|
No. 26 July September 1986
CIMMYT EASTERN AFRICA ECONOMICS PROGRAMME
INTERNATIONAL MAIZE AND WHEAT IMPROVEMENT CENTRE (CIMMYT)
P O Box 25171, Nairobi, Kenya, Telephone 592054/592206
10
Two asterisks indicate the actor or actors with major responsibilities in each particular activity. Several asterisks in a row indicate there must be cooperation between actors to do that activity effectively. Is the balance right here? Let us have your comments on the responsibilities as allocated to the actors for the sequence of OFR activities. Should these be changed?
Adding Community Level Variables to FSR: A Research Priority
This is the title of a recent paper by Barbara Grandin, an ILCA anthropologist, given at an IIMI-Rockfeller Foundation Workshop on 'Social Science Perspectives on Managing Agricultural Technology'. It adds a further dimension to the queries raised by Behnke & Kerven on household definitions in FSR (FSN No. 22 July-September 1985) making the case for adding a supra-household perspective to FSR. The paper highlights the community control of resources in rural societies, access to pasture, water, firewood and of course land can all be cited. It points out that even where resources are individually controlled, community level interventions can play an important role, if there is understanding and sensitivity about community organisation and factions. The potential importance of this type of intervention in a range of situations causes Grandin to urge FSR to move one level up the systems hierachy and embrace community variables. She summarises her paper:'FSR currently focuses its efforts within farm boundaries, thus limiting technology generation to inputs under the complete control of the producers. The initiallfocus on the farm-family has matured to include a sub-household focus with particular emphasis on the sexual division of labour and product. The focus remains
within the farm.
The applicability of FSR is currently limited by its lack of attention to community-level issues, particularly as they affect communal resource control and organization for community based development. The IARCs which have been in the forefront of developing FSR, particularly those working in Africa, have an urgent reponsibility to conduct the strategic research necessary to develop and test methods which will assist national programmes to incorporate community level variables and hence expand their scope for technology generation and dissemination.'
Practitioners would probably agree that while the 'social and cultural or community circumstances' are seen in FSR as an influence on farmers' decisions, it treats them essentially passively. We have to agree that, as with other circumstances of markets, extension services, and policies, community circumstances can be understood, and changed, for the benefit of farmers. Indeed CIMMYT itself is moving a step up the systems hierachy in the development of a new programme thrust termed 'Farmer based Policy Research' still however, with the farmer client square in the centre of the picture. Grandin's paper is another very useful advocacy for widening the role to be played by social anthropologists and rural sociologist in FSR and related research approaches.
THE NEWSLETTER ARTICLE
MANAGING BIAS: FARMER SELECTION DURING ON FARM RESEARCH
By Alistair J Sutherland *
1.0. WHY IS FARMER SELECTION IMPORTANT"
Farmers are the primary client group for FSR research activities. The majority of this client group is comprised of resource poor farmers who are severely handicapped in terms of taking their problems to research scientists. Anthropological and sociological studies have shown that rural communities have influential members, sometimes called "brokers", who tend to monopolise relations with representatives of the outsidee world", such as development project officials and indeed farming systems teams. The
influence of such brokers is often beneficial to the community, especially when they can mobilise local support and lobby outside the community for assistance for local projects (such as self help schools. clinics, dams etc.) However, brokers also have their own interests to serve when dealing with outsiders,, and when it comes to agricultural development their requirements and interests are likely to differ from the interests of the community at large. Unless the FSR team members take definite precautions, there is a high probability that most o4 their dealings with farmers will be directed through local brokers. The danger is that they may end up serving the interests of a small group of farmers, perhaps at the expense of the larger intended target group. For this reason it is important that careful attention is paid when selecting which farmers to communicate with during FSR activities. Farmers who are selected become, in effect, "spokesmen/women" for a particular target group.
* The author is senior rural sociologist with the Adaptive Research Planning Team (ARPT), Zambia's farming systems programme which is incorporated into the Research Branch of the Department of Agriculture, Ministry of Agriculture and Water Development. A shorter version of this paper was presented at the ARPT annual review as a basis for a discussion on farmer selection. The author wishes to
acknowledge the stimulus from other ARPT staff during discussion and also written comments from individuals
including Richard Edwards, Stuart Kean and Mike Collinson. The ideas presented are the author's and do not necessarily represent those of ARPT.
12
2.0. FARMER SELECTION WITHIN THE FSR SEQUENCE OF ACTIVITIES
Selecting farmers as spokesmen occurs at a number o different stages during the FSR sequence of activities:1. During the zoning of farming systems, a limited number o farmers may be interviewed in addition to extension staff anr local leaders. Such farmers are used as "key informants" usually being drawn from areas adjacent to agricultural camp they are expected have an overview sufficient to provide general picture of the farming system in the local camp are and possibly in adjacent areas too.
