A socio-economic survey of Lake Victoria-Uganda fishing communities was undertaken as part of the work programme of the FAO/UNDP FISHIN Project UGA/87/007. The objective was to create an information base which would complement the Bio-Statistical data also being collected and analysed by the Project team. Taken together, the two information bases are intended to serve as a foundation for development planning and management of the Lake Victoria fisheries. The socio-economic survey was also designed to serve as a model for subsequent periodic surveys needed by the Uganda Fisheries Department (UFD) to update baseline information on the national fisheries in general. In the original work programme it was planned to extend survey coverage to the Kichwamba Region in the western part of the country, to take in the fisherfolk communities of Lakes Edward and George and the Kazinga Channel. However, a comprehensive socio-economic survey of these communities had just been completed in 1989 (Infield 1989), and the resultant data set was available for further analysis to the members of the FISHIN Project team. It was decided therefore that it would make more sense to concentrate efforts on Lake Victoria, aiming at a fairly extensive survey coverage in terms of geographical area and sample size.
In order to fill the gap in information on the socio-economic aspects of Lake Victoria fisheries the survey exercise was designed to probe various features and activities at three different levels of local fishing settlement life -- namely, those of the community, of the household, and of individuals and their enterprises.
Priority areas for investigation were identified through a review of relevant literature, preliminary field trips by the socio-economic (SEC) working group, and through informal discussions with local fisherfolk.
Topics for investigation were established as follows:
Overall sizes of fishing settlements and the numbers and types of services/amenities associated with them.
The presence and significance of such institutions as fishing committees, co-operative societies, and the post of Gabunga (Head Fisher).
Participation of local community members in fishing as owners of boats and gear, or as labourers, and the availability of alternative income earning activities.
Mobility of the community residents, including reasons for migration, timing, and destinations involved.
Qualitative and quantitative information on fishing-related enterprises, and their cost/earning performances.
With these topics in mind, plans were made to organise and implement a survey covering a sample of landing communities drawn from each of the five fisheries regions into which Lake Victoria-Uganda is divided (see map, Fig. 1, Appendix I).
After consultation between members of the SEC group it was decided to design three different questionnaire forms. These are reproduced in Appendix II.
SECSURVI - COMMUNITY FORM was intended to elicit background information on various features of sample landing communities including population by gender and total households, recent changes in community size, principal development constraints, inventory of amenities and services, social organisations, women's income generating activities, and major health problems. The form was used in interviews with key respondents in each community surveyed, i.e. such people as Resistance Council (RC) officials, Gabunga (Head Fishers), Fishing Committee members, and elders and other long-term residents. One form was completed for each site visited.
SECSURV2 - HOUSEHOLD FORM consisted of questions to provide background information on residents of households selected as sub-samples of survey communities. Additional features to be enumerated ranged from basic household facilities and assets, dietary patterns, and crop and livestock production. The participation of household members in community activities, their perception of welfare problems, and their plans for the future were other items probed.
SECSURV3 - FISHER FORM questions were directed at those individuals in the household sub-samples involved with fishing or any fisheries-related activity. Basically the form was structured to yield information on the economic enterprises of these individuals.
Initial drafts of the three questionnaires were circulated to other members of the FISHIN Project team, and some revisions were carried out on the basis of the comments and suggestions that were made. The questionnaires were then tested before implementation of the survey at three different landing sites located within the Entebbe Fisheries Region. These particular landings were chosen because, collectively, they represented the diversity of circumstances expected to be encountered within the actual survey sample.
Ggerenge is located 9 kms. down a murram track that leaves the main Kampala road about 5 kms away from Entebbe. It is a small site whose fisherfolk reside both at the landing and in dispersed homesteads within a radius of about 2 km. It has a population of some 150 people who normally work as fishers, fishmongers, boat supervisers, and petty traders. All the structures at this landing are of a very temporary nature. According to local informants the landing was only about three months old at the time of the visit.
Kigungu is a long-established landing situated close to the southern tip of the Entebbe Peninsula. Though included within Entebbe Municipality boundaries, it forms a distinct settlement within itself. There is a dense collection of permanent and semi-permanent houses and shops sheltering a population estimated at around 500 people. The settlement is made up of many individuals who are not involved in fishing activities.
Kangulume Landing lies about 35 km from Entebbe. Access is by both land and over water from other mainland points. There is only one very temporary structure at this site. The community of people associated with Kangulume numbers about 150. None of the fisherfolk actually reside at the landing, but stay rather in homesteads scattered in the vacinity. Farming is an important complementary enterprise to fishing.
Testing confirmed that the three questionnaires were basically suitable to the task at hand. Only some minor changes in format were made, and some rephrasing of certain questions effected.
