Agricultural Development Support II (project A.D.S.-II)

MISSING IMAGE

Material Information

Title:
Agricultural Development Support II (project A.D.S.-II) midterm evaluation
Portion of title:
Project A.D.S.-II midterm evaluation
Physical Description:
178 leaves : ; 28 cm.
Language:
English
Creator:
Locher, Uli
Farming Systems Support Project
United States -- Agency for International Development
Publisher:
University of Florida, Farming Systems Support Project
Place of Publication:
Gainesville, FL
Publication Date:

Subjects

Subjects / Keywords:
Agriculture -- Research -- Haiti   ( lcsh )
Genre:
bibliography   ( marcgt )
non-fiction   ( marcgt )
Spatial Coverage:
Haiti

Notes

Bibliography:
Includes bibliographical references (leaves 174-178).
Statement of Responsibility:
Evaluation team: Uli Locher,... et al..
General Note:
Cover title.
General Note:
Typescript.
General Note:
"Port-au-Prince, Haiti ; May, 1986."

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
oclc - 767835292
ocn767835292
Classification:
lcc - S542.H2 A4 1986
System ID:
AA00011404:00001


This item is only available as the following downloads:


Full Text










AGRICULTURAL DEVELOPMENT SUPPORT II


(PROJECT A.D.S.-II)

MIDTERM EVALUATION












Evaluation Team: Uli Locher
Jan Broekhuyse
Mimi Gaudreau
John Lichte






Port-au-Prince, Haiti
May, 1986















University of Florida Farming Systems Support Project
United States Agency for International Development















CONTENTS


1. Executive summary 6

1.1 Purpose of evaluation 6
1.2 Summary of project description 6

1.2.1 Project history and implementation arrangements
1.2.2 Implementation actions

1.3 Recommendations 8

1.3.1 General recommendations
1.3.2 Specific recommendations

1.3.2.1 Socio-economic analysis
1.3.2.2 Agronomy
1.3.2.3 National level data collection

1.4 Summary of major findings 12
1.5 Major lessons learned 15
1.6 Evaluation methodology 15

1.6.1 Data gathering instrument
1.6.2 Itinerary and logistics
1.6.3 Detailed itinerary of team members
1.6.4 Composition of evaluation team

1.7 Comments on the evaluation scope of work 17

2. Recommendations 19

2.1 The general thrust of the recommendations 19
2.2 General recommendations 20
2.3 Specific recommendations 22

2.3.1 Socio-economic analysis
2.3.2 Agronomy
2.3.3 National level data collection


3. Findings 24

3.1 Major evaluation findings 24
3.2 Linkage between project and mission strategy 29
















4. Lessons learned


5. Project description

5.1 The development problem
5.2 Project goal and purpose


5.2.1
5.2.2
5.2.3
5.2.4


Project paper
Project paper supplement
Project agreement
Arkansas/Winrock proposal


5.3 Project inputs and outputs

5.3.1 Project inputs


5.3.1.1
5.3.1.2
5.3.1.3
5.3.1.4


Project paper
Project paper supplement
Project agreement
Arkansas/Winrock proposal


5.3.2 Project outputs


5.3.2.1
5.3.2.2
5.3.2.3
5.3.2.4


Project paper
Project paper supplement
Project agreement
Arkansas/Winrock proposal


5.4 Project assumptions


5.4.1
5.4.2
5.4.3
5.4.4


Project paper
Project paper supplement
Project agreement
Arkansas/Winrock proposal


5.5 Analysis of problem areas
5.6 Current state of implementation and outputs

5.6.1 Documentation

5.6.1.1 Characterization of agricultural
information systems
5.6.1.2 Characterization of existing farming
systems











5.6.1.3 Descriptive studies of regional and
community systems
5.6.1.4 Detailed description of farming and
cropping systems



5.6.2 Agricultural information collection instruments 51
and procedures

5.6.2.1 Survey instruments for routine
collection of production and marketing
data
5.6.2.2 Detailed procedures for farming systems
research
5.6.2.3 Sampling frames and procedures for
production and marketing information

