CENTRO INTERNATIONAL DE MEJORAMIENTO DE MAIZ Y TRIGO
INTERNATIONAL MAIZE AND WHEAT IMPROVEMENT CENTER
fT Londres 40, Apdo. Postal 6-641, Mexico 6, D.F. M6xico
DATA COLLECT ON, SITE SELECTION AND FARMER
PARTICIPATION IN ON-FARM EXPERIMENTATION*
Working Paper 82/1
* The views expressed in this paper are not necessarily those of CIMMYT
** CIMMYT Economics Program
N C 0 2
LIST OF AVAILABLE CIMMYT ECONOMICS WORKING PAPERS
81/1 Kwasi Bruce, Derek Byerlee and G. E. Edmeades, "Maize in the
Mampong Sekodumasi Area of Ghana; Results of an Exploratory Sur-
81/2 Derek Byerlee and Donald L. Winkelmann, "Accelerating Wheat
Production in Semi-Arid Developing Regions: Economic and Policy
*81/3 Edith Hesse de Polanco and Peter Walker, "A Users Guide to
FASAP- A Fortran Program for the Analysis of Farm Survey Data".
*81/4 Alan Benjamin, "An Agro-Economic Evaluation of Maize Production
in Three Valleys of the Peruvian Andes".
*81/5 Derek Byerlee, Larry Harrington and Paul Marko, "Farmers' Prac-
tices, Production Problems and Research Opportunities in Barley
Production in the Calpulalpan/Apan Valley, Mexico".
81/6 Larry Harrington, "Methodological Issues Facing Social Scien-
tists in On-Farm/Farming Systems Research".
*82/1 Larry Harrington, et al., "Maize in North Veracruz State, Mexi-
co--Farmer Practice and Research Opportunities".
*82/2 Larry Harrington, "Exercises in the Economic Analysis of Agron-
**82/3 J. C. Martinez, "Desarrollando Tecnologia Apropiada a las Cir-
cunstancias del productor: El Enfoque Restringido de Sistemas de
82/4 Robert Tripp, "Data Collection, Site Selection and Farmer Par-
ticipation in On-Farm Experimentation".
82/5 Robert Tripp, "Including Dietary Concerns in On-Farm Research:
An Example from Imbabura, Ecuador".
82/6 Derek Byerlee and Edith Hesse de Polanco, "The Rate and Sequence
of Adoption of Improved Cereal Technologies: The Case of Rainfed
Barley in the Mexican Altiplano".
* Available in English and Spanish
** Available in Spanish only
In cooperation with researchers in many national agricultural re-
search programs, CTIMYT has sought to develop procedures which help to
focus agricultural research squarely on the needs of farmers. The pro-
cess involves collaboration of biological scientists and economists to
identify the groups of farmers for wham technologies are to be devel-
oped, determining their circumstances and problems, screening this in-
formation for research opportunities, and then implementing the resul-
ting research program on experiment stations and on the fields of re-
CIMMYT's Economics Program has emphasized developing procedures for
the first stage of this process, through to establishing research oppor-
tunities. The evolution of the procedures, now synthesized in a manual
"Planning Technologies Appropriate to Farmers: Concepts and Procedures"
has been strongly influenced by collaborative research with many nation-
al programs and with CIMMYT's wheat and maize training programs. Our
efforts with national programs began in 1974 with Zaire's national maize
program, then moved to work in Tunisia, Pakistan, and Egypt. The pace of
work accelerated notably in 1976 with assignment of regional economists
stimulating similar work in Kenya, Tanzania, Zambia, Ecuador, Peru,
Bolivia, Panama, El Salvador, and India. Cooperation with still other
national programs is now underway. We believe that the resulting pro-
cedures offer cost effective and robust guidelines to national programs.
We are now preparing reports that illustrate the implementation of
these procedures in various national programs. While not all such work
can be reported, we take this opportunity to thank all of those who have
collaborated with us.
This paper describes work undertaken with the Production Research
Program of INIAP, Ecuador's national agricultural research institute. It
reports some of the experiences of researchers -- Ecuadorian and CIMMYT
professionals -- in on-farm experimentation. It focuses on the collec-
tion and organization of data during this phase of on-farm research and
as such contributes to CIMMYT's concern for developing and refining re-
search procedures that are useful to national programs.
Donald L. Winkelmann
Director, Economics Program
2. Selection of Sites and Collaborators
3. Communicating with Farmer Collaborators
4. Recording Data
5. Further Data Collection Possibilities
1.1 On-Farm Research Methods
This paper attempts to contribute to the development of methods
for the conduct of on-farm research. Its focus is the activities associa-
ted with on-farm experimentation. It presents some guidelines for site
selection, communicating with farmers and data collection during experimen-
tation. These guidelines were developed and used in an on-farm research
program in Ecuador.
When considering the organization of on-farm research which leads
to the promotion of improved technologies it is convenient to think of four
1) the planning phase, in which target groups of farmers are identi-
fied and information is collected about their circumstances;
2) the experimentation phase, in which trials designed to develop and
test new technologies are planted with representative farmers;
3) the analysis phase, in which data from the planning and experimen-
tation phases are combined in order to derive recommendations for
4) the assessment phase, in which data obtained on farmers' experien-
ces with the new technologies are used to make judgements on the
promotion of recommendations.
Good guides are now available for planning experiments- and for
1/ Byerlee, D., M. Collinson, et.al."Planning Technologies Appropriate to
Farmers Concepts and Procedures". CIMMYT, Mexico, 1980
the economic analysis of trial results- but little has been written
about the experimentation phase. This is unfortunate, because it is
the phase of on-farm research that accounts for the majority of resear-
chers' time and effort. The following pages offer some suggestions for
research activities during on-farm experimentation.
