• TABLE OF CONTENTS
HIDE
 Introduction
 General issues
 The farmer as the researcher's...
 Acquisition of information on farmer...
 Design, testing and evaluation...
 Institutionalizing social scientists...
 Annex I
 References














Title: Methodological issues facing social scientists in on-farm / farming systems research
CITATION THUMBNAILS PAGE IMAGE ZOOMABLE
Full Citation
STANDARD VIEW MARC VIEW
Permanent Link: http://ufdc.ufl.edu/UF00080813/00001
 Material Information
Title: Methodological issues facing social scientists in on-farm / farming systems research
Physical Description: Book
Language: English
Creator: Harrington, Larry.
Publisher: CIMMYT
Publication Date: 1980
 Record Information
Bibliographic ID: UF00080813
Volume ID: VID00001
Source Institution: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
Resource Identifier: oclc - 184986071

Table of Contents
    Introduction
        Page 1
    General issues
        Page 2
        Page 3
        Page 4
        Page 5
        Page 6
        Page 7
        Page 8
    The farmer as the researcher's client
        Page 9
        Page 10
        Page 11
    Acquisition of information on farmer problems and circumstances
        Page 12
        Page 13
        Page 14
        Page 15
        Page 16
    Design, testing and evaluation of new technology
        Page 17
        Page 18
        Page 19
        Page 20
        Page 21
        Page 22
        Page 23
    Institutionalizing social scientists in agricultural research systems
        Page 24
        Page 25
        Page 26
    Annex I
        Page 27
        Page 28
        Page 29
    References
        Page 30
        Page 31
Full Text

Nov./17/80
/S. 30

0/, O^Jr




METHODOLOGICAL ISSUES FACING SOCIAL SCIENTISTS IN ON-FARM/
FARMING SYSTEMS RESEARCH*


1.0 Introduction

There has recently been a substantial effort on the part of practitio-

ners of Farming Systems Research (FSR) and On-Farm Research (OFR) to'pool

their knowledge and share their experiences. It is possible that this effort

was sparked by the appearance of the TAC Stripe Review on FSR (Dillon, 1978)

or the Gilbert, et al paper (Gilbert et al, 1980). Equally, it may merely be

the case of an idea whose time has come.

One example of this recent effort to share experiences was the Workshop

held at CIMMYT on April 1-3, 1980, and attended by practitioners of FSR/OFR

from national agricultural research programs and international agricultural

research centers (IARC's).I/ This workshop focused on a number of methodolo-

gical issues that confront social scientists in their effort to participate

in FSR/OFR. These issues include:

Issues of general relevance (the role of the social scientist, the

use of research results for alternative purposes, the cost-effective of FSR,

assumptions on the policy context in which research is conducted, definitional

issues)



* Report prepared by Larry Harrington on a Workshop held at CIMMYT, April 1-3,
1980. Only for use at the CIMMYT-IDIAP meeting. Although not all workshop
participants were consulted in the writing of this report, an effort was
made by the author to include divergent opinions on the methodological issues
herein discussed.

1/ See Annex I for the list of Workshop participants and their institutional
affiliation.










Issues relative to the farmer as the researchers' client (criteria

for selection of target areas, criteria for stratifying farmers into relatively

homogeneous groups, research on the whole system versus research on a target

crop in the context of the system, formulation of technological "packages"

versus "pieces").

Issues relative to the acquisition of information on farmer problems

and circumstances (use of background information, informal surveys, formal

surveys, observations in farmers' fields).

Issues relative to the design, testing and evaluation of technology

(methods of pre-screening technological alternatives, issues in on-farm experi-

mentation, economic analysis of experimental data from both private and social

viewpoints, farmer assessment of technology).

Issues relative to the institutionalization of FSR/OFR in national

research programs.


In general, workshop participants achieved a remarkable degree of consensus

on many issues, at least when the decision-context was carefully specified and

the relevant alternatives clearly spelled-out. The following discussion will

identify such areas of methodological agreement, but will also note areas of

disagreement and the different positions taken.


2.0 General Issues

2.1 Terminology

The terms "farming system research" and "on-farm research", and

other terms used in such research, often mean different things to different

people. To insure fruitful discussion on common topics, it was first necessary

to deal with a few issues of terminology.









Workshop participants were generally in agreement with the defini-

tions presented by Norman (Norman, 1980). A distinction between "farming-systems

research" and "on-farm research" was accepted by most workshop participants.

This stemmed from a realization that not all farming-systems research need be

conducted on farmers' fields (e.g., prototype solutions to broad problems may

be developed on experiment stations) and that not all research conducted on

farmers' fields is FSR (e.g., traditional fertilizer trials conducted on-farm

but in isolation from farmer problems and circumstances).


Workshop participants accepted the following characterization of

FSR: 1) The farm as a whole is viewed in a comprehensive manner. 2) The choice

of priorities for research reflects the initial study of the whole farm.

3) Research on a farm sub-system is legitimate FSR provided the connections

with other sub-systems are recognized and taken into account. 4) Evaluation

of research results explicitly takes into account linkages between sub-systems.

5) As long as the concept of the whole farm and its environment is preserved,

not all factors determining the farming system need to considered as variable

-- some may be treated as parameters. Therefore, FSR may be called FSR-in-the-

small (low ratio of variables to parameters) or FSR-in-the-large (high ratio of

variables to parameters).


