ZIMBABWE ELECTION SUPPORT NETWORK Quality Control and Data Analysis • ZESN also conducted spot checks to assess whether there was proper identification of EAs, sampling and other challenges • The database was constantly monitored and enumerators and supervisors were requested to verify any inconsistent data. Quality control In order to ensure quality control, the following measures were taken: • Each team was allocated a supervisor who had to assist with identification of EAs. The Supervisor also had to randomly back-check households where interviews were conducted. This proved useful as a means of ensuring that enumerators comply with sampling procedures for both households and individuals. • Teams conducted daily de- briefing sessions to discuss field challenges, what went well, lessons learnt and to plan for the next areas. Supervisors had to submit a daily field work report at the end of the day informing ZESN on any field work related problems and challenges. Data Collection and Analysis Field enumerators where deployed for 8 to 12 days to conduct field work. Data collection was done using Open Data Kit (ODK) Collect. 7 Data was analysed using R , Statistical Package for Social Sciences (SPSS) v 24, Excel, QGIS and ARCGIS 10.3 to produce graphs, tables, charts and maps. Limitations and Challenges During data collection, ZESN faced a number of constraints that posed difficulties in the data collection exercise. These included: all communities were receptive to the ZESN enumerators, for example Mudzi in Mashonaland East. i. Locating voters'- list-to-people It was difficult to locate registrants who had migrated/moved to other areas, as it was difficult to obtain details of their new location. It was also difficult to locate female registrants, especially those residing in the rural areas, who are primarily called by their husbands surnames or by names of their first-born children, as a customary practice, yet they are registered in their maiden names. In most cases the ZESN team of enumerators would seek the assistance of the village head, with a register of people resident in his/ her village. The ZESN enumerators, also encountered incidences were respondents requested them to seek permission to conduct the interviews from Village heads, chiefs and ward chairpersons of political parties. Hence the enumerators ended up spending more time than planned, in the affected areas seeking for approval. iii. Limited time The data collection was conducted within a limited timeframe, therefore piloting could not be conducted and a smaller sample was selected for both field tests. ii. Political Context Whilst, space has opened up in the past months, the data collection was carried out during an election period. In places where campaigns and rallies were being conducted by political parties it was difficult to locate both people-to-list and list-to-people respondents, as they would be attending political party campaign and rally events. Additionally, not 7 iv. Missing data The absence of data effectively reduced the sample for analysis. ZESN, had targeted 1,544 and managed to obtain responses from 1,518 for the people-to-list test. For the list-to-people ZESN had targeted 1,200 and managed to locate 1,063. R is a free software environment for statistical computing and graphics. It compiles and runs on a wide variety of UNIX platforms, Windows and MacOS, < https://www.r-project.org> 17 FINAL REPORT OF THE 2018 VOTERS' ROLL AUDIT