9.2 Types of trials

Contents - Previous - Next

Once research priorities have been identified within the research programme and suitable hypotheses have been developed for testing, trials work can begin, The first requirement is for researchers -- and participating farmers, if appropriate -- to define the trial objectives as specifically as possible. This then will influence the type of trial that is most suitable for fulfilling the objectives. However, as indicated earlier (Section 2.4), FSD teams have multiple clients namely, farmers, station-based researchers, extension and development agency staff, and sometimes planners. Similar types of trials do not have equal appeal to all the clients.

One somewhat simplistic way of classifying trials undertaken on farms is differentiating them on the basis of who manages and who implements them (i.e., researcher (technician) or farmer (Table 9.1)). Thus, three major types of trials are possible:

• Researcher managed and researcher implemented (RMRI).
• Researcher managed and farmer implemented (RMFI).
• Farmer managed and farmer implemented (FMFI).

TABLE 9.1: EXPECTATIONS OF DIFFERENT TYPES OF TRIALS^a

a. There is a degree of subjectivity m some of the entries in the table, but they generally reflect what is the case. In a sense, these expectations also reflect the reasons why the different types of trials are undertaken. In the literature, RMRI trials sometimes are called exploratory trials, whereas RMFI trials sometimes are termed refinement trials.
b. Standard multilocational trials are also RMFI. Multilocational trials are differentiated from trials undertaken in a farming systems context in that they are not designed, implemented, or evaluated with a farming systems perspective in mind.
c. In a sense, there are two types of FMFI trials, those done through research- oriented farmer groups

A careful study of Table 9.1 indicates that the three trial types can be differentiated in terms of research objectives, methods, experimental design, types of data collected, methods of analysis, evaluation criteria, and likely appeal to the different 'actors' in the agricultural development (equivalent to validation trials in the literature) and the more widespread testing done through extension-oriented farmer groups (equivalent to verification trials in the literature). process. Generally in FSD, technology design work begins at the RMRI level; then when a technology is thought to be appropriate, it is passed on for testing by the farmer, first at the RMFI level, and then at the FMFI level, Obviously the farmer becomes progressively more involved as one moves from experimentation at the RMRI to the FMFI level. Farmer groups - which will be discussed later (see Section 9.8.6) -- are particularly useful for implementing trials at the FMFI level and to some extent at the RMFI level. In RMRI type trials, undertaken on-farm, the researcher input is much greater. RMRI trials on farmers, fields are most similar to those conducted on the experiment stations. Therefore, the level of testing achieved meets the standards demanded by station-based researchers. However, FMFI trials are the most satisfactory for the farmer and provide the most practical test of the technology. Because of management and resource constraints, yields or returns will diminish from the RMRI to the FMFI level. The information in Table 9.1 notes the major differences between RMRI work -mainly the preserve of experiment station researchers -- and RMFI and FMFI trials that emphasize on-farm work. One such difference, for example, is the tact that cause-effect relationships and 'hard' (quantitative) objective-type data are obtained more easily from RMRI work, whereas farmer attitudes and inputs into the research process are obtained more readily from RMFI and FMFI work undertaken on farmers' fields, However, data collected under such formats, particularly at the FMFI level, are likely to be 'softer' (more qualitative) and more subjective in nature. Once one understands the purposes of the different types of trials, it is easier to recognize the complementarities that exist between them and? therefore, easier to adopt appropriate criteria in evaluating their worth.

All three types of trials are vitally important in contributing to a well rounded FSD programme. Yet surprisingly, a survey of 41 FSD projects undertaken a few years ago indicated that only 32% undertook all three types of trials, 12% percent undertook only RMRI trials, and only 46% undertook any FMFI trials [Barker and Lightfoot, 1986].

There is also another type of trial, called superimposed, which has considerable potential that often is not exploited fully. Superimposed trials include elements of management by both farmers and researchers, with implementation primarily the responsibility of farmers. These trials tend to be single-factor experiments such as looking at the response of fertilizer superimposed directly on a farmer's own plot where he/she is providing both labour and management. Although, not strictly correct, these can be considered a type of FMFI trial.

Mainstream FSD, in spite of the iterative principle that underlies it, often has been implemented in somewhat of a linear mode. For example, often technology design work begins at the RMRI level; then when a technology is thought to be appropriate, it is passed on for testing with the farmer, first at the RMFI level and then at the FMFI level. Obviously the farmer becomes progressively more involved, on a daily basis, as one moves from experimentation at the RMRI to the FMFI level.