There was less agreement among workshop participants on the proper

terms for different types of FSR programs. This occurred in spite of a consensus

on the need to distinguish between types of programs. Norman used the term

"upstream FSR" for that FSR that seeks prototype solutions to major constraints

to crop or agricultural improvement. This type of FSR is not meant to provide

results for immediate adoption by farmers, but rather contributes to a general

body of knowledge upon which "downstream FSR" may draw. This "downstream FSR"









is research for the purpose of formulating technological improvements useful in

the short run to target groups of farmers.


Workshop participants agreed on the conceptual distinction, but

were unhappy with the "upstream versus downstream" terminology. Several alter-

natives were offered: "support research versus applied research", "generation

research versus adoption research", "developmental research versus adoptive

research", "source research versus adoptive research". No consensus, however,

was evident.


The remainder of the Workshop dealt with methodological issues

facing social scientists in"downstream" (or applied, adoption or adoptive)

farming systems research.


2.2 The Role of the Social Scientist

The role of the social scientist in FSR, especially that of the

agricultural economist, came under close scrutiny. All workshop participants,

agronomists and social scientists alike, agreed on the need for participation

by social scientists in FSR.


The social scientist was seen as responsible for incorporating the

"human element", or socio-economic problems and circumstances that affect farmers'

decisions, into the design and evaluation of new agricultural technology. As a

member of a multi-disciplinary team responsible for FSR activities, he must take

as much responsibility for research decisions as an agronomist or a plant breeder.

Indeed, some argued that the agricultural economist was particularly well suited

to coordinate the work of an FSR team, because his discipline ("the science of

allocating scarce resources between competing ends to maximize utilities")

parallels the activities of a farmer in decision-making on resource allocation

(Collinson, 1980A).









In research planning or technology design, the agricultural econo-
mist was seen as using his acquaintance with farm survey techniques and his

sensitivity to opportunity'cost and the economic value of time to aid the agrono-

mist in answering such questions as: 1) What is a representative farmer? 2) At

what level should non-experimental variables be set? 3) What are high-priority

problems in production? 4) Which possible solutions to production problems seem

most feasible in the context of farmer goals and circumstances ("pre-screening")?

5) How may resource constraints be eased or system flexibility be employed to

increase farmer income and meet other farmer goals?


In evaluation of research results, agricultural economists were
seen to examine similar issues: Is the new technology profitable? What are

the constraints to adoption posed by input or product markets, or by inadequate

infrastructure? How do farmers themselves assess the new technology? Does the

new technology meet social as well as private goals?


Some workshop participants pointed out that the character of indi-
vidual team members is as important as their disciplinary training. Team members

must be willing to engage in problem-solving research to meet a common team
objective: the development of new technology usable by farmers. That is, they

must be "task-oriented".

In this connection, it was noted that some training in inter-disci-
plinary research for prospective team members could be useful. This training

would focus on the questions that one disciplinarian may fruitfully ask of another.

(Bartlett and Akorhe, 1980).










Finally, the role of the non-economic social scientist, especially

the anthropologist, was examined. It was felt that the anthropologist could

play an important part through his experience with informal methods of data

acquisition.

2.3 Alternative Users of FSR Results: Farmers, Policy-Makers and
Experiment Stations

"Downstream" FSR is normally couched in terms of its usefulness to

farmers, that is, its primary objective is commonly considered to be the delivery

of new, usable technology to farmers within a relatively short period of time.

Workshop participants pointed out, however, that farmers are at times not the

principal direct users of downstream FSR results. Downstream FSR activities

often lead to a realization that previous recommendations are not usable by

farmers and that none of the "improved" inputs or practices currently

available offer any improvement,'in the context of farmer goals and circumstances,

over the current farmer practice. In such a case, it must be determined how

either upstream FSR or traditional, reductionist research may be oriented to

provide new, useful inputs or practices. Downstream FSR can usually provide

some detail on the required characteristics of a new input (e.g., a white maize

variety with less than 110 days to maturity that has good husk cover and can

support climbing beans). Many Workshop participants pointed out, however, that

the feedback linkages between downstream FSR and either upstream FSR or reduc-

tionist research are rarely well defined. There appears to be an under-utilized

potential to use downstream FSR to orient more "basic" research.

Policy-makers were also noted as potential users of downstream FSR

results. This is, again, a case where the use of research results is potentially

significant but where linkages are currently not well developed.









FSR results should be of interest to policy-makers because they

can often be used to measure the costs (in terms of foregone agricultural pro-

duction) of current policies and to predict the impact of alternative policies

on farmer decision-making. This is not to say that FSR results can be used to

decide on correct agricultural policy. Rather, research results provide neces-

sary (but not sufficient) information upon which to base such decisions. Examples

of such a use of downstream FSR are, however, rare.


A related issue discussed in the Workshop was to what extent the
policy environment should be viewed as fixed versus variable. When should new

technology be designed to fit current policies and when should it be designed

to fit expected changes in policy? No consensus was apparent, although partici-

pants were unwilling to restrict FSR activities to either of the two alternatives.


2.4 Cost-Effectiveness of Downstream FSR

The issue of the cost-effectiveness of "downstream" FSR will likely

gain in importance in the near future, as donors begin to examine the impacts

of past expenditure on FSR. Although few cost-effectiveness studies are available,

there was a general consensus among workshop participants that "downstream" FSR

can be a cost-effective research tool. This consensus was reached despite the

fact that FSR is commonly viewed as costly to implement.


Three arguments were forwarded in support of the cost-effectiveness

of "downstream" FSR. First, given that most FSR activities are conducted on

farmers' fields, recurrent expenses increase for such items as travel and per

diem. Such recurrent expenses impinge more severely on the current budgets of

research administrators and cash flow problems arise. Research administrators,

however, find it relatively easy to ignore the opportunity costs of resources










embodied as fixed capital on experiment stations. That is, FSR may not be more

expensive than traditional research but the structure of costs, recurrent versus

fixed, may increase the administrative burden.


