However, we did note some anomalous findings:      Urban Zimbabweans are more risk averse than their rural counterparts; Older Zimbabweans are more risk taking than younger Zimbabweans; The unemployed are as risk averse as the employed; Better educated Zimbabweans are more risk averse than the less educated; Finally, ZANU-PF supporters are less risk averse than the supporters of MDC-T. These are findings that might be not be expected and deserve further attention, for the explanations are not immediately obvious. An earlier study found that a simple classification of the Zimbabwean citizenry found four main groups: ZANU-PF supporters, MDC-T supporters, “active citizens” (those not admitting political affiliation, but evincing voice and participation), and “disconnected democrats” (urban, educated and employed) (RAU.2015). This did suggest that a disaggregated view of Zimbabwean citizenry was warranted. For example, the middle-class, “disconnected democrats”, usually the defenders of democracy, were evidently not, a finding replicated at least for Zimbabwean women (RAU. 2016(a); RAU. 2016 (b)). These findings led us to speculate that the critics of the Afrobarometer might well be “armchair critics”, relying mostly on personal experience and anecdote, but basing their criticisms on theory rather than fact. The Afrobarometer offers “fact”: quantitative, survey fact it is true, but nonetheless one of the very few sources of “fact” on the views of Zimbabwean citizens other than their voting preferences in elections. And given the enormous and continuous dispute about the validity of Zimbabwean elections, voting patterns seem unreliable evidence, but it can also be pointed out that careful analysis of elections using survey data can be useful in understanding citizen choice (Bratton et al, 2016). Thus, we attempt an analysis of a group that might be very influential in shaping public opinion, but perhaps less qualified to do so than might be assumed. We term this group an “elite”, but it may equally be termed middle-class. The simple question we ask: are the views of the elite grounded in their experience of engagement in the general political life of Zimbabwe? Methods Constructing a measure of “elitism” In order the answer this question, we first needed to determine whether we could construct a measure of “elitism”. Studies elsewhere have undertaken this exercise, but explicitly in trying to analyse the middle class (ref). Here, we draw on this previous work (Cheeseman. 2014), as well as a previous study of middle-class women (RAU. 2016), but extend this by reference to other aspects of what might define an “elite”. In the previous study, using Cheeseman’s index for middle class in Kenya, we defined middle class through an index based on questions from the AB around wealth, employment, education, and lived poverty. Whilst this seemed to give a useful measure for class, it did not seem wholly useful for the present study for several reasons. Firstly, the economic situation in Zimbabwe has become so severe that “lived poverty” might well be too inclusive, and here we wished to find measures 6