In the high income, or so-called developed, countries the conventional
experiment station-based agricultural research system has generally
served farmers well. However, the issue of increasingly limited research
resources together with rising concern about ecological sustainability and
recognition that the needs of certain farmers (e.g., smaller and more
diversified farmers) are not being sufficiently addressed, has prompted
questions to be asked about the potential of greater use of on-farm, and
hence farmer-based, research.
The relative lack of success of the conventional station-based agricultur-
al research system in many low income (i.e., developing) countries led to
the evolution of a farmer-based research approach popularly known as
farming systems research and extension. The extensive literature on this
subject, combined with our own experience in its development and ap-
plication, provides the impetus for this paper, which also draws on the
burgeoning literature on farm-based research in the developed countries.
After briefly charting the rationale and recent evolution of farmers' in-
volvement in agricultural research in low and high income countries, w\e
justify why on-farm research should be encouraged in the high income
countries such as the USA. We finish by considering four issues that
need addressing if the full potential of the complementarity between
station-based (i.e., on-station) and farm-based (i.e., on-farm) research is
to be exploited.
Should on-farm research in the USA be encouraged?
We believe that conventional stationl-based research and on-farm
research have different roles, strengths, and weaknesses. The experience
of others, and ourselves, has convinced us of the complementarity be-
tween the two approaches to research. Consequently, we believe that the
two should be integrated (i.e., be part of a continuum) to a greater ex-
tent than is currently the case. Both constitute critically important com-
ponents of a well-balanced research program. Station-based research
provides a convenient milieu for the continued application of the reduc-
tionist approach that has been so important in identifying ways to im-
prove the productivity of agriculture. On the other hand, the efficiency
of, and the return from, such research probably will be improved if it is
complemented by on-farm research that can provide an ideal environ-
ment for involving the farmer in the research process, for placing
research in a systems context, and for addressing location specificity,
particularly with respect to ecological sustainabilty.
Full scientific article from which this summary was written begins on page 124 of this issue.
J. Prod. Agric., Vol. 7, no. 1, 1994 31
Agricultural Research in Developed Countries:
Past, Present, and Future of Farming Systems
Research and Extension
D. W. Norman, T. R. Frankenberger, and P. E. Hildebrand
What are the critical issues that must be addressed in exploiting the in-
teractive linkages between station-based and on-farm research?
There are at least four critically important issues with references to on-
farm research that need resolution if there is to be truly effective, and
mutually beneficial, interaction between those engaged in station-based
research and those involved in on-farm research. The first involves the
greater acceptance and incorporation of farmers in the research and the
use of an interdisciplinary approaching eliciting their opinions and im-
proving the potential accuracy and return from a more holistic systems
perspective. Techniques developed in low-income countries, such as
rapid rural appraisal, provide a cost effective starting point for address-
ing this issue. The second issue needing resolution involves carefully
defining the type of on-farm trial to undertake based on considerations
such as the objectives of the trial and who the major clients of the trial
results are likely to be. Answers to such questions will determine the
trial design and who manages and implements it. Such considerations
will also have an influence on determining the costs of implementing
trials on-farm. Resolving the issue of costs with respect to on-station
and on-farm research requires further examination as does the issue of
using experimental designs and analytical techniques in on-farm research
that are acceptable to researchers based primarily on experiment sta-
tions. Experience in low-income countries has convinced us that the
"them" and "us" mentality of experiment station-based and on-farm
based researchers needs to be blurred deliberately through increased dia-
logue and direct collaboration between the two groups. Nevertheless, we
recognize that this is likely to be a long and slow process.
32 J. Prod. Agric., Vol. 7, no. 1, 1994
In the developed countries, the conventional experiment
station-based agricultural research system generally has served
farmers well. In developing countries, however, the same model
has had less success in addressing needs of many farmers, par-
ticularly those operating diversified farming systems in resource-
poor and risk-prone environments. Hence, in the 1970s, the
farming systems research and extension (FSRE) approach-
which has a high degree of complementarity with experiment
station-based research--evolved to address their needs. Prior
to this was the development of some analogous farm-based ap-
proaches in the USA (e~g., balanced farming). In the USA, in-
creasing concerns about himated research resources and
ecological sustainability have sparked interest in harnessing the
power of the farm-based and systems-oriented FSRE approach
in a way that is complementary to other approaches and in a
manner that produces results acceptable to all stakeholders (e.g.,
farmers, station-based researchers, extension/development staff,
and planners of policy/support systems).
T~HE OBJECTIVE of this paper is first to chart briefly
Sthe rationale and recent evolution of farmers' in-
volvement in agricultural research in low and high income
countries. This is followed by a more detailed discussion
of the reasons why farmer participation in the research
process should be encouraged and a consideration of four
issues that have to be addressed if the potential com-
plementarity between station-based (i.e., on-station) and
farm-based (i~e., on-farm) research is to be truly ex-
RATIONALE AND EVOLUTION
Conventional Research System in High Income
With the advent of the scientific revolution, there has
been considerable success in the USA and other high in-
come countries with an agricultural research approach
involving the development of improved technologies,
primarily on research stations. These are then dissemi-
nated to farmers via the extension service, which main-
tains close interactive two-way links with both parties.