Similarly, "downstream" FSR was seen as more efficient than tradi-

tional research in terms of "technology adopted per unit of money spent". Work-

shop participants held a general belief that traditional on-station research in

developing countries had led to little adoption of new technology by small farmers.


Finally, it was observed that much can be done to reduce the costs

of "downstream" FSR. The acquisition of data on farmer problems and circumstances

provides one area of possible cost reduction. Informal, non-probabilistic surveys

and well-focused, single-visit, small-sample formal surveys were seen as generally

preferable to large-sample surveys or frequent-visit surveys in this connection.

The criterion for the selection of survey instrument should be that of "the lowest

possible cost commensurate with the degree of understanding that is necessary"

(Norman, 1980).

Another area of possible cost reduction is that of expanding the

universe for which downstream FSR results are applicable. This may be performed

by determining the transferability of one set of results to other similar envi-

ronments. In this fashion, some (but rarely all) of the steps in downstream

FSR may be skipped.

Finally, it was pointed out that it is unnecessary for FSR to pro-

duce the "best" new technology for farmers. Insofar as it discovers anything

"better" than the current farmer practice, it will be useful. That is, FSR need

not engage in the fine tuning of the farming system, but rather may concentrate

on seeking the best of readily available solutions to important problems.










3.0 The Farmer as the Researcher's Client

The farmer is normally considered the primary client of "downstream"

FSR, although both policy-makers and "upstream" FSR may also be regarded as

clients. Insofar as research is aimed at helping farmers, some insight into

how to deal with the farmer-client is needed in planning research.


3.1 Target Groups of Farmers: Recommendation Domains

One such issue on "dealing with farmers" that was discussed in the

workshop was that of the identification of target areas and target groups of

farmers. A target area is merely a geographical area selected by policy-makers

and researchers to be a priority area for FSR. Workshop participants recommended

that policy-makers use several criteria in selecting a target area, including

research costs, likelihood of successfully developing new technology, and likely

impact.on such national goals as income distribution, employment, or savings of

foreign exchange.


Within a target area, it was pointed out, the next necessary step

is the delineation of target groups of farmers or "recommendation domains".

This step is necessary because both possible alternatives (conducting research

for individual farmers, or conducting research for all farmers in the target

area, heterogeneity in circumstances not withstanding) are clearly unfeasible.


There was less consensus, however, on the appropriate.criteria for

delineation of recommendation domains. Some argued that major agro-climatic

differences be used to identify "land-types" and that research be conducted in

the context of the known characteristics of a given "land-type". Others argued

that the current farming system itself be used to distinguish between recom-

mendation domains, that is, that farmers currently operating similar farming









systems be included in the same recommendation domain. This is because the

current farming system has evolved via a series of decisions made by farmers as

they weigh and balance all circumstances -- agro-climatic, socio-economic and

institutional -- that impinge on production decisions. Furthermore, the current

farming system provides the basis upon which new technology will be added by

farmers (Collinson, 1980 B).


The appropriate size of a recommendation domain was also posed as

an issue. Workshop participants agreed that the number of farmers to be grouped

into a single recommendation domain varies inversely with the size of the research

budget and with the heterogeneity found in farming practice in the target area.

Size, then, may vary from only a few thousand farmers to tens or even hundreds

of thousands of farmers per domain.


3.2 Focus on the Whole System Versus on a Pre-Determined Enterprise

The phrase "farming systems research" carries with it a connotation

of wholeness, a feeling that everything in the farming system must be considered

simultaneously. As was pointed out in section 2.1, however, "farming-systems-

research-in-the-small", in which the ratio of variables to parameters is low,

may be considered valid FSR because planning and evaluation of research on a

limited number of variables is conducted in the context of the whole system.


The question was raised regarding the extent to which the ratio of

variables to parameters should be minimized. Specifically, when is it wise to

focus research on a pre-determined commodity rather than on the whole farming

system, or on the crop sub-system?


Two conditions favorable to concentration on a pre-determined enter-

prise were noted in the workshop. First, when research is planned in a region










where the pre-determined enterprise in question is a major absorber of farmer

resources, that enterprise will frequently offer the best leverage on such

system problems as deficient income, excessive risk, and seasonal variability

in the use of farmer-owned resources. Second, agricultural research in devel-

oping countries is frequently organized by crop. The use of the pre-determined

enterprise focus allows an easier introduction of FSR concepts into on-going

research programs, especially (as is often the case) when there exist poor

linkages among agricultural research institutes, and between them and policy-

making and farmer-contact agencies (Collinson, 1980 B and Norman, 1980).


From the above, it should be clear that a focus on the broader
cropping system may be advisable when there is little scope for improvement in

the farmer's major crop activity and when the organization of research adminis-

tration allows this more ambitious approach.


3.3 Technological "Pieces" versus "Packages"

The wisdom of recommending complete packages for LDC farmers has

been considered an issue for some time. Five years ago, the "package of prac-

tices" approach was criticized in the Indian context (Ryan and Subrahmanyam,

1975). More recently, a sequence of adoption of practices was favorably compared

to the traditional package approach (Mann,


Surprisingly, Workshop participants found the "package" versus
"pieces of technology" question to be a non-issue. It was felt that the focus

should more properly be placed on the scaleand complexity of new inputs or

practices. A package of scale-neutral and simple innovations should be more

usable by farmers than a single but complex technological "component" which is

furthermore unusable on small farms. In addition, insofar as a "package" is

recommended (say, due to strong complementarity between package components),










each component should be individually tested for acceptability in terms of scale

and complexity. Otherwise, the rejection of one component will likely lead to

the rejection of the whole package.