1.2 On-Farm Experimentation
On-farm experimentation places the researcher in direct contact
with a farming system that is often quite complex. It is the researcher's
responsibility to learn as much as possible about this system and the ways
it affects the crops with which he is working. This requires research me-
thods much different from those utilized in traditional experiment station
Information about the farming system is collected during the plan-
ning phase of on-farm research, but this is only the beginning. During the
experimentation phase the researcher has access to the farmers and their
crops, and the opportunities for collecting information during the course
of the season are considerable. Through observations and conversations the
researcher picks up small pieces of information which are gradually woven
together with data obtained from the trials themselves and from the earlier
surveys. The results of this process are sometimes firm recommendations,
but are often hypotheses that must be tested by further trials and data col-
lection. There are ways of planning experiments, organizing research acti-
vities and recording information that make on-farm experimentation more ef-
Although this paper concerns the experimentation phase, its purpo-
se is not to suggest what kinds of experiments should be planted or what
types of agronomic data should be collected. Instead, the paper has two
2/ Perrin, R.K. et.al. "From Agronomic Data to Farmer Recommendations: An
Economics Training Manual". CIMMYT, Mexico, 1976
1) Discuss ways of organizing the experimentation process through improved
site selection to increase the availability and reliability of the data,
both agronomic and socio-economic.
2) Propose that the collection of information about farmers' circumstances
not end with the initial survey, but rather that it continue into the
experimentation phase. In this regard, the paper discusses:
a) Communicating with farmer collaborators.
b) Collecting a broad range of data during experimentation.
c) Recording and preserving this data.
The experiences that form the basis of the following discussion
are those of the Production Research Program of Ecuador's National Agricul-
tural Research Institute, INIAP- Researchers of the Production Research
Program are stationed in various areas of the country with basic responsi-
bility for collecting data on farmers' circumstances, planning and conduc-
ting on-farm experiments to evaluate technological alternatives for the
most important crop or crop association in that area, and formulating
tentative recommendations for farmers. Procedures described here are those
utilized by one or two researchers with one vehicle who have responsibility
for all program activities in the research area. Data collection methods are
designed taking account of these limitations of personnel and resources.
1/ For a more complete description of this program, see Moscardi, E.
"The Establishment of a National On-Farm Research Entity in Ecuador",
CIMMYT, Mexico, 1982.
- 3 -
2. SELECTION OF SITES AND COLLABORATORS
2.1 Recommendation Domains
The most important concept in helping the researcher select sites
for on-farm trials is that of the recommendation domain. A recommendation
domain is a set of farmers who work land with similar features and who have
access to similar resources. The farmers of one recommendation domain are
thus tentatively eligible for the same recommendations with respect to the
target crop. Information gathered during the planning phase of on-farm re-
search is used to identify target crops and to delineate recommendation do-
Recommendation domains are defined by two sets of circumstances:
natural features such as soil, climate and topography; and agro-economic
features such as access to resources and marketing opportunities. The re-
searcher wants to make sure that the sites selected conform to the defini-
tion of the recommendation domain. Site selection may also be concerned
with certain subsets of the recommendation domain, such as fields at a par-
ticular stage in a rotation pattern.
There are a number of possible sources of data to aid the resear-
cher in site selection. One important tool for finding representative far-
mers is the survey that is carried out before beginning on-farm experimenta-
tion. A question can be included in the survey on the farmer's willingness
to have a trial planted on his farm and a selection of appropriate farmers
who have expressed interest in a trial can be made. As the researcher gains
more experience in the area he will come to know many other farmers, and
from the beginning of his work he will want to dedicate time to familiari-
zing himself with the zone and thinking about where and with whom he might
plant the following cycle. In many cases it may be possible to plant trials
with the same collaborators for a second year, but the necessity of provid-
ing wide coverage of the area, and the dangers inherent in developing "prof-
fesional collaborators", would indicate that two or three years should be
the maximum time with one farmer, except in special cases such as experi-
ments which study crop rotations, certain types of management, or long-
term effects of herbicides or fertilizers. Finally, the local extension
service is a possible source of ideas for experimental sites (See 2.2).
As he looks for farmers with whom to plant trials, the research-
er has firmly in mind the characteristics of the various recommendation do-
mains of his area. It is usually fairly easy to determine if a farmer ful-
fills these basic requirements and might be eligible for a trial. But re-
commendation domians are dynamic, and although the goal is to define them
by as few characteristics as possible, the researcher is usually in the pro-
cess of refining and revising the domains, and this often adds additional
criteria to the selection process. In one area in Ecuador a particular recom-
mendation domain was defined as being those farmers with less than five hecta-
res, whose fields were between 2,500 and 2,900 meters above sea level, and
whose maize was harvested principally for home consumption. An early-matur-
ing maize achieved equally good results within this domain for farmers who
had access to complementary irrigation and for those who did not. Thus ac-
cess to irrigation was not a defining characteristic of the domain. In a
subsequent cycle, however, experiments were conducted with a rotation of
early-maturing maize and peas. In this case, the variety and planting me-
thod for the peas varied, depending on access to irrigation, and the resear-
cher had to select his sites accordingly. In other cases, experiments with-
in one domain may demand fields with specific fertility features or rotation
histories, as the researcher works to see if these might serve to further
disaggregate the current recommendation domain.
The idea of a recommendation domain is that a few key characteris-
tics identify a group of farmers whose total constellation of practices and
resources is relatively homogeneous. But within any such group of farmers
there is always considerable variation, and it is well to take this into ac-
count in selecting sites. In the example above, neither the survey carried
out beforehand nor the first year's trials gave any indication that there
were differences in maize practices or results between plots with and with-
out complementary irrigation. But as it was obvious that irrigation might
be an important factor in future work, and as the information at hand was
not sufficient to be absolutely certain that there were no differences, it
was thought advisable to try to select sites for trials by taking account
of this factor, choosing some sites with access to irrigation and others
without. It will thus be possible to analyze the results of several years'
maize trials on irrigated and non-irrigated land to see if irrigation might
indeed serve as a distinguishing feature in delineating new recommendation
It is also advisable to be aware of possible biases in site se-
lection. The information from the survey, and other sources of data, give
the researcher an idea of the occurrence of many secondary characteristics
among the farming population. Although it is assumed that most of these
will be randomly distributed among the collaborators, it is best to watch
for any significant concentration. For instance, if the survey indicates
that less than 10% of farmers in the recommendation domain prepare their
fields with a tractor, but the majority of the trials are planted in fields
plowed with a tractor, it may be an indication that trial collaborators
have access to a greater than average quantity of resources. Even though
there might be no evidence to indicate any differences in results between
maize planted in fields prepared with a tractor or with an ox plow, it would
be wise to search for more collaborators who use ox plows. Similarly, if a
disproportionate number of trials are found with farmers who sell much of
their maize, belong to a particular ethnic group, or are community leaders,
it may be an indication that trials are weighted towards farmers with partic-
ular resources or preferences. It is impossible to balance all of these
factors in the selection process, but the researcher should keep an eye open
for obvious biases.
2.2 Planning trials with extension agents
As the purpose of on-farm trials is to arrive at recommendations
- 6 -
tl:hl xtelnsion .() eCllt (.I c( use wit-h confidence, maximum coordination with
local extension services is essential. The idea is to develop shared res-
ponsibility for the design and management of the trials between researcher
and extension agent, even though several years of work may be necessary to
develop recommendations that the extension agent can utilize.
Collaboration between on-farm researcher and extension agent is
most effective when their respective institutions share the same agricultu-
ral development goals and strategies. Although extension agents may be able
to provide suggestions for areas in which to work and contacts with local
farmers, the researcher often must acquaint the extension agent with the con-
cept of a target group of farmers. On-farm researchers in Ecuador have
learned to cooperate closely with those extension agents whose work involves
them with representative small farmers, and to avoid extension programs
whose approach brings them in contact with only a handful of so-called "pro-
gressive" farmers, or special "pilot projects" which require a high invest-
ment per participant, reach few farmers, and are not replicable over a wide
There are also sites which are legitimate for extension activities
but are not necessarily appropriate for on-farm research. Ministry of Agri-
culture extension agents often provide advice to villages on the use of com-
munal land to raise crops for sale for the benefit of the community, or use
these plots themselves for demonstrations. If a considerable proportion of
crop land in a given zone is communally managed, then some trials should be
planted on this type of land. In one area, for instance, large cooperatives
which grow wheat form one recommendation domain, while the individual small
farmers that are found in the same area constitute a separate domain with
different resources and management practices. But if communally managed
land makes an insignificant contribution to local farming then trials on
that land may be subjected to unrepresentative or uncertain management.
Apart from these reservations, the extension agent and on-farm re-
- 7 -
searcher should work in partnership to identify areas for trials and to
develop research problems. Although it must always be made clear that
on-farm research trials are primarily experiments, certain types of trials
may be quite appropriate for demonstrations. The extension agent may of
course have a separate set of criteria for planning his own demonstration
Another consideration in choosing sites is a logistic one. The
ideal distribution of sites would have on-farm trials scattered throughout
the recommendation domain. Questions of transportation and time often make
this impossible in practice. If trials are placed too far from one another
it may be impossible to get to all of them frequently enough. One compro-
mise is to choose several areas within a domain to concentrate on during a
given cycle, and cluster trials in these areas. A three-hour one-way trip
to visit one site is probably not indicative of an efficient placement of
trials, but if that same trip can be made to an area that contains three or
four trials, then it is much more worthwhile (and much more likely to be
made on a regular basis). Areas of trial clusters can of course be shifted
from year to year.
It is sometimes worth considering placing several trials in the
same community, especially if a number of different types of trials are
being planted. There is a multiplier effect by putting more than one trial
in a limited area because non-participating farmers are more likely to be
aware of the researcher's work and the researcher has an increased opportu-
nity to become acquainted with these farmers, to learn from their observa-
tions and their problems, and to get a more in-depth look at the farming
system practiced in a particular area.
It is not usually worthwhile, however, to consider planting sev-
eral trials with the same farmer. As a general rule it can be said that
- 8 -
two is the maximum number of trials that should be planted with one farmer,
and that all things being equal, one trial with each of two neighboring far-
mers is better than two trials with the same farmer.
When considering the total number of trials to be planted it must
be remembered that the establishment, management and analysis of an on-farm
trial, including the necessity of spending a good deal of time with the far-
mer, make considerable demands on the researcher's schedule. It is possible
to over-extend oneself and this usually detracts from the quality of the
work. The exact number of trials for a given program depends on the type of
trials planted, the size of the research area, the number of researchers and
the type of transportation available, and the optimum number can only be es-
tablished with experience. In Ecuador, the number of trials planted in one
year in each research area generally varies from 15 to 25, managed by one or
2.4 Contacting the farmer
Once areas are selected for the year's research and recommendation
domains are defined, farmers who meet the requirements must be approached to
see if they are willing to participate in a trial. There is no doubt that
small farmers are themselves experimenters, and that the concept of trying a
different variety or a new technique is quite acceptable to them. Neverthe-
less, in many places farmers have had little experience in working with gov-
ernment agencies and none at all in on-farm research, so that careful expla-
nation of the farmer's duties and expectations for an on-farm trial must be
In the first few years of on-farm research in an area it is not
always easy to reach a wide range of farmers. There are a number of factors
that may affect the researcher's ability to make completely unbiased choi-
ces for collaborators. One common experience is that community leaders
suggest that trials be planted in their own fields. It is often a worth-
while strategy, especially when working in a new area where making contact
- 9 -
with farmers is difficult, to plant the first trials with local leaders,
for it serves to increase the visibility of the experimental work and add
legitimacy to it. The sacrifice made is that such local leaders may be
atypical in certain respects, having perhaps access to more resources than
the average farmer.