Reasons for this success include:
D.W. Norman, Dep. of Agric. Economics, Kansas State Univ., Man-
hattan, KS 66506-4011; T.R. Frankenberger, Univ. of Arizona, and
P.E. Hildebrand, Univ. of Florida. Based on invited paper given at
the Farming Systems Symposium, 1992. ASA Annual meeting, Min-
neapolis. 1-6 November. The constructive comments of J. Siebert and
S. Freyenberger on an earlier draft of this paper are gratefully ac-
knowledged. Kansas Agric. Exp. Stn. Contribution no. 93-282-J.
Received 18 Feb. 1993. *Corresponding author.
Published in J. Prod. Agric. 7:124-131 (1994).
Agricultural Research in Developed Countries: Past, Present,
and Future of Farming Systems Research and Extension
D. W. Norman,* T. R. Frankenberger, and P. E. Hildebrand
* Researchers on experiment stations have tended to have
a good idea of the problems faced by farmers and,
therefore, have focused their research on relevant
themes. This knowledge and resulting focus have been
consequences of, among other things, the fact that
many, at least until recently, came from farm back-
. In addition to feedback to research from farmers
through extension, many farmers have had additional
avenues to get their needs addressed (e.g., commodity
groups, lobbying, research by commercial organiza-
tions, etc.), and an increasingly high degree of monitor-
ing/accountability has occurred in terms of using
research funds in public and private institutions.
* In the USA, the gap between yields on experiment sta-
tions and on farms is usually much smaller than in most
low income countries, because the equipment used and
levels of nonexperimental variables on-station and on-
farm are often more similar, than in low income coun-
tries. That is, environments on farms in the USA are
made to fit experiment station technology. In low in-
come countries, farmers are not able to do this.
This experiment station-based approach is likely to con-
tinue playing an important, if reduced, role in the fur-
ther development of agriculture in, the USA. Another
research approach, which involves farmers on more of
a day-to-day basis (i.e., more farmer-based and interdis-
ciplinary), has emerged in recent years, particularly in low
income countries but also to some extent in the USA.
Why is this the case? This approach has some of its roots
in the farm management emphasis that prevailed in U.S.
land grant institutions in the early to mid 1900s.
Farmer Research Involvement in Low Income Countries
The FSRE approach--also known as on-farm research
with a farming systems perspective, farming systems de-
velopment, farming systems research, farming systems
research and development, etc.-that evolved and quickly
became very popular in the low income countries in the
1970s was helped to some extent by the enthusiastic sup-
port of many donors. For example, by the mid 1980s,
about 250 medium- and long-term projects worldwide
were carrying out FSRE type work. As one of the major
donors, the U.S. Agency for International Development
between 1978 and 1988 funded 76 bilateral, regional, and
centrally funded projects containing either a farming sys-
Abbreviations: FMFI, farmer managed, farmer implemented; FSRE,
farming systems research and extension; RMFI, researcher managed,
farmer implemented; RMRI, researcher managed, researcher implement-
ed; RRA, rapid rural appraisal.
124 J. Prod. Agric., Vol. 7, no. 1, 1994
tems orientation or clearly focusing on FSRE. Forty-five
of these were in Africa (Brown et al., 1988, unpublished
data). Many believe this enthusiastic support was over-
done (Gilbert et al., 1980), because the methodologies for
FSRE were, and in fact are, still evolving. The high degree
of support was based to some extent on the increasing
realization, supported by empirical evidence in the 1950s
and 1960s, that:
* Farmers, no matter what their educational level, have
considerable knowledge and insights about their local
biophysical and socio-economic environment (Norman
et al., 1981); are sensible in the methods they use (e.g.,
the practice of mixed cropping [Norman, 1974]), and
are natural experimenters (Biggs and Clay, 1981). Ob-
viously, the experimental methods farmers naturally
use are those that appeal to them and are informal in
nature (Lightfoot et al., 1991) in the sense that they
are not usually amenable to formal statistical analy-
sis. An increasing respect for farmers' innovativeness
and knowledge gave rise to the conviction that they
could be productively harnessed in designing and test-
ing relevant improved technologies. Reallocation of
limited research resources could be done in such a way
as to improve the productivity and the rate of return
from their use. A necessary condition was a mechan-
ism for capturing farmer knowledge through diagnos-
tic surveys carried out by interdisciplinary teams.
* On-station trials are often implemented under very
atypical environmental conditions, compared with
those of farmers, not only in terms of levels of nonex-
ment, ability to optimize timing of operation, etc.
Consequently, the yield gap between on-station and
farm level production, created by the very different en-
vironments, is often very large. Thus, the probability
of achieving success by disseminating technologies
based purely on experiment station evaluation is severe-
* Farmers have tended not to have much influence in de-
termining research priorities. This is particularly un-
fortunate given that they often have farming
systems--consisting of crops, livestock, and off-farm
enterprises (Norman et al., 1982)--which are often
more complex than the specialized farming systems in
the high income countries (Norman and Collinson,
1986). Thus, in such situations without some farmer
involvement, the conventional station-based approach
is unlikely to result in relevant improved technologies
for limited resource farmers.