In light of the above, workshop participants largely felt that

individual components or "small packages" were more likely to pass the scale/

complexity test than large packages and that, therefore, research should gener-

ally avoid the formulation of large "packages" of practices.


4.0 Acquisition of Information on Farmer Problems and Circumstances

A broad consensus was soon reached on the need for information on farmers:

An intimate acquaintance with the natural and socio-economic circumstances and

problems of target farmers was seen as necessary to the proper design and eval-

uation of technology. Discussion focused on the advantages and disadvantages

of alternative data sources and data collection instruments in achieving the

desired degree of understanding.


4.1 Secondary Data

Such secondary data on soils maps, land use maps, rainfall records

and the agricultural census were generally felt to be useful in distinguishing

parameters from variables, or "the environment vector" from "the management

vector", at least in a preliminary way. Such data may lead to a tentative

delineation of target groups of farmers or "recommendation domains" for the

target area in question. Furthermore, some participants felt that extensions

of these recommendation domains into locations outside of the current target

area may be hypothesized, based on a careful description of environmental

characteristics.










However, it was felt that secondary data will rarely provide the
insight into farmer problems and circumstances needed to design new technology.

Such data is generally inadequate to list production problems facing farmers,

"pre-screen" possible solutions to these problems in light of current farming

systems, or even describe current farmer practices. It is almost always, then,

necessary to gather some information directly from the target farmers themselves.


4.2 Farm Surveys: Informal versus Formal

Three major alternatives for obtaining information from farmers
were discussed by Workshop participants: informal surveys (at times referred

to as "sondeos"), one-contact formal surveys, and multiple-visit formal surveys

or farmer panels. Farmer record-keeping was grouped with multiple-visit surveys.

Each of the three alternatives was strongly advocated by at least one participant.

The advantages and disadvantages of each data collection instrument were discussed

and a surprising degree of consensus was reached with regard to the conditions

under which a given instrument may best be used.


Informal surveys (informal conversations with farmers without the
use of enumerators, questionnaires, or a random sample) were felt to be useful

and even necessary under a wide variety of conditions. They provide a rapid

and inexpensive means of obtaining a qualitative understanding of the farming

systems of target farmers (Hildebrand, 1979). They represent the minimum in

data collection that is required for planning downstream FSR. That is, when

time is short and resources (especially skilled personnel and transport) are

very scarce, informal surveys are a usable method for identifying the most

important production problems facing farmers, and for describing the salient

features of the natural and socio-economic environment in which farmers make

decisions.










Furthermore, informal surveys were felt to be essential for the

planning and execution of subsequent formal surveys, if such are desired. They

guide researchers in hypothesis formation, questionnaire development and choice

of sampling technique.


If the advantages of informal surveys (speed, low expense) were

recognized by workshop participants, their disadvantages were similarly noted.

The lack of random sampling combined with a focus on qualitative data may easily

lead to credibility problems: It may be difficult to convince decision-makers

not involved in the informal survey process that the data are valid. Further-

more, when representativeness is an issue ("Just what are the characteristics

of a representative farmer?"), a careful answer is not to be had without data

based on random sampling. Finally, even proponents of informal surveys admit

that they are not good sources of data for the analysis of experimental data:

This is due to the lack of quantification inherent in the method. In short, it

was felt that occasions will frequently arise when it is wise to follow an in-

formal survey with a formal survey. When time and resources allow, and when

researchers are faced with the issues of "credibility", "representativeness" or

"quantification" noted above, some kind of formal survey is advisable.

In this correction, some participants proceeded to advocate the

use of farmer panels, or multiple-visit surveys. It was argued that multiple-

visit surveys provide superior time series data, especially with regard to such

variables as labor flow, cash flow and non-cash income, that farmers find diffi-

cult to remember. They build more rapport with farmer-respondents, facilitating

the acquisition of reliable data on sensitive questions. (Binswanger and Ryan,

1979). In general, they provide a "depth" of understanding at the expense of

"breadth". That is, repeated visits to the same farmers may lead to a superior










understanding of the complex details of a local farming system ("depth") but

it will prove relatively difficult to generalize and extrapolate the under-

standing that is achieved ("breadth") due to the necessarily small number of

farmers visited.

A number of questions were raised with regard to the use of multiple-

visit surveys in planning and evaluating downstream FSR. Some participants felt

that the precision in measuring flow variables that is achievable with farmer

panels is excessive for the purpose at hand. That is, the increase in precision

may not be worth the increase in expense. Indeed, the expense of multiple-visit

surveys was a source of concern to many. Participants noted the high staff-time

requirements, enumerator and travel expense, the time lag between survey initia-

tion and its termination, and the expense and difficulty in analyzing and re-

porting survey results. Some participants questioned whether the widespread use

of farmer panels by national FSR programs is at all practical in light of the

above difficulties. Farmer panels, they felt, may be more appropriate to inter-

national centers that conduct upstream FSR.

A final alternative was discussed: the single-contact formal survey.