5.6.3 Farming systems research instruments and 52
procedures

5.6.3.1 Farming systems characterization

5.6.3.1.1 Initial rapid appraisal
5.6.3.1.2 Constraints surveys
5.6.3.1.3 Farm monitoring

5.6.3.2 Screening alternative technology
5.6.3.3 On-farm testing of technology

5.6.4 Agricultural technology 54

5.6.4.1 Alternative farm technology tests
5.6.4.2 Adoption of appropriate agricultural
technology

5.6.5 Trained agricultural technicians, survey 55
analysts, enumerators, and farmers

5.6.5.1 Agricultural technicians
5.6.5.2 Survey analysts
5.6.5.3 Field enumerators
5.6.5.4 Farmers

5.6.6 Agricultural production and marketing 57
information

5.6.6.1 Periodic information about marketing
5.6.6.2 Production enterprises
5.6.6.3 Major land uses
5.6.6.4 Socio-economic characteristics

5.6.7 Institutions and institutional capability 58







5



5.6.7.1 Organization capable of FSR/E
5.6.7.2 Organization in charge of agricultural
information


6. Technical annexes 59

6.1 Theory and practice of FSR/E at ADS-II 59
(John Lichte)

6.2 Agronomic considerations concerning 96
ADS-II (Mimi Gaudreau)

6.3 National level data collection and 110
analysis at ADS-II (Uli Locher)

6.4 History of FSR/E in Haiti and at ADS-II 136
(Mimi Gaudreau)

6.5 A Farming Systems Research Model for ADS-II 141
(Jan Broekhuyse)


7. Administrative annexes 158

7.1 Evaluation scope of work 158
7.2 Definition of FSR/E 164
7.3 Project logical framework 165
7.4 Project implementation plans 168
7.5 Persons contacted during the evaluation 171
7.6 Bibliography 174












1. EXECUTIVE SUMMARY

1.1 Purpose of evaluation

The Agricultural Development Support II Project (521-0092)
has just completed its second year of operation. It has three
distinct components which demand separate evaluation: (1) its
Farming Systems Research and Extension component (FSR/E), (2) its
National Agricultural Statistics component and (3) its
Comprehensive Resource Inventory and Evaluation component
(CRIES). The evaluation will put more emphasis on FSR/E since
the latter components are relatively more recent.

A midterm evaluation is an occasion for redefinition of
project objectives based upon accumulated experience, and for
mid-course corrections if they appear necessary. The latter can
arise out of changing internal as well as external conditions.
ADS-II, e.g., will now have to define its relationship with
TWAMP, the Targeted Watershed Management project which has
acquired a strong presence in planning circles at AID long before
its projected 1987 starting date for operations in the Cayes
region.

Two aspects of this report may require an explanation.
First, in the absence of a clear consensus and an authoritative
definition of FSR/E, the team felt obliged to state its standard
of evaluation in some detail. This has in turn made it useful to
include a theoretical piece in the annex to state, not what FSR/E
is at ADS-II, but also what it might ideally be or become in the
longer run (See technical annex 1). Secondly, some parts of this
report are rather critical of either the original project design
or the implementation. The evaluation team wants to state
emphatically that these criticisms should be read as
encouragement for change and as expressions of the conviction
that ADS-II has good potential and a bright future if the
necessary mid-course corrections are made.

1.2 Summary of Project Description

1.2.1 Project History and Implementation Arrangements

The ADS-II project contract was awarded to the University of
Arkansas and Winrock International, with the University of
Arkansas as the prime contractor. In July of 1983, a Haitian
project director and technical personnel were assigned to the
project. This staff, in cooperation with the contractors, chose
the regions of Jacmel and Cayes as the sites for project
activity. A zone on the plain and a zone on the hillsides was
chosen in each region to represent agro-ecological differences.
The project effectively began operation with the arrival of the
technical assistance team towards the end of February, 1984.
This team was composed of an experienced farming systems
agronomist, an agronomist with specialized training in extension











and agricultural education, and an anthropologist with farming
systems experience as the team leader. During 1985 an
agricultural economist specializing in the national survey was
added to the permanent team.