The fundamental principle underlying the development
of FSRE was that giving farmers a voice would allow
them to help in identifying the appropriate path to
agricultural development. The FSRE approach, consist-
ing of descriptive and diagnostic, design, testing, and dis-
semination activities, evolved as a method to involve
farmers and to provide more of a systems perspective.
This approach helped to complement and improve the
payoff from the vitally important station-based research
work (Tripp, 1991; Shaner et al., 1982).
In the early 1990~s, FSRIE is still alive and weii.
Although there has been a measured withdrawal on the
part of some donors in supporting "'farming systems"
projects, the approach has, or is, being institutionalized
in a large number of countries with internal, as well as
external, funding. Also, a global professional association
for FSRE exists, and regional associations and networks
are operating or planned within Asia, Africa, and Latin
America, as well as in Europe and North America. In
addition, many of the techniques developed in FSRE, and
what is called the farmer participatory approach, are
being used to address other problems important to
donors. For example, the U.S. Agency for International
Development recently funded a Collaborative Research
Support Program on sustainability, which places con-
siderable emphasis on farmer participation (National
Research Council, 1991). Also, the soil conservation
group in FAO is advocating the use of many of the prin-
ciples of FSRE, involving farmer participation in the de-
sign of strategies to prevent and cure soil erosion
Farmer Research Involvement
in High Income Countries
At the same time, and to some extent predating the evo-
lution of FSRE in low income countries, an analogous
approach developed in the USA, which, as indicated
above, also has some resemblance to the farm manage-
ment techniques of the early and mid 1900s. Once again,
this evolved largely as a result of perceived problems of
the conventional approach that has been so successful in
addressing the production needs of the majority of farm-
ers. Some would challenge this statement by arguing that
farm policies, types of research, and emphasis on fossil
energy-based technologies have contributed to economies
of scale resulting in larger farms at the expense of smaller
farms--see Anderson (1992) for a list of references relat-
ing to these assertions. We do not intend to evaluate this
issue, apart from pointing out that this is probably in-
evitable, given the way the economic system operates
inside and outside agriculture. In this system, as it cur-
rently operates, the future is heavily discounted, result-
ing in present high levels of production and living for
larger farmers, probably at the cost of sustainability in
the long run. The alternative approach arose for a num-
ber of reasons including:
The notion that the~ needs of limited resource (i.e.,
small) farmers were often not considered adequately
in the new conventional approach of the higher income
Perceived problems in implementing the conventional
station-based approach including:
-- The feeling that the conventional approach was in-
trmnsically more suited to developing relevant tech-
nologies for highly specialized farms (i.e., the norm
in the USA) rather than the more diversified sys-
tems that tend to be more common on small farms.
The conventional approach, until recently, placed
J. Prod. Agric., Vol. 7, no. 1, 1994 125
Technical and social
Researcher or farmer
Researcher or farmer
Likely to be lower
Likely to be higher
Researcher or fanner
Table 1. Some differences between researcher-based and farmer-basedl research
Location of trial
Priority setting for trial: Researcher
Experimental design: Complexity
Degree of experimentation control
Evaluation of trial results--factors taken into account:
Expense of experimental program:
Fixed (overhead) costs
Variable (recurrent) costs
? Source; Modified from Worman et al. (1992).
Likely to be higher
Likely to be lower
considerable emphasis on production with less at-
tention to sustainability. In highly specialized
farms, sustainability issues are substantially differ-
ent from those on more diversified farms. This is
because, in the former, heavy reliance has to be
placed on external inputs (i.e., often related directly
or indirectly to the use of fossil energy), whereas,
in the latter, much greater emphasis can be placed
on internal inputs (i.e., constructively exploiting bi-
ological interaction through the diverse enterprises
present on the farm).
The farm family-based, balanced, farming program,
initiated at the University of Missouri in the 1940s (Burch,
1949), was an earlier manifestation of this farmer-based
approach and was, in essence, a precursor of what are
now popularly known as the farmer-focused organic
farming and sustainability movements. They often have
a strong advocacy orientation. Currently, a large num-
ber of nongovernmental and nonprofit organizations
(e.g., Rodale Institute and the Kansas Rural Center) are
very supportive of these movements. The advocacy func-
tion of such organizations has often given rise to the per-
ception on the part of outsiders--sometimes justifiably
and sometimes not--that they lack a degree of objectivi-
ty in their research approach (Anderson and Lockeretz,
1991). Nevertheless, in recent years, considerable interest
in the sustainability theme also has developed in land
grant institutions. Probably all would claim to have an
explicit interest in this theme, but those universities we
know that have centers or prominent individuals work-
ing on this include, but are not necessarily confined to,
Maine, Michigan State, North Carolina State, Oregon
State, California at Davis, Wisconsin, North Dakota,
Minnesota, Colorado State, Iowa State, Nebraska, Mis-
souri, Illinois, Ohio State, and Utah State. Increasingly,
various foundations (e.g., Kellogg, Jessie Smith Noyes,
etc.), prestigious organizations such as the National
Research Council (National Research Council, 1989), and
USDA under its Sustainable Agriculture Program (Sus-
tainable Agriculture Operations Committee, 1991) are
advocating addressing sustainability issues through farm-
er participation and on-farm research. As a result, bridges
are slowly being built between all the institutions involved
SHOULD ON-FARM RESEARCH BE
ENCOURAGED IN THE USA?