A manual was prepared for training enumerators and also for use as a reference during the actual field work. Each question was explained and, where necessary, examples of expected answers were given. The manual also provided guidelines to proper interviewing, along with regional maps showing the locations of sample communities and a timetable for the survey activity. The purpose behind the manual was to ensure a thorough understanding of the questionnaires and a standard approach by all concerned in the survey. (see Appendix III).
Sample selection for the Socio - Economic Survey was based on the sampling frame for the FISHIN Project's Catch Assesment Survey (CAS) of the Ugandan waters of Lake Victoria (Malvestuto 1990; Tumwebaze and Coenen 1991). This was done with a view towards standardising information collection to serve the needs of both bio-statistical and socio-economic analysis.
The earlier Project Frame Survey (Tumwebaze and Coenen 1991) established that there are 715 fish landing sites situated along the mainland and island shores of Lake Victoria-Uganda. These landings host a total of 8,674 active canoes (2,242 dugouts and 6,432 wooden planked; 1,250 motorised). This information provided the basis for designing the CAS.
Geographical stratification of the Lake Victoria region had earlier been done by scientists working with the old East African Freshwater Fishery Research Organisation (EAFFRO). The strata were defined on the basis of limnological zones and administrative and logistical considerations, as follows:
| Stratum Code | Administrative Region | Boundary Description |
| IA | Tororo | Kenya border west to the tip of Bwembe Pt. in south Busoga, approx. 34°31' E. |
| IB | Tororo | Islands off of mainland stratum IA, principally Buganga, Sigulu, Dagusi, Kaza, Nainaivi, Vumba, Sagitu, Lolui, Siro and Kama. |
| IIA | Jinja/Entebbe | From Bwemba Pt. west to an imaginary line splitting Nalumuli Bay at approx. 32°57' E. |
| IIB | Jinja/Entebbe | Islands off mainland stratum IIA, principally Buvuma, Bugaia, Buziri, Mpata, Bwema and Kiribi. |
| IIIA | Entebbe/Masaka | From imaginary line splitting Nalumuli Bay west to Mawokkota/Buddu county line south of Bunjako/Katonga Bay system. |
| IIIB | Entebbe/Masaka | Islands off of mainland Stratum IIIA, principally Lwaji, Damba, Kome, Kimi, Nsadzi, Bussi, Zinga and Bunjako |
| IVA | Masaka | Mawokota/Buddu county line to Tanzania border. |
| IVB | Kalangala | Ssese Islands. |
Malvestuto (1990) pointed out that although this stratification was logical with respect to socio-technical and environmental differences, it was necessary to stratify further for purposes of the CAS in order to allow for field staff and logistical limitations. He therefore further divided the 8 major strata into a total of 50 minor strata. A few of the very large landings were assigned strata all to themselves, meaning that there are some strata with only one landing. This was done in view of their obvious importance and because when staff must cover a very large site, it is virtually impossible to take on extra duties at other scattered and more minor sites.
For the Socio-Economic Survey, it was decided that one landing should be sampled from each of the 50 minor strata. This decision was based on the desire to achieve maximum geographical coverage while also allowing for limitations of time, funds, and enumeration personnel. The total number of landing sites sampled was therefore 50, giving a representation of approximately 7% of all the landings on Lake Victoria-Uganda.
Just as in the case of the CAS design, landings were selected through non-uniform probability sampling. A site from each stratum was randomly chosen in proportion to its size in relation to other sites in the same stratum, measured as the number of active fishing boats. This number was taken as a direct indication of the level of economic activity and the size of the population associated with a landing.
What resulted was a sample representing (a) all the areas around the lakeshore, and (b) a range of different sizes of landings, with the larger ones likely to feature more prominently. A list of all the landings on Lake Victoria identified during the Frame Survey is provided in Table 1 (Appendix I). For each landing details are given as to its location, stratum assignment, number of canoes, and “fishing power” (expressed as the number of active canoes it hosts as a percentage of the total number of active canoes in the stratum). Also shown is the way in which a landing's fishing power percentage was converted to a random number range. When selecting a sample using random number tables, a landing's probability of being included in the sample obviously is directly proportional to the size of the range. The larger the range, the higher the probability. Those landing communities actually selected for the socio-economic survey sample are indicated by an asterisk (*) in the rightmost column of Table 1.