This is only one instance of cases where the farmer chooses the
researcher rather than the other way around. Although there is an obvious
advantage to working with farmers who are articulate and aggressive, one
must constantly be aware of possible biases in the selection process; male
researchers ignoring female farmers, local elites monopolizing the resear-
cher's attention, and researchers who do not speak the same language as the
farmers, are only a few examples of possible problems.
The selection of trial sites is an activity which requires a great
deal of thought and effort and is not something to be left until the last
minute. The researcher should have definite goals for the sites that he re-
quires. Some of the more important criteria are listed below, and for each
site where a trial is planted the researcher should be able to make explicit
the rationale for its choice.
1. Make sure that the site conforms to the basic characteristics of the re-
2. Check the distribution of other characteristics which may be used to
redefine recommendation domains.
3. Assure that the trial site fulfills the requirements for the type of ex-
periment to be planted (rotation history, planting associations, etc.)
4. Arrange trial sites so they can be visited and managed during the sea-
- 10 -
son and so they provide opportunities for farmer participation.
5. Strive to extend work to new areas and new collaborators. Two or
three years should be the maximum with one farmer, except for special
6. If some trials may be used as demonstrations, solicit suggestions from
extension agents regarding site selection.
7. Identify and correct biases in farmer selection from previous years.
3. COMMUNICATING WITH FARMER COLLABORATORS
3.1 Preparations for planting
The importance of explaining the nature of the on-farm experiment
to the farmer cannot be overemphasized. If at all possible the exact mea-
surements of the plot should be marked off well before planting, so the far-
mer knows the area where the trial will be located. Some types of trials
require that data on the plot be obtained beforehand --soil analysis, crop-
ping history, the exact crop association that the farmer will plant, etc.
Trial sites are selected taking account of the requirements for
the year's experiments. Plots of a particular size, slope, fertility, etc.
may be needed for certain types of trials. Although it is necessary to take
account of individual farmer's interests and experience in placing the dif-
ferent types of trials, it is usually not advisable to ask the farmer which
type of trial he would like, for often other considerations make it neces-
sary to overrule his choice.
Most trials examine a small number of variables and all other
factors are left at the farmer's level of management. This is sometimes
more difficult than it sounds, however. It is not uncommon to find that
- 11 -
the farmer is planning a different type of crop association or density
than the researcher is expecting, and a decision must be made as to whe-
ther to include these "unrepresentative" practices in the trial, or to
ask that the plot be planted under practices more nearly equal to those
of other trials. This of course argues for the desirability of learning
the farmer's plans, early on, so that decisions can be made and discussed
with the farmer.
In some cases it is best to leave the planting of the trial to
the farmer. This is the case when farmers have very special planting prac-
tices which the researcher would have a difficult time duplicating. In
one very dry area where maize trials were planted, farmers had developed
special procedures to assure germination. They planted only very early in
the morning, placing the seed immediately after plowing the furrows and
then covering it so as not to lose any moisture. In this case it was found
best to have the farmer himself do the planting, although the researcher
was present to supervise. As a second example, it is widely known that far-
mers are better able to broadcast seed than are researchers.
It is best to plant trials with the farmer at the same time as he
is planting the rest of his crop. Even if the trial requires planting at a
different time (experiments on planting dates, for instance), the farmer
should be present. It will often be necessary to make several visits to the
farmer to arrange the planting date.
Finally, once the trial is planted the researcher must be sure
that the farmer understands his responsibilities. If the trial is to be
left completely under farmer management then it must be explained that the
farmer is to treat the experimental plot in exactly the same way he treats
the rest of his field.
3.2 Farmer-Researcher Interactions
It is very easy to plant a trial, make observations on it, har-
vest it and take yield data, all without ever talking to the farmer. This
is a great loss, for the observations and opinions of the farmer are one
of the most valuable types of data provided by on-farm research. It should
be remembered that recommendation domains are defined as groups of farmers,
thus collecting information from fields fulfills only a part of the resear-
Farmers who have volunteered to have a trial planted in their
field are usually eager to talk about it, but in order for the researcher
to take maximum advantage of this opportunity it will be useful to consi-
der the farmer-researcher interaction from the outside. In doing so, he
will probably be able to recognize that the farmer will be looking upon
this activity in a different way than the researcher does. For the resear-
cher, on-farm research is part of government service in agricultural devel-
opment; it also offers him the added incentive of an intellectually chal-
lenging type of experimentation under unique conditions. The farmer repre-
sents one of a series of points on a map that make up his research strategy.
For the farmer, on the other hand, the on-farm trial is the work of a gov-
ernment agency he may not have even heard of. His principal motivation in
planting the trial may be to see if he can get some of the new seed or other
inputs for his farm, or to avail himself of the advice or other services he
perceives the researcher may have to offer.
As the season progresses, as the researcher and the farmer parti-
cipate in the planting together, as they exchange views and engage in casual
conversation during the cycle, as the researcher offers the farmer simple
favors (a ride to town, advice on his crops) and the farmer reciprocates with
hospitality during the researcher's visit, they come to look at each other
differently. But the establishment of this type of partnership takes time.
In the meantime, the researcher must set about trying to get the
farmer to give his frank opinions of what he sees in his field. One of the
principal problems is that the farmer may treat the researcher with exag-
gerated deference. His only experience with government agricultural per-
sonnel may have been in contacts with an extension agent, where something
has been "demonstrated" to him. He has learned that his best strategy in
such cases is not to question what is told to him and to adopt a humble
posture. The researcher would only contribute to this charade by talking
down to the farmer (for Spanish speakers, by addressing him as "tu" rather
than "Usted", for instance). The likelihood of accomplishing any worthwhile
transfer of information in this situation is very low. Part of the on-farm
research strategy is the assumption that, with respect to the trial in the
ground, both the farmer and the researcher have important contributions to
make, and any behavior that will facilitate the honest interchange of their
viewpoints is to be encouraged.
