Chapter 2 Methodology

Contents - Previous - Next

2.1 Step I: vulnerability mapping of farming systems
2.2 Step II: selection of research areas
2.3 Step III: selection of national consultants and capacity building
2.4 Step IV: fieldwork
2.5 Step V: data analysis and report writing

This chapter describes the sequence and the method of the work. The procedure adopted is explained in a way that should enable similar studies to be repeated in any other country or area. Figure 4 shows the workplan of the whole exercise.

2.1 Step I: vulnerability mapping of farming systems

To examine the impact of HIV/AIDS on different farming systems the approach of Barnett & Blaikie (1992) was used. The underlying hypothesis is that not every farming system is equally vulnerable to the loss of labour. The degree to which a farming system is vulnerable to the HIV/AIDS pandemic depends upon a number of characteristics; most of these are linked to adaptations to the loss of labour. Farming areas with favourable agro-ecological conditions, which permit the growing of a wide range of crops and the keeping of livestock are less vulnerable than farming systems in marginal agro-ecological areas, where the range of crops grown is limited and which are already protein and energy deficient. Other factors of vulnerability are the seasonality of labour demand and the division of labour by gender and age. In addition, a significant degree of out-migration and the importance of soil conservation works, such as terracing, may also have an effect because such work is often labour demanding.

Figure 4: Workplan of the study

Barnett and Blaikie developed an algorithm to classify the farming systems of Uganda as "not vulnerable, "moderately vulnerable", "most vulnerable" and "very vulnerable" on the basis of protein/ energy deficiency, labour supply and substitution of crops with less labour demanding crops. As a next step, a map of HIV incidence was laid over the map of farming systems. Accordingly, areas of both high labour vulnerability and a high HIV seroprevalence could then be identified.

For Zambia, the major farming systems have been described and classified in approximately the same way as was done in Uganda on the basis of labour and drought problems and on existing coping mechanisms (e.g. the use of less labour intensive crops, spreading labour peak demand periods, etc.). This farming systems map has not yet been overlaid with a Zambian seroprevalence map, because the latter is largely incomplete.

In Tanzania, regions, whose status in terms of relative food and energy availability had already been classified in the Comprehensive Food Security Study (1992), were mapped with the national sero-data. However, in this case labour peaks or the substitution of less labour intensive crops were not taken into account.

The quality of the vulnerability mapping is only as good as the available data. Comprehensive farming systems data covering the whole country is often not available or incomplete. Farming systems can also change rapidly over time due to population pressure, agricultural policy, changes in agro-ecological conditions and other factors. For the method to be effective, it is desirable that sero-data should cover the whole country, preferably disaggregated by district.

In practice, in most of the cases such data are incomplete or outdated or contain certain types of bias. For example, they may over-represent urban areas - such areas being easier to survey than rural areas. It should also be borne in mind that the seroprevalence maps do not reflect the current impact of the disease. Due to the long incubation period of the epidemic, high HIV rates in a given area provide only very limited indications of current morbidity or mortality. Tanzania is a case in point, where high levels or seroprevalence, are - with the exception of parts of Bukoba - accompanied by relatively low current rates of morbidity. One way of avoiding this problem would be to map the farming systems data with actual AIDS cases. In reality the statistics on AIDS cases are even more unreliable than on HIV rates due to underreporting. Therefore present vulnerability mapping can only give a first orientation and a rough guide to the selection of research areas. This must be supplemented by ground-checking.

2.2 Step II: selection of research areas

The rationale for selecting the research communities was the identification of very vulnerable farming systems in combination with high HIV incidence. Apart from previous work undertaken in a specific area, security reasons, budgetary and logistic reasons were also taken into account. The farming systems of the selected areas in all three countries are briefly described below.

Maps showing the selected research areas can be found in Annex 1.

Three communities were selected:

- Gwanda village in Rakai District
Classification: " not vulnerable "
Farming system description: located near Lake Victoria; 1140 mm annual rainfall; staples are bananas, cassava, sweet potatoes; beans and groundnuts are used as sauces or stews; coffee used to be an important cash crop, but has been neglected due ton decline in coffee prices; limited livestock keeping (poultry, and some goats and pigs); fishing activities; soils are mostly loams and clay loams with medium productivity.

The community was selected due to its very high seroprevalence rate and because Barnett & Blaikie had carried out a similar study in 1989 in the same community.

- Nakyerira community in northern Mubende District
Classification: "very vulnerable"
Farming system description: 1297 mm annual rainfall; vulnerable on both labour and protein criteria; the soils are acidic and infertile; sweet potatoes, cassava and maize are the predominant staples, livestock keeping is an important activity (cattle, pigs, goats and poultry); coffee was a dominant cash crop until 1980, today it is almost nonexistent; Irish potatoes are important cash and food crops.

