David Broshenka and Alfonso Peter Castro
1. General considerations
2. Data collection
3. Selecting research sites and study participants
4. Field workers and study logistics
1.1 Topics on which to gather information
We present here some guidelines for carrying out social surveys in connection with wood fuel projects., Our aim is to indicate what techniques are available, what sort of information should be gathered. We generalize, because generalizations can be made, but we remind readers that there is much variety in communities, and each specific situation requires its own approach, and its own modification of general surveys. Moreover, the amount of manpower, financial and organizational resources available to fuelwood investigators inevitably varies and further constrains generalizations about how surveys "ought to be" carried outs.
Surveys can be conducted in many ways, from a small-scale, quickly organized examination of one specific local situation, to a year-round national survey on a large scale. In this paper we are primarily concerned with more restricted surveys, not with elaborate national surveys. Surveys vary not only in scale, but also in location, objectives and methods, We indicate which of these is appropriate for specific needs. No single source - written, oral interview, observation, questionnaire - should be regarded as sufficient in revealing all aspects of fuelwood use. A combination of methodological approaches and procedures is recommended in conducting any investigation of the sociological context of fuelwood uses The aim is to build up many sources, so as to cross-check for reliability.
An ideal survey would permit generous allocations of time, and people who were experienced in social surveys. The ideal period would be at least a complete year, which would permit the important seasonal differences (how is wood gathered in the rainy season?) to be noted. But we recognize that most readers will not have the resources to do an ideal survey. In many cases the survey will be done in response to an urgent request or order from a government ministry or an international agency. "The district administrator at X says that the people have a big problem getting firewood - go and find out what exactly the position is. We need a report by next month." Here a quick simple survey is all that is possible, but we believe that it is possible to get good and relevant information even in such difficult circumstances, and we suggest ways of doing this.
Sometimes officials will have the responsibility to resist pressure to do a survey in too short a time, where it would be impossible to collect the required information. Rapid surveys can be valuable, but they have their limits of usefulness and accuracy.
These suggestions are intended for people like forestry, agricultural and other government officials who are called on to make a survey of wood fuels and other fuels. We need to emphasize the importance of other fuels, for in some parts of the world, at least at some seasons, fuels like crop residues or animal dung may be the main source of cooking fuels A good fuelwood survey is really a social survey, because it deals not only with wood, but with how a particular society manager its resources, with factors such as control of, and access to, trees; division and organization of labour; patterns of using fuels. These reflect the basic values of that society, showing who is important and privileged, and therefore does not suffer when fuel is scarce or conversely, it will also show which social groups and individuals, such as the poor, women, elderly, and handicapped, suffer the greatest hardships from fuel scarcity. A proper fuelwood survey should always involve a two-way process: on the one hand is the resource - wood and other fuel sources - and on the other hand are the rules (and technology) that society uses for managing that resource.
We stress the need to be flexible, to use common sense and courtesy, and to respect the views, values and rights of the people themselves. They should not be viewed as a passive entity waiting to be developed, but rather as a people who have their own perceptions and their own technical knowledge, and who have much to contribute.
This is intended to indicate the more important topics for investigation. Exactly which areas are explored, and in what depth, depends on specific circumstances of location, other information available, and precise needs of the survey, as well as on funds, time and staff available. Much information can be gathered, at least in broad outline, fairly easily and rapidly. In all surveys there is need of a balance between obtaining too much and too little information. For fuel projects, the aim should be to be able to decide on appropriate questions that will provide useful answers and to present enough basic information so as to make the actual fuel data meaningful. Information about collection of firewood, or manufacture of charcoal, or cooking patterns, does not make much sense unless it is related to the social background, to the specific people and place being studied. Information should always be gathered with fuel in mind, and it is important to resist collecting interesting facts on every aspect of life, then not knowing what to do with the results. No single check-list adequately covers all informational needs. Rather, what follows suggests the range of information that may be relevant, depending on the objectives and limits of each survey. First the objectives must be established, then the next step is to work out principles of selection - what needs to be included? What sort of questions must be asked?
Demography. "How many people live in this area?", is clearly a major factor in determining the demand for fuel. Other questions concern the density (how many people per square kilometer?) and the age and sex distribution. How many elderly or sick people who may be handicapped in getting fuelwood? What are the rates for births and deaths (if known) and how do these compare to national figures? What are the rates and patterns of migration (is it seasonal or permanent? are the people moving in from other areas?). These comparisons are important, as we need to know how "typical" any particular area is.
Environment. What is the physical area that we are dealing with? Is this clearly marked or should-it be arbitrarily defined? What resources (especially land, soils, vegetation) are available? What is known about climate, water, slopes, drainage? This information provides some sense of adequacy of reserves (actual and potential) of fuel.
Historical. There are two good reasons for knowing at least the general outline of local history recognizing the significance of important places and major persons. First, the present can only be understood in relation to the past, even in such an apparently ordinary study such as fuel. Second, local people regard their history as important, so it is both expedient and courteous to know something about significant events. Patterns of fuel use should not be considered as being historically static, since incidents and changes such as population movements, the expansion of farming, the building of roads, wars, or the growth of commercial timber operations may have had important impacts on fuel usage.
Community. We are studying a community of people, the most important part of which is the relatively stable set of relationships between the people - relationships between men and women, old and young, neighbours, kinsmen, in-laws, landlords and tenants, rich and poor, and so on. The relationships involve both conflict and cooperation, and are expressed in specific activities, one of which is a set of activities that concerns fuel. So fuel activities will both help to clarify the social structure of a community, and also be more intelligible when the social structure is understood.
The community will usually focus on a central place, a village or a hamlet, and there will generally be clear patterns of settlement and land use, which relate, of course, to the environment.
Domestic. It should be possible to make a rough typology of "households", a term which although sometimes vague, is better than "family" or "farm". Household may usually be defined as a group of people that shares a common kitchen (or cooking place) and that recognizes one household head. Households can be distinguished by size, and numbers of members affects both demand for, and supply of, fuelwood. Households also differ in income, which is usually difficult to measure as Many people are unable or unwilling to give precise figures of income. So we look for 'proxy indicators' that will give a good idea of income and wealth. These may include land (ownership and use), livestock permanent trees, type and size of compound and houses (materials, roof, furniture), and selected personal and household possessions like jewelry, watches, bicycles, radios, clothing - it is easy to establish a simple index for a particular area, then select a sample of households (see below) and use a Guttman scale to show the wealth graduations (see field manuals recommended in subsequent sections of the paper for description of Guttman scale).
The purpose of this is to establish the range and extent of inequality and variation between households, and to relate this to patterns of fuel use, and fuel sale (especially for export).
Social and Economic. This category seeks further information on differentation - how is access to land defined and who controls/owns the land? How many are landless or nearlandless, how many rent land? These are crucial questions in considering access to fuelwood. Who owns trees, if this is separate from land ownership? What are the main production systems, for agriculture (subsistence and cash-crops) and livestock? What are the general patterns for technology, markets, credit and indebtedness, commerce, communications, occupations, division of labour by age and sex? What are the main social, ethnic, and religious divisions, and how do these relate to economic status? While it is not necessary to obtain detailed precise information on all these topics, they are all significant for fuelwood. For example, the presence of an established, regular and thriving market, connected to the outside by all-weather roads, will facilitate fuelwood sales.
Services. What services are provided by government or private agencies in the fields of education, health, agricultural extension, forestry, community development and commerce? Where are offices located? How often do officials visit? How do the services compare with those available to other communities? This sort of information would be important especially if the survey calls for recommendations such as tree-planting.
Political-Administrative. What are the formal and informal channels of authority? What are the links to the regional (and national) centers of power? What is the extent of-local participation in making decisions? What laws, regulations, and local informal, sanctions affect fuelwood - e.g. who is allowed to out which trees?
