Citizenship, Active Citizenship & Social Capital in Zimbabwe: a Statistical Study* Research & Advocacy Unit

Citizenship, Active Citizenship & Social Capital in Zimbabwe: a Statistical Study* Research & Advocacy Unit May 2015 1

Executive summary Citizenship in Zimbabwe is generally discussed in very narrow terms, as the rights and obligations enshrined in the Constitution and the laws of the country. However, there is too little discussion about citizenship in its wider sense: the ways in which Zimbabweans act in common enterprise ( active citizenship ), and the kinds of social institutions and initiatives that citizens generate ( social capital ) are issues that have theoretical consideration, but little empirical content. Here the Afrobarometer surveys can provide interesting insights into understanding active citizenship and social capital. Using the data from the fifth Round of the Afrobarometer survey on Zimbabwe in 2012, a series of measures was constructed for active citizenship, as political participation and political efficacy, and civic participation, as well as social capital, as institutional trust and intimate trust, with gender and residence (urban or rural) as cross cutting variables. These measures were constructed from 33 questions in the Afrobarometer data. Cross-tabulations were carried out with the cross-cutting variables, and correlations between the measures were calculated. These latter demonstrated a high degree of inter-relationship between many of the related measures, such as between trust family and trust neighbours, and thus a series of composite measures were derived to examine the entire data set, which reduced the 33 variables to a reduced set of 11. These are described as follows: Civic participation Political restrictions Basic freedoms Voting Satisfied with voting Political participation Gender inequality Institutional trust Political efficacy Intimate trust Political engagement Active Citizenship As regards political participation, rural respondents were more likely to state that they were close to a political party than urban respondents, but fewer respondents from both groups were willing to state which political party they were close to. Many more respondents were unwilling to express a political party preference, and this was more pronounced for urban than rural respondents. Additionally, more rural respondents expressed support for ZANU PF than urban respondents, which were reversed for MDC-T supporters. This trend in difference between rural and urban respondents was repeated for responses about the elections in 2008, with more rural respondents that voted in March and June 2008, and saw these elections as free and fair. 2

Political participation as voting is not matched as strongly by political participation in political parties. The highest rates of reporting being close to a political party came in 2012, after nearly five years of the Global Political Agreement (GPA) and the Inclusive Government, and when political violence and intimidation were greatly reduced on the previous decade. Men were more willing than women, and rural respondents more willing than their urban counterparts to state being close to a party. Political efficacy was defined as follows by Campbell, the feeling that individual political action does have, or can have, an impact upon the political process, i.e., that it is worthwhile to perform one s civic duties. It is the feeling that political and social change is possible, and that the individual citizen can play a part in bringing about this change. Rural respondents were significantly more likely to have been to a rally, to have campaigned for a candidate, and tried to persuade others. Urban respondents felt more unequal under the law, and were fearful of violence during elections, although high frequencies of both rural respondents reported both of this latter as well. Civic Participation, defined as both voice and participation, also showed interesting differences. This was measured initially by seven questions from the Afrobarometer, and then combined into one score for the factor analysis. In the initial analysis, activities that might not be public, such as having an interest in public affairs or discussing politics at home, are more commonly reported than public activities. Rural respondents are also more likely to speak to their local councillor or go to a meeting, but the direction cannot be inferred from this data, and going to meetings in rural areas might easily be a response to a demand to attend, especially during election periods. As perhaps expected higher percentages of Zimbabweans reported belonging to a religious group, and many more women than men, and the youth are clearly less engaged than their elders. They are less interested in public affairs, discuss politics less at home, see politics as too complicated, and are more careful about they say in public. When it comes to freedoms, rural respondents state that they are freer than their urban counterparts to say what they think or join the political party of their choice, but are equally careful about what they say in public. This contradiction deserves more investigation. So, as regards citizen agency, whether as political or civic, the major theme that emerges is the difference between rural and urban Zimbabweans, with both voice and participation reducing as the domain of engagement becomes more explicitly party politics. Social Capital Social Capital, measured both as institutional trust (trust in ordinary people) and intimate trust (trust in officials and agents of the state), showed some interesting differences between urban and rural respondents. As regards intimate trust, both groups most frequently trust their relatives, but thereafter rural folk have greater indications of trust than urban folk. When political party affiliation is factored in, the measures of trust are all significantly inter-correlated, suggesting that they tap a common core of trust. Additionally, rural trust is significantly correlated with affiliation to ZANU PF, whereas trust in all respects is negatively correlated with membership of the MDC-T, and that urban residence negatively correlates with all the measures of trust. 3

