Data were gathered through interviews with key informants, interviews with farmers with the help of questionnaires, and structured interviews with elders. Three different questionnaires were designed for farmers. One was for those who had expressed interest and started the construction of an individual pond1 One was for those who preferred to have a communal pond2 and, one was for those who did not show any interest in fish farming. For the first two questionnaires, a set of questions was included for those who have abandoned the construction of the pond.
From all the villages in the two areas information was gathered on the number of households and whether they were male or female headed.
All farmers who had shown interest and taken up fish farming, regardless if they had abandoned afterwards, were interviewed. From the remaining households a stratified sample was taken, taking the percentage of female headed households into account. For the villages with 30 or less households a 30% sample was taken among those who were not interested in fish farming. For villages with more than 30 households the sample was 20%.
Four key informants were interviewed in Rukuzye and in Magwero Area. A total of 68 interviews were carried out in Rukuzye. 34% of the respondents were female. In Magwero 71 interviews were conducted. 16% of the households are headed by females, while 14% of the respondents were female (see Appendix 3). The figures given in the following chapters are based on the results of these interviews unless mentioned otherwise. Structured group interviews were carried out with elders in the villages. Eight were held with men and six with women.
An incidental consequence of this study is, that by cross checking the outcome of the different methods of data collection, indications were obtained on the reliability of the methods. It can thus provide guidelines for future use by, for example, extension agents. In Appendix 4 the results of the different methods are discussed.
Three statistical independence tests were used to analyze the data: Chi-square test, the Mann-Whitney rank sum test for two independent samples and the t-test (two-tailed). When a significant difference (sign. level 5%) was found between two variables, the association coefficient was calculated. Two association coefficients were used, depending on the nature of the data. For nominal data Jule's Q was calculated and for the ordinal data Goodman and Kruskal's gamma.
When significant differences were found between the socio-cultural and economic environment of Rukuzye and Magwero Area, separate statistical tests determining the difference between adopters and non-adopters were carried out for each area. Consequently, the influence of the environment on the factors distinguishing adopters from non-adopters, was eliminated as much as possible.
1 An individual pond is a pond owned by one household only.
2 A communal pond is owned by two or more households.