Proponents of this data collection instrument noted that a small-sample, random-

sample, single-contact formal survey can be used to quantify and "verify" the

results of a preceding informal survey. This formal survey is focused on topics

of importance, as determined by the informal survey. This combination of informal

survey plus small-sample formal survey overcomes many of the problems associated

with exclusive reliance on informal surveys ("credibility", "representativeness",

and "quantification") while avoiding the cost-related problems associated with

farmer panels (CIMMYT, 1980). Subsequent discussion brought out several diffi-

culties in using this approach. In the context of a complex farming system, it






16.


is difficult to obtain precise estimates of flow variable with a single-contact

survey. (There was little discussion, however, on how precise the measurement

of these variables must be in order to plan and evaluate research on new tech-

nology.) Furthermore, it is difficult to gather information on sensitive ques-

tions because a single-visit does not allow much development of rapport with

the respondent. The use of random sampling may be burdensome if suitable sam-

pling frames are lacking. Finally, data processing and analysis, while far less

onerous than that associated with farmer panels, may still lead to unexpected

costs and delays.

In summary, all three data collection methods demonstrate advantages

and disadvantages. With specific reference to downstream FSR conducted by na-

tional programs, however, the following may be concluded: 1) An informal survey

represents the minimum data collection effort necessary for planning research.

2) Where time and resources allow, it is wise to follow the informal survey with

a formal survey. 3) In many cases, a small-sample, single-contact formal survey

will be sufficient to "verify" the results of the informal survey. 4) However,

when time and resources allow, and when flow variables must be measured with

some precision in the context of a complex farming system, researchers might

wish to consider using a farmer panel in addition to a single-contact survey.

4.3 Observations in Farmers' Fields

One further data collection method was discussed: observations in

farmers' fields. This method has at times been combined with informal or formal

survey activities and has been used to help identify high-priority problems for

downstream FSR. Workshop participants agreed, however, that field observation

by itself is insufficient for identifying and ordering in importance the pro-

duction problems with which target farmers must cope. Field observations are










time and location specific, and are regarded as relatively costly. That is,

to gain an idea of the variation in the incidence of problems in farmers' fields,

quite a few sites must be inspected. Yet even this procedure will not deal with

the variation in the incidence of problems over time. Nonetheless, observations

in farmers' fields may prove useful when combined with data from farm interviews.

For example, the combination of a physical weed count in a farmer's field and

data on the farmer's weed control practices for that field may provide a clearer

insight into the scope for improving weed control practices. More generally,

field observation can lead to the identification of some production problems

while data from farmers can help determine the frequency of incidence and sever-

ity of those problems, and what farmers are currently doing about them.

5.0 Design, Testing and Evaluation of New Technology

The social scientist's influence in applied FSR may be measured by his

effect on the decisions taken in the design, testing and evaluation of new tech-

nology for a target group of farmers, or recommendation domain. Armed with his

knowledge of farmer problems and circumstances, and the tools of economics, he

is in a good position to aid agronomists and other biological scientists in pre-

screening possible solutions to high-priority problems, choosing representative

farmers for collaboration, analyzing experimental data, and much more. Nonethe-

less, the social scientist must address a series of procedural issues in order

to fulfill his role.


5.1 Pre-Screening Alternative Technologies: Technology Design

Workshop participants tended to agree on the basic steps in tech-

nology design. First, factors that limit production of the target enterprises)

must be listed. Those limiting factors about which nothing can be done are

relegated to the status of parameter -- they cannot be used as experimental









variables. For each limiting factor that remains, a list of potential solutions

or treatments drawn up.

Each potential solution is "pre-screened" for feasibility. That

is, budgets are constructed to determine yield changes needed to pay treatment

costs -- agronomists then can estimate the likelihood of achieving these yield

changes. Apparently profitable treatments are then examined for consistency

with the farming system of target farmers. Input and product markets, cash flow,

labor flow, crop calendars, etc. are noted and the effects of alternative treat-

ments on each are estimated. Those treatments that appear profitable and that

mesh well with the current farming system (use under-employed resources, reduce

resource use during peak demand periods, etc.) receive priority in technology

testing.

Some participants used different terms to describe the above crite-

ria for technology design: "biological feasibility", "resource feasibility"

(resource availability), and "economic viability" or profitability (Zandstra,

1980). The net effect, however, is about the same.

Although there was widespread agreement on the basic steps of tech-

nology design and pre-screening, there was considerable variance among partici-

pants in the detailed steps which they tended to follow. Some advocated formal

meetings a month before planting, attended by the entire research team, in which

experimental treatments are defined in detail. Others suggested a more informal,

flexible approach. These differences in detail were largely attributed to such

factors as team size, scope of the experimental program (variable to parameter

ratio) and class of experiment for which treatments are being designed.









One issue related to the identification of limiting factors was

raised but, unfortunately, was not treated in the detail that it merits: Are

problems in crop and livestock production best identified by researchers,

through observations in farmers' fields and/or through inferences derived from

farm survey data, or directly by farmers as they comment on their "felt needs"?

There is undoubtedly a role for both approaches. The balance between them,

however, is not well defined.

5.2 On-Farm Experimentation

Some of the broader issues related to on-farm experimentation have

already been discussed (research on the whole system versus research on a pre-

determined commodity, complete "packages" versus technological "pieces", etc.).

Other, narrower issues still remain regarding this topic, however.

One accepted role of the social scientist in applied FSR is in

aiding other researchers in the choice of representative farmers as collaborators.

In practical terms, this can become a difficult task because no single farmer

ever precisely represents all other farmers in a given recommendation domain.

Two feasible approaches were noted. In the first, farmers are asked (during a

formal survey with a random sample) if they would be willing to collaborate in

on-farm research. Terms and incentives are described. Of those farmers willing

to-collaborate (and experience shows that most farmers are willing if the incen-

tive structure is adequate), a random sub-sample is drawn.

In the second approach, farmer-collaborators are selected purpo-

sively. Any farmer in a given domain may be chosen as long as he fits into,

say, the middle 50% of the frequency distributions for those variables consid-

ered to be especially important by researchers.