It is legitimate to be somewhat uneasy about this linear connotation. Because of this and other considerations, such as the research capacity of farmers, the comparative advantage of FSD, and the need to improve the efficiency of the technology development process combined with recent methodological advances, there are strong arguments for moving in the following direction with respect to on-farm trials:

• Farmers should be involved much earlier in the research process than often has been the case to date, Good examples of where this has been done involve farmer assessment of varieties on experiment stations and farmer input into designing and evaluating treatments (Boxes 9,1 and 9,2).
• Minimal emphasis should be placed by FSD teams on RMRI type trials, FSD teams have a comparative ad vantage in trials that maximize farmer participation and assessment (i.e., trials at the RMFI and FMFI levels).
• Methods that have developed in recent years should be introduced that will enable the needs of more than one client (i.e., researchers and farmers) to be better satisfied in FMFI type trials. Farmers' needs are more likely to be satisfied as a result of their being involved as much as possible in such trials, At the same time, researchers' needs can now be better met through:

• Using methods that can express farmers' assessment in a more quantitative manner -- which will more likely appeal to technically oriented scientists (e.g., using PRA techniques (see Section 8,4.4)).
• Making greater use of alternative statistical techniques such as regression analysis including modified stability analysis (see Section 10,4,3) [Hildebrand, 1984; Stroup et al, 1991],

Thus increasingly, the objective of ensuring farmer precision in assessing whether the technology fits his/her needs does not have to be accomplished at the expense of 'statistical' precision.

• Research resource efficient methods should be utilized in implementing on-farm trials, Clustering work (see Section 5,5) and using farmer groups (see Section 9.8,6), especially in FMFI trials, are good examples of such approaches. Also relating to maximizing the return from limited research resources, the trials should be sufficiently documented (e.g., contain site descripters) and potentially relevant to large numbers of farmers (i.e., the recommendation domain should be extensive).

BOX 9.1: INTEGRATE FARMER EXPERTS INTO ONSTATION RESEARCH

In Rwanda, Sperling [1991], Sperling and Ntabomvura [1994], and Sperling and Berkowitz [1994], have demonstrated that local commodity experts (i.e., farmers) can be brought usefully onto the experiment station to participate in the early screening of experimental bean varieties, Station based researchers learn farmer and user criteria, whereas farmers make selections for on-farm experiments. Thus, women farmers and research station scientists have collaborated in evaluating the new bean lines, Such collaboration recognizes farmers' knowledge of soils, seasons, arid planting practices. After four years of testing, farmers' and breeders' selections have been evaluated in terms of farmer acceptance and diffusion. The implication is clear: involve farmers as early as possible in the research process!

Use of superimposed trials should be encouraged not only because they provide a flexible approach in responding to unexpected experimental opportunities or issues, but also because the non-experimental variables reflect farmers' actual situation.

BOX 9.2: USE LOCAL KNOWLEDGE IN DESIGNING TRIALS

Local knowledge about soils, paddy types, and fertilizer resources available to farmers helped in the design of on-farm trials for the Hibunawan barrio on the west coast of Leyte, Philippines [Perrot Maître and Weaver, 19921. Results from these trials led to the design of technology in the form of strategies that are more appropriate for farmers than current recommendations.

From their experiences, farmers thought that rice response was most related to the soil and paddy types and to the timing of application, especially with respect to the type of fertilizer used. None of the Hibunawan farmers applied a pre-harrowing basal application, as was done on research stations, because of the high risk of crop failure that can not be discerned at that early period in the season. Farmers also observed several rice variety and fertilizer application interactions that would be important. Their foremost concern was on how to make the most effective use of set amounts of fertilizer available to them.

Farmers and FSD workers were able to design trials that incorporated these specific questions, including trial sites, as part of the experimental design. From these trials, participants hoped to complement and adjust the farmer's own rules-of-thumb to engineer helpful fertilizer recommendations.

9.3 Functions of different trial types

In designing all trials the objectives should be specified clearly, which, in turn, will help indicate the most appropriate trial type. In order to make this determination, it is important to understand the functions of the different types of trials. These are as follows:

• RMRI Trials. This format is used most commonly in FSD during the design stage, when the objective is to determine whether a new technology will function, in a technical and biological sense, in the physical environment of a farmer's farm. Because all of the work. at least with reference to the treatment variables, is handled by researchers, the trial design can be as large and complex as is needed to answer all the relevant questions. Generally, formal, replicated designs are used in these trials. Such trials, which are designed to answer cause-effect relationships, should be carried out on experiment stations whenever possible. The reasons for this include: lower implementation costs (e.g., in terms of logistics or time) and potentially better control (e.g., in terms of easier supervision or easier maintenance of ceteris paribus conditions). However, there are occasions when conducting such trials on farmers' fields is highly desirable and sometimes even essential. Such situations arise if the special environmental situation of the experiment station does not seem to provide a realistic environment for testing a technology (Box 9.3).