For institutional trust, this was defined by reference to the questions dealing with trust in three key institutions: the police, the army, and the courts of law. There were significant differences between rural and urban respondents, with rural respondents expressing significantly greater trust in all three institutions than urban respondents. Institutional trust was greater for the rural respondents and for those supporting ZANU PF, which this is perhaps unsurprising given the history of the partisan nature of these institutions over the past 14 years, and the support given to ZANU PF s key support base in the rural areas. Since gender can be an important differentiating variable in social and political action in Zimbabwe, 5 questions examined gender as a variable in active citizenship. The results were perhaps obvious urban women saw greater discrimination against women than rural women but not so the views about whether men or women make better leaders. The major differentiation came between rural and urban respondents, and, as can be seen from Table 11 (over), urban respondents were significantly more likely to see women as being discriminated against. This would seem to reflect the effects of traditional culture, but it is startling that both groups agree, in very high numbers, that men make better leaders than women. Constructing composite measures of citizenship, social capital, & political efficacy It is clear that many of these measures are related to each other. For example, women s views about inequality seem to tap a common factor, and this seems the case for many of the other variables included in this analysis. Hence, it is sensible to aggregate variables that were highly inter-correlated into a single measure. Accordingly, the 33 individual variables were clustered into 11 new variables. A Principal Components Factor Analysis was carried out with these 35 variables (and the seven marker variables), using SPSS (Version 20). The initial solution revealed five main components with a total cumulative variance of 60.2%. These five components were not correlated on bivariate analysis, and hence an orthogonal rotation using the VARIMAX rotation was carried out. This then gave five orthogonal factors, with the cumulative variance of 55.5%, but one of these seemed purely methodological, loading only on Age and Primary Education. These two variables were removed and the rotated structure revealed four main components with a cumulative variance of 45.1%. The first two factors were defined most strongly by expressed support for ZANU PF and MDC-T respectively. This suggests very strongly how polarized Zimbabwean society has become, and how complex it will be for any civic strategy to create non-partisan civic action amongst Zimbabweans. The third factor seems to identify active citizens, 30% of respondents. They participate in politics by attending rallies, campaigning for political parties, and trying to engage others to support their political party. But they also have interest in public affairs, discuss politics, belong to church groups, belong to community groups, and contact both local councillors and Members of Parliament. The final factor was described by having a three distinct measures; having tertiary education, being an urban resident, and in some form of formal employment, but interestingly was negatively related to age, and, hence, was not confined to youth. This group represented only 8% of respondents. 4

Conclusions Throughout this analysis, whether looking at various hypothesised indicators of active citizenship or social capital, the divide between the rural and the urban in Zimbabwe looms large, and was strongly confirmed by the factor analysis. The various iterations consistently indicated two main factors: one supporting ZANU PF, one supporting the MDC-T, and a third that seems to suggest active citizens. When the various response categories were combined into eleven measures, the same two factors emerged ZANU PF and MDC-T supporters mostly defined in opposition to each other. The third factor suggested a group of Zimbabweans who are active citizens: they participate politically, and not merely by voting, but also attend rallies and forms of political participation. They also report many aspects of civic participation. The fourth factor, Disconnected Democrats, seem to indicate yet another aspect of the polarised society that Zimbabwe has become? Caught between the two powerful political parties, and a political terrain in which violence and intimidation not only rule elections, but permeate ordinary civic participation and pollute the possibility of conventional social capital, has the middle class retreated altogether: they expound the rhetoric of democracy but avoid the responsibility of active citizenship? Some very interesting speculations do emerge from this data and point to the need for much more sensitive, cross-sectional and qualitative research in order to more deeply understand citizenship, active citizenship, and social capital in Zimbabwe. It does also suggest that there will be profit in pursuing strategies for enhancing the already existing active citizens agency: * Start locally. * Build on elections. * Provide access to information. * Make representatives listen. * Aggressively pursue an anti-corruption agenda. Finally, there does seem a pressing need to understand what has happened to the middle class Zimbabweans. *This report was prepared by Tony Reeler, Senior Researcher. 5

Contents Executive summary... 2 Background... 7 Methods... 9 Findings... 9 Active citizenship:... 9 Political participation... 9 Political Efficacy... 11 Civic voice and participation... 12 Social Capital... 14 Trust (intimate)... 14 Trust (institutional)... 15 Gender... 16 Constructing composite measures of citizenship, social capital, & political efficacy... 17 Factor analysis... 18 Conclusions... 20 References... 23 Appendix 2... 29 Correlations between clustered variables.... 29 6

