante 40 Namnam (Nabdom) 15 Kwahu 41 Nankani and Gurense 16 Nzema 42 Nanumba 17 Sefwi 43 Waba (Wala)

101

Value Label Value Label 18 Wasa 44 Kasena Paga 19 Dangme 45 Mo 20 Ga 46 Sisala 21 Ewe 47 Vagala

22 Akpafu, Bowiri, Buem, Likpe, Lol 48 Other Grusi

23 Avatime, Logba, Nyangbo, Tafi 49 Busanga

24 Awutu, Efutu, Senya 50 Wangara

25 Cherepong, Adukrom, Anum, Larteh 88 Don't know

26 Gonja 99 Missing V4o: ‘To what ethnic group do you belong?’ – Other V11: ‘Let us suppose that you had to choose between being a Ghanaian and being a member of your ethnic group. Which of the following statements best expresses your feelings? Value Label 1 I feel only a member of my ethnic group 2 I feel more a member of my ethnic group than Ghanaian 3 I feel equally Ghanaian and a member of my ethnic group 4 I feel more Ghanaian than a member of my ethnic group 5 I feel only Ghanaian 8 Don't know 9 Missing

102

Codebook T2 Q28: The NSS has several objectives; indicate for each of them to what extent you think the NSS program makes a substantial contribution.

Not at all

Very

much Q28_1: Improving national unity 0 1 2 3 4 5 6 7 8 9 10

Q28_2: Diminishing of hunger in Ghana 0 1 2 3 4 5 6 7 8 9 10

Q28_3: Diminishing of illiteracy in Ghana 0 1 2 3 4 5 6 7 8 9 10

Q28_4: Diminishing of unemployment in Ghana

0 1 2 3 4 5 6 7 8 9 10

Q28_5: Delivering essential amenities to the rural areas in Ghana

0 1 2 3 4 5 6 7 8 9 10

Q40: On a 0 to 10 scale, how much do you personally trust each of the following institutions?

Not at all

Very

much Q40_1: parliament 0 1 2 3 4 5 6 7 8 9 10 Q40_2: legal system 0 1 2 3 4 5 6 7 8 9 10

Q40_3: traditional leaders 0 1 2 3 4 5 6 7 8 9 10

Q40_4: police 0 1 2 3 4 5 6 7 8 9 10 Q40_5: politicians 0 1 2 3 4 5 6 7 8 9 10 Q40_6: NSS administration 0 1 2 3 4 5 6 7 8 9 10 Q40_7: president 0 1 2 3 4 5 6 7 8 9 10

103

Q42: The Ghanaian government takes on certain responsibilities. Can you indicate for each of the following items to what extent you think the government is effective or not?

Not at all effective Very

effective Q42_1: In combatting poverty

0 1 2 3 4 5 6 7 8 9 10

Q42_2: In creating employment opportunities

0 1 2 3 4 5 6 7 8 9 10

Q42_3: In providing education

0 1 2 3 4 5 6 7 8 9 10

Q42_4: In supporting poor and deprived people

0 1 2 3 4 5 6 7 8 9 10

Q42_5: In promoting national unity

0 1 2 3 4 5 6 7 8 9 10

Q59: Some people say that the following things are important for being truly Ghanaian. Others say they are not important. How important do you think each of the following things are?

104

Not important at all Very

important Q59_1: have been born in Ghana 0 1 2 3 4 5 6 7 8 9 10 Q59_2: respect Ghana’s political institutions and laws. 0 1 2 3 4 5 6 7 8 9 10

Q59_3: have lived in Ghana for most of one’s life.

0 1 2 3 4 5 6 7 8 9 10

Q59_4: be able to speak English 0 1 2 3 4 5 6 7 8 9 10 Q59_5: feel Ghanaian 0 1 2 3 4 5 6 7 8 9 10 Q59_6: To have Ghanaian ancestry 0 1 2 3 4 5 6 7 8 9 10 Q59_7: To speak a local language 0 1 2 3 4 5 6 7 8 9 10 Q59_8: To be married to a Ghanaian 0 1 2 3 4 5 6 7 8 9 10

104

105

Q75: How much trust do you have in people with the following characteristics?

