APPENDIX 2
SAMPLING METHODOLOGY

1. SURVEY 1: FISH EATING HABITS AND ATTITUDES

Definition of Household

A total of 343 households at the seven selected dams were interviewed. For the purposes of the survey the household was defined in terms of people considered resident in the lolwapa, or family compound and included not only people currently living there, but also household members working elsewhere. It is important to point out that households in Botswana are actually partial households because members who belong to the economic and social unit known as the household may be scattered around different locations.¹ A typical household has a basic core, usually situated at the place of primary residence (usually the village), with different segments revolving around the different economic zones: village, lands, cattlepost, Towns in Botswana and South Africa. Each of these zones is characterized by a major economic resource, such as water at cattle post, land at land areas and social services and employment in villages and towns. For this reason, absent members were included as part of the household.

Questionnaire Pre-Testing

Survey 1 was pre-tested at Mathububukawane, a village on the banks of the Marico River. After discussions with enumerators, adjustments were made to the questionnaire. Time taken for each interview was recorded to enable estimates to be made of the number of households that could be interviewed in the time available for the survey, with allowance made for a decrease in interview time as enumerators became familiar with the survey form.

¹ For an excellent explanation of households in Botswana, see C. Kerven in Field, R., Patterns of Settlement at the Lands: Family Strategy in a Variegated Economy, Report to the ALDEP Team, Ministry of Agriculture, Botswana, 1980

Selection of Sample Size by Area

The Pilot Project Workplan laid down the following guidelines for minimum sample size based on 1981 Population Census figures and taking into account resource and time constraints:

Population and household counts collected during 1981 Population Census therefore formed the basis of selecting sample size. However, the population in the southern and eastern areas of Botswana are expanding very fast in terms of total population as well as rural-urban drift. It has been estimated that the national population growth rate is 3.5% p.a.; this growth rate was used to estimate 1989 populations at the selected sites.

Attempts were made not only to adhere to the minimum sample size set out in the Workplan guidelines, but to increase sample size where possible. In the case of Gampudi and Gakgatla, there were not enough occupied households to enable minimum sample size to be maintained.

Sample Selection

A quasi-random sampling method was chosen for selection of households in Survey 1 as it was considered the fastest and most appropriate sampling method given the time available.

Household counts from the 1981 Population Census were used together with a rough visual counts of the households in the smaller villages or lands area. Once the number of households had been estimated the following sampling fractions were used:

Households were counted from left to right along a track. Where a household was unoccupied, it was agreed that the next household on either side would be approached. Any member of the household over 16 years could be respondent, but preference was given to the household head or their spouse, if either was present.

For the larger villagers of Kanye, Molepolole and Moshupa, each day enumerators were located in different areas at varying distances from the dam. This enabled a broader coverage of households in these villages, which are spread out over a wide area. It was felt that confining enumerators to one area of each village would not have been representative because distance from household to dam varied greatly.

Enumerators

The services of three trained enumerators from the Ministry of Agriculture were employed for the duration of the survey work.

Time Taken For Interviews

On average, the interview lasted 15 minutes for households that did eat fresh fish, and between 8 and 10 minutes for households that did not eat fresh fish.

2. SURVEY 2: Fishing Activity

Questionnaire Pre-Testing

No fishermen were found during the pre-testing, so the questionnaire could not be pre-tested. However, after discussions with the enumerators, the questionnaire was altered slightly.

Sample Selection and Size

As no information was available on the fishing population in the pilot project area, it was not possible to select a sample. All fishermen seen fishing at any of the dams during the visit of the survey team were interviewed, or if other fishermen or members of the village identified people that caught fish, they were interviewed at their home.

Leisure fishermen were interviewed at weekends at the dam or at their place of business during the week. They were identified by asking villagers as well as other leisure fishermen. As the survey was carried out in summer when there was a great deal of rain, the number of leisure fishermen actually fishing at the dam was small.

The total number of fishermen interviewed was 37, of which 7 were leisure fishermen.

Enumerators

One enumerator was chosen to interview at the dam site during the survey visit. All enumerators were also responsible for interviewing fishermen at home, if they were not at the dam and had been identified by other members of the village. The Consultant socio-economist interviewed leisure fishermen.

Time taken for interviews

On average the interview lasted 15 minutes.