In each region a Haitian agronomist was assigned to
coordinate project efforts with the help of one of the technical
assistance team agronomists. An agronomist was assigned to
supervise each of the zones and provided with a supporting staff
of three resident agronomists, and 3 monitors for the testing
program. Superimposed on this farming systems component structure
there were a statistician with 3 monitors and an enumerator in
each zone to conduct survey activities.

The project was originally located with the Centre de
Recherche et Documentation Agricole (CRDA) and under the
responsibility of FAMV (The Faculty of Agriculture and Veterinary
Medicine). Politically motivated changes in Ministry personnel
led to a serious conflict of personalities. In September 1985,
the project was physically and organizationally moved out of the
faculty and into the Ministry of Agriculture. The project remains
an independent unit attached to the Ministry, outside the
Ministry's normal organization.


1.2.2 Implementation Actions

The first field trials in Haut Cap Rouge, Jacmel were begun
in February, 1984, through the initiative of the regional
coordinator. During April and May the farming systems component
(SP) field teams conducted reconnaissance surveys in Cayes and
Jacmel in order to characterize each of the zones and identify
constraints to agricultural production. Simultaneously, formal
surveys were conducted by the rural economics and statistics
component (ERS) to verify the results of the reconnaissance
survey and to begin the process of creating an agricultural
information system, on which to base a national agricultural
survey. The combined results of these surveys were used as the
basis for designing and -conducting second season field trials in
July and August, 1984. During this second season, ERS personnel
also conducted a series of more directed surveys, e.g. on
cultivated land area and livestock ownership.

The original project prepared in 1978 contained a major
component orientated towards collecting and analyzing
agricultural statistics at the national level. This was largely
eliminated from the project as redesigned in 1983, except for
viewing knowledge gained from the farming systems surveys as a
basis for establishing a national survey. By the end of the
first year, pressure from the Ministry supported by USAID,
reoriented the project towards establishing a national
agricultural survey. Although not necessarily a change in
substance from the project papers, it was in fact a major change
in the contract scope of work.












By early 1985, an area frame survey format was chosen and a
permanent technical assistant was brought in to help organize and
supervise the national survey effort. The Comprehensive Resource
Inventory and Evaluation System (CRIES) was added to the national
survey effort to strengthen the Ministry's capability to store,
access and analyze regional and national statistics.

The SP field trials have continued with two varieties
receiving wide acceptance by farmers by the second season
(July-December, 1985). Tamazulapa black beans are now widely
planted in Jacmel and Cayes, and Amina rice in Berault, Cayes.
As of the second season, 1985, SP activities have expanded to
working with farmers on construction of erosion control
structures on hillsides in conjunction with either tree planting
or grasses for stabilization. The teams have also constructed
pig pens and secured pigs for future studies of the reintegration
of these animals into the farming system.

1.3 Recommendations

1.3.1 General Recommendations

Recommendation 1

As it is currently being implemented, the farming systems
component of ADS-II is not performing as an effective
interdisciplinary farming systems unit. The project should be
encouraged to continue its operations while at the same time
restructuring its farming systems component.


Recommendation 2

A truly interdisciplinary approach to FSR/E should be
adopted by the farming systems component. Each local team should
consist of agronomists as well as specialists in
socio-anthropology and rural economics who will jointly plan,
organize and evaluate project activities.

Recommendation 3

To facilitate this interdisciplinary approach to FSR/E, the
rural economics and statistics component (ERS) within the project
should be abolished. The rural economics section should be
disbanded and its staff and budget fully integrated into the
farming systems component. The national agricultural information
component (the national survey and CRIES) should remain within
the ADS-II project but be sufficiently autonomous in terms of
personnel, space and budget so as not to impinge upon FSR/E
activities.












Recommendation 4

Any significant geographical expansion of FSR/E activities
should not be considered until a model and methodology acceptable
to an external evaluation team has been developed.

Recommendation 5

Given the scope of the change recommended and the large
number of new personnel, ADS-II should schedule a 2-3 day retreat
to discuss FSR/E, the approach they will take and how to
establish effective interdisciplinary units.