The survey assumed the landing community as the primary sampling unit and, as described above, the SECSURV1 form was devised as the data collection instrument. However, since information on fisherfolk households and enterprises was also being sought, it was necessary to draw sub-samples from the communities to be visited. But this presented certain difficulties, in that there was no way to know a priori how many “households” -- units of people who normally live and eat together -- existed in the total community sample. Moreover, there were the inevitable financial and logistical considerations to bear in mind. The decision was therefore made to randomly and with uniform probability select five households to interview from each sample community. This would yield a total of 250 households to visit in all -- a number that seemed manageable in terms of the resources available for the exercise. Selection was to be carried out after compiling a list of all households within a given community immediately upon making contact with the residents. Each household was assigned a separate number which was written on a small slip of paper. After all the slips had been placed in a hat or box and thoroughly mixed up, five were blindly drawn out. The residents of each of the selected households were then interviewed with the help of the SECSURV2 form.
Information on individual fishing-related enterprises was obtained through the simple expedient of identifying those residents in each household who were engaged in such enterprises, and conducting an interview with each through means of the FISHER form, SECSURV3.
Obviously case weighting to correct for the selection probabilities assigned to the sample communities and sub-sample households and individuals will have to be done in the final analysis of the various categories of data collected. This will make it possible to use the survey findings to make broader estimates and inferences about socio-economic characteristics of Lake Victoria-Uganda fishing communities.
Before the actual implementation of the survey, household lists of all the sampled landing sites were compiled with the help of the Regional Fisheries Officers and their field staff working in conjunction with local leaders. These were the lists used in the sampling of the five households on each landing site. In cases where sites did not have a well-defined settlement associated with them, household lists were built up by noting down the names of all people known to be living in the immediate neighbourhood as well as those usually active in some capacity at the landing, whether as boat crew, owners, supervisors, fishmongers, processors, porters, petty traders, or the like.
Enumerators were drawn from the UFD field staff establishment, comprising Fisheries Assistants (FAs), Assistant Fisheries Development Officers (AFDOs), and Fisheries Development Officers (FDOs). Staff were nominated from the five regions according to their work performance records and the convenience of their posting locations in relation to sample communities. The Regional Fisheries Officers were relied on for such information. This ensured that the survey team included enumerators with good local knowledge who could assist both as guides and local informants.
Regions differ in size and the number of enumerators accordingly varied from place to place. Entebbe had 9, Jinja and Masaka each 8, Tororo 4, and Kalangala 3.
One-day Training Workshops for the enumerators were organised at each Regional Fisheries Office. Training was done by the Project Socio-Economic Advisor and the National Survey Consultant, with the assistance of one of the National Project Economists. The regional workshop approach was considered a better alternative to one overall national workshop as it meant that the trainers could work with smaller numbers of people and there would be no minimal delay between the training session and commencement of the actual fieldwork.
The first workshop was held at UFD Headquarters on 25th March 1991 for Entebbe Region Enumerators. This group assisted in making the final corrections in the questionnaire and the Enumerators' Manual (Appendix II). The other workshops were scheduled over the course of subsequent weeks, exact times and venues being arranged with each RFO and announced over Radio Uganda beforehand (Masaka on 22nd April; Jinja/Tororo on 23rd April, and Kalangala (Ssese Islands) on 23rd May). During the workshops participants were briefed on their respective areas areas of responsibility. The enumerators were teamed in pairs and assigned three to four communities to cover.
It was decided that enumerators in Tororo Fisheries Region which basically comprises of Iganga District should be trained together with those of the Jinja Region. This was because most of the Tororo region staff involved in the survey found it more convenient to travel to Jinja than Tororo. The joint training also helped to avoid a break in the actual field work after Jinja Region.
Fieldwork was carried out in a sequential manner. It had originally been estimated that work on each landing would take about a day, and the timetables for each region had been set with this in mind. In some cases, particularly along stretches of shoreline not easily accessible by road, considerable logistical arrangements involving Project canoes, camping equipment, fuel and food provisions, and so on, had to be made before each field trip. The survey team usually consisted of at least four people, including the National Consultant as leader, a National Project Economist and two field officers as enumerators trained specifically for the area in which the team was working. The Socio-Economic Advisor joined in the field visits whenever possible. The Project coxswain assisted wherever it was necessary to move between landings by boat.
Survey work started in late March 1991 with Entebbe Region, since this was close to the Project office at UFD Headquarters. The intention was to give the supervisors a chance to develop their administrative routines and sort out any unforseen interviewing problems before venturing out to the more remote regions.
On arrival at a sample landing the survey team would first introduce itself to community leaders (Resistance Council officials, elders, and fishing committee members, if any) and explain the purpose of its mission. Very often the team was then introduced to the community as a whole, through a meeting called by the leaders. Again the objectives of the survey were explained and any questions about the work answered. Care was taken to ensure that residents clearly understood sample selection procedures and the purpose behind sampling -- i.e. using a smaller number of fishing communities and households to represent the whole.