2. During informal surveys, farmers interviewed are th principal spokesmen for target groups identified during th zoning of farming systems. As such they are expected t( represent a cross-section of the target cJ roup undet investigation, each presenting their own specific farming
practices and views of agriculture related problems, I total., their views provide the main foundation for a listing and prioritisation of research problems and priorities.
3. During. formal survey work farmers selected for interview are spokesmen in relation to a set of issues identified by, previous spokesmen. As such they have less control over the identification and prioriti nation of problems, and are expected instead to assist in quantifying the scale of problems and further pinpointing the causes.
4. During the prescreening of technological improvements farmers are rarely involved as spokesmen, largely because their opinion is rarely sought by FSR teams. While the full value of seeking the farmers opinion at this stage has not been effectively put to the test, there are good reasons for arguing that this would be beneficial. Discussion at the
community level prior to trial implementation permits farmer and the community to be actively involved in t trial assess-ment and to react to ideas being tested in the tri is l farmers' reactions area effectively ga uied and nec eary adjustments made, this will help th t i den it he
trials and the problems these address
5. During the running of on-farm tric-'ls sl, 1 e d f armer
... "u ... nq the... rJ i '2
cooperators"' become spokesmen wit h the Po-.t. =ial r f evaluating the relevance and future of tie technology undL er test. They may al-so be involved in helping the agronomit to test hypotheses relating to biological r el ationshis in
problem identification trials. T h Pp )to e of he
farmer cooperator offers the gre-atest -- nfluen e n the 73 agronomists and social scientists In elat ion tC re ineme t of research problems and the r e a S- of ;- et r
nri orities both on-farm and on-station.
13
6. Duri ng + i el d days and field visits farmer cooperators selected for vi sits also have a considerable potential influence as spokesmen communicating with a small number of influential decision makers in agricultural research and extension.
7. Finally, when a new technology is ready for release, local farmers who decide to try out the new technology on a full scale become spokesmen with a new role of encouraging others in their target group to adopt the technology. If extension demonstration plots are used then again farmer selection becomes a crucial activity.
3.0. BIASES LIKELY TO INFLUENCE FARMER SELECTION AND
COMMUNICATION WITH SELECTED FARMERS
When farmers are being selected for the above activities, a number of biases are likely to influence the selection process, and also subsequent communication with those selected. These biases are both conscious and unconscious. Some, especially the conscious ones, may have a positive effect on the success of the programme. What is important is that FSR teams are aware of the potential biases involved at different stages in the sequence and take appropriate action. Biases which are undesirable should be minimised, while those which are useful should be harnessed in a way that increases the relevance of on-farm research activities. Taking action requires indentification of the main sources of bias. Sources of bias can be classified into three basic kinds: bias arising from the local area under study through the use of middlemen; bias arising from the FSR team's internal characteristics, and biases arising from logistical constraints within the programme.
Three main types of bias from middlemen are:
a) bias from extension staff (field staff and specialists) who tend to favour more progressive farmers, cash croppers/credit receivers, and friends,
b) bias from local leaders and "brokers", who may favour their own family, interest group, or ethnic group,
c) bias from heads of households/household groupings, who may present their individual interests as identical with subordinate households/family members.
Biases arising from the FSR team's internal characteristics include:
d) bias towards male farmers as spokesmen (especially when the team is all male),
14
e) bias towards articulate farmers (especially when team members -don't speak the local language or want information fast),
f) bias towards "volunteer" farmers (especially when quick and easy cooperation is a high priority,
g) bias towards more wealthy farmers (especially when team or its guide values farmer hospitality),
h) bias towards more progressive and innovative farmers (especially if major changes in technology are being considered, or i-f team shares a model of new technology trickling down from the more to the less progressive),
i) bias towards farmers speaking the same language (especially when interpreters are mistrusted or team's language skills are limited),
j) bias arising from team's methods and criteria for defining what constitutes a household/farmer (especially when this has been given little thought or left up to middlemen),
k) bias arising from the desire to maintain the same farmers over several seasons who may have become a unique group due to teams interventions (especially in areas where cooperation is a problem or longitudinal case study data is being collected).
Biases arising from logistical factors and constraints include:
1) bias towards farming communities nearby roads and administrative centres (especially when roads are poor, time is scarce due to size of programme, support staff are seconded from existing institutions, and transport
resources are few),.