Recommendation 6

Detailed descriptive and analytical work on economic and
social aspects of farming systems, including relevant rural
institutions, should be incorporated into the interdisciplinary
research activities. This will often require in depth studies
which cannot be conducted using questionnaires. Case studies,
anthropological techniques, and informal but guided diagnostic
and monitoring surveys are appropriate tools for collecting this
type of information.

Recommendation 7

FSR/E should include research on extension techniques and
develop methods of working with and through appropriate farmer
organizations. Structures which promote farmers' capacity for
change should be identified.

Recommendation 8

Project FSR operations should continue in both the Cayes and
Jacmel regions. The subdivisions into mountain and plains zones
should be maintained and, where feasible, should be linked so
that activities are concentrated into managing a single
watershed. While irrigation is important for increasing
production in the plains, direct involvement of ADS-II personnel
in major construction and maintenance of irrigation systems is
discouraged. Project activities in Jacmel should continue. The
feasibility of managing one watershed completely, including
erosion control on the hillside and irrigated agriculture on the
plain, should be investigated. The present organizational
advantage and easy communication between zones should become a
model for other zones. Activities in the Cayes area need better
geographical focus and concentration. Work on hillside
environments should increase and preferably be concentrated in a
watershed contiguous to that part of the plains where activities
will take place. Organization in a manner which allows direct
communication between the plain and hillside sites is strongly
encouraged.









Recommendation 9

The project needs to establish a monitoring and evaluation
program on at least 4 levels in addition to that provided by the
agricultural data collection: (1) technical innovations the
adoption, adaption and impact on agricultural production; (2)
farm/farm family changes in farm/farm family activities which
indicate the impact of a technology or intervention on the
farm/farm family as a system; (3) social organization changes
over time; (4) project-documentation of activities and knowledge
acquired, performance of separate components.

1.3.2 Specific Recommendations

1.3.2.1 Socio-economic analysis

Recommendation 10

The technical assistance position for a farming systems
socio-economist, along with the Haitian counterpart, should be
located in Cayes. Their responsibilities should include working
closely with both the Jacmel and Cayes farming systems teams.

Recommendation 11

A major effort should be made to find Haitian
socio-economists to be included in the farming systems teams at
the zone level. Graduates from the University's new Rural
Economics and Development program will not be available for about
two years. In the interim, several resident agronomists should be
given training in a program like CIMMYT's socio-economic training
program.

Recommendation 12

Given the lack of qualified Haitian socio-economists, the
project should provide degree training for several candidates in
agricultural economics and sociology or anthropology.










1.3.2.2 Agronomy

Recommendation 13

Continued contact and exchange with existing Haitian
research and development programs as well as the International
Agricultural Research Centers should be maintained and where
appropriate, expanded. ADS-II researchers should also establish
contact with other research organizations in the Caribbean
region, e.g., INRA, Guadeloupe; CARDI in the Anglophone West
Indies; the University of the West Indies, Trinidad; and the
University of Puerto Rico.

Recommendation 14

Agronomic analysis of on-farm trials should be done at the
local level. Field agronomists should receive formal training in
experimental design and analysis. They should be providing
increased input into planning the research agenda based on their
knowledge of the local farming systems.

Recommendation 15

The emphasis on varietal testing within the on-farm research
program should be reduced. More interventions relating to system
interactions, tuber crops and other economically important crops
within the system should be included in the on-farm trials.

Recommendation 16

Site characterization for agronomic field trials needs to
improve. Agronomic field data collection should be modified to
include some simple site characterization information and general
observations, which will be useful analyzing and interpreting
results. Farmer evaluations should be added to the 12-point
write up of trial results.

1.3.2.3 National level data collection

Recommendation 17

The project should use its influence to achieve a
substantial improvement in the design and implementation of the
national agricultural survey. The most urgent changes pertain to
(1) the selection and training of local enumerators, (2) the
design and full analysis of protests of instruments as well as
the survey organization and (3) the planning of small data
quality control surveys on various aspects of land use, land
tenure, animal ownership and cultivation practices.











Recommendation 18

Expansion of the agricultural survey work to a national
level should be postponed until the improvements proposed in
recommendation 13 have been made and the existing data for the
department of the South have been fully analyzed.

