Tuesday, July 14, 2015

Survey: Perceptions of NGOs and the electoral process in Côte d’Ivoire

One of the major components of my research project is carrying out a public opinion survey about the electoral process, perceptions of NGOs and democracy in general. Due to budget restraints, I decided to limit the survey to Abidjan only, but with a representative sample coming from the 13 communes. The team has finished the interviews and are now working on data entry. Here’s how it went down:

First, I had to find a survey firm to carry-out this work. I selected the Afrobarometer implementing partner, the Centre for Research and Training on Integrated Development (CREFDI). I chose them because they came with high accolades from other researchers, but also because they are very familiar with Afrobarometer’s sampling and interviewing protocols as they have been working with Afrobarometer for two rounds (2013 and 2014, rounds 5 and 6).
CREFDI staff, 4 June 2015
Second, I developed the survey (well, over the course of the year during my classes, specifically a survey design course) and with input from COSOPCI. I then had six Ivorian university students pilot the survey online, to help with the response suggestions and survey flow (special shout out to my husband as well who corrected my French).
Piloting the survey, 27 June 2015
Then it was onto training the enumerators. I worked with a great team of eight enumerators, two supervisors, two data entry folks, and of course the director of CREFDI. On training day we went through the survey question by question, responding to any problems or comprehension issues, and distributing the 520 questionnaires to the teams. The questionnaires were printed, because, although ideal, tablets or phone surveys were outside of my budget.
Enumerator training 3 July 2015

Me and the team 3 July 2015
The questionnaire was administered between July 5 and July 12. Each team of four was overseen by a supervisor, and divided up the 65 enumeration area between the two teams (at random). An enumeration area, ZD here, are determined geographically by the Statistics Institute (INS) in charge of carrying out the census. The ZD are comprised of 1,000 to 1,200 individuals on average. Following Afrobarometer protocol, 8 respondents would be interviewed per ZD, thus, I needed 65 ZD for this study in order to reach a 520 respondent sample. The distribution of the ZD was determined after the 13 communes of Abidjan were stratified based on population; thus Yopougon, the largest commune, was allotted 15 ZD, while Treichville and Plateau, the two smallest, were allotted one ZD each. The ZD were then selected at random by CREFDI. Each team, then, would complete 40 respondent interviews per day, meaning five ZD per day, 10 interviews per enumerator per day. Because the ZD were chosen at random within each commune, sometimes the teams would start in the north part of the commune and then have to take a Gbaka or Woro-Woro to get to the other side of the commune, especially in the two largest communes of Yopougon and Abobo.
Abidjan, Communes not included on this map: Anyama, Bingerville, and Songon. http://data6.blog.de/media/070/5498070_544c450bb6_m.jpeg
I went out with each team for a day to see how things were going and to observe any difficulties they encountered. To determine which household to interview, the team followed the Afrobarometer sampling protocol: they would meet at a randomly selected intersection, one interviewer would walk East, another West, another North, and another South. They would select the fifth household that they encountered. If there was someone home (which was more often the case than not), they would give a brief recruitment speech explaining the purpose of their visit, show their CREFDI badge to demonstrate their credibility, and ask to make a list of the members of the household in order to randomly select the person to be interviewed. They alternated by household whether they would target a woman or a man for the interview (all respondents had to be over the age of 18). From the list they would have the member of the household draw a number, and that person was the person selected to be interviewed; if they were not home or if they refused, the enumerator was to leave the household and continue counting. If they agreed to participate, they had to give verbal consent, and the survey could commence.

Interview in Yopougon, 7 July 2015

Interview in Koumassi, 7 July 2015
From witnessing this process, I learned a few things:
  • A lot of people live in households in Abidjan; sometimes the people who lived there were uncertain about the number of people who were residing in a home. 
  • People do not particularly like divulging their age. It was the first question (because if they were not an adult, they could not complete the interview), and the enumerators were good at convincing people to give their age (“it’s so we can make sure you are of voting age”). People thought that asking age was not anonymous, which I still don’t completely understand. 
  • Most people like to talk, and one of the downsides of public opinion surveys with mostly closed questions is that their thoughts are not necessarily recorded. One respondent in Koumassi said he would not participate in the elections because he was afraid. The enumerator coded this response on the survey and continued onto the next question, but the man asked “wait, don’t you want to know why I am afraid?” and then told us stories of what had happened to him and his family in 2010-2011. 
  • On the other hand, there are some people who do not want to talk. Women in particular were reticent and often responded with I don’t know to many of the questions. 
At lunch with the first team, they asked me if in the U.S. there was a tradition about inviting someone into your home. I said no, not really, I mean I guess you could offer them something to drink? The enumerators were surprised: “if you don’t offer someone water and ask about the ‘nouvelles’ (news), it is clear to the visitor that they are not welcome.” They said that outside of Abidjan the welcoming experience varies, with some ethnic groups offering all you can drink liquor or Kola nut to guests in addition to water. At each interview, we were welcomed in, even before the household member agreed to participate, offered water and offered a seat. 
Discussing welcoming protocol over peanut sauce lunch, 3 July 2015
The data entry team is now busy at work inputting the responses from the 520 interviewees. I expect to be able to analyze the data, at least preliminary data, this weekend; I will hopefully give a dissemination presentation of the descriptive statistics to COSOPCI and CREFDI before my departure; two more weeks, so a lot of work to do!

No comments: