rapid rural appraisal methodology applied to project planning and implementation in natural resource...

13
Rapid Rural Appraisal Methodology Applied to Project Planning and Implementation in Natural Resource Management Augusta Molnar This chapter reviews Rapid Rural Appraisal methodology as it is ap- plied to land-based natural resource management project planning and implementation. Rapid Rural Appraisal (hereafter RRA) is a set of tech- niques adapted from social science interview and survey methods to meet the needs of farming systems research diagnosis and, more recently, to provide comprehensive sociocultural, economic, and ecological assess- ments of a project area for the purposes of project planning and imple- mentation. There are a number of promising RRA methods being pro- moted by social scientists and others for use by experts and local project staff, but, as yet, there is little consensus about a number of methodolog- ical issues and controversies involved in the use of RRA. The reaction to RRA in more strict, academic, social science circles has been lukewarm, due to its rather unorthodox and untested combina- tion of a variety of techniques and methods. RRA has been best repre- sented in articles by researchers associated with the various International Agricultural Research Centers. 1 The reaction has also been mixed among non-social science trained—but field-experienced—technical develop- ment experts, either because many of RRA principles appear to be pure common sense or because the more narrowly defined goals of the meth- odology as applied to farming systems research seems less relevant to broader planning concerns. Discussion of these techniques is quite im- portant, however, because increasing numbers of development decisions are being made on the basis of information collected by such quick tur- naround methods and through a wide range of strategies employed in a multidisciplinary context. The field is moving in a direction of participatory planning at a community level and the techniques to foment this kind of planning draw heavily from Rapid Appraisal methods. RRA techniques are in general characterized by the following: (1) they are rapid, so that results can be quickly made available to decision- makers; (2) they are eclectic, tailoring diverse interview and survey tech- niques to meet the needs of the specific information gathering task; (3) they aim to capture a holistic picture of the local situation, drawing on a multidisciplinary perspective; and (4) they allow for valuable interaction between investigators and target population during the course of the in- terview. 2 11

Upload: rightsandresources

Post on 06-Jan-2023

0 views

Category:

Documents


0 download

TRANSCRIPT

Rapid Rural Appraisal Methodology Applied to ProjectPlanning and Implementation in Natural ResourceManagement

Augusta Molnar

This chapter reviews Rapid Rural Appraisal methodology as it is ap-plied to land-based natural resource management project planning andimplementation. Rapid Rural Appraisal (hereafter RRA) is a set of tech-niques adapted from social science interview and survey methods to meetthe needs of farming systems research diagnosis and, more recently, toprovide comprehensive sociocultural, economic, and ecological assess-ments of a project area for the purposes of project planning and imple-mentation. There are a number of promising RRA methods being pro-moted by social scientists and others for use by experts and local projectstaff, but, as yet, there is little consensus about a number of methodolog-ical issues and controversies involved in the use of RRA.

The reaction to RRA in more strict, academic, social science circleshas been lukewarm, due to its rather unorthodox and untested combina-tion of a variety of techniques and methods. RRA has been best repre-sented in articles by researchers associated with the various InternationalAgricultural Research Centers.1 The reaction has also been mixed amongnon-social science trained—but field-experienced—technical develop-ment experts, either because many of RRA principles appear to be purecommon sense or because the more narrowly defined goals of the meth-odology as applied to farming systems research seems less relevant tobroader planning concerns. Discussion of these techniques is quite im-portant, however, because increasing numbers of development decisionsare being made on the basis of information collected by such quick tur-naround methods and through a wide range of strategies employed in amultidisciplinary context. The field is moving in a direction of participatoryplanning at a community level and the techniques to foment this kind ofplanning draw heavily from Rapid Appraisal methods.

RRA techniques are in general characterized by the following: (1)they are rapid, so that results can be quickly made available to decision-makers; (2) they are eclectic, tailoring diverse interview and survey tech-niques to meet the needs of the specific information gathering task; (3)they aim to capture a holistic picture of the local situation, drawing on amultidisciplinary perspective; and (4) they allow for valuable interactionbetween investigators and target population during the course of the in-terview.2

11

12 SOUNDINGS

This chapter presents state-of-the-art review of RRA methodologyfrom the perspective of 40 development practitioners in land-based nat-ural resource management, mainly social scientists, interviewed with re-gard to their own field methods. Given that RRA is eclectic and that thereare a wide range of guidelines, the objective of the interviews was to iden-tify the range of techniques in use and the consensus of opinion regardingparticularly controversial areas of methodology as these affect the validityof the information collected and analyzed. A related objective was to iden-tify the range of promising innovative methods developed by these prac-titioners as well as to determine the extensiveness of their use in the de-velopment field. Given these ambitious goals, I attempt here only to sum-marize some general issues and to introduce a short discussion of thefindings with regard to key controversies, including sampling, team inter-action, least visible target groups, and the use of minimum data sets. Dueto space constraints, I have completely omitted the findings regarding avery important part of recent rapid appraisal work, namely the participa-tory planning aspect.3

The interviews were carried out as part of a work program on RRAmethods for the Forest Department in the Food and Agricultural Organi-zation of the United Nations (FAO) and for the Asia Environment Unit of theWorld Bank. The interviews were conducted subsequent to a previous lit-erature review and together with that literature review will form the basis ofpreparation of guidelines for the systematic application of RRA methodsto natural resource management.

Content of the Interviews

In the pre-interview stage, the background review of RRA methodsbrought up the following unresolved methodological questions:

1. What are the criteria that separate an RRA from "rural developmenttourism"?

2. How much time should the social scientist spend apart from a multi-disciplinary team to ensure that he or she is collecting quality infor-mation on social and sensitive issues yet interacting sufficiently withother specialists?

3. What are the key concepts shared by practitioners mentioned in a va-riety of RRA guidelines and reports?

4. What are valid indicators and proxies available for use in communityforestry projects in specific geographical regions?

5. What are the optimal uses of group and individual interviews?6. How is a set of information best selected and which specific sampling

techniques are most useful for RRA purposes?7. How can the best use of categories be made to distill relevant infor-

mation without introducing biases?8. How can information about the least visible target groups, such as

women, be collected in a typical RRA?

NAPA BULLETIN 10 13

9. What areas do not lend themselves to RRA investigation?4

General Results

The results of the interviews, in the first place, provide evidence thatpractitioners share a consensus on the important problem areas which ap-pear during the application of rapid appraisal methodologies to short-terminformation gathering in the field, although their solutions may vary signif-icantly. On the other hand, the sample expressed a wide range of opinionsregarding the validity of different approaches, and these differences wereoften linked to that practitioner's training, geographic expertise, and, forsocial scientists, his or her theoretical bent. The respondents agreed thatmoving in the direction of more systematic use of RRA in natural resourcemanagement planning and evaluation is a highly desirable and neededstep. The advantages of systematization are that it improves the quality ofinformation being generated; clears up the expectations of managementstaff, both donor agency and government counterparts, as to what a con-sultant on a sensible field visit can provide; identifies the training needs ofproject staff; and indicates how best to make use of information thatemerges from short, field assessment visits.

One statement that emerged in response to the question about therespondent's use of RRA guidelines was "most of the recommendationsare obvious and something a good interviewer already does." Everyoneused a form of triangulation, cross-checking information from a variety ofsources before proceeding with the analysis of the situation, and all re-spondents showed awareness that the situation "on the road" differedconsiderably from that "off the road." But from there the consensus fadedaway. Some practitioners found the guidelines unnecessary for skilled in-terviewers and social scientists, and, further, dangerous for an inexperi-enced audience, since RRA principles are easy to misuse if the intervieweris unlearned in the dynamics and pitfalls of participant observation andformal surveys. Others found the guidelines a useful tool for even the ex-perienced researchers on the premise that it was important to be re-minded of the areas of potential bias in short-term work—a heavy work-load led one toward sloppy interviewing that regular reflection and prod-ding helped to reverse. Others felt there were a number of creative ap-proaches in the guidelines that could greatly improve the quality ofplanning and evaluation in most of the agencies in which they worked.Some found it most relevant to use in farming systems research (FSR),where the problem was narrow, the process certain to be iterative not aone-time exercise, and the researchers the same individuals as those in-tending to carry out the actual research (an argument that pushes stronglyfor planning exercises to be conducted by the implementing project staffas well as by expatriate advisors). Others were concerned by the lack ofmethod in quantitative data collection during a rapid appraisal and arguedfor the introduction of more formal sampling in the process.

14 SOUNDINGS

Others were wary of the whole concept of RRA, since in their experi-ence it led to a false sense of security in a situation that was, more oftenthan not, characterized by "rural development tourism" and sometimesled to a disdain for formal surveys that were essential companion exer-cises to confirm or disconfirm planning assumptions made glibly over afive-day period. One set of practitioners criticized RRA guidelines forglossing over the complexities and efficacy of specific participant-obser-vation techniques which are used by social scientists in a shortcut as wellas long-term research situation, including prior knowledge of theories ofsocial organization, context interviewing, situational analysis, and thickdescription of historical case studies and conflicts over rights and obli-gations in land management. Some of what I refer to as "package ap-proaches"—such as agro-ecosystems analysis and minimum data sets—were criticized for missing important complexities of the social/cultural dy-namics.

A common consensus regarding RRA was that "short is never better,"but that it is a necessary evil given the development planning schedulesin which large-scale projects operate (and many small-scale projects aswell). It is this perception that perhaps explains why some respondentsfound the application of RRA techniques, developed in FSR for narrow in-vestigative purposes, dangerous when applied to broad purpose plan-ning exercises in the field. They feared that users of RRA techniques areoverly complacent in making judgments from limited data, without a clearjudgmental sense of how that data was limited.

The majority agreed that more systematic use of shortcut informationgathering methods was a possible direction in which to move for all plan-ning and evaluation purposes. Again, however, there was firm agreementthat agencies should avoid dictating methodology. A team too con-strained by pre-established terms of reference and list of questions waslikely to miss central issues that could only emerge through a flexible RRAapproach.

More consensus on what RRA can accomplish was seen by some so-cial scientists as an important administrative step. One respondent ar-gued

One problem I face is the lack of precedent from which to negotiate the timeallocated to RRA and companion surveys with my employer. Ideally, I shouldbe able to argue concretely about the kind of product I can provide within agiven time frame from a solid base, especially for slippery social issues thatare more often avoided by project planners who fear any amount of time de-voted to them is fruitless. If you give me three weeks, I should be able to say,I will be able to provide you x confidence level of information, if you give mefour or five weeks, I can provide you with y. And if you want more conclusiveevidence of the viability of a particular approach to social issues, give me fourto five weeks in country for RRA work, and funds to oversee and contract amore formal companion survey to be carried out during my own field visits.

NAPA BULLETIN 10 15

Instead, the RRA process is highly idiosyncratic, the length of time allocatedto look at social issues tends to pattern that allocated to any technical issue,and more inputs are hard to justify within the organizational norms at work.

Controversial Areas

Sampling

Sampling is a controversial area in RRA methodology. Practitionersrepresented quite a wide range of opinion on the value of introducing moreformal methods of selecting the set of individuals interviewed during thecourse of field visits. While recent literature on RRA has argued that theresults of a systematic informal survey are very similar to those of a formal,statistical survey, practitioners were not all in agreement with this point.

As could be expected, the key controversy was not whether the pur-pose of an RRA was to formally sample the population—a clearly impos-sible task—but whether some kind of more formal sampling in the contextof an RRA was a useful means of correcting bias. None felt that all of theindividuals or groups interviewed in an RRA should be selected by "ran-domizing" criteria, but a number stated that this should be used in com-bination with the more opportunistic or purposive selection of informants.Some argued for more follow-on of RRA exercises with small, formal sur-veys to substantiate or follow up on the RRA results. One practitioner atthe World Bank has developed a computer package for sampling designof such follow-on surveys for use by the non-statistician.

The range of opinions expressed on this issue were as follows:

1. Formal sampling has no place in an RRA, which is a "scan" of therange of situations in the project area. There is neither time to carry outformal sampling, nor is the objective of gathering information aboutgeneral issues/hypotheses commensurate with a need for formalsampling.

2. The biggest constraint on formal sampling in an RRA is time. While itcould contribute substantially to validity of information, there is nottime to devote to constructing a random sample and interviewingthose in it.

3. More mixture of methods is needed in RRA. Subjective samplingmethods (purposive or opportunistic sampling) are too open to bias,even for those who follow Chambers's checklist of possible biases(near the road, elite, dry season, male, project staff's choice of site).What is often missed is the existence of a key recommendation do-main/type of situation, which some kind of randomizing can help theteam avoid.

4. Randomizing has a distinct value in preventing interviewer bias, butshould not be the basis for deriving quantities or frequencies (for ex-ample, 60 percent of those interviewed planted fruit trees). It shouldjust be a method of avoiding bias.

16 SOUNDINGS

5. Tabulating quantities and frequencies emerging from the set of thoseinterviewed is quite useful for the purposes of analysis. It should notnecessarily form part of a report, but it is definitely of value.

6. There are a number of RRA-appropriate means of random samplingwithin time limits. These include:a. taking random interview lists from surveys already carried out for

the area you are visiting, reviewing the original schedules forquantitative information, and choosing some of those respon-dents for in-depth interviewing during the RRA;

b. introducing random sampling halfway through field visits, onceyou have some hypotheses that need better testing;

c. select every 10th person from a tree-distribution register. Alter-natively, select a random sample from health campaign registersfor the village you are visiting;

d. before your RRA starts, list a few questions for which you wantinformation from a random sample, and have the in-country donoragency office contract a survey research firm to collect informa-tion for a small random sample of 100 or so. You can also do thisby drawing a sample with the "quick and dirty" statistical sam-pling program developed by World Bank for monitoring and eval-uation purposes, none of the respondents had done this, but sev-eral suggested it and many thought it was an excellent idea;

e. take the results of formal surveys conducted for the area and usethese as a basis of group discussion in a village. "A survey wasconducted for this district and it concluded that 40% of the farm-ers did not plant trees due to lack of water facilities. Is this a prob-lem you face? Why do you think so many farmers were con-strained by this factor?";

f. the null hypothesis—while you often cannot talk to enough peopleto test the frequency or degree of accuracy of a hypothesis, youcan find out what it is not, and be statistically correct. A variety ofhypotheses can be tested. Is there an assumption that nomadicgraziers are not a problem for plantation protection? Check to seeif five plantations have been damaged in this manner. Likewise, ifthere is an assumption that women do not plant trees for cash, thelikelihood of meeting five women in an RRA who are planting treesfor cash is very low, and confirms without any need for a formalsurvey that something different is going on.

Alternative Methodologies to Representative Sampling

One set of social scientists focused not on the trade-offs between ran-dom and purposive/opportunistic sampling, but on the need for RRA prac-titioners to become more conscious of the value (and limitations) of qual-itative data collection methods that are traditionally applied in a longertime frame, but that can be adapted to the RRA setting. These include use

NAPA BULLETIN 10 17

of "situational analysis"—following your nose regarding a situation or pat-tern of interest rather than sampling a range of "individuals." Another cen-tral technique is collecting information on the situation over time—changes in resource use, changes in the resolution of conflicts, orchanges in the value of products or crops. Conway's Agroecosystemsanalysis includes collection of information for a 10-year period for cropyields, tree harvests and prices, and market access, so that these can bejuxtaposed when evaluating the economics of an intervention from thefarmer's perspective.5 The qualitative approach generally requires a so-cial scientist on the team who has a first-hand understanding of qualitativeresearch methodology and staff with a first-hand understanding of thecommunities with which they plan to work, so that they can correctly eval-uate what values or norms underlie a particular person's expressed opin-ion or action.

Team Interaction

The sample included both team leaders and team members, leadingto some interesting discussions of the optimal team interaction for plan-ning and evaluation. In farming systems research methodology, it is as-sumed that the team will work as a multidisciplinary unit, dividing into ro-tating subgroups at times, but not working independently. This is not thecase with natural resource management planning. Respondents, partic-ularly social scientists, were quite divided on this issue. Some arguedstrongly that the role of the social scientist in a team setting was threefold:to collect relevant socioeconomic information for recommending a courseof action; to introduce a social perspective to team members from tech-nical disciplines and project officials, and to provide training for officialsand extension workers in fruitful methods of interviewing beneficiaries forthe RRA exercise itself. Leaving the team for several days of independentinterviews, perhaps with a translator or junior project staff member wasseldom a wise decision unless a special target group was likely to be ex-cluded from the sample. What was important to these social scientists wasthat other team members and officials have confidence in their observa-tions—something only possible if other team members and planning offi-cials participated in the interviews themselves. This group concurred withan interesting article by Jamieson (1986), who argues that good RRA isrevolutionary for project planning because it leads planners to talk genu-inely to target participants—both host-country planners and expatriates.

How the team split in the field was a question of logistics, particularlythe availability of vehicles and how "visible" the visit was. With more thanone vehicle in high profile situations, one subgroup could take the "offi-cial" route and the other an informal/considered route. With proper teampreparation, subteams could ask questions for each other and comparenotes in the evening (though with big teams and big agendas, this seldomhappens for teams only have time to complete their own checklists).

18 SOUNDINGS

The training function of field interviewing was also considered central.Spending time interviewing with extension staff provided these staff withinvaluable training for carrying out subsequent sensitive interviews and forbecoming aware of the importance of previously unnoticed socioeco-nomic factors. It also reinforced the donor agency's concern that a projectbe "participatory" to field staff, a must in plantation or soil and moistureconservation projects.

A second group argued for adequate time apart from the rest of theteam and high-profile officials. Only then could they have enough confi-dence in their findings to argue strongly for a particular course of actionwhen the report was being written. They understood the trade-off ex-pressed by the other group, but felt that reliable and comprehensive in-formation gathering on socioeconomic issues was more important. Theexclusive team approach meant that some team members often shortcutnecessary introductory warm-up questions and dialogue in an interview,which made it impossible to grasp subtle social dynamics. The individualswho desired more intensive interviewing spent half or more of their fieldtime on their own. One practice of team leaders to meet both objectiveswas to contract the social scientist and maybe another technician con-cerned with local conditions a week or two prior to the rest of the team'sarrival. The extra time allowed them to gather the field information felt nec-essary.

The interviews pinpointed some perceived ambiguity in the expec-tations of team leaders regarding the role of social scientists at the laterstages of design. When a geographic area specialist was used on theirproject, some team leaders expected less time for the collection of primarysocioeconomic data and more time dedicated to the assessment of de-sign alternatives from a social perspective. When, however, the informa-tion base on that project area was not adequate, the possibility of frictionemerged due to misperceptions over the social scientist's scope of work.

Least Visible Target Groups

As might be expected, the general consensus among practitioners isthat the typical RRA did not provide a very representative view of the leastvisible target groups, such as landless, rural poor, women, and tribalgroups. Some balked at the callousness that sometimes appeared whentime constraints became severe: "I hate to ask village officials to cough upa few landless poor for me to interview." A majority of male respondentsrelied upon women interviewers—team members and local counter-parts—to collect information about women. Interestingly, discussions withspecialists in women in development revealed no solid conceptual orpractical basis for this methodological decision. One specialist who reg-ularly encouraged male team members and extension workers to interviewwomen while they were in the field with her found little difference in the

NAPA BULLETIN 10 19

quality of information they could gather in such discussions from what shewould have gained.

Respondents had a number of valuable suggestions for reachingsuch groups, such as interviewing women about stove efficiency as anacceptable means of entering their hearth and becoming an "insider" withwhom women would more readily talk. Other strategies included stoppingto interview laborers encountered on the way to pre-set interviews, con-centrating on very degraded resource areas and smaller, marginal farms,dividing larger group meetings into subgroups based on occupationalroles and economic status and holding smaller, simultaneous discussions"chaired" by different team members, and most important, spending atleast one night in a village where there was time and favorable circum-stances for seeing a wider range of patterns.

All, even the experienced social scientists, agreed that the low profileof the poor was a major challenge for RRA, though somewhat less of aproblem in the natural resources management context. For example,when irrigation management by landed farmers was the issue, landlesslaborers were not a necessary interview category. More problematic, how-ever, were tribal groups—since in a broad-based project these comprisea small percentage of the total population and are often inadequately sam-pled in favor of more generalizable problems.

Minimum Data Sets and Minimum Indicator/Criteria Sets

Much work is being done on both generation of minimum data setsand generation of minimum indicators as a framework for RRA interviewcontent. These "sets" have been developed in response to a recognitionthat RRA fails most often due to missing important aspects of a particularproductive relationship (e.g., allocation of labor resources or market ac-cess) which could have been alleviated by more exhaustive questioning.While a researcher with a fairly long time frame is likely to hit upon suchaspects because of the frequency with which they discuss these topics,the short-term field practitioner is likely to miss an equally important topicby dwelling too extensively on another topic, for example, markets ratherthan labor.

One solution to this problem is to ensure that the informational topicsare directly relevant to the problem at hand. Farmer income levels are oflittle interest when the RRA target is on smallholder adoption of tree plant-ing. First of all, it is very difficult to measure income with accuracy, sincefarmers are reluctant to report their income in a limited interview frame,and sample size will be too small to make conclusions. Furthermore, theRRA team estimates average project area income figures from secondarysources, rather than from field data. Finally, income is not as important anindicator of tree-planting potential as resource availability—access toland, labor availability, marketability of produce, water availability, andseedling availability, which may or may not be linked to income levels.

20 SOUNDINGS

Holtzmann (1986) has developed an excellent grid framework for or-ganizing a minimum data set for the evaluation of agricultural marketingsystems (see Table 1). Applying this grid to an agro-forestry project, oneinformational area might be organized as shown in Table 2.

Based on such an analysis, the team may decide to focus on commonproperty management factors, price fluctuations of trees and crops, em-

Table 1

Areas ofInvestigation

(Ex.) CommodityCharacteristics

Areas forInvestigation

Common PropertyManagement

Components

a) Different gradesand end uses;

b) Degree of bulki-ness, perishability

c) Physical han-dling requirements

d) Degree/type ofprocessing.

Method ofInquiry

1) Review commod-ity manuals, studies

2) Observation ofhandling and pro-cessing3) Developing com-modity calendarsshowing periods ofproduction andtransformation

Table 2

Components

a) types of commonlands, and uses bydifferent types ofindividuals,

b) history of andinpiripnpp ofII lOIVJd 100 v/1

conflicts over use ofI anHcIdl lUo.

Method ofInquiry

1) Review ofethnographicinformation,

2) Discussions withvillagers

3) Review ofreports, documentson land tenure,holdings,legislation.

4) Court records

5) Discussions withdistrict officers.

Reasons forInvestigation

a) Commodity char-acteristics can influ-ence operation offunctions per-formed, how per-formed and relativecost

b) Nature of pro-duction process in-cludes timing andmagnitude of pro-ducer sales andmarketed flows.

Reasons forInquiry

a) Use of commonlands is oftendeceiving sincerights are notnecessarily equalfor all types ofvillagers, usagevaries (private landbecomes commongrazing land duringfallow period;outside herders usevillage landsseasonally.

b) Different inds.may be treateddifferently in theiraccess to lands—women vs wealthylivestock owners—and have differentaccess to productsgrown on thoselands.

NAPA BULLETIN 10 21

ployment of farmers in livestock-based enterprises (dairying, rentingoxen), outside employment of household members (and labor con-straints), group decision-making situations and labor arrangements (irri-gation management, grazing management, road/trail repair), seasonal la-bor requirements (whether people will have time to water, tend trees), gen-der-based labor inputs and gender-based income channeling in house-hold, traditional agro-forestry practices and local knowledge of trees.Once this information has been determined, existing minimum data setsensure that the range of questions will be pertinent to key hypotheses andfindings. For example, a Cornell manual on farming systems analysis (Gar-rett et al. 1987) lists the following questions on non-wage labor, which havebeen annotated with "reasons for investigating" in order to determine if aproject tapping village exchange labor for constructing contour bunds isrealistic:

Component1. Are there times when a group of

people from the community allwork together? What tasks dothey do?

2. Why do people accomplish taskA, B, C, . . . with exchange la-bor?

3. What kind of person organizesor requests the work party?

4. What kind of rewards do workersreceive?

5. How do people feel about beingcalled to a work party?

Reason for Investigating1. Is there a limited scale of situa-

tions in which such labor can beused (i.e., maybe not consid-ered appropriate for your pur-poses)

2. Is this task labor intensive? Doneat a labor-scarce time of theyear?

3. Must you be influential to organ-ize the group? Are groups anyvillagers or only members of kinor economic role group?

4. Is it expensive to pay workers?Could a poor group of villagersafford it?

5. Is it something dying out or stillconsidered important/accept-able?

The RRA investigator may ask all these questions of many informantsor only a few, depending on the gaps in their knowledge of the answers.If there is good secondary information on this aspect of village life, theymay only discuss this in one village to confirm what that secondary infor-mation said. If there is hardly any information to base an analysis on, theymay ask this information in every village and to a wide range of individualsto ensure data on all the tasks for which exchange labor is used and toidentify all age groups which still use this type of labor rather than wage

22 SOUNDINGS

labor. The data set is an effective tool that reduces the risk of an incom-plete vision of exchange labor groups thus placing too much or too littleweight on their appropriateness for project interventions. From a sociolog-ical perspective, it prevents interviewers from wasting valuable time ask-ing a host of questions about exchange labor that may be interesting forsociological reasons, but have little bearing on project design. But itequally prevents non-sociologically trained team members from missingthe importance of a comprehensive interview regarding a key social issue.

Conclusions

The above findings reveal a number of important controversies re-garding the appropriate application of interview and survey methods tothe needs of rapid rural appraisal as well as the emergence of a numberof "promising methods" that deserve more careful consideration and eval-uation within anthropology. The intention of this article is to generate morediscussion and work on these issues.

Notes'Hildebrand (1981) and Rhoades (1985) represent just a few of a range of articles that

have appeared in Agricultural Administration and Human Organization over the past eight ornine years.

2There are a number of manuals for Rapid Rural Appraisal methodology which de-scribe the basic principles. A few basic works include: Chambers (1985), Khon Kaen Uni-versity (1986), and Kumar (1987)

3There are a number of approaches to which the reader can refer, including Banerjee1987; Suelzer and Sharma 1986; Ministere de I'Environment et du Tourisme 1975; Rocheleau1985.

4On the basis of this list, an interview schedule was designed and administered to anopportunistic, yet purposive sample of practitioners based in the United States working forUSAID, FAO, the World Bank, and a number of international NGOs. Geographic location wasa must for budgetary reasons and I elected to interview practitioners only in person or bytelephone, rather than by mail, since the complexity of the interview topics required a moreface-to-face relationship between interviewer and respondent to generate meaningful infor-mation. Interviews were geared to a 40-minute period, though this period was extendedwhen respondents were able or willing to devote more time to the discussion. Not all inter-views covered all questions. I tried to maximize the time to glean as much about the prom-ising methods for controversial areas the individual deemed to be important. My own con-ceptions of RRA methodology have undoubtedly biased the interviews, but I have tried tocapture the range of opinions to the extent possible within those limitations.

5There are a large number of agroecosystems publications that include both manuals,discussion papers, and training materials. See for example Conway and McCracken 1988.

ReferencesBanerjee, Ajit

1987 Microplanning: A Tool for Social Forestry Implementation. New Delhi: NationalWastelands Development Board, Government of India.

Charoenwatana, Terd, ed1985 Proceedings of the 1985 International Conference on Rapid Appraisal. Bangkok,

Thailand. Khon Kaen University, Rural Systems Research and Farming Systems Re-search Projects.

Conway, Gordon1986 Agroecosystem Analysis for Research and Development Arkansas: Winrock Inter-

national Institute for Agricultural Development.

NAPA BULLETIN 10 23

Conway, Gordon, and Jenny McCracken1988 Rapid Rural Appraisal and Agroecosystems Analysis. In Agroecology and Small

Farm Development. M. A. Altieri and S. B. Hecht, eds. Boca Raton, FL: CRC Press.Garrett, Patricia, et al.

1987 Interview Guide for the Regional Analysis of Farming Systems. Cornell InternationalAgriculture Mimeograph 113. Ithaca, NY: Cornell University.

Hildebrand, Peter1981 Combining Disciplines in Rapid Appraisal: The Sondeo Approach. Agricultural

Administration 8(6):423-432.Holtzmann, John

1986 Rapid Reconnaissance Guidelines for Agricultural Marketing and Food SystemsResearch in Developing Countries. Working Paper No. 30. Ann Arbor: University ofMichigan.

Kumar, Krishna1987 Conducting Group Interviews in Developing Countries. A.I.D. Program Design and

Evaluation Methodology Report, No. 8. Washington, DC: Agency for International De-velopment (Office of CDIE).

Ministere du L'Environment et du Tourisme, Burkina Faso1974 Pour Une Pedagogie de L'Autopromotion. Burkina Faso: Unite Bois de Village BP,

Bobo-Dioulasso.Raintree, John

1985 An Introduction to Agro-Forestry Diagnosis and Design. Nairobi, Kenya: Interna-tional Center for Research on Agro-Forestry (ICRAF).

Rhoades, Robert1985 Informal Survey Methods for Farming Systems Research. Human Organization

44(3):215-218.Rocheleau, Dianne

1985 Land-Use Planning with Rural Farm Households and Communities: ParticipatoryAgroforestry Research. Working Paper No. 36. Nairobi, Kenya. International Center forResearch on Agro-forestry.

Suelzer, R., and K. Sharma1986 Working with the People: Some Experiences with the People-Centered Approach

(PDPP) in the Tinau Watershed Project 1983-1986. Tansen, Nepal: HMG/SATA TinauWatershed Project Paper.