This list is not comprehensive, and needs to be modified for particular circumstances. It would be easy to decide, for example, how much information was necessary on education, apart from basic essentials such all details on schools (location, grades or standards, male and female pupils, teachers) and literacy. In some areas residential schools may be major buyers and consumers of fuelwood, or teachers may be the main buyers of charcoal; in other areas, a high proportion of children in school may mean fewer children available for help in fuel collection. The aim should always be to relate the enquiries to the specific goal, fuels Thus, it might be important to ascertain how a fuel shortage affected diet and health. In arranging a social science survey, one needs to be constantly critical - "Am I asking the right questions? What have I neglected? How can I get information quickly on this subject?"
Fortunately, fuel in many areas is still a relatively open topics Unless an individual is engaged in illicit activity like theft of fuelwood, cutting trees in a forest reserve, or selling charcoal without a licence, he or she is likely to be cooperative and responsive, as most people already perceive fuel as a problem, and they welcome any prospect of alleviation. Even individuals who are technically involved in illegal aspects of fuel use have often provided valuable information to investigators. This depends, to a large extent, upon who the investigator is. That is to say, an "independent" researcher may at times have an easier time gathering information on some subjects than a forestry service officer, although this generalization is by no means always correct.
In all cases an attempt should be made to compare the community studied with others in the same locality, region or nation. In come countries (India, Kenya) much good information is available for comparative purposes. For most countries, however, there is little detailed information about fuelwood availability and use. Thus, it is advisable that when conducting a survey, some specific areas be selected for surveying.
Certain basic information on the community must be gathered, but it will seldom be necessary to collect detailed information on, for example, ritual, magical beliefs, or details of marriage. However, it will be desirable to know whether ceremonies affect fuelwood use, as does the Hindu practice of cremation, which creates a great demand for wood. It will be necessary, regarding marriage, to know how divorce affects a woman's chores and her access to firewood. Does a divorced women who lives alone, or with her children, face greater difficulties in ensuring an adequate fuelwood supply? The point is to determine which sorts of information are relevant, and to strike a balance between knowing enough about a community to have a good general picture, on one hand, and on the other hand, forever collecting trivial items because they may come in handy.
Most guides to anthropological fieldwork are not particularly relevant for our present purposes, because they apply to situations where much detailed and precise information must be collected, and long periods of fieldwork are involved. For many fuelwood surveys, there will not be time for elaborate and detailed investigations, and it will be impractical to become bogged down in fine questions of statistical significance and methodology, although surveys must aim for accuracy.
"Rapid rural appraisal", uses relatively simple techniques, and according to Robert Chambers "Simple is optimal" (Chambers, 1978). Here are some simple suggestions:
1. Walk, observe, listen and ask questions.
2. Remember that a quick visual survey can accomplish much.
3. Be flexible, use common sense and realise that no single approach is universally appropriate.
4. Always be aware of the range and variation of behaviour and attitudes as well as of the cluster in the middles. Try to discover norms or averages.
5. Involve the local people in the survey.
6. In interviews and conversations, develop your own style that is congenial to your personality, rather than attempting to use an imagined superior models
7. Don't ask more than you need: avoid collection of unnecessary surplus data.
8. Look at other, comparative communities.
We provide more details below of how to implement these suggestions.
2.1 Written sources
2.2 Key informants and informal informational networks
2.5 Group interviews
In this section we present an overview of the procedures used in collecting field data, to introduce the reader to basic methods that will be useful in conducting a fuel survey. Those readers interested in more in-depth treatments, can consult some of the standard books on field methods. (See, for example, Yang, 1955; Mbithi, 1974; Kearl, 1976; Connell and Lipton, 1977. There is also a good discussion of methodology in case studies such as Lewis, 1951; Hill, 1972).
The following section is divided into six parts. The first describes various written sources containing information on rural communities. Second, the advantages and disadvantages of using key informants are discussed. Part three considers the importance of observations, with participant and nonparticipant observation, an well as time allocation studies, being reviewed. In part four, interviews are examined. Questionnaires and how to frame questions are dealt with in part five. The final part is concerned with methodological pitfalls and biases and with their resolution. Usually information is gathered in several ways, one being used to supplement the others. For example, in collecting information on fuel used for cooking methods include measurement (of fuel used), interviews (what people say they do), observation (trying to decide what they actually do using ears, eyes, and nose!). No method is superior, all are necessary.
Information on particular rural communities or regions may be available from a variety of official, historical, anthropological, or other sources. First, reports from the central government, such as the census bureau, department of agriculture or the ministry responsible for forestry, or technical ministeries of roads, public works or water, may contain community level data. The quality, quantity, and frequency of publication of the data vary considerably nationally.
Local governments and local branches of national agencies may be able to provide useful information. For example, local tax rate records were instrumental in selecting stratified random samples for a fuelwood consumption survey in the Sudan (Digernes, 1977:51). Local forestry offices may have records on the sale of trees, or estimates of illegal cuttings. Agricultural officers are often a good source of information, as are officers such as Kenya's District Development officers.
The social science literature is another important source of written information. Farm management studies, ethnographic reports, and sociological surveys can supply valuable background facts on a particular community, region, ethnic group, or other socio-cultural grouping. Unpublished limited circulation reports can be very useful, These studies should be treated in the same way as any "fallible" historical source. At best, they provide a baseline against which present conditions can be compared and contrasted and are a good source of hypotheses to guide and direct the survey. In terms of specific data on fuel use, however, the social science literature generally contains little detailed information., Most major cultures of the world are represented by at least one outstanding ethnography which provides a reasonably up-to-date social and ecological account. A good example is provided by the Ethnographic Survey of Africa, which covers all major ethnic groups, and which provides excellent basic information, and bibliographies, on social, economic, domestic, and ritual life. These volumes, in English and French, are published by the International African Institute.
Other written sources may shed light on a community, its history, and patterns of fuel and resource use. Scientific accounts of local resources, local histories, church records, travellers' accounts, and records of plantations or mines are important sources. Once again, caution must be employed when using such sources as some records might be out-of-date and incomplete.
Available written records help to provide the historical background to the particular unit of study and populations Moreover, they can help to establish the social, cultural, political, economic, and ecological context of fuelwood and other energy uses in the community. As Oscar Lewis' account (1951) of Mexican villagers has shown, local patterns of forest use change throughout time, and an -understanding of these changes and their implications requires an understanding of the local milieu and its relationship with larger socio-economic and political forces.
Local universities, technical colleges and research institutes may be good sources for unpublished research. Inquiries should be directed not only at the social sciences, but also at natural sciences. A botanist, zoologist, ecologist or geologist, for example, who had worked systematically in an area, could be an excellent source for certain sorts of information.
A key informant is a person who is knowledgable, who has contacts and who is willing to talk. Sometimes key informants are self-selected, in that the local "Big Men" (as they are called in many parts of the world) are often the first to present themselves to a field investigator. The key informant is often a prominent, articulate person (usually a man) of some consequence, such as a village headman, merchant, landlord, large farmer or headmaster of the local school. Important key informants may also include people who are not members of the community but who have valuable experience and insights, such &a an agricultural or administrative officer, or a social scientist who has done research in the area. In a fuel survey, it might be advisable to choose a specialist such as a charcoal-maker or a woodcutter to supplement the views of the leaders.
Two disadvantages of using key informants are the problems of bias and representation. In selecting - or, as sometimes happens, being selected by - a key informant, it is well to remember that no one individual can ever represent all perceptions, all viewpoints, of his/her community. People always differ according to sex and age, and usually other differences are the result of wealth, status, occupation, or particular sorts of knowledge and insights. Sometimes an outside researcher will be adopted by an apparently well-informed and friendly person who later turns out to be marginal to that community. He may be marginal because of his unorthodox views, or because he has spent many years outside the community, or because of his personal history; this means that he will not be a good source of typical ideas and attitudes of that society, although he may help in analyzing behavior, from his unusual 'outsider' viewpoint. In general, key informants should not be relied on for information without other means of data collection that will check their accuracy. In spite of the disadvantages, key informants can be useful in fuel surveys. Where there are cost and time constraints, use of key informants may be a useful and rapid method for collecting data. Indeed, the only effective way of getting certain kinds of information quickly may be to find one well-informed and talkative person. At the preliminary stage of research, key informants can help the investigator to become familiar with the area, can explain local aspects of the particular topic of investigation, and can also help to develop preliminary hypotheses to guide the study. These hypotheses can be used in developing pilot questions, to be tested on other people. It is essential to develop a network of informants, extending out, perhaps, from the original key informant. One advantage of making extensive enquiries is that one builds up a network of people to see. "Oh, you must talk to Father Damiano, he lived in X for many years. "Haven't you seen Señor Rodriques yet? He used to sell goods in that village every week. "What about Kofi? You know he was the driver of the 'Sea Never Dry' lorry, and he really knows those places." Eventually, then the network starts closing - i.e., no new names are presented - one can feel reasonably confident that major sources have been contacted.
Key informants are unusual people, and thus do not necessarily provide typical or valid sociological information. But they do help the outsider to appreciate the following points: there is no such thing as a simple question; everything is related to everything else; answers depend on questions; there are usually apparently endless levels of complexity. Given this state of complexity, the exact nature of questions asked, and who is asked, are clearly of great importance. Even such apparently simple, straightforward questions as: "Where do you collect firewood?" "Who owns this forest?" "Do you pay fees to out wood?" "Do you sell charcoal at the market on Fridays?" "Who takes meals regularly in your household?" - may prove to be extremely complicated, and answerable only in situation-specific terms. This means simply that each question must be related to specific set of circumstances, or a particular person, because if these are changed, the nature of the question will be changed. It is essential to search for the right question, the one that will be meaningful to respondents and that will produce the information and explanation desired. A good key informant can be especially helpful in aiding the fieldworker to ask appropriate questions in the clearest and most productive manner.
2.3.1 Participant observation
2.3.2 Nonparticipant observation
2.3.3 Benefits and problems with participant and nonparticipant observations
2.3.4 Time allocation studies
The foundation of data collection in social research is observation. In field research we use several different modes of observation, three forms of which are discussed below: participant observation, nonparticipant observation, and time allocation studies.
Participant observation has been a major research strategy of anthropologists, involving a combination of informal interviewing data being collected in a relatively unstructured and flexible manner. The investigator usually decides beforehand what particular sorts of activities are to be observed and recorded, and even the form of the record. Fieldwork generally occurs over a prolonged period, during which time the researcher builds up sufficient rapport with the study population. To some extent, participant observation allows the observer to become an "insider". That is, the investigator directly participates in the study population's activities, perhaps accompanying and helping household members gather fuelwood., This permits an understanding of the study population and their activities from their own perspective. At the same time, the difference between what people say and what they do can be checked.
We stress the advantage of knowing some vernacular terms because knowledge of the local language is important in participant observations. Short terms in the field, coupled with pressures of other obligations, seldom allow for mastery of new language. Yet, depending on interest and commitment, and to a lesser extent on linguistic ability, it is not difficult to acquire a basic vocabulary, say twenty to a few hundred words, of vernacular terms. In the fuel domain, for example, one should know, and use, even when with an interpreter, local words for firewood, charcoal, crop residue, cow-dung, tree, stove, man, woman, child, axe - and so on. A knowledge, and correct pronunciation, of major localities, personalities and main species of trees helps greatly in persuading indigenous people that the enquirer is interested in understanding theme It is a courtesy to them, and an acknowledgment of the value of their culture, it eases social intercourse and it helps an outsider to appreciate local perceptions.
In nonparticipant observation, the observer remains separate from his study population's activities, and attempts to be unobtrusive. It may be possible to structure the observation, at least to the extent of deciding to focus on a particular activity such as wood-collecting, cooking, tree-planting, or fuelwood sales. There may be a conscious structuring of observation in the sense of developing hypotheses to be tested, or following up unclear relationships. But the observer must be careful not to impose preconceived notions, and must remain flexible, and open to new interpretations. Although the observer attempts to be unobtrusive, we do not recommend any attempts to deceive. For both ethical and practical reasons, honesty really is the best policy in fieldwork, and the people who are being observed have a right to know the scope and purpose of the study.
Participant and nonparticipant observations share several benefits and problems, both being useful fact-finding methods when qualitative data is sought. This is especially true when the investigator wants to describe a cycle of events. One example is the description of cutting of trees, construction of hornos (earth ovens), transporting, and the street sales involved in charcoal production and distribution. To describe such a cycle of events, it is beat to combine interviews and observations.
Both observation techniques are also useful where no records of previous studies exist. These techniques are important in studying selected members of the community, such as the aged or the handicapped, who might not be selected in a random or even stratified sample. The techniques, especially participant observation, allow the investigator to develop empathy for the study population through listening and participating. Among the problems are constraints of cost and times.
In particular, participant observation requires a good rapport between the observer and the observed. This may take a long time to develop. With studies of fuelwood, such as descriptions of production techniques or marketing, this may be less of a problem than with more sensitive issues such as village politics or wealth differences, though the usage of fuel may be part and parcel of these more sensitive issues.
The difference between participant and non-participant observation may parallel that between a resident and a non-resident observer. That is, participant observation can seldom be achieved without the sort of acceptance and rapport that comes from being at least a temporary resident of the community. Therefore, in deciding on the objectives of the survey, this is a question to be considered - does the survey need the sort of detailed information that can only be gained by a resident observer, or will the costs (in time, and money) outweigh the benefits? If so, can use be made of local assistants, and how?
No observation is "unstructured", in a strict sense. That is, the observer's preconceptions, biases and "knowledge" all colour what is observed. Anyone who has made observations in another culture - or even in an unfamiliar sub-culture in one's own society - can recall, with embarrassment, the sense of "seeing and not seeing", because of lack of understanding of the society. It is surprising how one can consciously and effectively enlarge the vision, so that what was once a blur of vague shapes soon fits into meaningful categories, For example, many first-time visitors to African savanna country remark on the apparent sameness of the social and physical landscape, and exclaim "How dull this is!" But those of us who have spent some time in the savanna, and who have opened our eyes, see it as a marvellously changing scene, where human activities are fitted, in complex ways, to natural resources, at least until disturbed by population pressure and technological innovation. The discovery of "the hidden savanna" - as has been experienced by many fieldworkers who have walked and listened and truly observed - is an exciting experience, reminding us how blind most of us are until we open our eyes. Fieldwork means constant questioning and seeking new perceptions. And this is especially important in studying a complex topic such as rural energy patterns.
In observation, then, one should have a clear idea of the main goal of the enquiry, which is to gather information on fuel for instance; but it is also important to be responsive to new stimuli, to be flexible and to follow new paths. Here is a delicate dilemma: how far should one stray along new paths that might eventually lead to new illuminations, and to what extent should one follow a systematic schedule? Ultimately, the observer will be guided by a sense of what is appropriate to the specific situation and by his own hypotheses.
An important aspect of any fuelwood survey concerns the amount of time spent in tasks such as collecting wood, preparing charcoal, cooking and other energy-related activities. It may be difficult to get precise information, because local people perceive and evaluate "time" in a different way from the investigator, so there may be a cultural misunderstanding in questioning them; there are important daily and seasonal variations in time spent, and there is also variation in proportion of time spent by different households, as well as by individuals within the same households. How can one determine a meaningful "average" for time spent on fuel activities?
One way (which has been found successful when investigators have enough time) is to make random spot observations on what community members are doing at particular times, The investigator (and preferably come research associates) note basic details (when? where? what? how? with whom? how long?) of what particular people are doing, over a long enough period to generalize. The random spot checking technique has several benefits. It eliminates the problem of representativeness of the data. Variability between sex, age, and, depending on the criteria used in selecting the sample, wealth differences, is indicated, The position and proportion of energy-related tasks within the daily work rhythm can be ascertained. The latter is particularly important where intervention in the local energy system is being considered. Finally, the survey can be combined with other investigation in the community of such topics as nutrition, employment, or farm management studies.
The biggest problem with the random spot checking technique is that of data over-collection, recalling one of our initial warnings on avoiding collection of surplus data. Considerable time is often spent in non-energy-related tasks. For example, in Erasmus' study (Erasmus 1955: 324-328) only about 3 percent of the observations made of men were related directly to energy tasks such as collecting or chopping wood or food preparation, although such tasks took up at least 28 percent of the women's time (including ironing). In some areas, where household members venture far to gather wood, direct verification of their labour time may be impractical, and other tasks performed on the fuelwood trip may go unrecorded. Nonetheless, random spot checks should be employed, where possible, in studying fuelwood-related labour time allocation.
Interviews are an integral part of social research. We shall postpone discussion of the selection of interviewees (see the section on Sampling), with one important exception. Fuelwood collection, cooking, and other energy-related tasks are usually gender-defined. In most cases, such as in cooking, women perform the work and are the ones directly consuming the fuelwood. Moreover, women may have a greater knowledge of consumption patterns than the men. Thus, women should be involved in most aspects of fuelwood and household energy-use surveys. At the same time, strong sexual stratification and customs may restrict access of interviewers, particularly males, to women. The importance of sexual stratification should not be underestimated in surveys. In Muslim areas, it will be essential for a woman to interview female household members, and even in non-Muslim regions a woman interviewer often has better success in getting the women to talk about their fuel tasks.
Where should the interview be conducted? In many cases the best place will be in the respondent's house or compound, although for information on commercial activities, an interview at the office or factory may help to get greater accuracy. Sometimes there is no choice, and the interview has to be held in a public place, or on a farm, or walking to do a chore.
In an interview, it is usual to have at least a set of questions to ask, and in most circumstances we recommend an open-ended type of interview, which allows the conversation to be directed to some extent by the respondent. Closed-form questionnaires, consisting of entirely preselected questions, are seldom the best approach, unless the interviewer already has extensive, accurate and up-to-date knowledge of the community in general and the energy system also. In general, a format that allows flexibility is best. Having a schedule (a set of pre-arranged questions to ask) gives interviews some structure and comparability. Having at least come open-ended questions on a questionarie allows for flexibility and the inclusion of respondent perceptions.
Questionnaires and a completely open-ended interview do not exclude each other, because questionnaires should never be used alone. They should be combined with relatively unstructured case-studies of specially selected households or villages (even a few will be helpful). These case studies can serve (a) to develop preliminary hypotheses which help to focus and define the limits of the survey, (b) to check on any initial tabulation and analysis of results and (c) to follow up, and find explanations for, any puzzling or unclear relationships or associations.
In both interviews and questionnaires, questions must be carefully phrased so that they can be understood by the local people. Questions should follow in a logical order so as to obtain the maximum amount of information, and to remind people of aspects on which they might comment. Questions ought to fit with indigenous knowledge systems, and with local perceptions. One way of ensuring this fit is by pre-testing the questions, using key informants and others who can be critical. This should remove ambiguities and also sharpen the focus of the questions.
It will often be impossible to conduct an interview in private with one person, as friends, neighbours and passers-by are likely to listen and even join in the conversation. Rather than resenting this as an interruption, a skilful interviewer can bring in the other people and create an informal group interview, although in a strict random sample the extra volunteered information should be excluded, so as not to skew the sample.
How does an audience affect the interview? Some people may become boastful, or try to present their actions in a favourable way; others may be suspicious and silent lost they give away valuable information. Obviously it depends on the specific culture and situation, but an audience can often be used to good advantage.
Field investigators should not be surprised or discouraged if they initially encounter an unwillingness among local people to be interviewed. "Why don't you see the other people first? They are much better off than I and have better things to talk about", may be a typical response to a question. Besides a general distrust of outsiders, people may be reluctant to speak about their landholdings or other assets. Such caution in dealing with outsiders is obviously a justifiable mechanism of self-protection. Some people may give answers designed to please, rather than reflecting a true state of affairs, also as a self-protective device. Some cultures promote more suspicion than others, and within cultures some individuals are likely to be more anxious than others, about giving away potentially incriminating information. Other reasons for reluctance to answer questions can be that people simply are not accustomed to being interviewed, or people may be reluctant to be interviewed because they already had been saturated with surveys and no effort had been made by previous researchers to convey the results or significance of the surveys back to the people.
Even where people are not suffering from survey fatigue they may be reluctant to be questioned because they view the interview as an imposition from outside forces. That is, the interview is something in which they had little choice of participation, and something from which they will receive few, if any, tangible benefits. In many cases their perceptions are entirely correct. Interviewing people in itself need not imply "local participation". Efforts should go in the planning and fact-finding stages towards identifying and incorporating the desires, needs, and aspirations of community members and social groups. And, importantly, local individuals and groups must be involved in analyzing the collected information and in suggesting policy alternatives.
Many instances exist where people may be reluctant to talk, especially where government regulations have curtailed free cutting and a black market in fuelwood exists. In Nepal, where fuelwood is in critically short supply, people are reluctant to answer questions on forest utilization (Donovan, 1979). Given the frequent reference to illicit cutting or wood theft in areas as diverse as the Sudan, Kenya, India, the Philippines, Mexico and Guatemala, the interviewer should recognize the problem of dealing with sensitive matters. The interviewer and investigator should ensure the confidentiality of the respondents.
Frequency of interviews depends upon the research design and the various time/cost constraints often only one interview will be possible. Memory recall of respondents depends upon the frequency, regularity, and significance of events. Twice weekly visits have been recommended, since daily visits may seem like pestering, and gaps of more than three or four days lead to increased recall errors. Seasonal differences should be accounted for.
A short-cut method of rapidly gathering data is to interview groups rather than individuals. This method presents problems of representativeness, since any group chosen is unlikely to represent a true cross-section of the local population, though attempts should be made to include individuals of different socio-economic status. The knowledge and experience of several individuals may serve as checks on information given by each others.
There is nothing specialized about a group interview. In fact, any one who interviews, i.e. asks questions, is likely to find that an impromptu group interview situation develops. As many individual interviews are conducted in public places, and also because of differing conceptions of public and private space, other community members often stop, from curiosity, suspicion, desire to help, or officiousness, when they see a stranger talking to one of their groups Instead of resenting what westerners (or western-educated people) may see as an interruption, it is often advantageous to accept and use the situation. This is especially so with fuel, where an individual alone is unlikely to divulge "secret" information that could not be publicly disclosed, so that the presence of many people should not be an obstacle to obtaining good information. But if the interviewer is a forestry official, his police duties may have alienated him from the villagers, who would then be cautious in answering questions. The response will in large part be determined by existing cultural and historical factors. For example, in parts of India, women (who have been hidden behind a partition) have been known to interrupt an interview and contradict the man being interviewed, thus constituting an unusual form of group interview.