The researcher should try to remain as open as possible in col-
lecting information, and refrain from giving the farmer the impression that
he already knows the answers. The question, "You weed this field twice,
don't you?" will likely be answered in the affirmative, no matter what the
farmer's practices may actually be. An attitude of honest curiosity on the
part of the researcher is more likely to give the farmer the confidence to
fully express his opinions and experiences.
Visits to trials should be organized so that there is a good pro-
bability of finding the farmer. If, for instance, the researcher is accus-
tomed to visiting in the morning, when the farmer is usually away working
in distant fields, he will want to plan visits for late in the afternoon.
The more often the researcher takes the farmer with him to visit the trial
the more likely it is that the farmer will realize that he, as well as the
researcher, has responsibility for observing the trial's development, and
the more likely he will be to carry out his duty.
Farmers appreciate visits from the researcher (this is one of
- 14 -
their greatest complaints against extension agents) and the information
gained from a few minutes of casual conversation may at times be as va-
luable as the results of the trial itself. It is a good idea to make
visits to farmers with specific questions in mind, although the farmer
should be encouraged to talk about whatever concerns him. Farmers' opi-
nions and comments should be noted in the field book (4.1)
The farmer is the researcher's partner in on-farm experimenta-
tion, and there are a number of rules that should be observed in order to
facilitate communication between them:
1. Before planting, make sure the farmer understands the nature of the
trial, and ensure that the specific planting practices to be used
are agreed upon.
2. Ensure that the exact site for the trial is agreed upon and that com-
munications are established with regard to planting date, so that the
farmer will be present.
3. After the trial is planted, arrange visits so that the farmer is en-
countered, and always take the farmer to observe the trial.
4. Address the farmer in the locally accepted polite manner; do not talk
down to him.
5. Develop the habit of including a wide range of topics in conversations
with the farmer as a way of learning as much about the total farming
system as possible.
6. Encourage the farmer to express his opinions rather than simply affirm-
ing what the researcher says.
- 15 -
7. Try to have in mind topics or questions to discuss with the farmer on
4. RECORDING DATA
4.1 The field book
The data collected during visits to trials must all be recorded
in such a way that they can be used in analyzing trial results and forming
hypotheses to be tested in following cycles. The data may be used not only
by the researcher in charge of the trials, but by others as well, even in
There are many ways of recording data from trials. Most involve
the use of some sort of field book for noting biological data. The follow-
ing is a description of a more comprehensive type of field book used in an
on-farm research program working with maize and beans (see Appendix). It
consists of a series of mimeographed pages for each trial, placed in perma-
nent binders. There are nine different types of pages for recording data.
The pages are mimeographed before planting, and their format can change from
year to year. The idea is to decide what information is of importance for a
particular season and then design forms to aid in its collection. Much of
the information is most easily obtained if the researcher develops good rap-
port with the farmer. The pages of the field book should not be administered
to the farmer in the form of a questionnaire; if they are they will assume
the same limitations as any formal survey. The idea is to take advantage of
informal conversations with the farmer and observations in the field to fill
in the field book during the course of the season. It is often a good idea
not to write too much in the farmer's presence, for this may inhibit him
from being completely open.
1) Planting data
This includes the date of planting, a description of the type of
trial and its design, dimensions of plots, including factors under the
farmer's control (such as width of furrows or planting density), the va-
rieties and inputs employed, and soil moisture and other conditions at
the time of planting. It is useful to note the sources of inputs used;
at times experimental varieties for trials come from more than one source
(experiment station supplies, seed multiplication institutions, etc) and
agricultural chemicals may have varying histories as well (insecticide re-
cently purchased or from old stocks, for instance).
2) Field plan
This is a careful map of the trial layout, completely labelled.
It includes enough identifying landmarks so that anyone visiting the trial
can orient himself at once and identify the various treatments. This page
and the one on planting data are duplicated so that extra copies are avail-
able for other researchers who may want to visit the trials, such as the ex-
tension agents who collaborate with the program.
3) Characteristics of the plot
This describes the plot where the trial is planted, and much of
this information is best collected before planting, in order to aid in de-
ciding what sort of trial would best be planted at this site. Relevant da-
ta include the cropping and fertilization history of the plot, land prepara-
tion before planting, type of soil, results of soil analysis, altitude and
4) Management of trial
This page is filled in during the course of the year and records
two types of information. It is first a record of the work of the resear-
cher on the trial, including any replanting, applications of fertilizer,
insecticides and other inputs during the course of the year. Second, and
of equal importance, it is a record of the work the farmer has done on the
trial, including all weeding, irrigation, etc.
In one maize program which studied the effects of various her-
bicide treatments against local weeding practices, an accurate record of
the timing and labor involved in the farmer's weedings was essential. In
another maize area, farmers had the custom of throwing household refuse
and manure on their fields, and some estimate of the quantity and distri-
bution of this extra fertilization was helpful in interpreting results.
5) Observations on the farmer's crop
In the case of the maize research program, a field of the farmer's
own maize, usually close to the trial, is selected for observation, and data
are taken throughout the year. These may include variety, associations,
planting density, management practices, most common insect and weed problems,
These data serve three purposes. First, they are valuable in com-
paring with the data on trial management. The researcher must conscientious-
ly study the plot, and not assume that he will find the same practices as in
the trial. In some cases farmers devote extra care to the trial, and this
must be discovered and discouraged if data are to be obtained under truly re-
presentative conditions. On the other hand, cases have come to light where
the trial has been weeded less than the farmer's own maize, either through
misunderstanding or a feeling that the trial maize really didn't require
that much attention.
A second use of data on farmers' practices is the chance to inves-
tigate in a semi-formal manner some questions that may have escaped the ini-
tial survey. The sample of collaborators is probably a non-random one, but
is designed to be fairly representative. Questions can be addressed to this
sample that may aid in understanding the farming system. In one case, use
of information available in the field book was helpful in understanding dif-
ferences among farmers in planting density, an issue that was difficult to
study in a formal survey.