Background Citizenship, active citizenship, and social capital are either highly contested in, or wholly absent from, the socio-political discourse in Zimbabwe. It can even be argued that the whole of the history of this country, from early colonisation to the present can be argued to be a clash between citizens and subjects, as Mahmood Mamdani (Mamdani.1996) would have it. Whilst it would be hard to refute the idea that the liberation war was not a struggle to refuse the notion of subjects by the black majority of the country one man, one vote was the fundamental cry it is also hard not to see that citizen status rapidly became a privileged status, increasingly reserved by the predatory state that Zimbabwe has become. For example, in the aftermath of the shocks to ZANU PF s hegemony in 2000, the government redefined citizenship in ways that would exclude a whole range of minorities, but especially commercial farm workers, thereby disenfranchising hundreds of thousands of voters (ZHRNGO.2001). And, although citizenship is clearly defined in the new (2013) Constitution, and is supported by a liberal Declaration of Rights, full clarification of citizenship will require future legislation to give effect to some of the ambiguities created by previous limitations placed on citizenship. Furthermore, now having a two-thirds constitution-amending majority in Parliament, the ZANU PF government can block progressive legislation and even amend the Constitution in retrogressive fashion, as it has done during the 33 years of the previous constitution. It is also evident that whatever is meant by active citizenship, this is a quality that has been increasingly diminished by the political violence of the past 14 years. As Eldred Masunungure has put it, the sensible response of citizens to a risk-taking state is to become risk averse (Masunungure.2006). This is graphically demonstrated through the five rounds of the Afrobarometer surveys since 1999. As these surveys show Zimbabwean citizens show very low levels of participation, and very high levels of fear about participation. In Round Two (2002), less than half said that they trusted Robert Mugabe, but this was an increase on Round One, but significantly very large numbers were unwilling to state their political affiliation, with more than 80% stating that they had to be careful about expressing political opinions. This latter trend has worsened over time: prior to the violence that began after the 2000 Constitutional referendum, Zimbabweans felt free to express themselves, but that rapidly changed, and has continued to date. Thus, citizen voice has been dramatically reduced, and particularly because those not expressing overt support for ZANU PF are seen as the supporters of opposition political parties, with the youth as the primary targets for the state s repression and coercion (SPT.2003). It is also appropriate to note here that Zimbabwe, perhaps more acutely than other Southern African countries, is bedeviled by the legacy of the Liberation War, and increasingly is described as a predatory state, and, furthermore, a state in which concepts such as failed or fragile are confusingly applied. Most recently, Michael Bratton has applied the predatory state concept to Zimbabwe in a very detailed analysis and it seems evident that the existing political power will not be easily transformed, with the implication that an active citizenry will not be fostered without resistance from the state (Bratton.2014; Bratton & Masunungure.2011). As pointed out above, it is the case that participation has significantly reduced over the past decade, except for voting. It is hardly surprising that participation would reduce when even having voice was felt to be risky. For example, Zimbabweans have rarely felt that participating in a demonstration was a popular form of political participation, but this has worsened markedly 7

over the past decade. According to the Afrobarometer, and in answer to the question have you ever attended a demonstration or protest march, in all the five rounds to date, more than half of all respondents state that they would never do this: 50% in 1999, 59% in 2002, 65% in 2005, 57% in 2008, and 76% in 2012. However, too simple an analysis of active citizenship in Zimbabwe as mere involvement in politics, and minimally as voters - may blind us to many ways in which Zimbabweans are actually behaving as citizens. It can be argued that the state survives in many ways because of the immense inventiveness and resilience of Zimbabweans in manufacturing livelihoods in the informal sector, additionally supported by the remittances from the millions of family members in the diaspora. This conforms to the risk averse, as opposed to the risk taking, view of being a citizen: faced with a highly risk taking State, citizens have elected to avoid conflict by similar risk taking (or revolutionary) civic action. What the risk averse conception of civic life and trust means for our understanding of social capital is problematic. Zimbabweans are chary of political affiliation, but not of voting, but there is little evidence to suggest in what other ways they engage associational life. Although there are hundreds of non-governmental organisations and other civic initiatives, there are few reports that analyse citizens activities from a social capital perspective per se. However, from one study, it is evident that the polarised and terrorised space has produced a divide in how many organisations view themselves, with many organisations choosing the softer issues; socioeconomic issues, health, or humanitarian assistance (Dorman.2008). Thus, a fundamental issue for most political commentators is the way in which virtually every aspect of potential associational life has been captured by party politics. However, there is evidence that this is not uniform: a recent investigation of the effects of remittances in Africa, a potential resource curse - that included Zimbabwe - suggested that remittances might act to facilitate more active citizenship in the recipients of remittances. Although those receiving remittances were less likely to vote, they were more likely to engage state officials and participate in protests or demonstrations (Dionne et al. 2014). It is also possible, as has been found for other countries, that the diaspora does not merely support livelihoods, but may be an important source of importing democracy (Perez-Armendariz, C., & Crow, D. 2010). With at least one million Zimbabweans in the diaspora other estimates put the number at two, or two and half million - exposed to the very different socio-political processes of the United Kingdom, South Africa, and Botswana, this is a potential variable in developing an active citizenry that deserves more than the brief attention it has had to date. For women, the underpinning of the totalitarian state by a dominant patriarchy results in their being almost wholly marginalized from political discourse, and, more seriously, have become frequent primary victims (and the overwhelming majority of secondary victims) in the political violence that has been endemic in every election since 2000. Youth have become dual victims: the major victims of the massive unemployment that has accompanied Zimbabwe s severe economic decline, and are also victims of the dominant patriarchy that undervalues youth in most respects. Thus, gender was explicitly included in the analysis. Given the paucity of studies examining active citizenship or social capital in Zimbabwe, the Afrobarometer provides one of the few sources of useful information here. Tracking 8

