In June and July of 1994, semi-structured interviews were carried out with 10 men and 10 women in each of the four sites targeted by the KDICP to inform the questionnaire, as well as to provide information on the content of conversations within social networks.

The sampling procedure for the interviews is discussed below. Transcripts of the semi-structured interviews carried out with women (with respondents’ names and other indentifying information removed) are all available for downloading.

Since there was no list of inhabitants to use for selecting couples, an ad hoc procedure was developed. Because social networks were a focus of the study, we attempted to ensure that respondents would on average be distributed equidistant from the places of more intense social interaction, i.e. the village center and the road (if there was one). The interviewer team began at the center of the village, and counted dwellings in a pre-determined direction and for a pre-determined number of dwellings. Only houses within the boundaries of the sublocation were counted, and if the interviewers reached the outer edge of the sublocation before they got to the predetermined house, they would go back and take the other direction. (The sub-chiefs told the interviewers when they had reached the sublocation boundary, as it was not evident visually). Only structures which the interviewers could visually, from a distance, classify as a building in which people did not live were not counted; questionable cases were rare. If, in the process of going to the house, another house within 50 meters of the road but not visible from it which would have been between the last and the penultimate houses became visible, that one was taken instead. If, when that dwelling was empty or had no one who met the sampling criteria (a woman of reproductive age, her husband or another man of an appropriate age), other people within the compound were questioned as to whether the eligible person(s) would be back soon, and if possible, someone was sent out to look for them or a future interview was arranged with them through another family member. If a house did not qualify, or no one in it could be contacted, interviewers would move to the next dwelling. At regular and pre-determined intervals interviewers returned to the road or to the intersection at the center of the village, and then set off in a different direction, following the same procedures. Interviewers tried to make appointments for interviews, but sometimes interviews were carried out on the spot. The study director introduced the interviewers and explained the aim of the project, usually in terms of research on health and family planning.

Eligible respondents rarely refused an interview, but one or both members of a couple were often away. This probably led to some underselection of the busy and socially active.

Interviews were conducted in Luo and taped. Shortly after the interview, the interviewer simultaneously transcribed and translated the interview, which was then immediately typed. The rapidity with which the interview was available made it possible to refine the questionnaire guide, especially during the first week in the first site. To check the quality of transcriptions and translations, five interviews were transcribed and translated by someone not connected with the project. Although there were occasional differences in wording, the translations were very close. The interviews were coded for analysis by Susan Watkins and Naomi Rutenberg. Tapes are available from Susan Watkins.