- 18 -
A third use of this page is to give the researcher an idea of
just how representative his sample of farmers is. Data on farm management
practices can be compared to those of the initial survey to see if there
have been any outstanding biases in sample selection. It should not be as-
sumed, however, that the original survey data is always valid. In one case
the survey indicated that farmers used the weeds in their fields for animal
feed, but closer questioning of the sample of collaborators showed that
only weeds that appeared in the later part of the cycle were so utilized.
This was important because it opened possibilities for herbicide trials.
6) Characteristics of the farmer
The purposes and nature of this page are very similar to those of
the previous one. It asks questions about the farmer, his landowning, mar-
keting practices and other economic activities. Like the page on farmer's
management practices, it gives the researcher a chance to compare his sam-
ple characteristics to those reported for the farmers of the area in the
This page also gives him the chance to study various matters in
greater depth than would be possible in a survey. For this reason, the
questions on this page (and on the previous one) are subject to change con-
siderably from year to year, as new research interests appear. Work in one
maize-growing area, for instance, indicated that there was considerable dif-
ference among farmers in the use they made of maize leaves for animal feed-
ing, and in the time at which they were cut, so.that several questions pur-
suing this matter could be included for the sample of collaborators to try
to understand the reasons for these differences.-
1/ It should be emphasized that neither the page on the farmer's crop nor
the one on farmer characteristics pretend to be anything like a farm
management record-keeping system. The present Ecuadorian program does
not have the personnel to carry out such an exercise, and it is felt
that a relatively few well-directed questions about collaborators and
their practices constitute a manageable alternative.
7) Agronomic data from the trials
There is a wide range of agronomic data that may be collected.=-
The important thing is for biological scientists to decide beforehand
which information is necessary for which trials and to plan the data col-
lection process accordingly. Much of the data require frequent visits to
trials and careful, time-consuming observation, so that it is essential to
identify priority topics for each trial. If a new variety is being tested,
for instance, station scientists may want to know days to flowering, but if
the data are already established, then there may be no need to collect them.
Similarly, if a variety of maize is being tested specifically for its stalk
strength, then percentage lodging of this and other varieties in the trial
should be carefully recorded. But if the research does not concern lodging
resistance, then the researcher may not have to spend his time taking exact
data from all the trials. Because the agronomic observations required may
vary from trial to trial, and from year to year, it has been found that a
relatively "open" form is best for recording this type of information.
8) Harvest data
This form includes spaces for date of harvest, harvest weight, per-
cent moisture, number of plants harvested, size of plots harvested, and other
9) Observation forms
Unlike the other eight pages, this is a form that is filled out on
each visit, so a large number of these forms are included for each trial. The
form is divided into areas for general observations. Besides spaces for the
date of visit and whether or not the farmer was contacted, there are spaces
for observations on insects, diseases, weeds, climatic conditions, etc. If
2/ As mentioned in the introduction to this paper, the purpose here is not
to define what sorts of agronomic data are to be collected, but rather
to present methods that aid in their collection and interpretation.
these observation pages are filed in the field book, at the end of the
season the researcher has a record of the number of visits made to the
trial and can quickly reference the behavior of the trial with respect
to insects or any other particular concern.
Copies of these forms are distributed to extension agents and
other researchers who may visit the trials so that their observations can
be included as well. A number of blank pages are also included in the
field book for each trial so that additional observations and data can be
The field book should be regarded as a tool which is to be used
during the management of the year's trials and as a source of information
in future years for reinterpretation and analysis of results. It is most
effective if it is conscientiously redesigned each year to meet the needs
of the research. It is only of value if it is used--if it is taken along
on each visit to the trials and if the data are recorded in the field.
The amount of data available to the on-farm researcher during the
experimentation phase is considerable, especially if care has been taken
in selecting trial sites and establishing rapport with participating farmers.
This information is useful not only in evaluating the technologies being
tested, but for further understanding the farming system and developing new
hypotheses to be examined. But the information is of little use if it is
not collected according to a fixed plan and recorded so that researchers can
use it. The data collection process should conform to the following guide-
1. All data should be recorded immediately, in some type of a permanent
2. Agronomic data required from each trial should be carefully identified
- 21 -
beforehand and arrangements made for their timely recording.
3. The format of the field book is redesigned each year in order to cor-
respond to increased knowledge of the farming system and changes in
4. There should be space in the field book for noting farmers' observa-
tions and opinions.
5. A brief record of each visit to the trial should be made.
6. Data recorded should not only come from the trial itself but from the
farmer's field as well.
7. Data should be recorded so that other researchers can understand it and
use it, even in future years, in interpreting trial results.
8. Most data can be recorded without administering questions to the far-
mer as in a formal survey; casual conversation with the farmer usually
5. FURTHER DATA COLLECTION POSSIBILITIES
5.1. Other data from the area
Since the researcher has the responsibility of formulating recom-
mendations, he must have at hand good data on crop prices and input markets.
Every couple of months he will want to visit local markets and traders to ob-
tain current prices for target crops, transportation costs, discounts for qua-
lity, etc. At the same time he will want to ascertain prices and availability
of the various agricultural chemicals and other inputs appropriate for his
crops. Such data should always be placed in a permanent record book, so sea-
sonal and long-term patterns can be analyzed. Conversations in both the
towns and countryside will also provide information about the characteristics
of farm labor markets and sources and rates of credit for farmers.
Of equal importance to the quantitative economic data, and also
deserving a place in the records, are the many qualitative observations
that the researcher makes throughout the year. In formulating insecticide
recommendations in one maize research program, for instance, observations
on farmers' use of insect-damaged maize as animal feed were every bit as
important as yield data and insecticide prices. The researcher should take
every opportunity to talk with not only trial collaborators but also with
their neighbors, to take a peek over the fence to see what other people's
fields look like. Each season the researcher may have in mind a series of
questions that he will want to ask farmers with whom he comes into contact.