A group of people can be highly informative, in modifying, supplementing or even contradicting individual statements. A group of Indian, Nigerian or Colombian peasants (men and/or women) might engage in a conspiracy to deceive an outsider over enquiries about fuelwood, but, because of their perceptions of the fuelwood crisis are at least as likely to impress the visitor with their problems, perhaps even exaggerating the situation. However, one thoughtful individual will often say "No we do not walk ten kilometers to collect wood, it is only as far as from here to x," ensuring a degree of reliability. Properly handled, group interviews can provide a useful check on reliability. There are limitations as some individuals will be reluctant to speak on certain topics, or they may wait for a senior or more powerful person to speak, then be unwilling to make any open contradiction, even when privately disagreeing. This happens in all communities, and provides a careful observer with a chance to observe patterns of hierarchy and influence.
There are no universal rules for proper handling of a group interview, except to say that it requires courtesy, and common sense, combined with two other qualities - a firm sense of what information is required, plus an appreciation of how flexibly the conversation can be guided. One does not wish autocratically to out off at once a speaker who wanders from the main topic, but who may be providing useful information, nor can one afford to let the conversation be dominated by the village bore, who may drive away more sensible villagers.
In this, as in other field situations, it is important not to raise false hopes, but to make clear what specific changes are possible or likely, and what cannot be done. One interviewer reported the result of a group interview of unemployed Nigerians in Lagos; the men turned on the interviewer with hostility and said in effect, "If you cannot offer us jobs, don't waste our time with stupid questions". Group interviews in urban situations however, offer many advantages, as they are easy to set up because of the population density.
Questionnaires are a popular method of data collection. The advantages of using a questionnaire are well-known: data can be collected quickly on specific items; these data can be easily transferred into forms allowing quantified and computerized analyses; and data collection tasks can be delegated to less expensive field staff. Questionnaires also compel the adoption of some "organized structure" upon data collection, but will be most effective when used by someone who can support and test the questionnaire findings with personal observations and insights and knowledge.
Using questionnaires is one means of recording data, but it is not the only means and it is not adequate to not cover all the information required. As we said earlier, no one method of information gathering is adequate for all purposes - all should be supplemented and checked.
However, several problems can arise when using a questionnaire. This is especially true where a questionnaire is the primary means of collecting information. A questionnaire can impose a rigid, preconceived idea of reality which may be inappropriate for the particular situation. If field enumerators are not supervised properly, errors in recording data can occur. Problems arise from respondents concealing, misreporting, or misunderstanding questions. Recall errors often happen, especially with regard to seasonal activities. The design and preparation of a questionnaire are extremely important, as they will influence the type of information collected, in somewhat the same way as the mesh-size of a fish-net determines the fish that are caught. Careful thought, then, must be given to the selection and phrasing of particular questions. Sometimes it is good to start with a general question, "what do you think about X?" followed by increasingly specific questions. First, one must have enough basic knowledge of the community to know which questions would be meaningful, and how, exactly, they should be framed so as to minimize the possibility of creating ambiguity, embarrassment or resentments The purpose of questions is to discover what people know, not what they do not know. Thus, the second stage should be a brief pre-testing, which allows for refining and clarifying the questions so that they really do elicit, in precise form, the information required.
Third is the administration of the questionnaire, which - as noted elsewhere requires constant supervision (see below), for even at this stage ambiguities are likely to occur. One perceptive interviewer may point out that Question 14 could be interpreted in two ways, for example, and a quick resolution of the problem could be communicated quickly to other interviewers to ensure consistency and accuracy. Fourth, there is the analysis of the results, to see what sort of picture emerges. And finally, there is a consideration of whether any supplementary questions are desirable. Throughout all stages, constant close supervision and cross-checking are necessary, combined with repeat interviews, supplementary observations, and, most important, regular discussion with interviewers. This both provides a check on the accuracy of the answers, and also encourages the interviewers to be conscientious.
Although the questionnaire ought to cover all questions needed, it should not be too elaborate nor too long; an hour is usually the maximum time period for any one questionnaire to be administered. Most rural people, especially women, have many demands on their time - collecting firewood and water, cooking, washing, cleaning, looking after children - and cannot spend too much time in answering questions. Whether to use a closed form (with itemized answers) or an open-ended form questionnaire depends on the researcher's own needs and requirements. If the closed form is used, a space for comments by the respondent and the interviewer should be included. In framing questions, there are several simple alternatives such as: yes/no; checking a scale (of incidence, or preference, or quantity) of 0-5. Some questions invite discursive answers, as in the open-ended questionnaire; while others ask for a straightforward factual answer - "how many times each day are cooked meals prepared?". Depending on the nature of the survey, a simple rapid survey might be the best.
Where computer facilities are available, it is advisable to frame and to code questionnaires so that computer analysis is possible. When a large number of sample households are involved, such as in a national or other macro-level survey, the use of computers is almost essential. By computer, we refer not only to macro-computers but also micro-computers, and even some calculators that can be programmed for regression analysis may be appropriate. But the use of computers is by no means indispensable in all surveys, particularly in more micro-level analyses, at least in the initial stages, Manual analysis often can be done quickly and cheaply so that a preliminary idea of results is obtained in a few days, instead of waiting at the mercy of the computer for months. When computers are used, it is recommended and probably necessary that a member of the computer staff be part of the research team.
In a perceptive article (Chambers, 1979; see also Chambers et al., 1979; Chambers, 1980) Robert Chambers has described many methodological shortcomings associated with field research and project evaluations. These shortcomings are important to note because they bias data collection and lead to an inaccurate analysis.
Field research tends to be conducted during the dry season. In part this is because accessibility to many rural areas is easier during this season. Academic summer vacations also coincide with this period of relative prosperity, after the crops have been harvested, In contrast, the wet season, before the harvest, is generally a harsh time in rural areas. Sickness and hunger prevail, particularly among the poorer households. The wet season is also the worst period for fuel collection, with wet wood, slippery paths, more illness, less time for collection and inadequate drying and storage facilities. Thus, research often has a seasonal bias, with conditions appearing more prosperous than they are over the balance of the year.
Researchers tend to visit the field for short periods, and seldom stray from the roads. In this respect, the Land Rover or Toyota Land Cruiser have been mixed blessings for development planning and research: while they do provide greater mobility, they offer the impression of "roughing it in the bush," when in fact what can be visited on rural roads is hardly representative of rural society. This is especially important if one is concerned, as are most development agencies, with seeing that benefits accrue to poor people, sometimes referred to as "'the lower 40 percent". For these people often live in remote localities, inaccessible except on foot. Few poor old widows live on a roadside; the people most ill, children and elderly especially, are more likely to be suffering inside some decrepit house rather than walking or sitting outside in any visible way.
To summarize, there are a number of biases typically found in field research that cause the worst rural poverty to go unperceived. Dry season, spatial, elite, male and project and adopter biases operate so as to hide the poorest sectors of the population, especially at the times they are the poorest. What emerges is an inaccurate portrayal of rural society and rural property.
What can be done to overcome such biases? We have mentioned several promising approaches, and here we summarize:
1. Go to the field well prepared, aware of what has been written, and of work being done by social scientists, especially local people.
2. Cooperate with officials, but obtain unofficial views, too.
3. Travel alone or in small groups, not as part of a large official party.
4. Tell the people you meet about yourself, your aims. Share information.
5. Spend longer in the field, preferably overnight, when it is easier to talk, and when peripheral people may appear.
6. Walk, Listen. Be silent when silence is appropriate.
7. Concentrate on one topic (fuelwood), but be aware of related topics.
8. Never rely on questionnaires alone. Always supplement with direct observation and participant observation.
9. Find out when the stress periods - of seasonal hunger and shortages - occur.
We have not attempted to provide a thorough examination of all methods, and their associated problems. Instead, we have been highly selective and, we hope, realistic, in recognizing that most readers will not be able to arrange elaborate prolonged investigations. So our emphasis is on what can be done quickly and on how to make the best use of limited resources. Although the resources available, the study population and the informational needs will differ for each fuel survey, we recommend that a combination of methodologies be used. This is to ensure against the biases and limitations that are inherent in any manner of data collection, so that a reasonably accurate description can emerge.
