What about undeclared voters? Important, too, are the 20% of citizens who did not declare a voting intention. This group is made up of the 15% who refused to answer the voting-intentions question and the 5% who said they did not know whom they would vote for.8 Yet the choices of undeclared voters at the ballot box could have a decisive effect on the outcome of the presidential election. So how might they vote? Some speculative estimates are possible, as follows: If undeclared voters split their votes in the same proportions as declared voters (i.e. 40% for the ZANU-PF and 37% for the MDC), then we estimate that the incumbent candidate could win on the first round. But what if, as seems more likely, undeclared voters lean toward the opposition? What proportion of these reticent citizens would have to vote for the MDC to ensure the opposition a majority of votes (50% plus one) on the first round? According to our calculation (see Box 1), the MDC would have to obtain about twothirds (66%) of undeclared votes in order to secure a first-round victory in the presidential election. In our opinion, this prospect lies within the realm of reasonable possibility. Box 1: How many undeclared voters would swing the election? To win on the first round of the presidential election, any party needs a minimum of 50% of the votes plus one. Out of the July 2018 survey sample of 2,400 citizens, 2,044 reported that they were both registered to vote and likely to vote. Within this subsample, any party would therefore require the votes of at least 1,023 respondents (50% of 2,044 plus one). The combined number of respondents who declared that they intended to vote for the MDC (party plus Alliance) was 750 (36.7%, or 37% when rounded). To achieve a total of 1,023, the MDC would need to attract an additional 273 vote commitments from among the population of undeclared voters. The total number of undeclared voters in the survey sample was 413 (304 “refused” + 109 “don’t know”). Therefore, in order to win the presidential election on the first round, the opposition MDC-T/A would need 273 out of 411 additional votes, or 66.1% (rounded to 66%) of all undeclared voters in the sample. Voting intentions: Parliamentary election How about the race for the House of Assembly? Again only the stated voting intentions of registered and likely voters were considered. In this race, a slightly wider gap separated the parties at the national level. As of early July, ZANU-PF candidates attracted the votes of 41% of survey respondents, and MDC-T (party and Alliance) candidates drew 36% (Figure 10). Since this 5-percentage-point margin lies outside the survey’s confidence interval, it seems likely that the ZANU-PF was on course to maintain a parliamentary majority. The distribution of intended votes by province was roughly the same for the House of Assembly as for the presidency, with one exception: In Matabeleland South, the ZANU-PF, rather than the MDC, appeared ahead in the parliamentary race. 8 Going forward, we exclude the 1% of registered voters who said that, after all, they would not vote. Copyright ©2018 9