No trust at all

Not very much trust

Some trust

A lot of trust

No answer Missing

Q75_1: People from your ethnic group

1 2 3 4 7 9

Q75_2: People from different ethnic groups

1 2 3 4 7 9

Q75_3: People of your religion

1 2 3 4 7 9

Q75_4: People of a different religion

1 2 3 4 7 9

Q75_5: People from your region of origin

1 2 3 4 7 9

Q75_6: People of different regions

1 2 3 4 7 9

Q77: In what type of environment did you grow up? Value Label 1 Small village 2 Medium sized town 3 Big city 9 Missing

106

Q81: What is the highest level of education your father has completed? Value Label 1 No formal schooling 2 Primary school 3 Secondary school / High school

4 Post-secondary qualifications, other than university (e.g. a diploma or degree from a polytechnic or college)

5 University 6 Post-graduate 8 Don't know 9 Missing Q83: What is the highest level of education your mother has completed? Value Label 1 No formal schooling 2 Primary school 3 Secondary school / High school

4 Post-secondary qualifications, other than university (e.g. a diploma or degree from a polytechnic or college)

5 University 6 Post-graduate 8 Don't know 9 Missing Q88: To which religion do you belong? Value Label 1 Catholic 2 Protestant 3 Pentecostal / Charismatic 4 other Christian

107

5 Muslim - Sunni 6 Muslim - Shia 7 Traditional religion 8 No religion 9 Other religions, specify below 99 Missing Q88_Text: To which religion do you belong? – Other (text). Q90: How religious would you say you are? Not religious

at all Very religious

0 1 2 3 4 5 6 7 8 9 10 Q92: To which political party do you feel closest? Value Label 1 NPP 2 NDC 3 CPP 4 Other 5 None 9 Missing Q92_text: To which political party do you feel closest? – Other.

108

109

Appendix 2: descriptives of key variables

Variable n Min. Max. Mean SD Dependent variable Political trust 3022 0 10 3.09 1.92 Independent variables Background variables Gender (female = 1) 3264 0 1 .32 .47 Age (in years) 3183 18 46 23.21 2.59 Town size 2951 1 3 2.34 .67 Religion (Muslim = 1) 2938 0 1 .07 .25 Religiousness 2939 0 10 7.36 2.22

Ethnicity (reference group: Akan)

Da-Dange 3154 0 1 .10 .30 Ewe 3154 0 1 .15 .36 Guan 3154 0 1 .03 .17 Gurma 3154 0 1 .03 .17 Mole-Dagbani 3154 0 1 .07 .26 Grusi 3154 0 1 .01 .11 Mande 3154 0 1 .01 .09

Political affiliation (reference group: NPP)

NDC 1401 0 1 .26 .44 NPP 1401 0 1 .03 .17 Complex citizenship 3004 0 10 5.70 2.47 Q59_1 2995 0 10 5.99 3.14 Q59_2 2992 0 10 6.41 2.99 Q59_3 2987 0 10 5.02 3.00 Q59_4 2993 0 10 5.02 3.58 Q59_5 2986 0 10 6.35 3.23

110

Q59_6 2986 0 10 5.93 3.18 Q59_7 2988 0 10 6.39 3.21 Interpersonal trust 2983 1 10 6.53 1.49 Q83_1 2611 1 4 2.74 .73 Q83_2 2584 1 4 2.49 .69 Q83_3 2669 1 4 2.93 .77 Q83_4 2586 1 4 2.44 .74 Q83_5 2591 1 4 2.62 .74 Q83_6 2547 1 4 2.44 .70

General policy performance 3019 0 10 3.28 2.03

Q42_1 3012 0 10 2.72 2.15 Q42_2 3013 0 10 2.50 2.11 Q42_3 3009 0 10 4.13 2.44 Q42_4 3007 0 10 2.92 2.26 Q42_5 3003 0 10 4.14 2.69 NSS Contribution 2928 0 10 3.76 2.21 Q28_1 2901 0 10 4.54 2.92 Q28_2 2895 0 10 3.05 2.35 Q28_3 2889 0 10 3.84 2.69 Q28_4 2890 0 10 3.79 2.60 Q28_5 2891 0 10 3.67 2.78

111

112