M) seasonal bias (especially in single season cropping systems when surveys and selections are carried out during the dry season),
n) bias to "home centred -Farmers" (those normally at home), and
o) bias to farmers who normally' attend public meetings called by extension staff/local leaders (especially when this is the common practice during informal surveys and selection of farmer cooperators) ,
The next step is to suggest ways of dealing with these biases in the context of the occasions- for -farmer selection within the sequence listed above. The suggestions below arise
largely from experiences in Zambia, but it is anticipated that they will be applicable elsewhere in Africa.
4.0. STRATEGIES FOR DEALING WITH BIAS
4. 1. Zoning.
- preselect extension workers for interview who have at least five years experience-n the area under study
- select local leaders who have lived in an area for ten years or more, and concentrate on older leaders with
traditional authority in the area (such as local chiefs and their advisors/retainers)
- select local villagers without office who were born in an area and are 45 or more years of age. Ensure older women are interviewed, even in preference to men as the other
informants are likely to be men. Traditional doctors are likely to be good informants as they are intellegent and tend to travel widely.
- conduct interviews in the local language with the aid of an interpreter if necessary. This will help avoid selecting more literate and travelled members of the local community.
- if a range of ethnic groups occupy the area being zoned, ensure informants are representative of the different groups.
4.2. Informal Survey.
- a frequent unconscious bias arises from the timing of the survey. Seasonal variations can give rise to finding
different people at home at different times of the year. For example in shifting cultivation systems combined with permanent villages nearby semi-permanent sites will give a bias to those doing more semi-permanent cultivation,
especially if the village is visited during peak periods of activity in the shifting system. Moreover, the time of -day is likely to influence who will be at home, especially whether men or women will be at home. Avoiding such biases requires a prior understanding of seasonal movements between permanent villages and distant lands and also a prior
understanding of the daily routine for both sexes at
different times of the year.
- avoid reliance on extension staff for the selection of
survey areas, villages, and individual farmers within villages. This implies having good maps and secondary data and using field skills to obtain the cooperation of the local extension worker as a provider of information and introductions, rather than as a broker and navigator.
16
-The same general principles apply to using other local leaders, but with the recognition that it may be "politic" to interview and visit the field of the leader as a means to ensuring smoother cooperation and greater control of subsequent selection.
- When selecting heads of household for interview, it is important to be aware of the local kinship and residential organisation. For example, interviewing only the family head in situations where residential groupings are based on extended families should be avoided. In such situations it is better to interview all adults in the residential grouping, rather than singling out one and then moving onto another residential group to-do the same and so on. If the
usual informal approach to villages is used it is important also to remember that household heads most likely to come forward for interview are those most senior in age and rank and also those most used to dealing with outsiders (often the more educated).
- While not essential to seek out farmers in the most r emot e areas, it is important to make the extra effort to interview farmers staying further from the roads and service centres, and establish in what ways their priorities differ, if any, from farmers staying close by. It is also useful to estimate the proportion of the farmers in the target group living more. than -a given distance from a main road/service centre. Aerial photographs can be used for this. If very substantial differences appear then it may be necessary to establish onfarm trials in a more remote area.
- During informal surveys male farmers usually serve as the main spokesmen, even though they are not always the main providers of labour and decision takers (usually both labour and decision making result from informal/private dialogue between male and female within the household). This tendency arises from two related factors. Firstly, most FSR teams are male dominated, and secondly adult males are the ones who deal with outsiders in most rural areas. Selecting women for interview requires a conscious effort from the FSR team to seek women out and to ensure that wives attending contribute to the discussion. This is not always easy. Women beinq busy people, will often stand up and leave an interview, or cut it short, especially if they regard the questions as irrelevent, while men are more likely to continue to answer politely. For this reason it is particularly important to get women involved in the interview as quickly as possible, as well as to anticipate the kinds of questions they are most likely to find relevant.
- The most articulate farmers. are likely to come forward during an informal survey and select themselves. Moreover, being arti cul ate they are li kel y to provide more information and have more influence on the teams assessment of priorities
17
and problems. While such farmers are valuable sources of information, it should always be kept clearly in mind that their opinion on a particular issue is in fact no more important than any other farmers. Indeed silence on an issue under discussion may indicate that it is a non issue for the farmer (even though it may be an issue for a previously interviewed articulate farmer and/or FSR team member).
- Farmers who volunteer themselves for interview may often have other motives than to be helpful. Often individuals who are marginal in a community may try and enhance their status by being seen talking to apparently influential outsiders. Or they may believe the team is not simply seeking information but providing resources such as credit or inputs.
- In rural communities, more wealthy farmers may also be the more articulate ones who volunteer information. Such farmers are more likely to speak a similar language to research scientists (use similar agronomic terms) to ask for technical advice, and more inclined to provide refreshments for a group of weary interviewers, and so hold their attention for longer.
- More progressive and innovative farmers are often more articulate and wealthy. They are likely to be more
interesting to the interviewer and more likely to engage in dialogue and ask questions as well as provide answers. The team should not forget that most farmers in the target group may be relatively conservative and reluctant to take risks. While innovative farmers may be very useful as cooperators during trials, they should not be over-represented during surveys.
- Language biases are important. When expatriates are involved there is great scope for misinterpretation of questions and answers, and this may lead to selection of more farmers with a knowledge of english and a more extended dialogue with these farmers. This is common when there is a lack of faith in interpreters being used. When nationals are involved, the bias towards english speaking farmers is probably less likely to take place, but may still occur if the interviewer has a preference for discussing agronomic issues in the technical language of his or her training. Also, when nationals are working in areas where a minor language is spoken, there may still be a bias towards selection of farmers who speak the main language in the province. Such biases can be avoided by ensuring that there are a sufficient number of people on the interview team who speak the local language well, and also that those who don't are prepared to make an effort to learn and record relevant local terms relating to local practices, and local ideas
relating to causal relationships between variables under investigation. In short, the bias away from the local language generates a bias away from trying to access
18
i ndi genous techni cal k.nowl edge; a great furd of valuable information in missed.
4. 3.~m~ SUrv e
Because a formal survey is~ a more structured a.cti vity, there is more scope for controlling bias. Most bias enters during the sampling stage. Only in the most exceptional case would it be possible to attempt a random sample irom a target population. Lists of farmers are rarely complete, often they don't exist and those that do exi st have biases. For example in Zambia village registers do not indicate adults as househol d heads. and are o-ften both i ncompl1etc and out of datti. Extension workers have farmer registers which are usually incomplete. The registers are usually biased towards
-farmers staying closer to the camp to male farmers, and to farmers growing cash crops and receiving credit. For these reasons a multi-stage sampling procedure is usually the best bet. In the Zambian context, the first step is to sample from a list of wards (the political unit immediately below the constituency) which make up a recommendation domain. These can be randomly sampled, but it may make more sense to purposely select one or two wards which represent the most typical features of the domain and also take in a fair measure of the internal variation within it. Data on these characteristics should be available in the zoning report (e.g. CIMMYT, 1979?). This will avoid the possibility of accidentally choosing the one or two of the least typical wards and will also allow selection to be considered in relation to transport and other logistics. The next step
is to obtain a list of villages in the selected wards (from District Council village listings) and use a similar procedure to select from these. This will involve a visit to the ward and interview of key informants to obtain An overview of the size of the villages on the list, any missing ones, and the degree of variation between villages. At this stage care should be taken to establish whether or not registered villages are in fact discrete residential units, or whether they are scattered units with overlapping boundaries. If fields are to be visited as part of the survey, a further point to establish is the distance from house to fields and the general relationship between dwelling places and fields. In areas where shifting cultivation is practised this exercise should ensure distant fields are included to avoid a bias to semi-permanent plots close to settlement. In semi-permanent systems if neighbouring households also have neighbouring fields this makes the logistics of field visits during survey much more feasible. Generally speaking, a good understanding of local residence and land use patterns will allow more flexibility to be built into farmer selection at a later date with less risk of bias.
Using the estimated size of the villages to be selected ir
19
rel ati on to the target number of interviews, the next stage of selection of farmers for interview can proceed. 1+ villages are purposely selected for representativeness, the risk of geographical bias is greatly reduced, and logistics made easier. Purposely selecting villages involves noting the main characteristics of each village on the list (size, location, main crops, soil type, and other important productive specialisations), and then choosing a small number with representative characteristics, appropriate size and convenient location.
Having selected villages, the next stage is to obtain a complete listing of households in each. Starting with the farmer register, going on to the village headman's register, and then interviewing the headman (or his 'representative) a full list of farm households in the village can be made. The next step is to decide whether 'to interview all households in the village or. a proportion of these. The
advantage of interviewing all households is that the smaller marginal households (particularly female headed) are more likely to be included, and that linkage- between households in the village will be brought out during the survey. The disadvantage is that a narrower geographical area is covered, but as villages have been purposely selected to avoid uncontrolled geographical bias, the benefits of covering household linkages outweigh the small geographical bias.