The questions may be asked in the course of informal conversations, but a
simple record-keeping page may be designed to record the answers. Another
important activity is to build a glossary of the local farming vocabulary
which may be of a quite regional and specific nature.
Weather conditions should be noted during the growing season.
Some programs put rain gauges with a few of the collaborating farmers and
provide mimeographed pages with facsimiles of the gauge face on them, so
the farmer only has to draw a line at the level of rainfall and mark the
date. In some areas the national meteorological service has stations, and
data can be obtained from them.
The on-farm experimentation is often not the first research that
has been carried out in the area. The on-farm researcher should acquaint
himself with the results of any past or present work done with his target
crops. Equally important to obtaining this type of secondary data is the
necessity of sharing trial data and conclusions with all interested parties
working in the area.
At times in the course of the work a formal survey or other spe-
cial study may be called for. In one program looking at maize, the initial
survey provided sufficient information about maize practices to begin work.
By the fourth year of research recommendations regarding new early-maturing
varieties of maize were being produced, and work became more oriented to-
- 23 -
wards possible rotations and associations with this type of maize. The
first survey had not provided much detail on other crops, so a small sur-
vey was designed and carried out in a series of sites known to be repre-
sentative of the research area, and this served to orient work for the fol-
This illustrates the balance that exists in on-farm research
between surveys and trials. 'Formal data collection should not run too far
ahead of experimental capabilities. Initial surveys should provide enough
data to begin trials on those technologies for which there exists research
capacity. As work progresses, questions and hypotheses are formed. Some
of these can be tested in trials, while others can be explored through in-
formal questioning of collaborators and other farmers in the area. After
several years of work, however, there may be a list of questions which are
best answered through a carefully designed formal survey.
5.2 Observation trials
As an intermediate step between formal on-farm trials and simply
following farmers' behavior with a new technology, informal observation
trials are sometimes useful. In these, farmers are given the test variety
or input and are asked to use it in a part of their fields. Although this
type of trial may not require the researcher's presence for planting or
even perhaps for harvesting, it still demands a good deal of his time, in
carefully selecting farmers and in visits during the growing season. In
few cases is the simple distribution of a new variety or input to farmers
and the reliance solely on their evaluations a worthwhile strategy. The
researcher must visit the field at least several times, see under what con-
ditions the innovation is being used, and make his own evaluation of the
results. Therefore observation trials should be undertaken only with the
realization that they are also trials, and that they will require some
amount of the investigator's attention.
Contrasts between observation trial results and those of formal
trials are sometimes striking, and may lead to new hypotheses. In one
area, a new maize variety grown in trials did rather poorly, while in ob-
servation trials planted with local farmers it did quite well. Investi-
gation showed that the primary difference in management between the two
types of trials seemed to be manuring, which was not practiced in the
fields offered for the formal trials. It is not unusual to find that poor-
er than average land is offered for a trial, while observations with small
quantities of a new variety are planted by the farmer with special care. At-
tention to the results of both types of trials helps to establish a range
of data and to identify critical variables which may limit the performance
of the variety.
5.3 Follow-up on adoption of practices
In the second, and subsequent years of on-farm trials the resear-
cher will want to do some follow-up on the activities of previous collabo-
rators as an aid in assessing technologies that have been tested, whether
or not they have yet become part of recommendations.1-
If a farmer has grown a new variety as part of an on-farm trial
the previous year, it is most worthwhile checking to see whether he has
planted the new variety again, on his own. In one maize area, many collab-
orators in one recommendation domain grew a new variety the following year,
while few in another recommendation domain did so. The difference turned
out to be the amount of insect attack that the variety suffered in storage
in the latter domain. In another case, a number of farmers who had planted
an early-maturing maize as part of a variety trial used it the following
year in their own "experiments", in which they planted small quantities of
the variety at various times of the year. Their experience provided valua-
ble data on the range of planting dates possible for the new variety.
1/ Methods for formally assessing farmers' experiences with recommendations
fall outside the scope of this paper. (See L. Harrington, "Farmer Assess-
ment of Maize Recommendations in Northern Veracruz State, Mexico."
Unpublished Ph.D dissertation, Michigan State University, 1980.)
- 25 -
Changes in management practices should also be noted. In one
maize area, most farmers who had seen insecticide used to control ear worm
in trials adopted insecticides the following year, on their own. This
helped verify the importance of subsequent trials which concentrated on re-
fining insecticide recommendations.
In the case of new varieties, it is necessary to follow their
acceptance in the kitchen and in the market as well, to see if they are
consumed by the farmer's family in the same way as local varieties, and if
they can be sold as easily as local types. In a cassava program, neighbor-
ing farmers and merchants were invited to the harvest of variety trials.
Their opinions as to the market potential of the various types proved va-
luable in selecting varieties for further testing. Samples of the varie-
ties were also distributed so that they could be prepared in the farmer's
homes and opinions on the palatability of the new varieties were then col-
5.4 Other activities with farmers
The more contact the researcher has with collaborating farmers the
better.- It has been found valuable to take farmers to visit each other's
trials, for they are particularly sensitive to small variations in farming
practices within their own environment and often make valuable observations.
One program has experimented with having several farmers from the same rec-
ommendation domain assist in the harvest of the various trials. The same
program found it useful to call together the collaborators of a particular
domain after harvest to discuss the results and propose experiments or im-
provements for the following year.
Taking account of what farmers believe to be important has led to
experiments with other technologies or even with crops which had not been
1/ For a detailed description of ways to involve farmers in an on-farm re-
search program, see Kirkby, R., P. Gallegos, T. Cornick, "On-Farm Re-
search Methods: A Comparative Approach. Experiences of the Quimiag-Pe-
nipe Project, Ecuador". Cornell International Agricultural Mimeograph
Series, Cornell University, Ithaca, New York, 1981.