Recommendation 19

The national agricultural survey should not be construed as
serving all needs for quantitative information on agriculture.
Standardized, periodic and longitudinal studies should now be
initiated of such critical issues as the volumes of fertilizer
use by region, market prices for basic foodstuffs, and overall
flows of several commodities.

Recommendation 20

The implementation of the CRIES system should be accelerated
by the addition of a large training component aimed at
appropriate interpretation of CRIES output. CRIES based
documents should be published and distributed widely, and
specifically aimed at an audience of technicians and policymakers
unfamiliar with specialized quantitative methodologies.


1.4 Summary of Major Findings

ADS-II has a number of accomplishments which make it a
worthwhile and, in part, successful project. It has established
a large-scale on-farm testing program which has already scored
two major successes. It has designed an agricultural information
system based on a national survey, and tested it at the
department level. This national survey model promises to provide
better agricultural information than any presently available in
Haiti. It has also established an information management system
(CRIES) which provides more information and greater location-
specific detail than any source available in Haiti. This should
be very useful for decision making and project design and may
have some direct application for other projects. But what ADS-II
has yet to become is a full-fledged, well-designed and
potentially exemplary FSR/E project.

ADS-II achieved a major accomplishment by moving from a
reconnaissance survey and its verification to establishing
on-farm trials in about a three month period. The extensive
on-farm testing program covered 293 researcher-managed and 285
farmer-managed trial replications in 1985. Two major successes
have emerged from this testing program, Tamazulapa beans and
Amina rice, as well as the extension opportunity offered by these
many farmer contacts.

The ADS-II project contains a flaw of either design or early
implementation which has seriously impeded the farming systems











component of the project. Socio-economists were physically and
organizationally separated from the farming systems component,
because they were lumped administratively and functionally
together in one 'unit called the Economie Rurale et Statistique
Section (ERS). This resulted in a situation where the farming
systems teams were not interdisciplinary and had little or no
socio-economic input. They were concentrating entirely on
agronomic testing even though the farming systems within zones
had not been fully characterized and little progress was being
made in the identification of farmer problems and constraints.
Structural changes which fully integrate socio-economists into
the farming systems component must be made. No geographical
expansion or substantive expansion into extension should be
considered until these shortcomings are resolved and a more
complete and effective farming systems research program is
established.

The project organization dictates that the basic farming
systems teams are found at the zone level (Haut Cap Rouge, Bas
Cap Rouge, Berault and Maniche). Interdisciplinary integration
should be promoted precisely at the zone level. Basic teams
should be provided with the necessary training and guidance to
design, implement and analyze their own agronomic tests as well
as diagnostic and monitoring surveys. This implies that
technical assistance in particular, and the project in general,
should take on a much stronger training function.

The agronomic testing program needs to pay more attention to
isolating factors which affect results. This includes improving
the characterization of test sites as well as designing tests so
that the effect of individual factors is clearly demonstrated.
Farmer evaluations should be systematically solicited on all
tests and made an integral part of the write-up of each test.
Test designs do not always correspond to the stated objective and
the analyses of some tests do not appear to be appropriate for
the design actually used.

The ADS-II interpretation of FSR/E has been flawed and
incomplete in several respects, particularly with regard to
institutional problems. It has not as yet begun to study rural
institutions and their effects on adoption of innovations.
Innovations have not been monitored after farmers had begun to
use them on their own, to see what changes take place in the
farming system as a whole. It is not clear that the researchers
are sufficiently aware of labor and resource constraints to
include such considerations in their selection of technologies.
The importance of different innovations and the size of the
target group to which they apply have not been specifically
evaluated.

The definition of extension has also been too narrow and
needs to be expanded to include the promotion of rural
institutions (organizations) which facilitate the adoption of
innovations by farmers. Although the scope of this project does











not include the possibility of a large extension program, it can
contribute to a better understanding of rural institutions and
how extension might involve them in the transfer of technology
process. The ability to identify and analyze this potential of
rural institutions needs to be built into extension training
programs.

The technical assistance team leader has three nearly full
time functions in administration, farming systems socio-economic
research, and the national survey and CRIES activities. He
cannot fill all these functions. The addition of an
administrative assistant would allow him to fulfill more of his
technical duties.