One issue that received considerable attention from workshop par-

ticipants was that of the proper level for non-experimental variables (NEV's)

or the "background matrix". Three alternatives were presented: the farmer's

level, some high and non-limiting level, or some intermediate level (chosen for

its expected profitability in a future technology package). The choice of NEV

level was found to depend on the research client and the level of research

("upstream" or "downstream"). Selected NEV's may be sit at high, non-limiting

levels for some kinds of "upstream" research or for demonstrations for policy-

makers (e.g., on the yield effect of an unavailable input). The consensus of

participants, however, was that NEV's should be set at the farmers' level in

trials aimed at formulating near-term recommendations for farmers, that is,

in "downstream" research. Some participants noted, however, that as research

continues, some NEV's may be raised to an "expected future recommendation level".

One issue upon which there was little consensus is the proper num-

ber of experimental variables to be included in a research program for a given

recommendation domain. Suggestions ranged from three to thirty. This is, of

course, merely a re-statement of the issue of the proper scope of research --

that is, the question of the variable to parameter ratio, which has already been

treated.

The question of farmer participation in on-farm experiments was

discussed briefly. Farmers were seen as playing many roles in on-farm research:

responding to surveys, loaning their fields, commenting on alternative experi-

mental treatments, completely managing other treatments, and sharing their

experiences with new technology. The extent of direct farmer involvement, how-

ever, is dependent on the kind of experiment being conducted. Farmer participa-

tion becomes increasingly crucial as one moves from small-plot, replicated










experiments to verification of recommendations for new technology. Farmer-

managed experiments in which farmers operate the new technology themselves and

are even responsible for input acquisition and risk management, were seen as

useful in testing recommendations flowing from small-plot, researcher-managed

trials.


A host of other issues related to on-farm experiments were raised

but not discussed in the workshop. For the record, they are listed as follows:

choice of experimental design, plot size, number of replications per site per

experiment, number of sites per experiment, number of experiments per site,

extension personnel involvement, highest C.V. acceptable, probability level for

the rejection of the null hypothesis, and assembly of site-specific data during

experimentation (Violic, 1980).


5.3 Economic Analysis of Experimental Data

The issues related to economic analysis of the results of on-farm

experiments were divided into two classes: private effects and social effects

of new technology.


The choice of analytical technique was the major issue regarding

the measurement of the private effects of new technology (that is, costs and

benefits that accrue only to farmers who adopt the new technology). Participants

agreed that kind of experiment (researcher-managed versus farmer-managed), choice

of target crop, level of research (upstream versus downstream), and location of

research (national programs versus international centers) influence the choice

of proper analytical techniques. Analysis of variance to determine the signifi-

cance of differences between treatment means was advocated, but is clearly re-

stricted to small-plot, replicated experiments. Partial budgets and, at times










enterprise or whole farm budgets, were proposed to examine the profitability of

new technology. Finally, linear programming and formal risk analysis were

advocated for use under special circumstances: when the target enterprise

interacts strongly with other farm enterprises, when there are no markets for fanner-owne

resources and when time and resources are available to correctly use these tech-

niques. At times, some participants noted, international centers become directly

responsible for downstream research because no national program is available to

accept this responsibility. In such a case, more sophisticated techniques may

be used to analyze experimental results. In general, however, it was felt that

national FSR programs should develop the capacity to use budgets with imagination

and flexibility rather than attempt the use of more complicated analytical tools.

A further issue in the measurement of private effects of new tech-

nology is that of the adjustment of experimental yields. Researchers normally

obtain higher yields than farmers, even when they simulate the farmer's practice.

When they occur, these yield differences are thought to be due to unconscious
*-
improvement in management (timeliness in weeding or insect control, better plant

population) or earlier harvesting (by virtue of which researchers avoid field

losses that farmers must face). Inflated yield estimates, however, are likely

to lead to inflated expectations of returns and over-stated estimates of prof-

itability. Workshop participants agreed that the problem is a common one, but

reached no consensus on how to handle it. A proportional deflation of yields

for all treatments was not found to be generally acceptable, as interactions-

between treatments and the causes of yield inflation are likely to exist.

In the workshop issues surrounding the use of survey and experi-

mental data to predict the broader consequences of new technology for society

were briefly discussed. As one participant noted, we would like to be able to









predict the effect of new technology on nutritional levels, employment, income

distribution, long run versus short run effects, the balance of payments, etc.

-However, even ex-post explanations of the impacts of new technology have been

generally unsatisfactory. Ex-ante prediction of these effects are likely to be

even less satisfactory because of difficulties in predicting such factors as:

farmer adaptation of new technology, adoption rates by farmer strata, yield

and employment changes over time and space, performance of support institutions

(e.g., input suppliers and product markets), and institutional change due to'

the effects of new technology (Flinn, 1980).

5.4 Farmer Assessment of New Technology

"Economic evaluation of experimental results", as discussed in the

proceeding section, is one step in selecting new technology useful to, and
usable by farmers. In conducting economic evaluation, researchers assume

that they are familiar with the important costs and returns associated with new

technology and that the decision criteria they use are similar to those used

by farmers.

