6.5 Research resource considerations

Contents - Previous - Next

Maximizing the return from the limited resources available for research must always be a major consideration. Two areas that are important in ensuring that research resources are allocated in an optimal manner are considering very careful the level of understanding that is required -greater accuracy often requires more research resources - and ensuring that the research priorities selected are the correct ones. These two topics are considered in a little more detail in the following sections.

6.5.1 Degree of Understanding Required

At first glance? it seems that going through all the steps described above could take a great deal of time. This is likely to be the case, if an effort is made to collect at first hand large amounts of accurate quantitative-type data. However, this is not always necessary and, therefore, is not desirable. After all, collecting descriptive information is not an end in itself but is used to provide an input into designing and testing work,

BOX 6.4: BOTH HIGH AND LOW LEVERAGE ACTIVITIES CAN BE USEFUL

In Botswana, FSD teams have worked with both high- and low-leverage interventions. For example, a good deal of the work done in collaboration with the commodity researchers on the experiment station has a low leverage characteristic -- particularly in the case of an enterprise that is not a major component in the farming system. However, adoption of the proposed changes can help improve the productivity of that enterprise, in particular, and possibly have some impact on the farming system as a whole. This occurs if the enterprise concerned uses a significant proportion of the resources involved in implementing the farming system as a whole.

FSD teams in Botswana have been willing to support work with low-leverage characteristics because of the difficulty, in this harsh climatic environment, of developing interventions with a high leverage characteristic that will be adopted readily by farmers. The difficult environmental conditions provide little flexibility in the farming system, because a key management factor is the ability of farmers to plough and plant when they are likely to optimize the use of water available for plant growth and maximize efficiency of water use. Therefore, researchers are faced with the challenge of trying to develop relevant technologies to break a major constraint, rather than having the easier task of developing technologies that will avoid the constraint by exploiting the flexibility that exists within the farming system.

Under the harsh climatic conditions of Botswana, major emphasis has had to be on developing technologies that break constraints. Adoption of such technologies by farmers involves major changes in their farming systems. For example, much of the FSD work has concentrated on changing the ploughing-planting operation to improve water availability. However, strategies to improve water availability to the plant at planting time require more draught than the traditional strategy of broadcasting the seed and ploughing it in.

Therefore, in summary, a good deal of the FSD work done in Botswana 'with a pre-determined focus, tends to have a low-leverage characteristic, whereas FSD work 'in the small' tends to have a high leverage characteristic. Unfortunately, it involves interventions that have to break the major constraint of limited water availability, rather than being able to exploit the little flexibility that generally exists in Botswana farming systems.

In deciding what quantitative-type data to collect, guidelines such as the following can be used:

• As much use as possible has been made of existing secondary information.
• Very careful consideration is given to just how accurate descriptive information needs to be in order to provide an adequate guide or input into the designing and testing stages of FSD. Emphasis is placed on collecting the required information in the cheapest possible way, with a method that allows attainment of the degree of accuracy that is required. Examples of cheap methods are those that emphasize qualitative rather than quantitative type of data and relative ranking of data rather than actual accurate measurement (i.e., ordinal rather than cardinal). Also, greater interaction with, and participation by, the farmer at all stages of the FSD process can provide many short-cuts in the research process,
• Informal interdisciplinary surveys are used whenever possible to collect information required to fulfil the point above that cannot be obtained from secondary sources, The major respondents for this information are the farmers themselves, If necessary, these findings are supplemented or confirmed with a formal one-time survey that can be analyzed quickly (see Section 7,2).
• Bearing in mind the three preceding points, the amount of time taken in doing the initial descriptive/diagnostic work is minimized so that design and testing activities are started at the earliest possible opportunity.
• Descriptive/diagnostic activities to verify the initial quick descriptive/diagnostic work, if necessary, are undertaken at the same time as the initial testing stage, These descriptive/diagnostic activities can be in the form of more formal surveys and studies such as monitoring fields, etc. In fact, results arising out of design and testing activities can help in refining future descriptive/diagnostic activities. This is an example of the iterative nature of FSD work,

6.5.2 Setting Priorities

During the descriptive and diagnostic work, it is likely that farmers will articulate, and researchers will observe, more problems in the farming system than can possibly be addressed, Researchers' time and resources often are limited, so they must consider carefully how much work can be done and restrict the number of activities undertaken, Setting research priorities is a complex process and will be influenced somewhat by information and analysis arising out of the steps that were outlined in Sections 5,7 and 5,8, Four general factors that are important in influencing the setting of the research priorities are as follows:

• Farmers' objectives have a direct effect on the types of technology that they will adopt. If researchers and farmers have different objectives, a good deal of research work may end up being wasted. Therefore, it is very important, to try and find solutions to the problems farmers feel are most important. They are likely to result in technologies that have high-leverage characteristics,
• Other clients for FSD teams are station-based researchers who play a major role in providing a 'bank' of potentially useful technologies that can be considered in the design stage and used in the testing stage. Although the farmer remains the major client for FSD work, FSD teams can help in developing a more useful 'bank' of technologies by responding to on-station researcher requests to test specific potential technologies on farmers' fields. Some of these are likely to have low-leverage characteristics.
• To benefit from the interdisciplinary nature of FSD, research priorities must be set by the whole team, and not just by discipline (see earlier discussion in Section 5.7,1 ). For example, if descriptive and diagnostic activities indicate that labour is the most limiting factor in the system, then the agronomist should give priority to this and work with the economist to develop labour-saving technologies or increase the return per unit of labour used during the labour-constrained periods, rather than addressing some other topic that may be only an agronomic issue, such as rice variety yield trials. Any rice varieties selected for testing should be based on the preferences of the farming households and take into account the resources (e.g., money, labour) that farmers have at their disposal.
• The mandate and the location of the research will have an influence on setting research priorities. For example, a team may be responsible for increasing crop production among limited-resource farmers working in a particular area. Thus, the team usually starts out with some target areas and specified objectives.

Other important issues to consider when setting priorities for research include the following:

• How common is the problem within the research area, and how many families does it affect? Higher priority should be given to problems that affect most families.
• How severe is the problem and how much does it affect the productivity of the system? For example, head smut in sorghum may occur in all fields throughout a region, but the actual reduction in grain yield may be very small. Higher priority should be given to problems that have the most severe effects on farm production.
• If there is major concern about addressing intra-household relationships via the technological route, considerations mentioned under the two previous points may need to be modified. For example, helping improve the position of women within the household may involve low-leverage types of interventions that will not necessarily greatly improve the welfare of households as a whole.
• What is the potential for solving the problem? It may not be appropriate for an FSD team to tackle a problem like bird control, even though all parties recognize it as a problem. FSD teams generally should give priority to problems they feel they have some chance of solving and try to feed back problems they can't solve to other groups that have a better chance of identifying a solution (e.g., station-based researchers).
• If solutions to problems are likely to result in technologies that require new, or changes in existing, policy/support systems, then it is important to ascertain the potential attitude and responsiveness from those responsible for developing/implementing such systems. There is little object in undertaking such research if there is no hope of any changes. An analogous idea applies to research that is likely to have a long gestation period (e.g., much livestock research). It would be important' for example, in pert-urban dairy research to know the priorities of government in supporting such systems both now and in the future.
• Are there areas of opportunity for increasing farm production that are being under utilized at present? For example, could certain types of fruit trees planted around the home compound provide useful income? Not all improvements come from solving problems. Some may come from starting new and profitable activities.

Part III - Methodology for the farming systems development approach

Descriptions and 'how to' guides for a number of data collection and analytical methods are presented in this part. The four chapters in this part, which contain information for both technical and socio-economic scientists, are devoted to:

• An overview of the data collection methods used in FSD (Chapter 7).
• A discussion of survey methods and their implementation in FSD (Chapter 8).
• A review of trial methods and their implementation in FSD (Chapter 9).
• An introduction to analytical and evaluation techniques used in FSD (Chapter 10).

7. Data collection methods

7.1 Objectives of chapter

The objectives of this chapter are to:

• Briefly outline and evaluate the different methods of data collection.
• Indicate the major methods of data collection desirable at each stage of the FSD process.

7.2 Common data collection methods

Obviously, both technical and socio-economic data are needed in the development of relevant improved technologies for farmers. For example, the time spent on ploughing a field is dependent on the type of draught used and the number of hours per day that it is used. However, ploughing time also is determined by a number of technical factors, such as moisture in the soil and the soil type. Unfortunately, data collected and analysis undertaken often reflect only socio-economic or technical data. Looking at only one aspect does not represent the whole picture. Collecting both types of data is not always possible, but the limitations of collecting only one type should be recognized.

The most common methods of collecting data in FSD work are the following:

• Surveys (interviews and questionnaires).
• Observation.
• Direct measurement.

Surveys, which involve interviewing, include many different types. One possible schematic breakdown, although not perfect in that it would probably not be universally accepted, is given in Figure 7.1. It is important to note that usually, in formal (i.e., structured) surveys, only one respondent is interviewed at a time, whereas in informal (i.e., unstructured) surveys, respondents may be interviewed individually' or quite often, as a group. Direct observation, because of its expense, has tended to be used sparingly in FSD work and, when utilized, has tended to be undertaken in conjunction with other methods. Direct measurement activities often are associated with monitoring and trial activities (see Section 8.6).