Zimbabweans views on democracy, politics, governance and many other factors, the data available allows the development of some understanding of how Zimbabwean citizens view active citizenship and social capital, albeit indirectly. This is the ambit of this report. Methods The data for this study is taken from the Round 5 Afrobarometer survey carried out in 2012. There were four previous surveys in 1999, 2005, 2009 and 2010. 2012 was chosen in order to have the most up-to-date perspective on the issues under examination. 1 The questions chosen were taken to reflect aspects about active citizenship and social capital, and are given in Appendix 1, together with the explanation of how each of these variables was calculated from the Afrobarometer data. There were five marker variables for cross-tabulations: age (under 30 versus over 30), gender (male versus female), employment (employed versus unemployed), education (none, primary only, secondary only, and tertiary), and residence (rural versus urban). In all, 33 additional content questions were selected in order to examine the various aspects of active citizenship and social capital. The aspects of active citizenship and social capital used in the study were as follows: Political participation; Civic participation; Trust (intimate & institutional); and Gender. An additional sort of the variables was undertaken as it seemed that there were a very large number of inter-correlations between the 33 variables and the 7 marker variables. Accordingly, a number of national grouped measures were developed (see later), and these were then analysed in SPSS (Version 20). Frequencies were calculated and tests of significance (Chi-square) also calculated. A Principal Components factor Analysis (PCA) was carried out in order to examine the correlations between the variables. Findings Active citizenship: Active citizenship was examined by means of two clusters; Political participation and civic participation. The rationale here is that citizenship, whether legally or theoretically defined, can be a wholly passive relationship between and individual and the state. For example, there is an endless discussion in Zimbabwe about rights and considerably less about responsibilities, and frequent discussion in civil society about the need to educate citizens about their rights. But the concern here is the reciprocal of rights, the responsibilities and duties of citizens: how they demand their rights; how they try to use their voice and participate; and how they engage each other (and not necessarily the state). Political participation The measures of political participation were derived from six questions: (i) Do you feel close to any particular political party; (ii) Which party is that; (iii) Did you vote in the March 2008 election; (iv) Did you vote in the June 2008 Presidential run-off; (v) How free and fair would you rate the freeness and fairness of the last national election, held in March 2008; (vi) How 1 The most recent Afrobarometer survey, Round 6, was carried out in November 2014, but the data is not publicly available at present. 9

free and fair how would you rate the freeness and fairness of the last national Presidential runoff election, held in June 2008. As regards political participation, rural respondents were more likely to state that they were close to a political party than urban respondents [72% v. 60%, p=0.0001], but fewer respondents from both groups were willing to state which political party they were close to, and many more respondents were unwilling to express a political party preference: this was more pronounced for urban than rural respondents [39% v. 29%; p=0.0001]. Additionally, more rural respondents expressed support for ZANU PF than urban respondents [34% v. 17%; p=0.0001], and this trend was reversed for MDC-T supporters: 33% were urban as opposed to 28% rural (p=0.002). Table 1: Voting in 2008, Rural v. Urban respondents [Source: Afrobarometer online data, Round 5] Rural Urban Voted in March 2008 election 73%* 60% Voted in June 2008 election 65%* 46% March election was free & fair 47%* 30% June election was free & fair 32%* 19% *p=0.0001 This trend in difference between rural and urban respondents was repeated for responses about the elections in 2008. Without exception, more rural respondents voted in March and June 2008, and saw these elections as free and fair. However, it is noteworthy that there were drops for both groups in their views from the March to the June polls: fewer rural and urban respondents voted in June, and there were large changes in their views on whether the June poll was free and fair. Table 2: Percentage of men & women reporting being close to a political party [Source: Afrobarometer online data base, 1999-2012] Year Male (%) Female (%) Rural Urban Rural Urban 1999 48% 33% 42% 30% 2004 49% 30% 41% 34% 2005 62% 42% 55% 44% 2008 51% 43% 47% 30% 2012 86% 87% 85% 87% It is also evident that political participation as voting is not matched as strongly by political participation in political parties. As can be seen from Table 2, the highest rates of reporting being close to a political party came in 2012, after nearly five years of the GPA and the Inclusive Government, and when political violence and intimidation were greatly reduced on the previous decade. Prior to that, it was rare that more than half of the respondents would indicate being 10