- Ndaiga community in southern Iganga District
Classification: "very vulnerable"
Farming system description: located near Lake Victoria; 1200 mm annual rainfall; vulnerable on both protein and energy criteria; soils are young and fertile, the area has only been populated since the late 1960s; maize is a major food and cash crop; cassava is an additional major staple; fishing is limited as is the keeping of livestock; there is only a limited range of crops grown.

The rationale for the selection of the study areas in Tanzania was as follows. Farming systems in Tanzania are conventionally classified into six broad categories: coffee-banana, maize surplus, pastoralist, sorghum/millet-livestock, rice farming, and cassava. Mbeya and Tanga regions are two of the most heavily HIV-affected areas of the country apart from Kagera. Kagera was specifically omitted from the present study because both the World Bank and UNDP were either planning or executing similar studies in these areas. Within the selected regions, together with Rukwa, it was possible to find representative areas for most of the major farming systems, and including the estate and horticultural sectors. On this basis the following farming systems were selected:

Region Farming System

Rukwa: Main production systems: maize/legume/pulses; apart from maize which is generally in surplus, rice farming is also undertaken. Some 90 percent of the region's economically active population is engaged in subsistence agriculture. Limited cash crops are grown notably tobacco, some coffee, sunflower and cocoa (these later recently introduced as part of an effort to diversify crop production base).

Mbeya: The region has 13 agro-ecological zones mainly as a result of which a large variety of crops are grown and intercropping is widely practiced. Several communities within the region like Kyela, Rungwe, and Mbozi have high population densities and face land shortages. The production systems are characterised by pastoralism, smallholder paddy production and estate production. There is a high level of mobility partially as a result of the colonial period when parts of the Western, Southern and Southern highlands were designated as labour reserves for the plantation sector, the most dominant tribe in the region, the Nyakyusa continue to be highly mobile, thus the potential for the spread of HIV/AIDS is considerable.

Tanga: The region comprises four agro-ecological zones; the coastal plains, the inland plains, the highlands and plateau land. The research focused on the horticulture and cassava production systems but other crops are also grown like sisal, cashew nuts, coffee and tea. Ninety percent of the population work in the agricultural sector but only a relatively small amount of the total available land for crop and livestock development has been cultivated (some 17 percent). The region's population structure is characterised by a high ratio of men to women (while nationally there is a surplus of women to men).

Two communities were selected. These were expected to have middle to high seroprevalence rates. The study was based on previous farming systems work carried out by the Adaptive Research Planning Team (ARPT). The communities were:

- Teta in Serenje District
Classification: "moderately vulnerable".
Farming system description: 1100 mm annual rainfall; hand hoe cultivation supplemented by oxen hire; finger millet as a chief staple; maize, sweet potatoes and cassava often supplement finger millet, maize meal is often purchased; chicken, beef and fish are purchased on occasion; maize is a main cash source, supplemented by beans and millet beer, Citemene is being replaced by hand-hoe ploughing.

The system, being based upon finger millet cultivation using the citemene system and thus being quite labour demanding would be affected by labour losses, however, cassava could be developed as a major staple. Hence the system was classified as being moderately vulnerable.

- Mpongwe in Ndola Rural District
Classification: "vulnerable"
Farming system description: 1300 mm annual rainfall; hand hoe cultivation supplemented by oxen and tractor hire; maize is the dominant starch crop with some sorghum which is also used for brewing beer; relish crops are grown together with maize, namely groundnuts and beans; cash sources are varied but include beer and fish, supplemented by maize, vegetables and chicken. The system is vulnerable to the effects of labour loss due to the insufficiency of draught power and the heavy labour inputs required to grow maize, the major staple. Large areas of land are required for maize sufficiency under this system.

2.3 Step III: selection of national consultants and capacity building

From the inception of this project, it was envisaged that national consultants and institutions from different disciplines would carry out the field studies. Full involvement of national consultants and institutions contributed to the success of the study in several ways. Although it was not the objective of the project to train nationals, in fact the research activities have contributed to a national capacity-building. As the link between HIV/AIDS and agriculture is a very new field of policy concern in all three countries, the national consultants who participated in the study are now among the first people in their countries with expertise and first-hand field experience of studying this problem. In addition to the background expertise which was gained, most of them (with the exception of the Zambian team which had considerable experience with the method) were using the Rapid Rural Appraisal (RRA) methods for the first time and thus extended their research capability. Some of the national consultants are working as lecturers at universities where they have a forum to disseminate their knowledge to a wider audience.

