2. SURVEY FRAMEWORK

2.1 Scope

The pilot survey addressed the following technical, social and economic aspects.

1. Technical aspects

   • fish farming production practices

   • pond construction

   • disposal of products

   • sources of technical information

2. Social aspects

   • activity profiles

   • access and control over production and benefits

   • food consumption and eating habits

   • attitudes and perceptions towards fish farming

3. Economic aspects

   • economic importance of fish farming for fish farming households

   • other economic activities of fish farming households

2.2 Unit of the inquiry

A fish farming unit (FFU) was selected as the unit of inquiry. A FFU is defined as an economic unit of fish cultivation under single management. Single management may be exercised by an individual, a household, jointly by two or more individuals or households¹, or by a juridical person such as a school, prison or any government or non-government institution (e.g. religious organization). A FFU can include one or more ponds.

¹ There are several Swahili terms for “household”(i.e. “familia”, “kaya”, and “jamii”). “Familia” stands for a nuclear family unit which is composed of one generation, while “kaya” means extended family from more than two generations. “Jamii” has wider concept including those who do not share living space and food together but participate in agricultural or other productive activities.

2.3 Enumeration frame

A complete list of active fish ponds as of 1990/91, was used as a primary enumeration frame. The list of ponds contained names of owners, names of villages, sizes of ponds and species stocked. The list was completed by the fisheries officers or villages leaders.

2.4 Sample design

A stratified two-stage Sample Design was chosen. Primary sampling units (PSU) were wards, which were stratified into three categories: urban, rural and mixed². FFUs formed Secondary Sampling Units (SSU). Sample size was set at 30% and FFUs were proportionately drawn from each district and type of ward. In selecting the wards, the following criteria were taken into consideration: agro-ecological zones, remoteness/proximity to the major roads or town centre, and whether fisheries officers were stationed at the wards. Where necessary, selection of villages were made within the ward. In order to avoid a possible bias towards villages with a concentration of ponds, both villages with the least and largest number of FFUs were selected. Table 2 summarizes the enumeration area of the survey.

² Classification of wards prepared by Bureau of Statistics was used.

Table 1: Summary of results of 1990/91 pond census

Table 2: Summary of survey enumeration

2.5 Survey methodologies

Two forms of precoded questionnaires (form I and form II) were prepared in English and translated into Swahili (English version of questionnaires are included in Annex I). Form I contained Fish Farming Unit (FFU) identification, pond particulars and the most essential technical information. It was administered to all FFUs. Form II includes more elaborate information on the land, labour, fertilizing and feeding practices of the last production cycle and the reasons for fish farming. Form II also contained selected questions about the fish farming households (FFHs), such as the major sources of household income, agricultural holdings, frequency of fish and other animal protein consumption. Form II was administered to 20 % of randomly selected FFUs. Administration of Form I took five to ten minutes and Form II 30 to 40 minutes. Form I was administered to 643 respondents and Form II to 169.

In addition to the precoded questionnaire, an interview schedule with open-ended questions was prepared (interview schedules are included in Annex II). The interview schedules covered more qualitative questions, particularly gender-related information, such as attitudes toward fish farming by gender, activity profiles of men and women in fish farming households and also questions put to farmers without fish ponds. The interview schedules were utilized during informal interviews of individual men and women, key informants and during group discussions with men and women. Respondents for interview schedules were identified on arrival to the villages. At least one interview was made in each village, either individually or as a group, with men or women from non-fish farming households. Whenever appropriate and feasible, separate group meetings with women were organized in order to obtain the women's view, while questionnaires were in most of the cases answered by men. Details of organization of the interviews are discussed in Chapter 5.

Prior to the implementation of field work, a three days' preparatory workshop was held in Songea between 27 and 29 August 1992. The workshop was attended by all members of the survey working group and the field survey team (see Annex VIII for the report of the workshop).

A final seminar held after the survey discussed draft proposals for follow-up projects. The proposals are found in Annexure VII.

2.6 Organization of the survey

At the DOF headquarters' level, a working group was formed which was composed of two aquaculturists, a fishery statistician from the statistics section, Regional Fishery Officer (RFO) and fishery socio-economists. The working group was responsible for overall execution of the pilot survey and analysis of survey results.

At the regional level, a survey team was organized involving all District Fisheries officers (DFOs), supporting staff and field staff. It was headed by a senior aquaculturist. Day-to-day activities of the survey team were coordinated by an RFO. Occasionally, teachers, agricultural extensionists or villagers were employed as interpreters and assisted in the conduct of informal interviews and open ended questions.

The composition of the working group and the field survey team are given in Annex III.

2.7 Data processing programme

A customized data entry programme was prepared and compiled with Clipper (Dbase III plus). Data were stored in three files: POND.dbf, FFU.dbf and FFE.dbf. POND.dbf contains data on individual ponds, FFU.dbf contains information on FFUs, and FFE.dbf contained data on fish farming households. A brief description of the data processing programme is given in Annex IX.

2.8 Limitations of the survey

Due to the “one shot” nature of the survey, seasonal variations were not reflected. The survey was conducted during the dry season and the peak season for fish farming activites after the main harvesting of agricultural produce. Consequently, data on feeding frequency, intermittent harvests and batch harvests seem higher than the annual mean. This seasonal gap must, therefore, be taken into account during interpretation of survey results.

Information on production was often difficult to obtain. Respondents were requested to recall the last production cycle of each of his/her ponds. Memory recall of earlier production cycles should also be read with caution.

The lists of ponds and pond owners of 1990/91 were not always reliable. In Naikeshi village, for instance, names of pond owners were not recognized by the village leaders so that a new list had to be made. The same names were listed under different villages, and some active pond owners were not included in the list. Accuracy of sampling was to some extent influenced by the accuracy of the list of ponds.