close to a political party, with men more willing than women, and rural respondents more willing than their urban counterparts. Table 3: Rural and Urban Zimbabweans expressing support for political parties and their voting preferences. [Source: Afrobarometer Online data, Round 5] Party Male (%) Female (%) Rural Urban Rural Urban MDC-T 27% 32% 33% 50% ZANU PF 37% 24% 51% 35% MDC-M 0 0 0 - MKD 0 - - 2% Refused to answer 25% 27% 15% 13% When actually asked which political party they supported, even in 2012, a quarter of respondents amongst males and slightly more than 10% of women refused to answer the question. This was the source of some dispute amongst political commentators, but the most reliable study suggested, on strong evidence, that political fear remained present in significant numbers of Zimbabweans (Bratton & Masunugure.2012). However, political participation is probably the most minimal citizen activity that Zimbabweans undertake for all the reasons outlined above. That so few felt that the elections were free and fair leads to the view that even in this minimalist activity, Zimbabweans do not feel that their participation in political events has much effect. Thus it is useful to look at political efficacy. Political Efficacy Although the term efficacy has had widespread use in the field of psychology (Bandura.1977), the term has been increasingly used in studies of political behaviour. As originally defined by Campbell (Campbell et al. 1954), political efficacy is the feeling that individual political action does have, or can have, an impact upon the political process, i.e., that it is worthwhile to perform one s civic duties. It is the feeling that political and social change is possible, and that the individual citizen can play a part in bringing about this change. It is in this sense the term was used in selecting the variables hypothesized to represent political efficacy. Five measures of political efficacy were derived from the Afrobarometer data: (vii) did you go to a rally; (viii) did you try to persuade others to vote for a certain presidential or legislative candidate or political party; (ix) during election campaigns in this country, how much do you personally fear becoming a victim of political intimidation or violence; and (x) are people treated unequally under the law? Rural respondents were significantly more likely to have been to a rally [73% v. 47%; p=0.0001], to have campaigned for a candidate [14% v. 10%; p=0.002], and tried to persuade others [25% v. 13%; p=0.0001 ], whilst urban respondents felt more unequal under the law [70% v. 54%; p=0.0001], and were fearful of violence during elections [65% v. 62%; p=0.02, n/s], although high frequencies of both rural respondents reported both of this latter as well. 11

However, as pointed in the main review, active citizenship means a wider engagement with one s society than the narrower domain of electoral politics or central government. It is suggested that active citizenship should reflect engagement with the wider society, and politics in local government and communities. So how do Zimbabweans measure on these wider standards? Civic voice and participation Voice and participation were defined as being outside of politics per se, and involved participation in the wider social life possible for a citizen. For civic participation, this was measured by reference to seven questions: (xi) During the past year, how often have you contacted a local government councillor; (xii) during the past year, how often have you contacted a Member of Parliament; (xiii) could you tell me whether you are an official leader, an active member, an inactive member, or not a member of a religious group that meets outside of regular worship services; (xiv) could you tell me whether you are an official leader, an active member, an inactive member, or not a member of some other voluntary association or community group; (xv) have you attended a community meeting; (xvi) How interested would you say you are in public affairs; (xvii) When you get together with your friends or family, would you say you discuss political matters? Table 4: Measures of civic participation and interest, Rural v. Urban [Source: Afrobarometer online data, Round 5] Rural Urban Contact councilor 29%* 19% Contact MP 13%* 8% Belong to church group 45% 43% Belong to community group 22% 22% Went to community meeting 88%* 52% Interested in public affairs 60% 65%** Discuss politics at home 20% 24%** *P=0.0001; **P=0.005 As can be seen from Table 4, activities that might not be public, such as having an interest in public affairs or discussing politics at home, are more commonly reported than public activities. Rural respondents are also more likely to speak to their local councilor or go to a meeting. However, the direction of contact - from respondent to councilor or vice versa cannot be assumed from this data, and it may be, given the coercive nature of broad politics in much of rural life, that these responses may reflect bi-directional influences: councillors and traditional leaders calling meetings and people are expected to attend, and, during election times, there might be an element of coercion involved. 12

As perhaps expected higher percentages of Zimbabweans reported belonging to a religious group, and many more women than men 2. This in itself is unsurprising in a country where the overwhelming preponderance of Zimbabweans is Christian in one form or another, but the question seeks a different understanding from mere religious affiliation: the question aims to find out what social engagement people have with their churches, for churches are an important source of social capital in Zimbabwe (as elsewhere). They provide social, economic and psychological support for their members, as well as the more obvious spiritual support. Churches would thus seem the primary base for social capital in Zimbabwe, which is probably very different to many Western countries. Youth suffer from a number of discriminatory practices, and, in particular, the traditional overvaluation of maturity as age: young persons views, and particularly young women, are not accorded the same value as those of men (foremost) and older men, especially. This is evident in the Afrobarometer data, where in response to questions about civic participation the youth are clearly less engaged than their elders (Table 5 over). They are less interested in public affairs, discuss politics less at home, see politics as too complicated, and are more careful about they say in public. Table 5: Measures of civic participation, Under 30 year s v. Over 30 years [Source: Afrobarometer online data, Round 5] Under 30 (%) Over 30 (%) Interested in public affairs 55% 63%* Discuss politics 72% 78% Politics is too complicated for me 70%* 68% Careful what you say 92% 88% *p=0.0001 For citizen voice, this was measured by reference to five questions extracted from the Afrobarometer data for Round 5: (xviii) In this country, how free are you to say what you think; (xix) In this country, how free are you to join any political organisation you want; (xx) In this country, how free are you to choose who to vote for without feeling pressured; (xxi) In your opinion, how often, in this country do people have to be careful of what they say about politics; (xxii) Sometimes politics and government seem so complicated that a person like me cannot really understand what is going on? There were some significant differences between rural and urban respondents as seen in Table 6 (over), but the findings were somewhat contradictory. Rural respondents state that they are freer than their urban counterparts to say what they think or join the political party of their choice, but are equally careful about what they say in public. This contradiction deserves more investigation. 2 Belonging to church group; Women v. Men: 50% v. 37%; p=0.001. 13