Figure 7.1: Types of surveys (interviews)

A number of factors influence what method or methods of data collection are used. Some of these are:

• The discipline of the researcher(s) and the objective of the specific research. For example, historically in FSD-type work, social scientists have tended to concentrate on surveys, economists have relied more heavily of formal surveys, and sociologist/anthropologists have emphasized more informal surveying and direct observation methods. In contrast, the major thrust of technical scientists has been, and continues to be, direct measurement. Attitudinal information is collected, for example, by informal survey methods, whereas technical information concerning performance of technologies requires the implementation of trials.
• Quality and quantity of research resources available.Limited financial resources obviously limit the use of direct measurement and formal surveys involving multiple visits. At the same time, the skills involved in undertaking informal surveys may inhibit their widespread use.
• Time available to undertake the study. Surveys, particularly informal, are clearly less time-intensive than activities involving direct measurement,
• Whether the type of farming system on which the study is to be undertaken is complex or simple (e.g., year-round versus seasonal farming, mixtures versus sole stands of crops, etc.). It is difficult to make categorical statements on the relative ranking of the data collection methods, although assuming all other things to be equal, direct measurement and observation are likely to yield the most accurate results in more complex systems.
• How important it is to reduce sampling error. Basically this means that a sample chosen from a population does not perfectly represent the population. For a number of reasons, sampling error tends to be a problem in much FSD work (e.g., because of imperfect sampling frames, use of small samples, etc). Unfortunately, the likelihood exists in FSD that sampling error, which generally means larger samples, increases the possibility of larger measurement errors.
• How important it is to reduce measurement error. This relates to mistakes in the enumeration and analysis of data. The type of data being collected determines whether a high or low potential exists for measurement errors. This is illustrated in Figure 7.2. There are two continua:

• The 'single point' to 'continuous' continuum (i.e., relating to an event or activity that occurs over a brief time span, as against one that has much longer duration).
• 'Registered' to 'non-registered' continuum (i.e., relating to an event whose occurrence and dimensions leave a written record or distinct and vivid recollection as against one less likely to be recorded or remembered).

Usually, measurement errors are lower with single-point registered data (i.e., when memory recall would be good) and higher with continuous non-registered data (i.e., when memory recall would be bad).

It is possible to subjectively evaluate the methods of data collection in terms of the above factors. In Table 7.1 the significance of each factor or operational constraint with respect to each method is given a ranking of I to 6, with '1' indicating a very favourable ranking and '6' indicating a very unfavourable ranking.

TABLE 7.1: EVALUATION OF DATA COLLECTION METHODS IN TERMS OF OPERATIONAL CONSTRAINTS^a

Source: Modified from Kearl [1976].

a. Evaluation of data collection strategies in terms of cost per unit with respect to various operations (1=lowest cost per unit, 6 = highest).
b. Evaluated in terms of ability to reduce sampling error.
c. Evaluated in terms of ability to specify sampling error.
d. Single-point registered data.
e. Continuous non-registered data

Because of different disciplinary needs, as well as needs to mesh together the use of limited research resources, a strategy consisting of a combination of methods normally is used. The results in Table 7.1 show that no one method of data collection is basically superior in minimizing operational constraints.

As indicated earlier, careful thinking is required about how accurately the data must be collected (Section 6.5.1). Keep in mind that the costs (e.g., financial and time) associated with a data collection method should be comparable to the type of information sought. For example, multiple-visit surveys for quantifying actual seasonal labour flows are expensive and require a good deal of time. A rough estimate based on a much cheaper and quicker, oneshot survey, in fact, may be sufficient for the purpose required. It is important to consider whether a more accurate measurement would improve the understanding enough to justify giving up other opportunities, such as working with more farming families, As indicated earlier, depending on the type of data being collected, measurement errors are reduced with more frequent interviewing techniques, whereas sampling errors are reduced by involving larger numbers of farmers. Because, in FSD work, research resources are invariably limited, a trade-off between these two errors is necessary.

Related to the above, and to efficiency, is the idea that cheaper, qualitative type data sometimes may provide sufficient understanding rather than expensive, quantitative type data. Qualitative information can include not only attitudinal information, but information on the relative labour requirements of different operations, etc. Limiting quantification to key characteristics reduces the cost involved in collecting data,

Surveys involving interviews are used a great deal in FSD, Figure 7.1 illustrates the different types. The most suitable type of interview will depend on circumstances. In general greater emphasis now is being placed on informal surveys to get the necessary information in a quick and efficient manner. All surveys, no matter how they are undertaken, need to be designed to facilitate quick processing so as to simplify the transfer of data to computerbased systems.