The national agriculture survey is an ambitious undertaking
which demands specialized training on all levels. A high level
of expertise in such an undertaking will be translated into high
quality data in the end. No one at present working within ADS-II
possesses all of the expertise necessary to assure that good
quality data will be produced in a national survey.

The project has carried out its survey in the department of
the South as a test for the national survey. It intends to move
on to a new department within weeks, and to survey the whole
country this year. This research is being pushed ahead despite
serious flaws in the technical aspects of survey work. The
questionnaire is written in the wrong language (French), it is
overcrowded, at times imprecise and conceptually unclear, and it
demands an excessive amount of interviewer training. Pretests
appear to have been skipped and/or analyzed only very
incompletely. Data verification is nowhere attempted, and
topical in-depth surveys to verify the range of survey answers
and to help in the interpretation and extrapolation of the survey
data are neither done nor planned. There should be an immediate
moratorium in the national survey work to allow more adequate
testing, preparation and training.

There are many aspects of agriculture which are critical for
FSR/E and for rural development but do not lend themselves to
coverage by large-scale questionnaire surveys. The project is in
a privileged position to carry out such work and should not
neglect it for the sake of a large-scale survey whose results
will be very approximate at best, and probably quite irrelevant
for FSR/E.

The CRIES system is a splendid tool and its introduction to
Haiti by ADS-II has been rapid, and carried out competently.
This system has great potential overall and is of particular
value in a context where what little systematic data exists is
frequently not accessible. But CRIES is so new and so complex
that its installation should be followed by a systematic campaign
to mobilize its potential for all future users. The CRIES-based
reports published so far have little advertising value. They
could be complemented by more accessible and more convincing
reports.















1.5 Major Lessons Learned

1. Farming systems research is not effective unless
integrated interdisciplinary teams are formed. The attempt to
have a separate, city-based socio-economic survey component
complement on-farm testing by agronomists was fundamentally
flawed and has largely failed. It did not permit the critical
amount of interdisciplinary interaction.

2. The choice of four different zones in two geographically
distant regions made it very difficult to establish
interdisciplinary local farming systems teams. It would have
been more effective to begin work in a single region in order to
establish a methodology and model -of FSR/E before expanding
geographically.

3. When a project does not include a full time
administrative position on the technical assistance team, the
team leader cannot devote adequate time to technical tasks
because he is constrained by administrative imperatives.

4. In order to be useful, quantitative survey work has to
follow its own rules, profit from specialized expertise and lead
to a particular mode of analytical work. It is rare to find a
team which combines the ethnographic, socio-economic and
agronomic expertise, and the style of work necessary for
successful FSR/E, with the talent and style typical of good
quantitative work. ADS-II has let the quantitative work dominate
most of its socio-economic analysis, to the neglect of the type
of analysis needed in FSR/E. Questionnaire data of uneven
quality has been substituted for a comprehensive analysis of the
farm system under study. This was the outcome of building two
entirely different tasks into one project. Such a tour de force
should not be repeated.

1.6 Evaluation Methodology

1.6.1 Data gathering instruments

Since ADS-II is above all a research project and since this
is only the midpoint of its projected four-year duration, the
evaluation team abstained from embarking upon new data collection
of its own. Instead, we spent time reviewing all data collection
instruments used by the project so far and evaluating the data
collections and analysis produced on their basis. Specific
comments and recommendations are included in the annexes of this
report.









1.6.2 Itinerary and Logistics

The team arrived in Port-au-Prince on Sunday, April 20,
1986, and was introduced to AID personnel on April 21, and to
Ministry (MARNDR) personnel and project headquarters at Damien on
April 22. A quick reconnaissance trip to project sites in Jacmel
and Cayes occupied the following three days. On Saturday, April
26, the team elaborated its work plan for the rest of the
mission. The whole first week had been planned and organized by
project staff. The team was increased by one member during this
week, Dr. Jim Jones of the University of Florida's Farming
Systems Support Project (FSSP), the agency responsible for the
evaluation.