Household sample selection was done in the presence of the assembly with one of the leaders assisting to pick blindly five numbered papers from a box or hat. This assured the sampled households that they had been chosen randomly and not with any premeditation on the part of the team. It was also clearly explained to the community that the household information was confidential and would be treated anonymously at a general level -- i.e. as part of a much larger collection of information to be analysed as a whole rather than on an individual basis.
On a few occasions the team had to allay initial suspicions about their work. This probably arose from the fact that a good deal of the interaction between local fisherfolk and outsiders from “Government” is in the context of law enforcement or tax collection. However, the team experienced no problems in any community once residents were convinced about the nature of the survey. Fisherfolk appeared to welcome the chance to talk about their lives and the problems they faced and the team found that discussions were generally quite frank and open.
In the early stages of the survey the procedure was followed of having one team member sit with local leaders to complete the community questionnaire, while the rest of the team split up to interview selected households. Later there was a change in strategy in that it was recognised that people whose households had not been selected wanted a chance of making their views known. The team therefore decided to keep the community-level interviews open to anyone who wanted to participate. This turned out to be a very fruitful strategy as many interesting ideas were gained from those not in leadership roles. Leaders sometimes appeared to be reluctant to mention things that would point to shortcomings with Government or fisheries staff. But this was not a problem with most other community members.
In order to save time after work was completed in Entebbe Region the team split into two, with the National Consultant moving into the Jinja Region to supervise enumerators and the National Economist doing the same for the Masaka Region. This arrangement also enabled the Economist a chance to gain more experience in the conduct of socio-economic surveys. The team reunited in Tororo Region and completed work there before moving on to cover the last region of Kalangala, in the Ssese Islands.
Although the survey work generally proceeded without serious problems or setbacks, a few difficulties were encountered that warrant mention here.
First of all, it sometimes was not easy to elicit information on individuals' incomes. In such cases interviewers had to probe through a series of indirect questions and also to rely on the local knowledge of the field staff. Correct information on sanitary facilities was also difficult to obtain in some instances. Particularly where such facilities were poor or nonexistent respondents sometimes gave misleading answers.
The absence of fisherfolk at some landing sites on the scheduled day of visit meant that plans had to be changed. Return visits had to be made in the cases of Kigugo in Entebbe Region and Nasangazi in Jinja Region.
Finally, as the supervisors worked with different groups every two to three days there was a need for very close attention to control for oversights and errors. It also proved necessary to train extra Fisheries Assistants as enumerators in two cases where the previously trained staff were not sure of the locations of sampled landing sites.
This report has provided an account of the organisation and conduct of the survey of Lake Victoria-Uganda fishing communities undertaken by the SEC working group of the FISHIN Project during March-May 1991. Three questionnaires were devised to elicit information at the community, household, and individual enterprise levels respectively. These were pre-tested at suitable sites around the Entebbe area before being finalised. A manual was prepared as a training guide for field reference for survey enumerators, who were selected from the UFD field staff establishments of each of the five Fisheries Regions that cover the Lake -- Entebbe, Jinja, Tororo, Masaka, and Kalangala. Training courses were held for the enumerators just before the survey exercise was commenced in each region.
Sample selection was based on a stratified approach which ensured full geographical representation of the lakeshore and islands. A total of 50 communities were included in first stage sample. These were selected randomly in proportion to the size of the landing associated with them. Five households from each community were randomly selected for the second stage sample, and there was complete enumeration of all individual fishing-related enterprises for the third stage of the exercise. Weighting factors will be applied to the case data collected for the final tabulation of survey results. The results will be presented in a separate report now in preparation.
In general the SEC team members are pleased with the outcome of the survey exercise. The team was accorded a good reception in all of the sample communities, and with few exceptions information was provided in an open and straightforward manner. In closing, the team would like to record its appreciation for the excellent co-operation and understanding shown by local community residents, and for the patience with which they participated in the interviews. Acknowledgement is also due to the fisheries field staff (Appendix IV) who worked so diligently as enumerators to ensure the success of the survey, and to the Regional Officers, UFD Headquarters personnel, and other FISHIN Project team members who provided effective backup support.
Infield, M. 1989. Socio-economic survey in the Queen Elizabeth National Park, Uganda. Report prepared for the EEC Conservation of Natural Resources Project, No. 4100.037.42.44. AGRICONSULTING, Rome.
Malvestuto, S.P. 1990. Catch assessment survey design for the Ugandan waters of Lake Victoria. Report prepared for the FISHIN Project, UGA/87/007. Entebbe. (mimeo).
Tumwebaze, R., & E.J. Coenen, eds., 1991. Report on the freame survey conducted in the Ugandan part of Lake Victoria. BIOSTAT Field Report No. 22, FISHIN Notes and Records. Fisheries Statistics and Information Systems, FAO/UNDP Project UGA/87/007.