Table 6: Views on freedom, Rural v. Urban [Source: Afrobarometer online data, Round 5] Rural Urban Free to say what you think 50%* 37% Free to join political 56%* 45% organization of choice Free to vote without pressure 63% 61% Careful what you say in public 88% 92%** *p=0.0001; **p=0.005 So, as regards citizen agency, whether as political or civic, the major theme that emerges is the difference between rural and urban Zimbabweans, with both voice and participation reducing as the domain of engagement becomes more explicitly party politics. Social Capital Trust is hypothesised to be a key component in social capital, and here two different sets of trust were examined. What might be called intimate trust, or trust in the ordinary people whom one interacts with in the community, and institutional trust, or trust in the various officials and agents of the state with whom one may come into contact. For the latter, we chose the police, the army, and the courts of law because these are all supposed to be neutral according to the constitution and their enabling legislation, but there are continuous allegations that they, in fact, operate in partisan fashion. Trust (intimate) Intimate trust was measured by four questions: (xxiii) would you say that most people can be trusted or that you must be very careful in dealing with people; (xxiv) How much do you trust your relatives; (xxv) How much do you trust each of the following types of people your neighbors; and (xxvi) how much do you trust other people you know? Table 7: Intimate trust, Rural v. Urban respondents [Source: Afrobarometer online data, Round 5] Rural Urban Trust most people 23%* 7% Trust relatives 80% 78% Trust neighbours 61%* 52% Trust others you know 36%* 29% *p=0.0001 There are marked and significant differences in intimate trust between rural and urban respondents. Both groups most frequently trust their relatives, but thereafter, rural folk have greater indications of trust than urban folk. This may be the artifact of the rural areas where 14

associational life has much greater importance than is the case for urban life, but it is interesting that this should be so when political violence and intimidation have been far more prevalent in the rural areas with few exceptions. Table 8: Intimate trust and political party affiliation Gender General trust Trust relatives Trust neighbours Trust others MDC-T ZANU PF Refused to answer Urban.014 -.191 ** -.043 -.125 ** -.104 **.064 ** -.176 **.091 ** Gender.008 -.038.015.010.046.039 -.082 ** General.206 **.312 **.422 ** -.074 **.170 ** -.088 ** trust Trust.501 **.289 ** -.086 **.077 **.007 relatives Trust.527 ** -.077 **.102 ** -.023 neighbours Trust others -.082 **.108 ** -.026 MDC-T -.381 ** -.532 ** ZANU PF -.550 ** **p=0.01 When political party affiliation is factored in, the measures of trust are all significantly intercorrelated (Table 8), suggesting that they tap a common core of trust. It can also be seen that rural trust is significantly correlated with affiliation to ZANU PF, whereas trust in all respects is negatively correlated with membership of the MDC-T, and that urban residence negatively correlates with all the measures of trust. Thus, there are very good grounds for concluding that social capital operates differently between the rural and urban areas of Zimbabwe, and also that political parties exert an enormous influence on how this social capital operates as well. Trust (institutional) Institutional trust was defined by reference to the questions dealing with trust in three key institutions: the police (Question xxvii), the army (Question xxviii), and the courts of law (Question xxix). Whilst generally respondents expressed trust in these three institutions, this was amplified when the effects of residence was examined. As can be seen in Table 9, there were significant differences between rural and urban respondents, with rural respondents expressing significantly greater trust in all three than urban respondents. Table 9: Institutional trust, Rural v. Urban respondents [Source: Afrobarometer online data, Round 5] Rural Urban Trust police 55%* 27% Trust army 60%* 42% Trust the courts 68%* 49% *p=0.0001 15

