that would distinguish an elite, and this might well be exclusive of “lived poverty”. Here we hypothesized that access to amenities such as piped water and toilets in the house might be a better differentiating factor between groups. For example, according to the 2012 Census, 18% of Zimbabweans have piped water inside their houses and 33% use flush toilets (ZimStat. 2012). Secondly, and since there are demonstrable differences in attitudes between rural and urban Zimbabweans, we wanted measures that might not be polluted by residence since we hypothesize that elites could just as easily be rural as urban, such as those now occupying commercial farms. Thirdly, we wanted measures that tapped into differences between elites and others in the ways that they acquired information, and here access to the internet, social media and other types of information. Fourthly, we hypothesized that forms of ownership might distinguish an elite. However, we did retain both education and employment as variables in constructing an index of elitism as we hypothesized that these were likely to be crucial features of an elite group. Thus we used nine questions in four groups from the AB Round Seven (2017) to construct the index (See Appendix 1). The data was converted into binary variables in order to contruct an overall score for “elitism”. Thus the final index was composed as follows: Elite = News(2) + Ownership (1) + Education (1) + Employment (1) + Amenities (2) This gave a maximum score of 7, with the mean score on the index for the 1,200 respondents to AB Round Seven (2017) was 1.91(sd. 2.37). Although this may have been arbitrary, we took a cut-off score of 5 as representing “elitism”. This gave us two groups: an elite group of 182 (18%), and non-elite group of 1018 (82%). This cut-off seemed satisfactory, given that the Census indicated that 18% of Zimbabweans had access to piped water in their homes The scale had average reliability (.620)1. Additionally, the items showed good correlations between them, except with going without income, but reliability was not improved by removal of this item. Constructing measures of political agency In order to test the research question, we constructed a number of indices of political agency. These were hypothetical constructs, derived from the questions in Afrobarometer Round Seven (2017), and based on variables frequently described in the literature (See Appendix 2):    1 Voice – the ability to express one’s views in public around political issues: feeling free to express one’s views and criticise officials; Community Participation – the ability to participate in public activities: belonging to organisations and participating in community activities; Political Participation – the ability to participate in political activities, with elections as the chosen vector: discussing politics, voting, and participating in electoral events; Cronbach’s Alpha. 7