Negotiating Reproductive Outcomes Survey 1995-1996

The NRO sample was designed to provide estimates for women and men in Lira and Masaka separately. It was also designed to allow estimation for urban and rural areas within each district.

Sample Eligibility
In order to complete a full interview, a woman had to pass three eligibility criteria. She had to be a regular resident of the household. She had to be between age 20 and age 44 in completed years. Finally, those women meeting the age and residence criteria were asked a series of introductory questions about marital status. Within the accepted age range, women who reported themselves to be "married" were automatically considered eligible to complete the full questionnaire. Unmarried women were asked to complete the full questionnaire only if they reported being in a conjugal relationship lasting six months or more. The rationale for the six-month cutoff was that non-marital, short-term relationships would be less likely to involve negotiations about long-term issues of family formation, family planning, and so forth. Teenagers were excluded on the same grounds; even in a young-marrying population, it was thought that the sample would yield a sizeable proportion of short-term, uncommitted relationships.

A different set of eligibility criteria were set for men. They were required to be partners of eligible women, either formally married or living with a woman. No age criteria were set. Residence criteria depended on marital status. Any married or unmarried partner living in the same household with an eligible woman was considered eligible to answer the male questionnaire. Husbands living in a different residence were still considered eligible, and interviews were attempted if the husband could be located within a reasonable distance of the survey area. If the woman was not married and her partner lived elsewhere, however, he was ruled ineligible (to protect the confidentiality of both partners), and no attempt was made to trace him. Men with multiple wives living in the same household and meeting the other eligibility criteria were administered separate questionnaires for each wife. In general, locating males for interview, whether they were resident or not, proved to be the most difficult and time-consuming part of the fieldwork, requiring multiple visits and visits at irregular times in the early morning or late evening.

Sample Design
The sample was selected in two stages. At the first stage, census enumeration areas (EAs) were selected systematically with probability proportional to size in the 1991 census. In order to take advantage of the household listings assembled for the recent Uganda DHS, all of the DHS EAs in each district were included. The selection proceeded as follows: if 5 EAs were selected in a district for the DHS survey with a selection interval I and the NRO sample required the selection of 10 EAs, then the NRO sample was selected by reducing the interval by half (i.e., I/2) and maintaining the first random selection as in the DHS sample. At the second stage, households were selected systematically within each EA.

A random stratified sample of 40 enumeration areas was selected from each district. Due to the tendency of Masaka EAs to be larger than Lira EAs, a higher proportion of the total sample was expected from Masaka compared with Lira. In order to obtain adequate representation of urban areas, urban areas were oversampled. In Masaka district, with a population that was 10 percent urban at the time of the 1991 census, 20 EAs--or half of the sample--were drawn from urban areas.

Urban areas in Lira also were oversampled. With 5 percent of the population categorized as urban at the time of the 1991 census, 16 out of the total 40 EAs in Lira were selected as urban. The selection procedure in Lira was altered to adjust for varying definitions of "urban" in Uganda. The Department of Statistics in Uganda defines urban in one of two ways. The first is based on a set of objective demographic criteria taken during every decennial census; these include a population of over 10,000 people, access to roads, water supplies, schools, and related "urban" amenities. The presence of such amenities is determined prior to each census during the mapping of enumeration areas. The second way to achieve urban status is for an area legally to register itself as a city or town. At the time of the 1991 census, many northern districts, including Lira, were never mapped due to local political instability. In the absence of mapping to establish demographic criteria for urban status, Lira town is the only officially recognized urban area in Lira district; its status is based on legal registration. Because Masaka was mapped prior to the 1991 census, the two districts have asymmetric definitions of urban areas.

To improve the comparability of the definitions of "urban" between the two districts and to avoid oversaturation of the one official urban site in Lira, a secondary set of potential urban sites was chosen. A list of the 12 largest trading centers outside Lira town was compiled using the 1981 census records. Six of these were selected at random and included in a kind of second tier, "small urban" sample. The remaining 10 urban EAs were drawn from Lira town.