Afrobarometer. The Afrobarometer data also shows significantly more females than males reporting affiliation with an Apostolic church.3 Table 2: Rank order of frequencies of Apostolics -Inter-Censal Demographis Survey (2017) compared with Afrobarometer Round 7 (2017) ICDS AB4 Manicaland 2 3 Mashonaland Central 1 3 Mashonaland East 3 10 Mashonaland West 5 5 Matabeleland North 5 6 Matabeleland South 4 6 Midlands 8 1 Masvingo 7 2 Harare 9 6 Bulawayo 10 6 This difference between the two data sets provides even greater problems when any extrapolation is made against the voting pattern in the 2018 elections. For example, on the ICDS data, it can be assumed that since there were 5,575,659 voters on the final BVR roll, this would mean that nearly 2 million (1,890,148) of these were “Apostolic”, and, if partisan to only one party, then a huge advantage for that party. If we take the lower, Afrobarometer, rate (12.2%), then the advantage is rather less, 680,230 “Apostolics”, but still a significant factor in a very close election. Of course, this all speculative in the absence of knowing anything about voter preferences in relation to religion, and even more knowing actually how people voted. This latter cannot be derived from the publicly available information on the 2018 elections, but we can know something about political party affiliation and other aspects of political participation using the Afrobarometer data. We decided to test these issues by examining the data from the Afrobarometer Round Seven (2017) survey. We used the Afrobarometer estimate for Apostolics in preference to the ICDS estimate as it seems more realistic. Methods First we developed a code book of relevant questions.5 All responses were re-coded as binary variables. A number of indexes of participation, with good reliability, were constructed as follows:6  Lived Poverty: how often have gone without the following: food, water, medical care, fuel for cooking, cash income;7 Apostolics versus other religions: χ2=176.6 (p=0.0001) There is a negative, non-significant correlation between the two data sets; Pearson’s r= -0.073. 5 The code book is available on request. 6 Cronbach’s alpha: Lived Poverty (.89); Social Capital (.94); Political Fear (.97); and Political Trust (.996). 7 Lived Poverty was included in order to check whether possible differences between rural and urban respondents might be confounded by poverty since Zimbabweans have suffered severe effects from the economic decline in the past decade or so. 3 4 3

Select target paragraph3