A good starting point in evaluating the future of on-
farm research in the USA is to review briefly when it is
justifiable. Obviously, it is important to understand clear-
ly the rules, strengths, and weakness of on-station and
on-farm research (Table 1). Consideration of these points
is important in deciding whether and when to do on-farm
trials, usually implemented by farmers. Such trials are
not always superior--it depends on the objective of the
trial, the conditions under which the trial must be un-
dertaken, and the criteria to be used in evaluating the
results. It is important to note that, however, although
Table 1 implies that researcher-based and farmer-based
research are two distinct entities, this does not have to
be the case. In fact, they should be integrated--be part
of a continuum--something that has always been empha-
sized in FSRE. In the USA, apart from the few bridges
that are now developing, institutions have been at one
end or the other of this continuum. Lockeretz (1987) is
correct in arguing that on-farm research should be a com-
ponent of any well balanced research program.
The continuum aspect of researcher-based to farmer-
based (i.e., usually synonymous with station-based and
farm-based, respectively) trials is illustrated by Table 2,
which is a simplistic way of differentiating trials accord-
ing to whether the farmer (F) or researcher (R) manages
(M) or implements (I) them. Thus, experiment station
trials are usually both managed and implemented by
researchers (RMRI). The RMRI trials also can be im-
plemented on-farm, where they are usually called multi-
locational trials. In a sense, such trials are farm-based
rather than farmer-based, although, in low income coun-
tries, farmer involvement is' increasingly being used in
RMRI trials by eliciting farmer opinions on proposed
treatments and their results (Siebert, 1993, personal com-
126 J. Prod. Agric., Vol. 7, no. 1, 1994
Table 2. Expectations of different types of trials (Source: Norman, 1989).t
Researcher managed and Researcher mannged and farmer im- Farmer managed and farm-
Item researcher implemented plemented er implemented
(RMRI) (RMFI) (FMFI)
Stage: DesignS 1st stage testing 2nd stage testing
Complexity Most Less Least
Type Standard Simple standard With and without
Replication Within and between sites Usually only between sites but can also Between sites only
Numbers/levels of treatment Most Less Least
Standardized level of non- Most Less Least
Plot size Smallest Larger Usually largest
Who selects technology? Researcher Researcher/farmer Farmer
Who shoulders risk? Mainly researcher Researcher/farmer Mainly farmer
Main discipline of researcher Mainly technical Technicallsocial Mainly social
Farmer Least More Most
Researcher Most Less Least
Number of farmers None Some Most
Farmer groups Least More Most
"Yield" Most Less Least
Measurement errors Least More Most
Degree of precision Highest Less Least
"Hard" (objective) Most Less Least
"Soft" (subjective) Least More Most
Degree of accuracy Least Less Most
Determination of causeleffect relationships Easiest Less easy Least likely
Incorporation into farming system Least More Most
Who by? Mainly researcher Researcher/farmer Mainly farmer
Nature of test Assesses technical feasi- Technical feasibility plus economic Validity for farmers--
bility evaluation practicality, acceptability
Researchers Most Less Least
Extension staff Usually least More Most
Farmers Least More Most
East of acceptance of results of trial Researcher Researcher/f armerlex tension Farmer
? There is a degree of subjectivity in some of the entries in the table, but they generally reflect what is the case. In a sense, these exceptions also reflect
the reasons why the different types of trials are undertaken. In the literature, RMRI trials are sometimes called exploratory trials, whereas RMFI trials
are sometimes termed refinement trials.
$ Standard multi-locational trials are also RMRI. Multi-locational trials are differentiated from trials undertaken in a farming systems context in that they
are not designed, implemented, or evaluated with a farming systems perspective in mind.
SIn a sense there are two types of FMFI trials, those done through research-oriented farmer groups (equivalent to validation trials in the literature) and
the more widespread testing done through extension-oriented farmer groups (equivalent to verification trials in the literature).
1 This means how accurately the resutis reflect farm reality.
portant that these scarce resources be used in the vari-
ous institutions to complement each other and to
maximize return to them (i.e., the results should be use-
ful to as many parties as possible--other researchers,
extension/development staff, farmers, policymakers,
etc.). Incorporating farmers as much as possible is im-
portant especially if, as has been claimed, farmers con-
duct 20 times more research than universities and
industry combined (Janke et al., 1990).
*There is unexploited complementarity between on-farm
and on-station research. The former tends to be very
systems-oriented, whereas the latter makes liberal use
of the reductionist approach. The advocates of the
FSRE approach have always argued that experiment
station-based research is vitally important in provid-
ing possible solutions to problems identified in the
descriptive/diagnostic stage of FSRE methodology. Ex-
periment station-based research is often a resource ef-
munication; Sperling et al., 1992). Farmer-based trials
constitute those implemented and managed by farmers
(FMFI) or by the researcher (RMFI). From our work in
low income countries, we believe that all three types of
trials are important in a well rounded research program.
Yet surprisingly, a survey of 41 FSRE projects undertaken
a few years ago in low income countries indicated that
only 32% undertook all three types of trials, 12% un-
dertook only RMRI trials, and only 46%/ undertook
FMFI trials (Barker and Lightfoot, 1986, unpublished
In addition to the desirability of having researcher- and
farmer-based components in a research program, a num-
ber of other reasons also exist for greater integration of
U.S. farmers in the technology development process.
Some of the reasons are as follows:
*Research resources are increasingly limited. It is im-
J. Prod. Agric., Vol. 7, no. 1, 1994 127
Thlcereore, there is a high degree of complementarity
between the conventionally based on-station trials and
those involving farmers located on-farm. This com-
plementarity could, and should, be exploited to a great-
er extent than it is at the moment.
ON-FARM RESEARCH IN THE USA
Although the logic of encouraging greater on-farm
research appears irrefutable, a number of practical issues
affect its general acceptance to all those involved in
research. Interestingly, many of them also are raised with
reference to FSRE in low income countries. These prac-
tical issues must be resolved, if the desirable interactive
continuum between on-station and on-farm research is
to operate effectively. Four frequently mentioned in the
literature are discussed in the following sub-sections.
Incorporation of the Descriptive and Diagnostic
Stage in On-Farm Research
Although social scientists based at agricultural experi-
ment stations carry out a number of different surveys,
very few of these are multidisciplinary in nature. The
FSRE researchers working in low income countries have
developed techniques for diagnosing farm problems and
solutions that allow interdisciplinary interaction to oc-
cur. One such technique is the sondeo (Hildebrand, 1981)
or rapid rural appraisal (RRA). The RRAs employ a set
of data collection techniques that provide a comprehen-
sive sociocultural, economic, and ecological assessment
of a given area for research planning and implementa-
tion. The RRAs are used to collect data on farmer values,
opinions, and objectives, as well as on biophysical and
The RRAs were developed because of the shortcom-
ings of other, more conventional, formal survey methods
(Molnar, 1989), which included:
* The time lag to produce results.
* The high cost of administering surveys.
* The irrelevance of many questions for specific im-
* The failure of such surveys to generate interdiscipli-
nary dialogue among researchers, planners, exten-
sionists, and farmers.
The major objective of RRAs is to gain maximum
knowledge of the target area with minimum amounts of
time and resources (Eklund, 1990). The major distin-
guishing features of such approaches include the follow-
ing (Franzel, 1984, unpublished data):
* Interviews are conducted by researchers themselves, not
by enumerators or by mail as in formal surveys.
* Interviews are essentially unstructured and semi-
directed with emphasis on dialogue and probing for in-
formation. Questionnaires are never used; however,
some researchers use topical guidelines to ensure that
they cover all relevant topics on a given subject.
* Informal random and purposeful sampling procedures
ficient way of ascertaining cause-effect relationships
and developing relevant improved technologies. Farm-
er-based research not only ensures that farmers have
a direct means of articulating their needs but also en-
abiles them to play important roles in designing and test-
ing solutions. When farmers are involved, a systems
perspective is likely to receive more attention. In an
era of increasingly limited research resources, there is
need to exploit any complementarities that exist be-
tween different research approaches.
*A number of other factors justify closer integration of
research-based and farmer-based research that result
from their potential complementarity and from the
limited research resources. They include the following:
-- As sustainability issues become more important,
relevant developments probably will require an ap-
proach that is more holistic, more interdisciplinary,
and more location-specific, involving more on-farm
research (Tripp et al., 1990). This was something
already noted earlier with reference to institutions
supporting sustainability work. Under such circum-
stances, links between on-station and on-farm
research will become vitally important.
On-farm trials enable technologies to be tested
simultaneously in a large number of different tech-
nical environments, rather than on only one or a
few experiment stations. This provides an oppor-
tunity for speeding up the testing period and lower-
ing the cost of research through large-scale testing
in a short time, rather than extended testing over
a longer time in a few locations. Superior results
over a large number of locations--which may some-
times act as proxies for different years--would in-
dicate a technology that is technically superior,
Much current multi-locational research is RMRI.
This could often yield more valuable results--that
is, in terms of economic evaluation and potential
adoption--if elements of the socio-economic en-
vironment are included by involving farmers in the
implementation or management (i.e., RMFI or
FMFI) of trials.
Limited research resources preclude land grant ex-
periment stations from investigating all possible en-
terprises and issues. Harnessing the potential power
of on-farm trials (i.e., farmer based) provides a
potential low-cost way of conducting research on
some enterprises and issues that otherwise might go
unexplored. This approach could be particularly
useful, for example, in research on low input
production strategies, fertility management tech-
niques, integrated pest management strategies,
minor crop and livestock enterprises, etc. Informa-
tion and ideas from many farmers may provide a
useful first step in further developing relevant tech-
nologies or, in certain circumstances, in transfer-
ring the same technologies to other farmers. Interest
in enterprise diversity is growing not only because
of the need to address sustainability issues but also
because of current legislation that limts farm areas
devoted to certain major crops.
128 J. Prod. Agric., Vol. 7, no. 1, 1994
are used instead of formal random sampling from a
* The data collection process is dynamic, interactive, and
sequential; that is, researchers evaluate the data col-
lected and reformulate data needs on a daily basis.
* Such surveys are generally conducted and analyzed and
reports completed over a relatively brief period-1i wk
to 2 mo.
* To deal with the accuracy/timeliness trade-off, a pro-
cess of triangulation (i.e., cross-checking through ask-
ing the question in more than one way) is used, whereby
diverse methods and information sources are used to
* Such approaches rely on interdisciplinary teams.
In summary, the major advantages of RRAs are that
they (Molnar, 1989):
* Are rapid--that is, results can be made available to
* Are interdisciplinary. .
* Are eclectic in techniques aimed at capturing a hohis-
tic picture of the local situation.
* Rely more on open-ended interview techniques that
reduce nonsampling error.
Allow for valuable interaction between researchers and
the target population.
Through the use of such interdisciplinary diagnostic
tools, U.S. researchers can gain a holistic, systems per-
spective of the farming situation in a given area. When
their opinions are elicited, farmers can articulate their
needs and problems more effectively. Such surveys also
enable researchers to determine what types of technolo-
gy or practices can be incorporated easily into existing
farming systems. This information will help researchers
to determine what are the most appropriate on-farm ex-
periments to carry out.
Deciding on the Client
It has been suggested that, in thinking about research,
farmers tend to move from the general to the specific (i.e.,
what will work on their farm), whereas professional
researchers tend to move from the specific to the general
(i.e., what generalized results can be achieved from specif-
ic trials) (Anderson and Lockeretz, 1991). Many have
identified areas for which there are particular advantages
to doing research on-farm. (Lockeretz, 1987). Several
clients interested in the results of on-farm research can
be identified. Perhaps most important, but not always
considered, are the farmers themselves. Extension per-
sonnel and researchers from private companies and public
institutions are others who can be considered clients. Each
of these stakeholders may have different information
needs and data requirements. As a result, each may have
different preferences regarding design. Obviously, the
higher the number of clients who benefit from a trial,
the more cost-efficient it is.
In terms of the types of trials delineated in Table 2,
RMRI trials, which are particularly useful for examin-
ing cause-effect relationships, have long been the preserve
of station-based researchers. It is in RMFI and FMFIvi
trials, which tend to be concentrated on-farm and require
farmer involvement, that the potential for misunderstand-
ing between the professional researcher and other clients,
particularly the farmer, is greatest. There will always be
a place for on-station research, but it is also apparent that
the recent upsurge in on-farm research has been due to
a desire for greater farmer participation in order to give
them a "'voice"' in the research process (Anderson, 1992).
As Krcil (1993, personal communication) has stressed,
there is no reason why trials cannot have multiple clients,
as long as the potential strengths and weaknesses of the
different types of trials are recognized. For example, this
is certainly likely to be the case in RMFI trials. On the
other hand, in FMFI trials, the major client is likely to
be the farmer and, therefore, the standard statistical de-
signs that professional researchers are comfortable with
may, and perhaps should, have less applicability. Thus,
it is important to bear in mind that trials can be differen-
tiated on the basis of research objectives, experimental
design, types of data collected, methods of analysis,
evaluation criteria, and client appeal (i.e., researchers,
extension/development staff, or farmers) (Table 2). Too
often, there seems to be too little consideration of such
issues in designing on-farm research and, consequently
there are unrealistic expectations about what specific trials
can achieve. It is important, therefore, to determine which
client or clients are to be taken into consideration before
designing any specific trial, or else the results may not
be of benefit to any of them.
Resolving the Cost Issue
A common issue that is raised is the concern that on-
farm research is expensive relative to on-station research.
The ratio of variable to fixed costs is likely to be higher
in the case of the former, but the conclusion with refer-
ence to total costs is much more debatable. Unfortunate-
ly, in low income countries, fixed costs are rarely taken
into consideration when such comparisons are made.
Although we have been unable to find more detail in U.S.
studies, Franzluebbers et al. (1988) concluded from their
analysis that, on average, experiment station trials are 3.8
times more expensive than researcher-implemented trials
on-farm. When researcher-implemented trials were com-
pared with replicated farmer-implemented trials, they
were on average 3.7 times more expensive. The average
cost of the replicated farmer-implemented trials was $800.
The Practical Farmers of lowa pay cooperators up to
$350 for one or two trials per year (Exner and Rosmann,
no date). Such comparisons should be interpreted cau-
tiously, however, because there are major differences in
the types of data collected, numbers of treatments, and
levels of control of nontreatment management variables.
Also, expenses of on-farm research may often be underes-
timated in the sense of professional researcher time re-
quired to coordinate the research, farmers providing labor
and equipment that are not compensated, etc. In low in-
come countries, the former problem is dealt with through
clustering work in specific geographical areas or en-
J. Prod. Agric., Vol. 7, no. 1, 1994 129
couraging networking through farmer groups (Heinrich
and Masikara, 1991, unpublished data). This is not a
problem however, because contributions of labor, equip-
ment, etc. not remunerated are an expression of the in-
terest and commitment of farmers in the trials.