2.4 Step IV: fieldwork

Various forms of the RRA method were used in all three countries. Although this is a qualitative method, efforts were made to collect some comparative quantitative information on production activities. Some of the methods used were participatory, but the exercise was essentially extractive rather than facilitative in nature, aiming at 'learning from outsiders' rather than the 'empowerment of local people' (Chambers, 1992).

The composition of the three teams varied considerably, from three people in Uganda and four in Tanzania to 25 in Zambia. But in each case, the team came from different disciplines, mostly agricultural, sociological and health fields. Apart from the Zambian team, the national researchers received a briefing in the conduct of RRA and the use of RRA tools.

During their fieldwork, the three teams used different techniques depending on the composition of the teams and their familiarity with RRA. The most common tools were:

- village walk and map
- the community transect
- farming calendar
- focus group discussions
- key informant interviews
- detailed case studies

In addition, the Ugandan team produced village demographic profiles and conducted social ranking exercises. The Zambian team elaborated a health time chart with the farmers. The amount of time taken for data collection ranged from 19 days in the case of the Zambian team to 40 days for the Tanzanian team.

While the approach used by each team is explained in more detail below, special attention is drawn to that used by the Zambian team because the ARPT has had more experience of RRA than the other two teams and has developed the RRA in some novel and effective ways.

The approach adopted by the Zambian team conforms with an intensive RRA approach which ARPT has been using in its analytical studies in the Central and Copperbelt Provinces. In all the farming systems work that ARPT has carried out in the last few years, the concept of cluster has formed the basis of the analyses undertaken. The reason for this is that rarely have households been found to exist as discrete social and economic units. Single households are usually connected to others in overlapping relationships which, if one is to make sense of them, require another analytical unit. This unit, the cluster, has been defined as:

"A group of producers between which there are multiple resource exchanges, usually based on the factors of kinship, labour and food exchange, and or common access to draught power." (Drinkwater, 1992)

Different terms are used to refer to different types of producer within a cluster. The primary producer household is that producer (and household) which produces the most food within the cluster and is primarily responsible for the food security of all the members. Secondary producer households have their own separate kitchens, but have lower production levels than the primary producer.

In various previous assessments completed by the ARPT, the following cluster types have been identified and were used in the Zambian study:

Type 1 - Small-scale commercial Primary producer grows 15+ ha of maize.
Type 2 - Successful small-scale Primary producer grows 5-10 ha of maize.
Type 3 - Vulnerable Primary producer grows 2-5 ha maize
Type 4 - Resource-poor Primary producer grows 2 ha staple

The ARPT has adopted several key principles in their approach:

- a large multi-disciplinary research team is used (some 25 individuals);
- the approach used is qualitative, but within case studies effort is made to collect comparable quantitative information on production activities. Since this information is collected within a case study framework its consistency can be checked;

the methodology is based on two types of interaction.

i) that occurring within the research team,
ii) that within the rural community.
Different scales of interaction take place. Much fieldwork takes place with the research team subdivided into smaller groups of three or four, but the initial planning and final synthesis of each of the activities occurs in a plenary session. Interaction with farmers takes place in meetings, and with smaller groups or individuals within the village environment;

- the methodology is iterative in nature, that is it proceeds through a series of cycles (or spirals). In this exercise the two main iterations were:

i) analysis
ii) validation of issues and elaboration
The research team, in addition, discussed follow-up options and issues in a final meeting. The suggestions made were in response to farmers' comments during the ultimate focus group meetings that were held, but this does not constitute a full iteration as no final discussion was held with farmers on this topic; and

- effort is also made to triangulate and cross-check information. Several means are used for doing this:

i) by using different methods to collect similar information;
ii) by having multi-disciplinary interview teams (each member should be checking that the interview is consistent from their particular perspective); and
iii) by following information (validating and elaborating data) through the iterative nature of the methodology.

With the help of the RRA tools the teams assessed the present situation as well as the future trends and the type of intervention which the affected households require.

2.5 Step V: data analysis and report writing

The research teams all commented that the most difficult part of the whole exercise was inevitably the data analysis and report writing. RRA generates a lot of information in a short amount of time. Much of this is qualitative and cannot be analyzed in a mechanical way as with quantitative data. The interpretative nature of this type of data analysis therefore requires intensive discussions within the research team if agreed conclusions which are focused on the study objectives are to be arrived at. As no template for the report was provided to the teams, the final reports followed different formats and therefore the results from each country team were not directly comparable in all respects. Instead, each report reflected the different make-up, experience and interests of the national teams as well as the national contexts within which each team was operating.

The final synthesis of the three country reports plus the studies of the major farming systems in Zambia and the Nakambala sugar estate were made by an international consultant. The present book takes the synthesis further and presents it in a form which it is hoped will be of use to others wishing to undertake similar studies.

Contents - Previous - Next