3.1 Defining "community"
3.2 The representativeness of a community
3.3 Choice of community
3.4 Households as the basic unit of study
3.5 Other socio-cultural and economic units of study
3.7 Sample size
This section is concerned with how to select a population to study. We first deal with the concept of community - how it is defined, the question of representativeness, and procedures for selecting a research site. Next, we examine the basic unit of study in most fuelwood surveys - the household, while also mentioning other possible study units. This section ends with a discussion of sampling procedures and their applicability to fuelwood surveys.
We have assumed that there will be some choice of which community to study, but we recognize that sometimes the researcher will be directed to do a study of a particular problem. However, governments and development agencies are unlikely to pinpoint a particular community or village, they will rather choose a wider area, so there may still be some degree of choice.
The terms "community" and "village" connote socio-cultural, residential, and administrative units with supposedly clearly demarcated boundaries, although in actuality these boundaries are seldom clearly defined. Instead, subsumed under these terms is a complex array of historical, spatial, social, economic, political and other relationships.
When you travel by road, you see, of course, only those people whose homes can be reached by road or jeep track, and you do not see hidden pockets of poverty that are often scattered about at some distance from road or track. In looking at rural fuel situations, it is vital to include the peripheries, and not to concentrate only on members of the core elite who are likely to be the first people one meets.
The ideal way to obtain a comprehensive picture of rural energy use would be to spend at least a full year in a community. This would allow the investigator to notice important seasonal changes and how these affect patterns of labour allocation, social relations, and energy usage. A long residence is also likely to lead to good rapport with the community, and consequently to a better understanding of their ways and an ability to ask the "right" questions. Unfortunately, many surveys and their researchers will have little opportunity or resources to conduct such detailed investigations. There are intermediate steps, though, in terms of research design and manner of field research.
On large-scale projects, such as a national energy survey, case studies of selected communities, chosen according to appropriate environmental, economic, socio-cultural and other variables, could be used to supplement and to round out the data gathered by questionnaires and less intensive methods. Specific consideration should be given in all projects to include data collection during the wet season, the pre-harvest period, and times of peak agricultural labor demand.
For the field researcher, a recommended method is the "walk and listen" approach, which can be adopted even on one-day field visits. Leave the vehicle, arranging to meet it later, and walk a few hours away from the road, checking (by observing, and questioning an interpreter may be needed) on what one "knows". For example, conventional wisdom may have it that people prefer certain species of trees for charcoal; that women go in groups to gather firewood; that certain trees are never out, because of a ritual prohibition that women do not climb trees; that men never carry firewood; that land-owners do not permit non-kinsmen to collect wood. All such statements may be true, or true in certain circumstances, or subject to considerable reservations, or simply false. It is important to distinguish between past norms of behaviour and past values, and present practice. There may well be some confusion in the minds of local people about this. There is no substitute for walking and listening to check on such universal generalisations, even, or particularly, when these are confidently made by community members. In all societies there is a difference between what people actually do, and what they say they do. People tend, if asked about social behavior, to present an idealized picture of what ought to happen. For example, in Ghana we were told that when a man died, his property was inherited by his father's younger brother's son: an investigation of actual cases of inheritance showed that only a minority followed the ideal pattern in practice.
The "walk and listen" approach is not without potential problems of bias. As mentioned in the key informant section, the rural elite tend to be the people who come forward, while the poor do not speak up. The researcher's own sense of politeness may inhibit him from probing into the lives of the poor. Unless one has some immediate benefit to offer, questioning the poorest people often appears to be an unwarranted and immoral intrusion. Another problem is that of male bias, since researchers tend to be males and their contact with women, especially poor rural women (who constitute a "deprived class within a class"), is generally limited.
A final type of bias centers around projects and innovations. Researchers tend to go or to be taken into a place where "something is happening", where a project or innovation seems to be having a favourable or the desired impact. The short-visiting researcher may tend to meet only those individuals who are the users or adopters of innovations, such as new stoves. This 'project' and adopter bias leads researchers to neglect those areas, communities and people who have not benefitted but who might have been affected in an indirect (or even direct) fashion by the project. For example, meeting with a landowner who had his cotton land converted to an eucalyptus plantation might reveal little about how this affected his labour force.
The terms also refer to a certain pattern or style of living typified by personalistic relationships, social solidarity, and a local orientation. Impersonality, factionalism, and a shrewd understanding of how to manipulate outside social forces have been found to typify rural life.
The delineation of the community or village as a unit of study depends on several local factors, including residential patterns, administrative divisions, land-use systems, and local people's perceptions. Formal administrative boundaries by themselves can be misleading indicators of community boundaries. For example, "census villages" in India do not always correspond with residential clusters: "it is not uncommon for two adjoining houses to belong to two different revenue villages".
A major concern about any survey is how applicable the findings of one area or community are to other areas and communities. To a large extent, the representativeness of a community depends on how comparable its internal composition is with other communities. Five key characteristics of a community should be identified in order to serve as a basis for inter-community comparison:
a. Population - including the total number of people and sex and age distributions. At a broad level this will indicate the aggregate local demand level for fuelwood and other fuel resources.
b. People/land ratio - with information on tenure relationships, types of agriculture, land quality and use patterns, land distribution, and ecological zones. Household access to trees and other fuel resources, as well as the condition of these resources, will be among the important variables considered here.
c. Urban contact - the nearness to urban centres, linkages with roads, the amount of farm, fuel, and village output sold outside the community, and the amount of employment community members find outside. A crucial variable to be considered is the amount of wood and other fuel sources imported into or exported from the community.
d. Resource distribution - how wealth and productive resources are distributed; thus, the extent to which the community is economically and socially stratified. Once again, household access to fuel resources (whether direct access to trees or its ability to purchase fuelwood) would be considered.
e. Occupational structure - the relative importance of agriculture, seasonal wage labor, tenancy, nonfarm income and so on. Of particular importance for fuelwood surveys is the scale of fuelwood sales and degree of wage labor involved in fuelwood enterprises within the community.
These headings correspond in some degree to our "Topics on which to gather information", presented at the beginning of this chapter. But these are selective, and chosen to be relatively easy to identify and to aid an effective selection of villages. Identification of these variables allows the investigator to see how typical the community is in comparison to other communities. The question of typicality or representativeness is meaningful if a good national survey exists, to allow for comparison. The variables will also help in eliminating a village that clearly is unrepresentative - e.g. by being located close to a sugar-mill.
In choosing a community to study, it might be desirable to consider the degree to which local people recognize that there is a fuel problem. Where there are clear perceptions of a crisis, that warrants serious attention, the response to the survey is likely to be good.
When available, village-level census data (if reliable, and up-to-date) lots the investigator check beforehand how representative a community is in these variables.
Studies of rural communities rarely mention why and how the particular community was chosen for study. When reasons are given, most choices are "purposive." For example, Frederick Conway explains why he selected a particular site in Haiti:
Fonds Parisien, east of Croix-des-Bouquest in the Plains du Cul du Sac, was chosen as the field site because it had been an important center of charcoal production and remains a center of charcoal exchange, because deforestation as a result of fuelwood collection is advanced there, and because the contractor had established rapport through previous research in the area. The last factor was essential in obtaining reliable information in a brief period of time (Conway, 1979:4).
Conway's quotation also demonstrates one of the advantages of having the investigator restudy a community. Scarlett Epstein has given an account of how she was able to gather remarkable amounts of information in a five-week visit to two south Indian villages that she knew well from an earlier study (Epstein 1978:128).