Once the listing of names has taken place, the team can choose whether or not to start the interviews on the same day or arrange a time in the future when they can return. If time is limiting it may be most economical for the team social scientist to make the final selection of households and set up times for interview prior to the arrival of other. team members (preferably one or two days before hand). However, it is better for all team members to be involved. This is not only because they will be more. likely to identify with the approach, but because the social scientist's
criteria for final selection may omit some previously
undiscussed yet important variables required by the natural scientist. If the natural scientist is unable to assist directly, the social scientist should make a point of discussing in advance the kinds of criteria (e.g. soils, cropping history, history of site, livestock situation)
regarded as important. To allow for absentee householders it is advisable to overselect by 10-15%. If more than this number are absent on the day then neighbouring households falling within the target group parameters can usually be easily selected to make up the total.
4.4. Prescreening technical options. The CIMMYT approach does not include farmers as an important part of- the group which sifts through problems and related technical solutions before coming up with an on-farm research
20
programme. However, there is considerable scope for involving farmers at this staqe. If the formal survey is carried out at a small number of villages, the same villages can be used to orcani se di scussi ons about research possibilities. Selecting farmers for such discussions requires a sensitivity to local patterns of public disussion. For example, if men QenerallV discuss matters of public concern as a group apart from women, seperate meetings may be required to get discussion and feedback from both sexes. For example, the Chewa of Eastern Province have a traditional male forum (the mphala), which can be used to obtain male opinion. However, if public meetings are generally associated with a lot of ceremony and oratory, it may be advisable to hold small informal meetings with three or four farmers rather than try and discuss details in a group of twenty or more adults. If the T and V system is operating it may be possible to use T and V groups as an easilly accessible forum for discussion.
4.5. On-farm Trials
Selection of farmers for on-farm trials can follow the same pattern as selection for surveys and prescreening discussions. By focusing on a small number -of local areas selection can proceed on the basis of an understanding of a community built up during survey work. Rather than call a large meeting and ask for volunteers, a more purposive approach gives a greater degree of control over the farmer selection process. By purposely selecting farmers over a narrow geographical area the logistics of trial management and monitoring can be made much easier and the potential for dialogue with farmers increased. A trial assistant along with FSR team members will be able to comfortably visit all farmers in the area on foot and also benefit from observations of fields and conversations with other farmers along the way. When selecting farmers in this way it is important to establish how the location of their fields relates to the location of their residence. This requires a brief survey of the village area prior to the final farmer selection.
Farmer selection can be made more purposive by matching types of farmer With particular types of trial. If the trial is researcher manaQed and implemented, aiming at problem identification, the most apropriate farmer is a larger and more literate farmer. The agronomist will probably find it easier to explain the purposes of the trial to such a farmer and because the farmer has more land and resources, he is more likely to be willing to sacrifice a larger part of his field and provide the necessary resources to ensure timeliness of operations. At the same time, when such a trial is spread across a broader range of farmers it may generate more information overall (even if CVs are higher).
21
For levels type trials the same type of farmer is also likely to be better due to the fact that timeliness and appreciation of the trial objectives are important. However, it is important to include a large proportion of the smaller/suboptimal management farmers in this type of trial and if necessary modify the trial design to fit the farmer's production system, rather than vice versa.
For verification trials, it is essential that representative farmers are used, and the trial design enables easy assessment by farmers. For this type of trial a community approach to selection providing many farmers with adjacent fields will enable easy comparison by farmers as well as trial assistants and researchers.
More insights into using a community approach to selecting farmers for trials can be gained from examination of a case study from Lusaka Province in Zambia.
4.6. Case Study of Farmer Selection in TRD2 Lusaka Province, Chinapa Target Area.
Farming systems research in Zambia is institutionalised through the Adaptive Research Planning Team (ARPT), a section within the Research Branch of the Department of Agriculture (Kean and Chibasa, 1983). Each province has a team comprised of a farming systems agronomist, economist, and research extension liaison officer, while sociologists provide an input on a regional basis. On farm research is conducted with the help of a "trials assistant" seconded from the
Extension Branch.
Lusaka Province team started work in 1980, and Chipapa was the first target area selected within Traditional Recommendation Domain 2 (TRD2). TRD2 is a domain characterised by ox cultivation of maize, subsidiary food crops, and limited cash cropping. Sharing of oxen was
identified as a common cause of late planting of long season hybrid maize, resulting in ver poor yields. On-farm research had focused mainly on addresssing this problem with the
introduction of short season hybrids and late planted alternative crops (eg. beans, soybeans, sunflower, and short duration sorghum).
After four seasons of trials in Chipapa it was decided to review the farmer selection procedure. One factor which prompted this decision was a transport problem. The trials assistant's motor bike had broken down the previous season with the result that trial sites more than walking distance from the agricultural camp had been neglected. As there was no certainty of a motor bike for the coming season it was decided to move the sites to within walking distance of the
22
trial assistant's house. Further factors were a change of staff. The new trials assistant was inexperienced, and did not have established relationships with the previous trial farmers. The new agronomist felt that farmers used in the previous season represented a biased group of farmers, and in addition was interested to understand how resources were shared around the community and felt a clustered community perspective would facilitate this. He decided to involve the rural sociologist in the farmer selection process in order to bring about more farmer participation and feedback in the experimental programme. The rural sociologist was keen to be involved as he had seen the need for a more systematic approach to farmer selection in order to introduce more of a community perspective in Chipapa target area, but had been unable to persuade the previous agronomists that this was necessary. Their reaction to suggestions had been that the value of having more representative and accessible farmers would probably not compensate for the loss of good relations with existing farmer cooperators. A change of staff
therefore provided the main opportunity for reviewing the situation.
4.61. Previous approaches to farmer selection in Chipapa
The original approach had been to hold meetings called by the local extension staff at which ARPT trial activities had been described and volunteers requested. in the first two
seasons, three sub target areas had been selected, one around the Chipapa agricultural camp, one about 10 km north at
Shantumbu, and the third further south at Chiyokoma. in 1984 it was decided to reduce the size of the area by
concentrating on three villages closer to Chipapa camp, the objective being to have three clusters of trials. Again, selection was on the basis of volunteers/village representatives selected at meetings. While reducing the distances to travel between sites, particularly for ARPT officers stationed at Mount Makulu, this did not significantly reduce the dependence on a motorbike for the trial assistant as the clusters were still relatively
scattered across the area.
4.62. Bringing a community perspective to bear on Chipapa farmer selection.
In order to bring both the trials assistant and the ARPT officers closer to the farmers it was decided to select
farmers within walking distance of the trials assistant's house (up to 45 minutes). This would enable both of them to get to know a smaller group of local farmers more intimately, reduce the time taken traveling between sites, and remove the trials assistant's dependence on a motorbike. Because much of the travelling could be on foot, the opportunities -or informal contact with local people could increase, as would
23
the opportunities for observations of agronomic and management variables in farmers' fields through the season.
The actual selection process took place within certain
critical parameters likely to be common to any on-farm trial programme:
. Nature of the on-farm trials. The on-farm experimental
programme for the season was to include three main trials, with a possibility of two more subsidiary trials. The main trials were; maize varieties (verification under farmer management), late planted alternative crops (verification farmer management), and maize/sunflower/cowpea intercrop (exploratory farmer managed) Subsidiary trials were a sorghum variety trial and testing of a new pumpkin variety.
While under pressure to reduce the scale of the trial programme and the number of sites involved, the agronomist was concerned to replace researcher managed split plot designs with a more simple style of trial which left most managment decisions up to the farmer. To give credible statistical results, this implied having more replications, and in turn more farmers. In fact the number of trial farmers required increased over the previous season, from three to six farmers per trial to a minimum of 24 (this figure assumed each trial would have a minimum of 9 sites/replications and 12 of the 24 farmers would have 2 sites, each for a different trial, whilethe other 12 would have one trial giving a total of 36 sites).
2 Soil topogrpb. y and climatic variables. The agronomist was concerned to spread the trials over the range of major soil types and topography predominantly used by farmers in the domain. This involved selecting sites on both hilly and flat areas and on both lighter gravelly soils and heavier clay soils. Rainfall was thought to vary considerably locally but a minimum distance between clusters of sites was not established in advance of the selection. It was felt that major differences would be related to management rather than climate.
J, ..'iv..L gro c h@rate i t: s.. It was known that a large proportion of the target group did not own oxen and depended on access to these by other means. Moreover, it was also suspected that female headed households const tuted a significant proportion of the population. In addition, not all farmers grew the crops planned in the trials every year (particularly sorghum and the late planted alternative crops).
4.60. Stages of farmer selection.
.. Sel. c-ti vi i 1g. g:.i A st of 15 vi I 1 ages within the Chipapa agricultural camp area was compiled. Of these 15,
24
five villages within easy walking distance of the camp area were identified for more detailed data collection and a sketch map made of their location. The trials assistant did this by interviewing camp staff and local people as the farm register was not complete. For each of the five villages a preliminary list of households was compiled and for each household, data on resource base and crops grown was
recorded based on information volunteered by village members.