Once again this is perhaps unsurprising given the history of the partisan nature of these institutions over the past 14 years, and the support given to ZANU PF s key support base in the rural areas. Table 10: Institutional trust and political party affiliation [Source: Afrobarometer online data, Round 5] ZANU Refused to Gender Police Army Courts MDC-T PF answer Urban.014 -.251 ** -.165 ** -.186 **.064 ** -.176 **.091 ** Gender.001.001 -.015.046.039 -.082 ** Police.681 **.531 ** -.259 **.361 ** -.087 ** Army.573 ** -.260 **.374 ** -.101 ** Courts -.204 **.266 ** -.045 MDC-T -.381 ** -.532 ** ZANU PF -.550 ** **p=0.01 This apparent bias towards a political constituency is further demonstrated by the correlations between the measures of social capital. Whilst trust in institutions is generally inter-correlated, all three are negatively correlated with urban residence and with support for MDC-T, as well as for those respondents that refused to answer the question about their political party support. Trust in institutions is however positively correlated with support for ZANU PF, and hence the difference between rural and urban respondents, seen in Table 10 (over), is explained as a feature of political party support. Again an aspect of social capital is differentially distributed in Zimbabwe. Gender Since gender can be an important differentiating variable in social and political action in Zimbabwe (Reeler. 2014), 5 questions examined gender as a variable in active citizenship. These were as follows: (xxx) In your opinion, how often, in this country are women treated unequally by traditional leaders; (xxxi) In your opinion, how often, in this country, are women treated unequally by the police and courts; (xxxii) In your opinion, how often, in this country, are women treated unequally by employers; (xxxiii) Men make better political leaders than women, and should be elected rather than women? The results were perhaps obvious, but not so the views about whether men or women make better leaders. The major differentiation came between rural and urban respondents, and, as can be seen from Table 11 (over), urban respondents were significantly more likely to see women as being discriminated against. This would seem to reflect the effects of traditional culture, but it is startling that both groups agree, in very high numbers, that men make better leaders than women. 16

Table 11: Rural and urban views about women [Source: Afrobarometer online data, Round 5] 3 Rural (No; %) Urban (No; %) Female Male Female Male Are women treated unequally by 69% 66% 78% 76% traditional leaders? Are women treated unequally by 27% 27% 41% 30% the police and the courts? Are women treated unequally by 37% 36% 50% 42% employers? Do men make better leaders than women 91% 80% 92% 87% A number of underlying factors may be operating here, although it does not appear that either age or gender is significant (see footnote 2). It may be speculated that the major factor is the dominant patriarchal nature of Zimbabwean society, especially seen when women agree with their male counterparts that the latter make better leaders, but this has been a common finding in other Zimbabwean studies (MPOI. 2002). This is obviously important when attempting to increase women s agency and participation. Constructing composite measures of citizenship, social capital, & political efficacy It is clear that many of these measures are related to each other. For example, women s views about inequality (see table 11 above) seem to tap a common factor, and this seems the case for many of the other variables included in this analysis. Hence, it is sensible to aggregate variables that were highly inter-correlated into a single measure. Accordingly, the 33 individual variables were clustered into 11 new variables. This was done by constructing an overall score for each grouping. For example, civic participation was composed of seven variables, which gave a range of scores for civic participation from 0 to 7. This was done for each of the hypothesised groupings, with an overall score for each of these. Additionally, the marker variables urban/rural residence, age, gender, education, and employment - were retained for the analysis. A further two variables were added to obtain a stronger measure of political efficacy, defined as the belief that ordinary citizens can hold government and policy makers to account. These were two other questions from the Afrobarometer 2012 data: Who should be responsible for making sure that, once elected, the president does his job? Who should be responsible for making sure that, once elected, the prime minister does his job? 3 There were no differences due to either age [under 30 v. over 30] or gender [male v. female]. 17

Table 12: Consolidated variables Interested in public affairs; discuss politics at home; belong to church group; belong to community group; attend Civic participation community meetings; contact councilor; contact MP. Politics too complicated; fear political violence; careful what Political restrictions you say. Basic freedoms Free to say; free to join; free to vote. Voting Voted in March 2008; voted in June 2008. Satisfied with voting March 2008 free and fair; June 2008 free and fair. Political participation Attended a rally; campaigned for party; canvassed voters. Traditional leaders discriminatory; police discriminatory; Gender inequality employers discriminatory. Institutional trust Trust the police; trust the army; trust the courts. Voters can hold President accountable; voters can hold Prime Minister accountable; voters can hold government Political efficacy accountable. Trust most people; trust family; trust neighbours; trust others Intimate trust Political engagement you know. Close to political party; support MDC-T; support ZANU PF; not willing to say. The correlations of the original (33 variables) data set supported this notional arrangement, as can be seen in Appendix A. For example, political restrictions are negatively correlated with freedoms [-0.22], participating in voting [-0.73], satisfied with voting [-0.22], and institutional trust [-0.22]. Again, civic participation was positively correlated with freedoms [0.13], participating in voting [0.25], satisfied with voting [0.86], and political participation [0.39]. This suggested that there be an underlying structure to the variables under discussion, and, accordingly, factor analysis seemed the method of choice for revealing a potential structure. Factor analysis Therefore, a Principal Components Factor Analysis was carried out with these 35 variables (and the seven marker variables), using SPSS (Version 20). Correlations and their significance were calculated, and a total of 20 factors extracted on initial solution. Five main components were identified with eigenvalues of 1 or more, and with a total cumulative variance of 60.2%. These five components were not correlated on a bivariate analysis, suggesting that the components were orthogonal, and hence an orthogonal rotation using the VARIMAX rotation was carried out. This then gave the five orthogonal factors, with the cumulative variance of 55.5%, but one of these seemed purely methodological, loading only on Age and Primary Education. These two variables were removed and the rotated structure revealed four main components with a cumulative variance of 45.1% (Table 13). 18

