After spending two weeks chatting with villagers and testing tentative questions, I have finalized my questionnaire and interviewed 13 members of the SACCO. Here are my reflections on field testing and surveying:
Field testing helps a surveyor weed out useless and ambiguous questions. A question that yields homogenous answers across the survey population is redundant and should be omitted. For example, a question about the source of drinking water will fail to convey any useful information because almost all households in this town get water from nearby wells. If many respondents have trouble understanding the same question, you should check if this question conveys what you intend to ask in an appropriate vocabulary.
Before conducting your survey, you should gather enough information about your intended interviewees. The more detailed contacts the better. Call the respondent before you visit his/her home, and set up an appointment when necessary.
If your translator asks the questions, you need to make sure he/she interprets the questions correctly and can convey the meaning in the local language. Ask your translator to translate every sentence the respondent says. A translator, especially one that is not trained in surveying, tends to omit a lot of information and keeps what he/she thinks is useful. Detailed translation will help you avoid missing information and allow you to ask clarification or follow-up questions in time.
A side note:
I interviewed a 28-year-old woman with 7 children. She started having children when she was 18 and had one child approximately every 1.5 years. This reminds me of what my colleague at the bank told me: as a Ugandan, if you have too few children, you are not blessed. I am sure this family is filled with blessings. Unfortunately, several of their children have displayed signs of malnutrition. The good news is the children seem to be independent and taking care of each other, especially the older ones.