The walking distance (in minutes) to each household was also recorded. At this stage of collecting information, it became clear that a sufficient number of farmers could be selected from three out of the five villages. The three villages chosen included two closest to the camp which had the biggest range of soil types and topography and the. most sorghum growers. These two villages also represented one of the main arable producing areas of the camp, a bias which matched the nature of the on-farm trials towards crops rather than non-arable enterprises. The third village, the largest in the cluster, straddled the road leading to the agricultural camp which ARPT officers used to enter the target area. This selection permitted easy logistics for visiting both farmers and sites.
2. Household Selection. For the three villages selected, an effort was made to visit each household and interview the household head to verify the information recorded in the preliminary listing of households. In the process of doing this it was discovered that a considerable number of households had been left out of the list, particularly female headed households and those of newcomers to the area. The reason for this was that the definition of what constituted a household had not been agreed in advance. Lack of clarity was compounded by the nature of the settlement pattern in the area. Households, defined subsequently as one or more adult persons with a seperate plot, grain store, and sleeping quarters tended to cluster in small groups around family heads. Often the family head had been recorded as the household head.
Through the process of visiting households, the initial total of 48 households in the two largest villages was increased to 68, and with the knowledge that this number would be increased further if other parts of the villages were
visited. However, it was decided to stop at this point as time was limited and a sufficient number of potential trial farmers had been identified.
While visiting households, household heads were asked if they would be willing to participate in the trial programme. All readilly agreed, with the exception of a cluster of female headed households who expressed reservations due to the fact that their oxen had died, and they didn't know how they would be ploughing. After assurances that they would be expected to plant the trials only when they were ready, they agreed
25
and all did host trials. At this stage an attempt was made to note which trial might be suitable for which particular household, although no definite arrangements were made. Farmers were told that the trials assistant would contact them if he required their help with trials.
This process of identifying households was accomplished in a short period of time, a morning (5 hours). In the first part of the morning, 16 potential households were identified from one village, grouped into four clusters on the basis of proximity of fields and dwellings. Most of these 16 were interviewed in person and agreed to cooperate in trials. The remainder were mentioned as possible cooperators by close relatives, and an arrangement made for the trials assistant to follow these up at a later date. In the second part of the morning, three more clusters totalling 18 households were identified. These three clusters were based solely on proximity of fields, as experience of the previous clusters had shown that sometimes proximity of dwellings and fields did not coincide (also time did not permit visiting each household individually). Some households in each cluster were visited and fields inspected in conjunction with a representative of the household. Sketch maps of the household's field and adjacent fields were made on the spot. The trial assistant was left with the task of making follow up visits to households not visited but identified as potential cooperators on the basis of field proximity, and of making sketch maps of fields of the 16 households in the first village. No major problems were subsequently
experienced by the trials assistant in obtaining the agreement of farmers not visited to cooperate.
At the end of the preliminary exercise, 34 potential cooperators were -identified. and 31 of these subsequently agreed to host trials. In addition, it was possi *ble to recruit nine more potential farmers at the time that guidelines for planting the trials were being discussed with selected farmers, giving a preliminary total of 40 farmers. Overselection of the provisional total of 24 was done on the anticipation that a proportion of these would drop out, and that perhaps less. than half would be willing to host more than one trial.
During the selection an effort was made to keep within the parameters relating to access to draught power, gender, and cropping history as indicated by Table 1. 40%. of households selected were headed by females, exceeding the target of one in three by 6.6%.. Approximately half of the households selected were ox owners, matching the target. With regard to reported cropping histories and plans, the proportion growing maize only was over-represented in relation to trial needs while the proportion growing sorghum was under-represented. A significant proportion were growing sunflower and other cash crops so that selection for the intercropping and
26
alternative late planted crops trials would be feasible.
Tle. 1 FChr a-er st icsE. of +arrmers selected
Type of HH. Draught Power Main Crops Other Ca-sh
Reported Crop
FHH 16 Ox Owner 16 Mze only 15 SF Only 5
MHH 24 No Ox.en 16 Mze & Sg 4 SF &/or
other 8
N.I. 0 N.I. 8 Staple + 14
Csh crp None 17
N.I. 9 N.I. 10
Tot. 40 40 40 40
Ma3. tchig farmers to tria.. Matching farmers to trials was achieved through negotiation with farmers, rather than by giving them a direct choice. Most of the farmers selected would have preferred to have the mixed maize variety, trial. However many of these were persuaded to accept other trials. In four cases farmers agreed to one trial in addition to the mixed maize trial, and in two other cases tr, ey areed to host both the intercropping and al ternati. ve late planted cr ps trial. Finding enough farmers for the sorghum trial was more difficult, refl ecting the importance (of the tr al from the farmers' point of view, Howev,.er, as the trial was regarded as important for strengteni. n I inks withI sorghum commodity researchers, an fort wa made to select extra -armers.- The matching of farmers to trials prior to planting was as follows; mixed maize trial 15, sorghum trial 11, intercrop trial 10, and late alt, crop trial 10 (giving a total of 46 trial sites and 40 farmers).
