(introduction...)

The first phase of the UNRISD/UNDP research consisted of collecting, reviewing and synthesizing available information relevant to gender and NTAE expansion. The results of this phase were summarized above.

It is clear from these findings that the interaction between gender and NTAE expansion is a complex one, and that, while some dimensions of this interaction are fairly well understood, data are currently insufficient for illuminating other areas, and a number of essential questions are just now beginning to be raised. There is little information available on the agricultural division of labour between women and men in different types of households, or on access to and control over production resources and benefits within the household. Many surveys do not distinguish between male and female-headed households, while some ignore female households altogether. While some of the national data sets reviewed (Balihuta, 1997) had data disaggregated by sex, planners at national and sector levels tend not to use this information. They often use the aggregated data and develop plans in terms of broad categories such as “people”, “communities” or “farmers” - rendering the sex disaggregated data redundant. The concept of gender remains foreign to many planners, who do not seem to be comfortable with programming using gender-disaggregated data.

The second and third phases of the research sought to shed further light on some of the questions raised in the first phase. They involved fieldwork in selected villages in two districts: Kitanyatta, in Masindi District, and Gonve, in Mukono district. First, a participatory rural appraisal (PRA) exercise was carried out in July 1997 in the two villages to explore the local assessment of local conditions and problems. The focus group discussion and preference ranking methods were used to provide insights into men’s and women’s conception of their livelihoods and the constraints that they face as farmers, their explanation for those constraints, and their means of coping with them.

The results of this qualitative part of the study provided both the indicators and the focus of the third phase, which was a questionnaire survey in the same villages, carried out in November-December 1997, in which 396 households participated. The survey was a rather narrow one, focusing on household characteristics, supply response issues, food security and workloads. The sample design endeavoured to include all types of households, which were stratified into low-, medium- and high-income categories. Data collection procedures at the household level were borrowed from Tibaijuka’s (1994) activity profile. A village sampling frame already existed from the chairman of the village council and the PRA village mapping exercise in the villages, which had classified the household types in the villages. The random sampling method was used to select households within each household type as in table 10, observing the proportion of each type. The child-headed households were so few that they were not interviewed. In the male-headed households, husband and wife were interviewed separately. The polygamous households which were sampled were handled like female-headed households, with each wife interviewed separately. However, the questionnaire had a question that required each respondent to state their relationship with the female or male head as well as the husband’s name, where applicable. Each respondent was asked to indicate the type of household they came from, as a separate question. A combination of these two questions made it possible to trace households that shared the same male head.

The use of the case study method - which was dictated by time and funding constraints - limited the generalizability of the findings, and the lack of baseline data was an added disadvantage. Unfortunately, some of the survey information suffers from a high level of missing data, non-response or internal inconsistency. Thus we were not able to use, for instance, data on field size. Acreage, sale and price data are problematic, presumably because respondents were being asked to remember details of the previous year’s harvest. The other data, including labour data, were judged to be more robust.