One further step in evaluation is possible, however, and was briefly

discussed in the Workshop. This step may be termed "farmer assessment" of new

technology or "monitoring farmer experience" with that technology. In farmer

assessment, those farmers with sufficient experience with the new technology

(such that they have made a decision for or against its use) explain how they

employ it, list its advantages and disadvantages, and describe why they decided

in favor or (or against) its use in commercial production. This normally

implies that farmers must have used the technology, possibly upon the request

of researchers, on their own fields on a substantial scale, incurring all costs,

benefits, and risks. Farmer assessment is likely to be useful: 1) when farmers









had little previous experience with a component of the new technology (and

therefore cannot be expected to provide useful information in an ex-ante

survey), 2) when a knowledge of the recommended technology is needed to sharply

focus questions regarding farmers' experiences (because there are many possible

solutions to farmers' problems), 3) when the farming system is complex (and

therefore costly to accurately represent in a formal model), and 4) when ex-ante

survey work was not well conducted (Harrington, 1980).


Two issues related to farmer assessment were brought out: Which

farmers should be chosen to assess a new technology, and which data collection

instruments are best used in assessment?


With respect to choice of farmer, it was argued that only "knowl-

edgeable" farmers should participate in assessment. It is unfair to include

in assessment those farmers who have not made up their minds. This can restrict

assessment to ex-collaborators and early adopters. Will they be representative

of target farmers? In any event, how does one determine whether a given farmer

is sufficiently "knowledgeable" that he may participate in assessment?


With respect to data collection instruments, it was pointed out

that a wide range of tools may be appropriate, depending on the number of target

farmers with previous experience with the new technology to be assessed. Purpo-

sive surveys, random surveys and formal farmer trials are all likely to play a

role at one time or another.


6.0 Institutionalizing Social Scientists in Agricultural Research Systems

As a final theme, workshop participants discussed the issues that

surround the establishment of social scientists in agricultural research insti-

tutions. The major issue was that of the organization of agricultural research










as such, and the place of the social scientist under varying organizational

arrangements. Interwoven into this main issue were sub-issues of incentives,

minimum qualifications, and training.

With regard to the organization of agricultural research, three alterna-

tives were presented and discussed. The first alternative stemmed from expe-

rience in E. Africa, where both "upstream" and "downstream" research are co-

ordinated through area-based experiment stations. Social scientists have re-

cently been introduced directly onto these stations as staff members. This

role is two-fold: Help agronomists design "downstream", on-farm research that

is more relevant to the needs of target farmers, and provide feedback on farmer

problems to help breeders re-orient "upstream", on-station research. The major

problem faced to date is one of status: The junior-level economists find it

difficult to deal with the relatively senior-level breeders and agronomists.

A second alternative stemmed from experience in Ecuador and Guatemala.

In these cases, a clearer mandate is given to a research team to conduct "down-

stream" research for defined areas. The tie to the experiment station still

exists, but is looser. The on-farm research team receives transportation faci-

lities with which to conduct its work and, in one case, receives a salary pre-

mium. That is, their wage is above that of their "on-station" counterparts.

The research team, which has a mandate to deal with all important crops, is

largely composed of agronomists. A social science support unit is available to

aid research teams in conducting and analyzing farm surveys, maintaining contact

with collaborators during experimentation, and analyzing experimental data. But

social scientists are not part of the research team as such.










This arrangement appears to ease many of the status-related problems

noted under the first alternative. It seems to have been effective in making

agronomists more sensitive to economic issues (and economists more sensitive

to biological issues). However, insofar as social scientists do not share

responsibility for research decisions, and insofar as they cannot become inti-

mately acquainted with a research area by continuously working there, their

input may be less effective and more easily ignored.

A third alternative for organizing agricultural research stemmed from

experience in Asia. Under this alternative, a research team is responsible for

"downstream" research for a defined area, with little formal contact with a

given experiment station. A social scientist is directly assigned to the

research team and is just as responsible as other team members for research

decisions. Most research decisions are taken in a formal meeting of the whole

team about a month before planting.































ANNEX I




ApAzi .1-3, 1980


WoAUhop on methodologicall Issune Facing SociaZ Scieitists
in On- Farm/FaAruin Sqytexmns .Reseac.h"


Lis6t o PaAicipan-ta


DA. C.D.S. Battkett
I ITA
Oyo Road, P.M.B. 5320
Ibadan, Nigeaia

pr. D/LekI BuleZee.
Econooincs, CIMMYT

Dr. MLchaeg Cot4znon
ClhMMiYT
P.O. Box 25171
Nairobi, Kenya

Mtl. Nabacui E1 Sayed
Inst. o6 AgricutWuAlt Econ.
ResedAch, Dokki, Caito
(CuArrentty in CIMM T Economica
TAaining Cowise)

Da. John Flinn
Economics, IRRI
P.O. Box 933
Manila, Philippines

Dt. Richzad Goldman
Ha-rvaAr TIntitute Tot InteA.nationeZ
Development
1737 Cambridge S-teet,
CambAidge, Mlss. 02138

DV. Larry HaAing.ton
Economics, CIMMYT

Drt. P.B.R. Hazetf
IFPRI
1776 Massachusetts Ave. N.W.
Washington, D.C. 20036

Dt. PeteA Hildebtand
Food Reouace Economic6 Dept.
Univeuaity o6 Florida
Gainezvieue, Flotida 32611

DL. Douglas Hoaton
CIP
Apartado 5969
Lima, PeAu


OD. FedeAico KocheA
Maize' Tuaining Proguwn, CIMMYT

DV. John Lynam
CIAT
Apartado Ae'eo 6713
CatL, Colombia

Il. P.J. Makundi
ARI llonga, Tanzania
(Cwutentty in CIMMYT Economic6
Training CouAbe)

pDr. Juan Caktos Mattinez
Economic, CIIMVT

D&. Philippe Masson
PAoject CereaZs CIMIYT/IDGC
B.P. 16 EZ HaAAach
AtgeA, Algeaie