4.64. Problems encountered and lessons learned
Overall the selection procedure as described above functioned according to plan. It enabled a rapid select .on of a rel atively 1 arge number of representative armers in a confined area over a short period of time. This 1all owed the new trials assistant, during his first year of work, to cover his area on foot and effectively monitor all trials planted. The main problem experienced was the familiar one of farmers who dropped out during implementation of the programme. 16 of the 40 farmers who had initially agreed, failed to plant the trial. The main reason was that they were relying on assistance from others in ploughingq, and an arrangement made to plough on a particular day fell throCugh, upset, ting the
27
arrangement made with the trials assistant who had agreed to be present at the trial (not to direct it but record what the farmer did). This was especially a problem when the hired/borrowed oxen came ahead of the expected time. Another reason was a prolonged wet spell which encouraged weed growth to the point where ox owning farmers were using oxen for weeding in order to save their crop, rather than assisting other farmers with ploughing. This finding suggests that usi ng resource poor households as trial cooperators in oxcuilt ivati ng systems may considerably reduce the researcher 's control over management of the trial. Also if the trial is, farmer managed, the farmer should not be asked to give a date when he or she will plant; this is unrealistic. After giving careful instructions, the trials assistant can request to be informed immediately the trial has been planted. This will not greatly inconvenience the resource poor farmer when a
community approach is used and the trial assistant stays nearby.
While the level of farmer "dropout" was a problem, and probably 'higher than if selection had been based on volunteers recruited at meetings, the community approach proved resilient and also educative on the system. The trials assistant managed, without difficulty, to select five more farmers from adjacent fields and so the number of farmers planting stood at 23. four above the original target of 24. The team learned more about the relation between ox
owners and non-owners, and the way this can vary according not just to social arrangements and economic opportunities, but also seasonal variations in rainfall.
On reflection more time could have been spent collecting background information, particularly with regard to location of fields in relation to dwelling, and the scattering of fields. More detailed information on previous cropping histories would also have been useful, for the intercropping and the other trials too. Moi~e information on competing activities engaged in by selected farmers, especially
charcoal burning and mushroom gathering, would have been usefu~l also. Further lessons from this approach to farmer selection will be drawn from further analysis of trial results, and farmer feedback this _season.
A final point is that using a community approach like the one descri bed above was probably made easier because ARPT had been operating in the area for a number of years, and so local people mostly understood what was involved with hosting on-farm trials. If a new target area was to be selected for such an approach it would clearly be necessary to call one or more meetings within the community in order to explain the purpose of the on-farm research programme in advance.
47. 4:ar t, t no sa d emon str-tio s
28
For this stage a different approach to -farrer selection may be required after the -First year of on farm tests. After test demonstrations have proved suCcessfLul in the target area, the clustering of farmers used in the previous staqes will need to be dropped in favour of an approach which covers a wider geographical area. One or two farmers from each cluster may be selected along with a good number from other parts of recommendation domain. Farmers selected should be located near to denser population areas and in the centres of these. If the T and V system of extension is operating, the contact farmers will obviously be involved. However, care should be taken to ensure that these are representative of the target group in relation to such factors as gender, wealth, age, language, and local origin. One option is to. persuade extension staff to use some of the former farmer cooperators as trial farmers. Interestingly, in the Chipapa area discussed above, the extension officer responsible -for demonstrations has selected some of the poorer ARPT farmer cooperators -for the coming seasons demonstrations.
5.0. Conclusion.
The above are some ideas and suggestions relating to selecting farmers as spokesmen and spokeswomen, both as informants and cooperators, during on farm research. These are largely based on experience from -farming systems
programmes in Zambia, but it i-s hoped that practitioners from neighbouring countries may find them useful. The author
welcomes comments and suggestions from other practitioners in the region.
References
CIMMYT (1979), Zoning in Central Province: Demonstrations of an interdisciplinary aproach to Qlanning ada2tive agricultural research PCogrammes Report No. 3. CIMMYT East Africa Economics Programme, Nairobi.
Kean, S.A. and Chibasa, W. (1983), "Institutionalising Farming Systems Research in Zambia" in Farming Systems Newsletter. CIMMYT East Africa Economics Programme, Nairobi.
|
