Key informants were most reliable in giving information on subjects such as:
physical geography and public buildings (e.g. “Is there a school/church here?” and “Are there places with water in this area?”);
institutions and institutional roles (e.g. “Do you have extension workers here; what is their specialization”? and “Do you have Marketing Boards here?”);
farming activities (e.g. “What crops do farmers grow for home consumption; which ones for sale?” and “When is the planting/weeding/harvesting season?”).
On the other hand, evaluative questions such as, “How do people look upon farmers having fish ponds” or “What percentage of adults work as day or seasonal labourers?” show a much lower degree of agreement (hence reliability). When key informants' answers on questions asking for quantitative data were compared with the results of the survey data in both areas, it appears that cross checking of their information is necessary. The results are summarized in table 1.
Table 1: Comparisons between key informants and survey data
ITEM | RUKUZYE | MAGWERO | ||||
---|---|---|---|---|---|---|
Key info. Survey | Key info. Survey | |||||
Adult literacy | 66% | (8) | 66% | 56% | (21) | 77% |
Female headed households | 31% | (13) | 37% | 26% | (18) | 16% |
Day & seasonal labourers | 41% | (30) | 8% | 45% | (26) | 21% |
Households keeping chickens | 89% | (10) | 75% | 97% | (5) | 70% |
Households keeping goats | 38% | (17) | 28% | 38% | (25) | 21% |
Households keeping cattle | 30% | (25) | 14% | 33% | (26) | 10% |
Standard deviation is given in brackets
It appears that cross checking of their information is necessary. The key informants used were the better educated, better off and more powerful persons in the community with presumably extensive knowledge of their own communities and fellow citizens. In fact, it appeared that many of them were “outsiders” to the communities and could as such not give much detailed information on social relations within villages and people's priorities.
Informal interviews with elders gave important information on the history of the village, important events in the area and the village's attitude concerning certain groups of people (e.g. those who beg for food, day labourers, those who hire labour, innovators) without identifying individuals. This information helped to form a picture of the socio-cultural environment. However, elders and especially headmen, would like to give a harmonious picture of their village. This introduces biases.
It is obvious that care should be taken when identifying informants. Efforts should be made to select representatives of all groups within the communities.
The survey gave valuable quantitative data on the relative importance of certain variables. Nevertheless, its major shortcomings are:
it is difficult to include time in a one-shot survey because of the recall problem of the interviewees. For example, it would have been better if the knowledge on and perception of fish farming had been measured at the time of adoption.
it can show the correlation between several factors, but can not prove causality. The explanation of how factors relate to each other is post hoc and based on information obtained during fieldwork.
There were indications during fieldwork that network analysis could be an important contribution to the understanding of the adoption process. However, reliance on mainly survey data produced inadequate information. A network analysis requires a lot more direct observation than was carried out during this study.