ZIMBABWE ELECTION SUPPORT NETWORK The table below shows the number of actual interviews conducted for each province. People-to-list The sample was based on the projected population of adult Zimbabweans aged 18 plus (see table 2 below), that is the voting population up to 2018. The 2012 population census data was used as the base population to project the 2018 voting population. The survey sample was designed to be representative of Zimbabwe and is stratified by urban and rural strata, province and district. The sample was allocated using PPS; the measure of size being the 2018 18+ projected population of provinces. Table 3: Number of interviews conducted Province Bulawayo Harare Manicaland Mash Central Mash East Mash West Masvingo Mat North Mat South Midlands Total PPS was also used to select enumeration areas (EAs) from Zimbabwe Sampling Frame (developed after the 2012 census). The measure of size for EAs during selection was number of persons as per the projected population census. The table below outlines the allocation of enumeration areas for the 10 provinces. Urban EAs 11 5 1 3 6 1 1 7 2 34 71 Rural EAs 0 19 16 17 16 9 9 16 18 2 122 Sampled Individuals 90 300 191 138 168 150 170 60 81 170 1,518 % Sampled Individuals 6 20 13 9 11 10 11 4 5 11 100 Primary sampling units (PSU) At the first stage the sampling frame is made up of EAs. EAs are the smallest geographical units that consist of about 100 households. Each EA has a unique 10-digit geo-code that reflects the province, district, ward and land use sector in which it is located. For the purpose of survey, the sampling frame excludes non-household population, people residing on state land (national parks, safari areas, etc.) and in institutions, and these account for less than one percent of the population. Table 2: Allocation of Enumeration Areas Province Bulawayo Manicaland Mash Central Mash East Mash West Mat North Mat South Midlands Masvingo Harare Total % Registrants 5 16 13 9 11 11 11 6 5 13 100 Total 11 24 17 20 22 10 10 23 20 36 193 Secondary sampling units (SSU) Two approaches were considered. The first one requires that, prior to every survey, a list of households in the selected EAs is established. This is used as the frame for the second stage systematic selection of households. However, because, there was no list of households for this stage of sapling, Systematic Random Sampling was used to select 8 households in each EA. The enumerators used a 5/8 interval to randomly select households in the urban areas. This entailed skipping the 5households, with counting starting from the right for the first house and the second house the enumerators would skip 8 households. In the rural areas because of distance and that the households are not linear, the enumerators skipped every 3rd house. Respondents were selected using simple random sampling or lucky draw. 193 enumeration areas across the country were selected at random. For each sampled enumeration area, enumerators were instructed to identify eight individuals for the people-to-list using the lucky draw method. For the people-to-list enumerators were instructed to identify 50% female and 50% male respondents. In total, the people-to-list sample should have included 1,544 respondents of whom 722 were women and 722 were men. In reality, the enumerators were able to conduct a total of 1,518 interviews for the people-to list test of which 55% of the respondents were women and 45% were men. 15 FINAL REPORT OF THE 2018 VOTERS' ROLL AUDIT