Da. Edgatdo Mosncardi
CIM!,YT/INIAP
ApaAtado 2600
Quito, Ecuadoa

DV. HiZmat Ncas
Wheat TAtaing PrLogina, CIMMYT

DV. Lu&s A. NavaAto
CATIE
TuwAiatba, Costa Rica

D&. David Norman
Dept. of Economics
Kansas State UnivvM ity
WateAz Hatt
Manhattan, Kan-sa 66506

Ing. Ramiun OArtiz
TechnicalZ Dviector, ICTA
Edijicio CoAtez, 5a. Avenida 12-31
Zona 9, 2 Nivet
Guatemala, Guatemnia

DLA. A.F.E. Palmet
Maize Training P:og.Lam CIMMYT


*








M[. Mu.jadi Pudjdo'umaAto
University o d Bawijaaya, Indonesia
(CwVAentfy in CITMWT Economics
TAaining Cous e)

Dr. DuuvvuAL RamakAulhnazah
G. B. Put Univelshity (IndCa)
(CuwAentt2 in CIMMYT Economic6
Training Co u- e)

Dt. RobeAt Rhoades
CUTP
Apattado 5969
Lima, Peru

Dt. Jawme Riyan
Dept. o6 Economics
Duke Unlzvesiity
ODuham, Notth CaAot&na
(on Leave. wom ICRISAT)

D0. W. Schmeah
School o EngineeAkng
Colotado State UrNveuit/t
Fott Co~-Uin, ColoAado 80523

Dt. W.S. ShaneA
-Schw-ooi o73 Enginee.ung
CoZloado State University
Fort Collinz, Colorado 80523

DO. Du0wtan SpenceA
Head, Devetopment Department
WARDA
P.O. Box 1079
MonAovia, LibeAia

Dr. RobeAt TAipp
CIMM T/ INTIAP
Apatado 2600
Quito, Ecuadox

DA. Antonio Turrent
Rama de Suetlo
CoZegio de Postgraduados
Chapingo, Mex.

D4. Atejandko Viaoic
Maize Traainig Ptogram CIMMYT

Mt. Seth VordzoAqbe
CRI Ghana
(CahrntU-y in CIMMYT Economicl
Training Cow ure)


Dp. V.S. V Ia Director
Indian In-titute o Mianagement
Vast.apuA, Almnedabad 380015

DA. PatAick Wael
Wheat PAogram CIMJVT

D4. Donald Winkelmann
i ectoh, Economics CIMMYT

rt. HubeAt ZandAtta
Head, Cuopping System, IRRI
P.O. Box 933
Manita, Phitippi.es


I










REFERENCES


Bartlett, C.D.S. and J.A. Akorhe, 1980.
to Identify Innovations for Small
mist". I.I.T.A.


"Interdisciplinary Co-operation
Farmers -- The Role of the Econo-


Binswanger, H.P., and J.G. Ryan, 1979. "Village Level Studies as a Locus
for Research and Technology Adaptation". ICRISAT.


CIMMYT, 1980. "Planning Technologies Appropriate to Farmers; Concepts and
Procedures".


Collinson, M.P., 1980A.
note presented at
Social Scientists


"The Role.of the Social Scientist". Background
the conference on "Methodological Issues Facing
in On-Farm/Farming Systems Research".


Collinson, M.P., 1980B. "Some Notes on the Farmer as the Client for Research".
Background note presented at the conference on "Methodological Issues
Facing Social Scientists in On-Farm/Farming Systems Research".


Dillon, J.D., 1978. "Farming Systems Research at the International Agricul-
tural Research Centers". Washington; Technical Advisory Committee,
Consultative Group on International Agricultural Research.


Flinn, J., 1980. "Analysis of
ground note presented at
Facing Social Scientists


Broader Consequences for Society". Back-
the conference on "Methodological Issues
in On-Farm/Farming Systems Research".


Gilbert, E.H., D.W. Norman and F. Winch, 1980.
Agricultural Development: A Review of the
come Countries". East Lansing, Michigan


"Farming Systems Research for
State of the Arts in Low In-
State University.


Harrington, L., 1980. "Farmer Assessment of New Technology". Background
note presented at the conference on "Methodological Issues Facing So-
cial Scientists in On-Farm/Farming Systems Research".


Hildebrand,P.E,, 1979. "Summary of the Sondeo Methodology Used by ICTA".
Guatemala: ICTA.


Mann, C.K., 1977. "Factors Affecting Farmers' Adoption of New Production
Technology: Clusters of Practices". Paper prepared for presentation
at the Fourth Regional Winter Cereals Workshop Barley








REFERENCES 2.



Norman, D.W., 1980. "General Overview of Farming Systems Research".
Background note presented at the conference on "Methodological
Issues Facing Social Scientists in On-Farm/Farming Systems
Research".

Ryan, J.G., and K.V. Subrahmanyam, 1975. "An Appraisal of the Package
of Practices Approach in Adoption of Modern Varieties". ICRISAT.

Violic, A., 1980. "On-Farm Experimentation". Background note presented
at the conference on "Methodological Issues Facing Social Scientists
in On-Farm/Farming Systems Research".

Zandstra, H.G., 1980. "Design of the On-Farm Research Program". Background
note presented at the conference on "Methodological Issues Facing Social
Scientists in On-Farm/Farming Systems Research.




University of Florida Home Page
© 2004 - 2010 University of Florida George A. Smathers Libraries.
All rights reserved.

Acceptable Use, Copyright, and Disclaimer Statement
Last updated October 10, 2010 - - mvs