Choice of location of, and farmer involvement in, a
trial should depend on its objectives and the type and level
of data required to fulfill the objectives. The point is that
on-farm trials are not by definition expensive to imple-
ment. Therefore, factors other than cost should be major
determinants of the type of trial to be undertaken. All
other things being equal, greater farmer involvement is
obviously desirable in the sense that it can stimulate
professional researchers' thinking in proposing research
suggestions and in evaluating the research results. There
is another side to the cost issue, namely, that of efficien-
cy, which relates partly to the results or end product.
Farmer involvement can pay off by stimulating their own
thinking or aiding in dissemination of, and giving
credibility to, research results through interaction with
other farmers (Anderson, 1992). It also usually enhances
the probability of achieving positive (i.e., from the farm-
ers' perspectives) results.
Design and Analysis Issue
This is perhaps the most critical issue and prevents
much constructive interaction between researchers used
to the on-station RMRI format and those involved in on-
farm FMFI research. Evidence convincing to one client
may prove to be very unconvincing to another. For ex-
ample, statistical significance of one specific indicator
(e.g., yield) is usually very important to professional
researchers, whereas farmers use multiple evaluation
criteria, which in the USA probably include, but are not
necessarily confined to, profitability and dependability.
Not surprisingly, the optimal trial for each of these clients
is likely to differ. In fact, farmers often use their own
experimental procedures in deciding whether to adopt a
new technology. These informal trials are unlikely to be
amenable to a form of statistical analysis. Therefore,
most problems in coming to an accommodation between
the different needs of the professional researchers and
farmers are likely to occur at the FMFI level. In a curso-
ry look at the literature, two major points have been very
obvious, particularly with respect to FMFI trials.
Size and Shape of Plots. There has been major con-
cern about the practical implementation of such trials,
with the general conclusion being that large plots in long
strips--to permit easy mechanization--give results as
satisfactory as the smaller, more square plots. This is
based on estimation of coefficients of variation (CVs) that
compare favorably with those obtained in experiment
station-based trials (Rzewnicki et al., 1988; Shapiro et al.,
1989 and 1990). This has been challenged occasionally
as an erroneous conclusion in the sense that the larger
plots can mask variation within them (Stucker and Hicks,
1992, unpublished data). Whatever the merits of this ar-
gument, it is not central to the purpose of the trial, that
is, giving an assessment of the treatments under actual
practical farming conditions, where farmers always grow.
crops in large plots. For the farmer, the average perfor-
mance of the treatment over the micro-environmental var-
iations in the field is likely to be more important than
how it performs under every micro-environment (Freyen-
berger, 1993, personal communication). The latter kind
of information is better gleaned in RMRI type trials.
Whether or not to replicate. Some literature empha-
sizes the need to replicate FMFI trials on farmers' fields
(Janke et al., 1990; Exner and Rosmann, no date;
Rzewnicki, 1990). This may reflect simply the training
of professional researchers or their desire to obtain in-
formation on technological responses on a limited num-
ber of individual farms. It could also be argued from a
scientific viewpoint that replication is important for in-
dividual farmers who want data that pertain specifically
to their own farms. When responses within specific en-
vironments are desired, then replications within each are
necessary. It is apparent, however, that many farmers
make decisions to change on the basis of a number of
criteria, which raises some doubt about the value of some
form of statistical analysis usually focused on one or two
indicators (e.g., yield). For example, Rzewnicki (1990)
found in a survey of lowa and Nebraska farmers, that
two-thirds of those surveyed were not interested in using
replications and randomization for their own on-farm
tests. There was a general belief, however, that experi-
ment station results should be confirmed on farmers'
fields. So it may well be that replicated designs provide
information that is more useful to the professional
researcher than the farmer..
.However, there is another type of on-farm trial that
is designed to help researchers, with farmer input, to
make recommendations for a number of environments
and, hence, a number of farms beyond those included
in the specific trial. In this case, whether using Modified
Stability Analysis (Hildebrand, 1984) or ANOVA, a large
number and wide range of environments are desirable,
but replications within each environment are not neces-
sary (Stroup et al., 1991). These trials may be considered
more RMFI than FMFI in nature, because inferences are
made beyond the boundaries of individual farms. These
nonreplicated trials also are effective for extension
demonstration purposes--hence, they are also very cost-
It is encouraging to see an increasing convergence be-
tween the U.S. on-farm research approach and the FSRE
approach, with the former adapting some of the metho-
dologies developed under FSRE, and the latter becom-
ing increasingly concerned about incorporating
sustainability issues. We are convinced that the input of
the farmer (i.e., consumer of the results of agricultural
research) is here to stay, not only in low, but also in high
income countries. The challenge is to continue aggressive-
ly seeking ways that the complementary relationships be-
tween on-station and on-farm research involving farmers
truly can be exploited.
139 J. Prod. Agric., Vol. 7, no. 1, 1994
Norman, D.W. 1992. Soil conservation: Using farming systems develop-
ment as an aid. FAO, Rome (draft).
Norman, D.W., and M. Collinson. 1986. Farming systems research in
theory and practice. p. 16-30. In J. Remenyi (ed.) Agricultural sys-
tems research for developing countries. Australian Council for Int.