Table 13: Principal Components on Afrobarometer measures Factor (variance) Factor 1 [17.1%] ZANU PF Factor 2 [12.8%] Component (loadings) ZANU PF (0.74); Institutional trust (0.68); Happy with election results (0.58); Positive views on political freedoms (0.56) Close to political party (0.84); MDC-T (0.71) MDC-T Factor 3 [8.7%] Active citizens Factor 4 [6.9%] Politically active (0.74); Participates in civic activities (0.69); Votes in elections (0.66) Tertiary education (0.88); Urban (0.55); Employed (0.53) Disconnected democrats The first two factors were defined most strongly by expressed support for ZANU PF and MDC-T respectively. The ZANU PF factor loaded on different measures of citizenship and social capital: these respondents expressed trust in the army, police and the courts, as well as being satisfied with the results of the 2008 elections, and felt that they had political freedoms. This represented 26% of respondents. This is unsurprising in the light of all the previous discussion and analysis. The second factor loaded on those who expressed closeness to a political party as well stating they supported the MDC-T, and thus is clearly a group of opposition party supporters. This represented 27% of respondents. As Bratton and Masunungure have suggested earlier, those expressing closeness to a political party may not be a distinct group, but merely those potential supporters of MDC-T that are fearful of expressing their political preferences (Bratton & Masunungure.2012), This suggests very strongly how polarised Zimbabwean society has become, and how complex it will be for any civic strategy to create non-partisan civic action amongst Zimbabweans. The third factor seems to identify active citizens, 30% of respondents. They participate in politics by attending rallies, campaigning for political parties, and trying to engage others to support their political party. But they also have interest in public affairs, discuss politics, belong to church groups, belong to community groups, and contact both local councillors and Members of Parliament. However, they do not express closeness to a political party or state affiliation to a political party. The final factor was described by having a three distinct measures; having tertiary education, being an urban resident, and in some form of formal employment, but interestingly was negatively related to age, and, hence, was not confined to youth. This group represented only 8% 19

of respondents. This group might more easily be described as middle class, and the group that John Gay defined as being part of a virtuous circle in African democracy (Gay. 2003). However, this group does not seem to have any connections to the variables hypothesized to represent active citizenship, social capital or political efficacy, and hence the term disconnected democrats. Thus, the factor analysis extends the discussion into active citizenship and social capital by suggesting that there may well be three, possibly four, very distinct citizen groups. This indicates that future investigations into active citizenship and social capital must move into a more nuanced approach, beyond the simple binaries of ZANU PF v. MDC-T, male v. female and rural v. urban. Conclusions Whilst it is always possible to criticise survey data, especially when concerned with voting, the Afrobarometer data does throw considerable light on the agency of Zimbabweans, and, as can be seen from these findings, their agency is wholly constrained by the deeply polarised nature of current Zimbabwe, split between political parties, and further split between rural and urban society. The major significant cross-tabulation emerged from the rural/urban comparison, although there were small differences emerging for age and gender. Examining the differences for the under 30 age group also revealed a small number of significant differences between the rural and urban groups. The demographic data revealed expected differences between rural and urban citizens. The rural group was significantly older than the urban group 4, were less likely to be in formal employment [26% v. 46%], and fewer had any form of tertiary education [5% v. 36%]. When it comes to political participation as an indication of active citizenship, it is evident that Zimbabweans vote and admit to having political affiliation, but markedly less are willing to actually state their political affiliation. Additionally, it seems that admitting political affiliation (but not saying what that is) is a function of how peaceful is the political environment: the highest rates in responding positively to this question came in 2012 after a prolonged period (5 years) in which political violence was largely absent. But, as pointed out earlier, just voting is a minimalist form of active citizenship. There are differences between rural and urban Zimbabweans in how they participate in the wider aspects of elections. Rural folk are more likely to have attended a rally, to have canvassed for a political party, to be free to say what they think, and to join the party of their choice. This seems in conflict with the high rates also reported of being fearful during elections and being careful in what they say in public. Throughout this analysis, whether looking at various hypothesised indicators of active citizenship or social capital, the divide between the rural and the urban in Zimbabwe looms large, and was strongly confirmed by the factor analysis. The various iterations consistently indicated two main factors: one supporting ZANU PF, one supporting the MDC-T, and a third 4 Rural [41 yrs.) versus urban (35 yrs.); p=0.0001 20