- 26 -
considered. In one case, although trials had focused on maize, the prin-
cipal crop of the area, farmers expressed great interest in research that
would deal with the diseases that were affecting their broad beans, and the
following year new broad bean varieties were screened in the area. In ano-
ther program, farmers' complaints about storage losses led to the development
of a series of on-farm storage trials.
Throughout the year the researcher will also want to make sure he
works closely with extension agents, not only in the management of the
trials, but in other activities as well. Demonstrations or field days can
be jointly organized. As the on-farm research progresses and recommenda-
tions begin to be produced, the researcher and the extension agent begin to
share another common interest: they both will want to measure the adoption
rates of recommended practices, to analyze the appropriateness of the recom-
mendations and the effectiveness of their communication methods.
The management of on-farm trials requires the researcher's full time
presence in the target area, and it is possible to take advantage of this
to collect data which is useful in interpreting results and planning future
work. Among the strategies that have been found useful are:
1. Decide what data (prices, meteorological, labor supply, etc.)
are necessary and design forms to record and store this information.
2. Be alert for ways of increasing interchange with farmers; talk to
farmers who are not trial collaborators and visit their fields.
3. Develop contacts with local merchants, traders and others who are
good sources of information on markets.
4. Learn about other research that is, or has been, done in the area.
5. Follow the experiences of former collaborators with new technologies.
b. Maintain good contracts with local extension personnel and plan
work wit h them.
7. Develop a list of research questions and decide whether trials,
less formal experiments, informal enquiry or survey methods are
required to answer them.
This paper has described various types of data that are available to
researchers in the course of on-farm experimentation. Although agronomic
data derived from the trials is the basic information sought in this phase
of on-farm research, no attempt has been made to suggest exactly what type
of biological information should be obtained, nor how it should be analyzed.
The object has been instead to point out that the way in which the field
research is organized can make an important contribution to the validity
and completeness of this type of data, and that a wide variety of informa-
tion is available to the researcher to help interpret the agronomic obser-
Most programs in on-farm research operate under severe budgetary cons-
traints, and the one described in this paper is no exception. Thus data
collection methods must take account of these limitations and take maximum
advantage of the resources that are available. The basic organization of
data collection requires: careful planning of the trials, including the ac-
comodation of each site selected to the overall research strategy; atten-
tion to developing participating farmers as collaborators and valuable sour-
ces of information; the establishment of means of recording and storing all
of the data so that they are available for analysis; a conscious effort to
collect supplementary information from as wide a range of sources as possi-
ble; and the coordination of research efforts with extension services.
Although it is possible to develop guidelines for managing data collec-
tion during on-farm experimentation, the experience and imagination of the
researcher are equally important, for no set of fixed procedures can subs-
titue for the ability to take a flexible, open-minded approach to problems
in the field. Much of the information obtained in the course of the re-
search is tentative, at times even contradictory.; It comes in bits and
pieces and requires much of the researcher's timelin trying to fit it to-
gether into logical patterns. Careful recording and maintenance of rele-
vant data is essential to the process. It becomes part of the dynamic of
on-farm research, where this year's uncertainties are transformed into next
year's experiments. The researcher is of course involved in data collec-
tion and analysis in order to produce recommendations as rapidly as possi-
ble, but an equally important product of the process is a set of hypothe-
ses to be tested in subsequent cycles.
Finally, it should be noted that the data collection procedures of on-
farm experimentation are complex and varied because the problems that they
treat are so difficult. The concepts of on-farm research have been developed
in response to the fact that many programs of technology transfer in ru-
ral development have failed because they have not taken account of the many
factors that impinge on small farmer decision-making. The approach des-
cribed here asks that the researcher collect a wide range of information
through a variety of techniques, in order to understand the place of his
target crops within the total farming system. His job is to use on-farm
experiments as a basis for establishing a research partnership with farmers
and extension agents which will improve the effectiveness of agricultural
- 29 -
Field book used in an on-farm research program investigating new tech-
nologies for maize and beans in Imbabura Province, Ecuador.
(1) PLANTING DATA
Location Farmer Trial
Type of trial Design
Size of trial
Length of rows Distance between rows
No. seeds per hole: Maize Beans
Method of planting beans
Source of seed
Fertilization of plot (including that of the farmer)
Other operations at planting
(2) FIELD PLAN
(3) CHARACTERISTICS OF THE PLOT
Year Crop(s) Fertilization
Preparation of the plot
Irrigation? Yes / No Frequency of irrigat
Soil analysis: N P K Zn
(4) MANAGEMENT OF THE TRIAL
ACTIVITY DATE METHOD
Weed control 1
(5) OBSERVATIONS ON THE FARMER'S CROP
Location of the field under observation
Year Crop(s) Fertilization
No. of Distance between
Crop Variety seeds per hole: Plants Rows
Activity Date Method
Weed control 1
Irrigation during year_
(5) OBSERVATIONS ON THE FARMER'S CROP (Cont.)
What are the principal insect problems in the field?
What are the principal disease problems in the crop?
What are the most common weeds?
Source of seed used in the field:
(6) CHARACTERISTICS OF THE FARMER
Number of hectares worked this year:
Principal crops: 1)
In which crops does he hire labor?
In which crops does he use fertilizer?
Use of other agro-chemicals
Which crops are sold?
Off-farm employment or activities:
Problems in storage of crops:
(7) AGRONOMIC DATA
Diseases Damage Other observations
o/o Days to
Date Classification Date Classification Date O/o Date o/o Date Date
(8) HARVEST DATA
Each treatment harvested is rows of meters.
Area of each treatment harvested: square meters.
(9) OBSERVATIONS OF THE TRIAL
(To be filled in on each visit to the trial)
Was farmer present? Yes / No
Work on the plot since last visit:
Development of the crop