NorA D., M. ewman, and 1. Ouedraogo. 1981. Farm and village
terr utco rnesyew o re ach ICIA A es Bul 1o.WT4vo 2.APat -
cheru, A.P., India.
Norman, D., E. Simmons, and H. Hays. 1982. Farming systems in the
Nigerian Savanna: Research and strategies for development. West-
view Press, Boulder, CO.
Rzewnicki, P.E. 1990. Farmer's perceptions of technology transfer and
on-farm research in agronomy. p. 29. In Agronomy abstracts. ASA,
Rzewnicki, P.E., R. Thompson, G.W. Lesoing, R.W. Elmore, C.A.
Francis, A.M. Parkhurst, and R.S. Moomaw. 1988. On-farm experi-
ment designs and implications for locating research sites. Am. J. Al-
tern. Agric. 3:168-173.
Shaner, W.W., P.F. Philipp, and W.R. Schmehl. 1982. Farming sys-
tems research and development: Guidelines for developing countries.
Westview Press, Boulder, CO.
Shapiro, C.A., W.L. Krantz, and A.M. Parkhurst. 1989. Comparison
of harvest techniques for corn field demonstration. Am. J. Altern.
Shapiro, C.A., A.M. Parkhurst, and W.L. Krantz. 1990. Update on
the statistical implications of experimental units for on-farm trials.
p. 30. In Agronomy abstracts. ASA, Madison, WI.
Sperling, L., L. Loevinsohn, and B. Ntambovura. 1992. Rethinking
the farmer's role in plant breeding: Local bean experts and on-station
on selection in Rwanda. Exp. Agric. (in review)
Stroup, W., P.E. Hildebrand, and C.A. Francis. 1991. Farmer partic-
ipation for more effective research in sustainable agriculture. Univ.
of Florida Food and Resour. Econ. Dep. Staff Paper SP 91-32.
Sustainable Agriculture Operations Committee. 1991. Guidelines for
grants programs on sustainable agriculture. USDA-CSRS, Washing-
Tripp, R. (ed.). 1991. Planned change in farming systems: Progress in
on-farm research. John Wiley, Chichester.
Tripp, R., P. Anandajayayasekeram, D. Byerlee, and L. Harrington.
1990. Farming systems research revisited. p. 384-399. In C. Eicher
and J. Staatz (ed.) Agricultural development in the Third World.
Johns Hopkins Univ. Press, Baltimore.
Worman, F., D. Norman, and J. Ware-Snyder (ed.) 1992. Farming sys-
tems research handbook for Botswana ATIP RP 3 (Rev.) DAR,
MOA, Gaborone, Botswana.
J. Prod. Agric., Vol. 7, no. 1, 1994 131
Anderson, MD. o9 2. 8Reas5 s fr new interest in on-farm research.
Anderson, M.D., and W. Lockeretz. 1991. On-farm research techniques;
Reoerp te aa wogrkhp St. Sre, r~MeNt, .6 Nov. 1990. Institute
Bigste logy ad rl'. lay. 1981. S Irce sf innovation in agriculture~
Burch, J.W. 1949. Missouri plan (balanced farming). J. Farm Econ-
Economic Development Institute. 1990. Rapid rural assessments for
sub-saharan Africa: Two case studies. World Bank, Washington, DC.
Exner, D.N., and R.L. Rosmann. No date. Collaborative on-farm
research and demonstration: An example from lowa. Iowa State
Franzleubbers, A., C. Francis, P. Rzewnicki, R. Thompson, G. Leso-
ing, and R. Elmore. 1988. Relative costs and efficiencies of on-farm
vs. on-station research. p. 126. In Agronomy abstracts. ASA, Madi-
Gilbert, E., D. Norman, and F. Winch. 1980. Farming systems research:
A critical appraisal. Michigan State Univ. Rural Devel. Paper no. 6.
Hildebrand, P.E. 1981. Combining disciplines in rapid rural appraisal:
The Sondeo appraoch. Agric. Admin. 8(6):423-432.
Hildebrand, P.E. 1984. Modified stability analysis of farmer managed,
on-farm trials. Agron. J. 76;271-274. .
Janke, R., D. Thompson, K. McNamara, C. Cramer. 1990. A farm-
er's guide to on-farm research. Rodale Institute, Kutztown, PA.
Lightfoot, C., N. Axinn, R. Chambers, R. Singh, D. Garrity, V. Singh,
P. Mishra, and A. Salman. 1991. Training resource book for par-
ticipatory experimental design. Int. Cent. for Living Aquatic Resour.
Lockeretz, W. 1987. Establishing the proper role for on-farm research.
Adv. Agron. 43:155-198.
Molnar, A. 1989. Community forestry rapid appraisal. FAO, Rome.
National Research Council. 1989. Alternative agriculture. National
Academy Press, Washington,
National Research Council. 1991. Toward sustainability: A plan for
collaborative research on agriculture and natural resource manage-
ment. National Academy Press, Washington.
Norman, D.W. 1974. The rationalisation of a crop mixture strategy
adopted by farmers under indigenous conditions: The example of
northern Nigeria. J. Devel. Studies 11(1):3-21.
Norman, D.W. 1989. Accountability: A dilemma in farming systems
research. Culture Agric. 38(Spring/Summer):2-8.