Conclusions The data thus lead us to a number of conclusions. The first is that our hypotheses were all rejected: Elites display much less agency than Nonelites, and suggests that our research question, about how grounded are the views of the elite, has been answered in the negative. Elites may have strong views on the personal level about the Zimbabwean polity, but it does not appear that they could obtain the evidence for these views from experience: they have low community and political participation, low activism, and little political trust. This does not mean that they cannot obtain information from sources other than personal experience, but there remains the question about what empirical sources of information there are about the Zimbabwean polity in general, other than the Afrobarometer. Secondly, and when the indices are disaggregated into their components, Elites show all the aspects of what we have previously termed “disconnected democrats” (RAU.2016):       They do not belong to a community organisation, do not attend community meetings, or join others to raise an issue; They are less likely to vote, less likely to go to a campaign rally, and to work for a candidate; They are less active, and contact duty bearers much less frequently; They are stronger supporters of democracy, but not happy with the democracy they have; They have very little trust in political institutions and offices; Finally, they seem less likely to support a political party. Thirdly, it is clear the Non-elite group is dominated by a large rural subgroup, and it is evident that the influence of this subgroup over-determines the direction of this group’s responses. As seen in Table 2 the rural group is significantly different to the urban group on every measure: it is rural residence that expresses agency in every way. It is also probable that this rural group are largely supporters of ZANU-PF for whom the “fear” factor” may not be relevant. As for the “fear” factor, critics of the AB Round 7 results were vociferous in the view that the survey would not get accurate answers because of this. This seems unjustified on closer analysis of the results. As was seen in Figure 8, non-elite respondents were significantly more likely to see the sponsor of the survey as the government (despite being told it was being conducted by the Afrobarometer, and this did not seem to inhibit their willingness to state a political party preference. By contrast, the elite group, who seem to be largely urban, believed the explanation that the survey was being conducted by the Afrobarometer, and, actually, hardly anyone believed it was being conducted by a political party in either group. Thus, we would discount the thesis that the results of the survey are inaccurate due to “fear”, and suggest rather that Zimbabweans as a whole can be trusted to give accurate reports of their views. As for the criticisms around the predictive power of AB in elections, we can only endorse what Howard and Logan said: 14