Often, convenience and practical considerations are the most important factors behind the selection of a community. Communities may be chosen on account of their accessibility, or their congenial political and social climate. Reasons for deciding on one community over another may reflect the investigator rather than the community, as the former may prefer (though seldom is this admitted) a place that has "friendly" inhabitants, or good access to medical facilities, or absence of mosquitoes, or a pleasing view., Representativeness is also important: in his survey of energy use in rural Southern Africa, Marc Best purposely chose three villages which represented different "biophysical environments" (Best 1979:5). One of the villages was selected because of its "remoteness" and its "traditionalness" (that is, Best considered the village to be not "strongly affected by modernization", although it, like the other two, was "purposely affected by the migrant labour system"). This purposive selection process has often resulted in bias, since communities with good transportation, communications facilities, and other amenities are generally more developed. According to Connell and Lipton (1979:99):
The relatively "developed" state of accessible villages is likely to give both those undertaking village surveys and those reading them an unduly optimistic view of rural conditions.
Meanwhile, the worst poverty goes undetected or unobserved.
Some researchers have recommended using area sampling to select a community. This involves randomly selecting a community from a large number of possible communities, each possessing an equal chance of being selected (see the discussion of sampling in the next sections.) Factors such as location, environmental setting, farming type, and population can be taken into account. But area sampling presents several problems, and is not feasible unless a large sample in used, together with extensive knowledge of many villages. Maps, aerial photography, and rapid surveys of a number of villages can help overcome the lack of a sampling frame, combined with an interval sampling technique. Sometimes accurate maps are unobtainable, and aerial photographs, if enlarged, can be effectively used as maps. Infra-red photography can be useful in high lighting the vegetation. In some parts of the world, good aerial photographic coverage exists, and is available; in others, either little aerial photography has been done, or it is not accessible, perhaps "for reasons of national security".
Overall, in selecting a community, the research should take into account the key variables mentioned (Population; people/land; urban contact; resource distribution; and occupational structure). Each variable must be regarded for its relevance to fuelwood, as indicated above. "Key" is used in specific relation to fuel problems, not in a general senses. The community, and adjoining communities, should be visited and looked at carefully before research begins. Factors such as local historical and ecological variables, and location which would limit the applicability of inferences made from the community to other areas, should be considered. Above all, the reasons for selecting a particular community should be explicitly stated.
Most surveys take the household as their basic unit of study. In some cases a household consists of a nuclear family - a man and woman with their children. However, it sometimes includes the extended family - the grouping together of several nuclear households; for example, a man with several wives and their children and/or his or his wives' parents. There also may be "fragmented" households, such as an elderly or young or divorced person who lives alone.
In general, a household may be considered as a grouping of people who share the same cooking facilities. Even with such a broad definition of household, problems can arise. One common problem is that households are constantly changing in composition, and some members may be away for various time periods, from two days to two years or more. This is especially so when migrant labour is common. At the same time there may be several visiting guests and relatives who increase demand for fuel. In many parts of the Middle East, and also in other regions, belief in the influence of an Evil Eye makes people reluctant to disclose exact numbers of children to strangers. In reckoning per-capita fuel consumption of a household, one obviously needs to know which members are physically present.
Household size and composition influences how production is organized and what are relative consumption levels. In most households family labour is the most abundant resource, so that the number of household members available for doing work is crucial. Family size and the availability of energy resources are often correlated. This is important for fuelwood; as the available forest stock diminishes and as the time required for collecting fuel increases, so does the value increase of children as gatherers. Information about household members' occupations helps in determining household size. This would appear particularly necessary where seasonal wage labour migrations are prevalent.
Besides the household, fuelwood surveys may concentrate on other socio-cultural and economic units, John Briscoe (1979) for example, is concerned with analyzing his studied households in terms of their socio-economic class. Several indices, such as landholding size, sources and amount of income, house size and composition, ownership of consumer goods, and so on, can be used in assigning households a relative wealth position. Local people are often able to identify and to rank, with a keen precision, relative household wealth positions (See Castro et al., 1981, for a discussion of indicators of inequality).
Other possible social groupings include ethnic groups, caste divisions, religious groups, charcoal producers, kinship groupings (such as lineages) and neighbourhood groups.
Relevant economic units include various shops, restaurants, industries, businesses, and public institutions that consume fuelwood, Donovan (1979 and 1980), for example, has studied small-scale industries in Nepal that consume fuelwood. Digernes (1977) surveyed the shops, schools, and similar units which consume fuelwood in Bara, Sudan. Surveys may also consider the contractors, dealers, and other entrepreneurs who are involved in the production and marketing of fuelwood.
Researchers are rarely able to examine every possible unit of study in a given population. It is necessary to choose a sub-set or sample of the population for investigation. Thus, the issue for researchers is to decide what kind of sampling procedures to use. The manner of sample selection is critical because it defines how representative of the population the chosen group will be and therefore the extent to which findings from the study can be applied to the population as a whole.
Sampling methods are generally divided into two major types: non-probability and probability sampling (Kearl, 1976:27). A non-probability sample means that the individual units of study are selected either purposely (on the basis of some judgemental criteria) or accidentally (on the basis of whomever is met or whatever is seen). Thus, a non-probability sample is "unique", but not necessarily unrepresentative of the population. That is, in a non-probability sample every unit of a population does not have an equal chance of being selected. However, a sample can be purposively selected so that it contains the general characteristics of a population (this will be discussed below in greater detail).
A probability sample is based on the principle that every unit of a population has an equal chance of being selected for study. This ensures, theoretically, that the chosen units of study are representative of the population. Probability sampling involves a random or systematic manner of selection, for example selection based on a table of random numbers, or based on every nth unit, such as every tenth household.
Selecting the appropriate sampling procedure depends upon the particular situation, and in all situations each manner of selection will have its own advantages and inherent limitations. Although we will describe each sampling method as an exclusive type, in practice they are often combined. Thus, a village may be selected purposively, on the basis of environmental, economic and other variables, while the sample selected in the village may be done randomly.
Non-probability sampling allows for considerable flexibility in selection, an important consideration when the study universe is undefined or unclear, or when there is a problem obtaining consent from those selected. But non-probability sampling has several drawbacks. Any group may be easily under or over representated in the sample. There is a tendency to select those individuals who are more articulate or approchable, which generally means the wealthier or more educated community members. The actual basis of selection may appear in retrospect as having been largely ad hoc even when specific criteria were considered.
Probability sampling is generally seen as more "scientific" because it supposedly will yield a sample that is precisely representative of the population. In practice, though, there are limits to the use of probability sampling. A random or systematic sample needs to be drawn from a defined universe, but frequently little census information is known about the population and its geographical distribution is undermined. Thus, a sampling frame has to be created. Sometimes maps and aerial photographs, perhaps combined with an overflight of the area, can help to construct a sampling frame. An area sample could be chosen, then one particular area randomly selected on the basis of agro-environmental and socio-economic variables. Households in the selected area can then be listed and a random or systematic sample chosen.
Not all researchers have had to start from scratch. Digernes used the local rates lists (which show the assessments for public services) in Bara, Sudan, to construct a stratified random sample. Fortunately, rates are based on such important socio-economic variables as house size, location, number of rooms, building materials, number of-latrines, number of household members, and their socio-economic position. Digernes, who carried out her study in 1976 and 1977, found that the latest list was from 1973, and it seemed "complete and fully representative". She stratified the households into three groups according to the levels of rates paid, selecting a 10 percent random sample from each group. Twenty-five of the 162 sampled households had moved or been dissolved and were excluded from the survey. No new households were added. A list of the twenty-five public institutions and private businesses using charcoal was obtained from the local government, and these were included in a fuelwood consumption survey (Digernes, 1977:57).
Another problem with probability sampling is that a high percentage of those selected must consent to being studied. In most cases some households will not consent and others will have to be chosen. Conversely, households that were not selected will want to be studied. When possible in such cases the data ought to be collected, although it should not be included with the sample.