When RMRI trials are undertaken on-farm, it is desirable to have station based researchers play a major role in implementing them on behalf of? or in collaboration with, FSD team staff.

• RMFI Trials. The primary purposes of RMFI trials are to determine whether farmers can implement a particular technology using their own resources and to determine how much benefit the technology provides when this occurs, The benefit may be measured in many ways, including grain yield, net profits, cash returns to farmers' labour, net production per unit of ploughing time, etc. The participation by the farmer is essential in order to get a realistic picture of whether farmers will be able to implement the technology with the resources they possess, to develop a realistic economic picture, and to obtain initial farmer assessment of the technology,

At the RMFI level, researcher management still is required, because researchers need to collect basic data on what happens when farmers implement a new technology and on the performance of the technology, or technologies, versus the farmers' own production system. Accurate descriptions of technology performance and accurate estimates of values required for partial budget analysis (see Section 10.5.2) both require well designed and properly implemented trials and a good deal of fairly detailed data collection -- both technical and socio-economic in nature.

The involvement of the farmer has a major impact on the design of RMFI trials. It is necessary to keep in mind that the farmer is not a paid or unpaid labourer. Rather, the farmer is an independent person who is trying to run a business (i.e., the farm). Hence, the researcher should not take any more of the farmer's time than is absolutely necessary. For this reason, a good rule of thumb might be that trials should not require more than a single day for planting. If the farmer is using draught animals for ploughing, this will limit the size of the trial. The trial design should be simple enough that the farmer will understand clearly what the treatments are and how they need to be implemented.

• FMFI Trials. The primary purposes of FMFI trials are to determine how a technology will perform under farmer management and to gather extensive farmer assessments of the technology, so that it can be modified, if necessary, before it goes into the dissemination stage. Farmer assessments generally are collected through discussions (i.e., informal surveys) and through one-shot, formal, structured surveys. The more farmer opinions collected in this activity, the better, One good way to handle these types of trials is through the use of farmer groups (see Section 9,8.6). Because these trials are both farmer-managed and -implemented, they need to be extremely simple in design so that they can be implemented without any guidance from researchers.

BOX 9.3: RMRI TRIALS CONCENTRATING PRIMARILY ON TECHNICAL OBJECTIVES SOMETIMES NEED TO BE ON FARMERS' FIELDS

Much work relating to weeds needs to be implemented on farmers, fields. This is because the weed complex is likely to be very different from that found on experiment stations. Examples of such work are as follows:

• Field tests were conducted on-farm in Illinois, USA, to evaluate herbicide control of bigroot morningglory (BM) [Horak and Wax, 1992], BM is a weedy perennial vine with a deep taproot and occurs in corn and soybeans, particularly on sandy bottomland. Populations of this weed occur in patches on infested fields though these fields can be scattered widely. Such patches are identified readily by farmers but are not represented adequately on research stations in the area.
• Field tests to evaluate genetic resistance to herbicide effects among populations of Palmer amaranth (PA) were conducted at numerous on-farm locations in South Carolina, USA [Gossett et al, 1992], These tests were on natural populations at their original locations, because researchers were fearful that harvesting and importing seeds to research stations might spread the genetic resistance to other populations of PA in the area.
• Leafy spurge (LS) is a serious economic problem on rangeland of the northern plains of the USA [Lym and Carlson, 1994]. Research has focused on use of biological control agents because of the long-term resurgence of this weed following treatment with herbicides. Technology development work in North Dakota, USA, was carried out on farm range sites. This work lead to a strategy of combining leafy spurge gall midge, a biological control insect that is most effective at LS control in wooded areas, with herbicide treatments to knock back LS populations in open areas of the range. The midge was able to exert biological control in open areas once the LS populations were reduced by the herbicide.
• Another weed technology is being designed through on-farm research. Common crupina (CC), a fall germinating, annual range weed, was first recorded in the northwestern USA in 1969 [Zamora et al, 1989; Zamora and Thill, 1989]. Its distribution is still limited, and a plan is being designed to eradicate this weed before it becomes a major economic threat. Seedling emergence and seed bank longevity were monitored in farm range areas during a four- year period to understand the biological characteristics of CC necessary to plan its eradication. Farmers on whose land CC is growing are participating in assessing human health risks and potential environmental disruptions from the eradication treatments before the plan is implemented.

Contents - Previous - Next