Structured interviews, using schedules with pre-coded, closed-ended questions (see Section 8.5), are potentially more efficient in this regard and should be used whenever possible, especially il.:

• The researchers themselves are not conducting the interviews.
• The enumerators are not particularly well-qualified.
• Comparative analysis, such as cross tabulation, etc., is to be undertaken.
• Statistical analysis is required.

7.3 Major data collection methods by FSD stage

Table 7.2 is an attempt to indicate the major data collection methods now considered by most FSD practitioners to be best at each FSD stage. Once again, there is a degree of personal subjectivity about the degree of importance attached to each method in each stage of FSD. Indeed, some of the weighting reflects what is considered to be desirable rather than what is currently done. As examples, multiple-visit formal surveys used to be very important in the descriptive/diagnostic stage, and a one-visit formal survey still is considered to be important by some practitioners in the same stage. Also impact/adoption studies still are not undertaken in the dissemination stage to the extent that would be desirable. Therefore, it is important to note that changes have occurred in priorities given to different methods over the years. Reasons for this relate to evolution in methodology and increasing concern with maximizing the return from limited research resources. Related to the latter have been three considerations, namely:

• Not aiming for near perfect understanding and great accuracy but doing enough to get the job done. Attainment of such goals would involve considerable expense.
• Moving through the different stages of FSD as quickly and efficiently as possible. Obviously, this helps in improving the return from limited research resources.
• Development of better technologies but not necessarily the perfect solutions to problems. All farmers are different in some respect, and the perfect solution, or near perfect solution, is an impossible goal idealistically and practically in terms of the limited resources available for research.

TABLE 7.2: CURRENTLY USED DATA COLLECTION METHODS BY FSD STAGE^a

a. Those efforts not involving primary data collection are separated out in the table. Also within each stage, a subjective evaluation is given of the importance of each method with 1 indicating most important. A value of 5 indicates that it is generally undesirable and should be undertaken only in extreme situations. Values between 1 and 5 reflect the gradient between important/desirable and unimportant/undesirable. Cells with no entry reflect little or no reliance on that method/approach.

8. Survey methods (indirect measurements)

8.1 Objectives

The major form of data collection through indirect measurement is via surveys. Therefore, this chapter concentrates on surveys, The objectives of the chapter are to:

• Provide guidelines on eliciting the cooperation of the respondent,
• Describe different types of informal and formal surveys.
• Discuss some of the issues relevant to the implementation of informal and formal surveys.

8.2 Obtaining farmer cooperation

Without doubt, in order to benefit from any type of collaboration with farmers, it is important to develop a good relationship with them, The initial approach to the farmer and the relationship that develops over time, are critical elements in creating an environment conducive to collaboration, Two points are important in developing a good relationship:

• When working in the village for the first time, the objectives of the survey, work programme, and the proposed approach should be explained to, and approval for undertaking the survey should be obtained from, the village leadership. If desired by the leadership, a presentation can be made at a village meeting. Such meetings provide an opportunity to obtain general information about the village, institutions, etc.
• It is important to respect the customs of the village. Related to the issue of respect is the need for a collegiate or partnership type of relationship to develop between the FSD workers and farmers. So often, because of differences in educational levels, a paternalistic or master/servant type of relationship develops.

There are a number of tips on executing interviews in a way that will maximize the interaction between the farmers and team members and the value of information that is obtained. Four stages in the interviewing process are [Rhoades, 1982]:

• Approach. Keep a low profile; interview when it is convenient for the farming family; whenever possible, conduct the interview at the farm,
• Warm-Up. Use a polite form of address; if necessary, make appointments to discuss the survey; take time to approach the survey topics (i.e., be prepared to talk on other topics of interest to the farmer); if desirable, indicate you are there to learn and he/she is the expert,
• Dialogue. Be natural and relaxed; let the discussion flow and be flexible with the ordering of the questions; use plain understandable language and terms farmers can relate to; make sure the questions take into account the cultural setting, avoid sensitive questions at first, and, if possible, obtain such information through indirect questioning; if the farmer can't answer a question, try rephrasing it; don't ask questions that are too abstract; observe the farmer's reactions to questions, because these may reveal a great deal about his/her concerns or reservations; remember that what people say and do may be different; record the information in writing during the interview only if the farmer does not appear to be inhibited or suspicious; don't let the interview last more than 30 to 45 minutes unless the farmer is talkative, etc.
• Departure. Bring the conversation to an end when the topics have been discussed or the farmer can spare no more time, thank the farmer for his/her time, and depart respectfully.

8.3 Formal (structured) and informal (unstructured) surveys compared

Surveys are obviously very useful and efficient means of collecting data. Table 8.1 presents, in general terms, some of the distinguishing characteristics of the two major types of surveys, namely informal (unstructured) surveys and formal (structured) surveys. These characteristics highlight the different uses and means of implementing them.

Contents - Previous - Next