All communities, even the poorest, have gradations of poverty and wealth. Stratifying the sample is usually essential. This means dividing the population into different strata or groups, based on income or other relevant variables. In probability sampling, study units can be selected randomly or systematically from each strata. With non-probability sampling, a quota might be allocated to each stratum, and case studies might be selected from each for intensive study.
Once again, there is often a paucity of data on income, land ownership and other critical variables necessary to stratify a sample. Proxy indicators of income, such as housing type, condition and construction materials, and possession of consumer goods, can be used to rank households. Perceptions of local people can also be used in constructing a stratified sample. Hill asked several "key informants" to rank village households according to their ability to cope with seasonal food shortages. A random sample was taken of the resulting strata and several measures, including land and capital equipment ownership, were used to test the perception. Hill found that the local people were extremely accurate in their ranking, in that their subjective assessment correlated closely with other objective measures (Hill, 1972:59).
Several factors need to be considered to determine the appropriate size of the sample. Sample size is related to costs that can be afforded by the researchers. The most critical consideration, though, is that the sample be large enough to be adequately representative of the study population. Too small a sample can undermine the validity of months of research.
In populations that are socio-economically homogenous, a small sample is sufficient because describing one unit describes them all. But such populations are increasingly rare, indeed it is doubtful if they were ever as prevalent ass once believed. Some analysts recommend that surveys include at least 10 percent of the households with the proportion increasing with bigger, more heterogeneous communities (Connell and Lipton, 1977:106).
A final consideration is that sample size affects techniques and questions. For example, a recent survey in Nigeria covered 11,000 households, so detailed techniques were out of the question. Where a sample is very small, such as fewer than 10, they more resemble case studies than a sample survey.
4.2 Training and supervision
Especially among the rural agricultural peoples with whom we are primarily concerned, there will be little difficulty in finding suitable assistants, or "associates", to use a more exact term. We consider some aspects of selection, training, supervision and logistics.
Depending on the particular local circumstances, one obvious initial source of likely recruits is among local unemployed high-school graduates, or high-school students or teachers - who are on vacation. We have found that many have the appropriate level of education: what is needed are adequate skills in reading, writing, record-keeping, and interpretation of questions, together with qualities of reliability, honesty and ability to ask questions in a polite, respectful yet firm manner. Those who have only primary education are unlikely to manage the written records, nor will they be able to write their own observations. On the other hand, people who have an advanced education are often unsatisfactory because they regard the job as dull, routine and unglamorous; or they complain constantly about the rigours of long walks along hot dusty paths, or about the mosquitoes or fleas, or whatever is locally an irritant. This is not an anti-intellectual argument, for university students, both undergraduates and post-graduates, can play very important parts in village surveys; it is simply that they regard themselves as over-qualified for the rather routine and relatively low paying task of gathering basic information.
Government departments are a good source for survey associates, especially such departments as forestry and agriculture. But where forestry officials have law-enforcement duties, they may have such negative images in the community that people distrust them. In selecting associates, certain qualities are desirable - some degree of literacy and sensitivity to people being interviewed, for example. Age, religion, gender, ethnicity, and experience may all be relevant. Above all, people selected should have a real interest in the survey and an understanding of its aims. It may be possible to ask ten potential associates to do some simple interviews and observations for a day, and to report verbally and in writing. One or two will emerge as "natural researchers" in that they are interested, competent, sensitive and conscientious; in short, they have the necessary "sociological imaginations".
High school students are not the only people to be considered, Given the poverty in most communities, and the lack of non-farm opportunities, there are likely to be many applicants. Sometimes there will be so many that the choice of individuals becomes difficulty especially when, as often happens, locally important people put forward their kinsfolk or dependents. In some cases this can lead to problems because applicants or their "sponsors" are aligned with factions within the community.
One selection device, which worked quite well in Central Kenya, was to ask about one hundred people to do a rapid four-day survey of their villages, suggesting that they add certain types of supplementary information. On the basis of what they wrote, combined with checking their forms for consistency, it was easy to select five or ten of the best, who were invited to continue in helping with the study for one or two months. Who are the "best"? It is easy to identify those who have a sociological imagination, which is not necessarily correlated with a high level of formal education; it is also usually easy to determine who has completed the questionnaires in an accurate and careful fashion. Some fieldworkers "fudge" results, by deciding, for example, that it is not necessary to walk five kilometers to see one household, preferring to guess from their general knowledge. One soon develops a sense of when this is happening, because of inconsistencies, irregular patterns, and erratic or unusual responses to questions.
A major aspect of selection is the inclusion of both men and women, as emphasised above. Women should be incorporated because so many aspects of fuelwood, from collecting to cooking, fall mainly within women's sphere of activities,
We combine these two categories, as it is rarely possible to organize formal training sessions, beyond perhaps one or two initial meetings when the project is explained, and questions are answered, so supervision becomes doubly important.
Here are some guidelines:
(a) Arrange a definite time and place for supervisor and fieldworker to meet, preferably at least once a week.
(b) Because of the prevailing difficulty of communication, if it proved impossible to meet - for example because of illness, poor road conditions, or vehicle breakdown -arrange a contingency "Plan B", making certain that there was some way of communicating.
(c) Allow enough time for a thorough check of the fieldworkers' records, questioning any unusual or dubious information, and praising any special contribution. This is important, for supervision has both negative and positive goals. As well as serving as a cheek against sloppy or inaccurate performance, it should be used for positive reinforcement. Fieldworkers, like most people, respond well to interest in their work and to appreciation of problems.
(d) Engage, if numbers warrant, "assistant supervisors", so that it is possible to check adequately on all workers. University students, where they are available, have been found to be good in this role.
(e) Constantly and critically assess the survey instrument - are the questions providing the sort of information required?
(f) Eliminate problem questions, especially if they create too many negative reactions to the interview situation.
Research associates should be made to feel that their job is not simply one of mechanically recording responses on a questionnaire. Such an attitude leads the fieldworker to emphasize the rapid collecting of data instead of accurately collecting relevant data. They should fully understand the nature and importance of the survey, and of their own role. Moreover, without any incentive or interest in the research, the fieldworker has "little reason or incentive to make or record or induce a new line of enquiry". Field staff should be paid adequately and they should be given full credit for their contributions. Moreover, their perceptions and observations should be constantly invited by the senior researcher. For these reasons, we recommend that research should be regarded as a co-operative endeavor, with the professional investigator, the associates, and the local people all participating in a common enterprise. There are compelling ethical and practical reasons for adopting such an approach, which gives proper credit to associates and villagers and which is likely to lead to better (i.e. more comprehensive, more reflective of actual conditions) reports than does the usual top-down survey. Where associates and villagers are literate it is a good idea to ask them to read and comment on draft reports. If some are illerate, it is easy to organize a "bush seminar" to present and discuss the major findings. These can be regarding exercises.
Even though the fieldworkers will presumably be from the same region, and will speak the local vernacular, they will often be working outside their home communities, which results in two distinct sets of problems. First, they may come from a group or faction or lineage that is not really acceptable, or which is not highly regarded, and this should be investigated. Second, there are often problems of logistics - of transport, of housing, food, even of details of finding a bed, or a place to sleep, or a lamp, or a table. We need not dwell on these, except to say that it is preferable to face these potential problems squarely at the beginning, clearly stating what will and what will not be provided, and also specifying precise duties and responsibilities of each fieldworker. The prospects for misunderstanding, as in any relationship, are limitless!
We conclude by reiterating three points that should be considered in planning fuel surveys of any scope:
(1) using a combination of methods and procedures will probably give the most complete and accurate portrait of energy use;
(2) be aware of local knowledge, expertise and perceptions; invite local participation in the planning, implementation and analysis of the survey; and
(3) beware of the numerous biases - dry season, elite, male, roadsides, project, etc. that can skew the description and analysis of local fuel situations.