In order to make maximum use of the disciplinary
specializations and linguistic skills of the team members, and to
do justice to the very diverse aspects of the scope of work, the
team members separated for certain periods of time. In the
following section there is a detailed itinerary for each team
member.

While in Jacmel and Cayes, the team members stayed at the
hotels "La Jacmelienne" and "Concorde". In Port-au-Prince, part
of the team could be housed at Turgeau apartment kept by ADS-II
and Winrock International. The team used two cars, one of which
was a project Jeep, and made use of a portable computer and
printer allowing the production of texts anywhere in the field.















1.6.3 Detailed Itinerary of Team Members


Broekhuyse
arrive
P
J
C
P
P
J
J
P
P
C
C
P
P
depart


Gaudreau
arrive
P
J
C
P
J
J
C
C
P
C
C
C
P
P
depart


P Port-au-Prince
J Jacmel
C Les Cayes
G Gainesville, Fla.




1.6.4 Composition of evaluation team

Jan Broekhuyse, Anthropologist and specialist in
institutional analysis
Mimi Gaudreau, agronomist
John Lichte, agricultural economist.
Uli Locher, sociologist and team leader

Broekhuyse, Gaudreau and Lichte have made FSR/E their specialty;
Locher has done research in Haiti since 1972.


1.7 Comments on the Evaluation Scope of Work

The scope of work for the 1986 evaluation appears to be the
product of new priorities and new thinking of AID-Haiti rather
than of the original project document. The focus is rather
different from that of the 1978 project paper and the 1982
amendment. We find it close to the 1983 Arkansas/Winrock


June 20
21-22
23
24-25
26-27
28-3
May 4-5
6-8
9
10-11
12
13
14
15-23
24
25
26-30
June 1


Lichte
arrive
P
J
C
P
J
J
C
C
P
C
C
C
P
P
G
G
depart


Locher
arrive
P
J
C
P
P
J
J
P
P
P
C
P
P
P
P
G
depart










contract in spirit but we cannot determine whether this applies
to the letter as well, since we were given a truncated copy of
the contract only, which excluded the relevant sections on
evaluation.

The evaluation team has clearly chosen sides between the
somewhat contradictory foci of the project documents. We
have established a standard definition of FSR/E, and have
evaluated the project in terms of this standard, as well as the
binding contractual arrangements of 1983. Our standard
definition of FSR/E is included as Administrative annex 7.2.

The scope of work asked for an evaluation along both the
institution building focus of 1978 and 1982 and the FSR/E focus
of 1983. This resulted in an evaluation document composed of
several parts which are less than perfectly integrated. This
lack of integration was the inevitable consequence of the
conflicting directions taken by the various project documents,
the project implementation, and the imperatives of a mid-course
evaluation as stated in our scope of work. The scope of work was
comprehensive and rather ambitious. It was our intention to do
justice to all of its parts, even at the risk of imperfect
integration of the various parts of our report.

At the beginning of its work the evaluation team was handed
additional instructions concerning the format of reporting. We
were surprised to find that several new topics were in fact added
to the scope of work in this way. They were the so-called
"cross-cutting issues" of (1) women in development, (2)
sustainability/replicability (of project activities), (3)
environmental impact, (4) privatization, and (5) democratization.
While we found the prescribed reporting format acceptable, we
have not been able to build the five "cross-cutting issues" into
the report in a systematic way. All we could do was to address
these issues when they appeared critical within the general
context of project activities.










2. RECOMMENDATIONS

2.1 The general thrust of the recommendations

The following recommendations should be read in the context
of all subsequent chapters of this report, and particularly in
connection with the major findings of the evaluation. It is only
in that context that recommendations can be fully appreciated,
and hopefully also accepted. There are four general concerns,
however, which are underlying what is being recommended here, and
they deserve to be stated explicitly at this point.

First, the evaluation team is supportive of ADS-II and
wishes it to continue and evolve into a full-fledged farming
systems research project. We think this can be achieved if part
of the research activity is restructured and redirected.

Second, many of the weaker aspects of the project were in
some way "built into it" from the very beginning. Whether they
were flaws of design or of the earliest implementation phase is
immaterial. What does matter at this point is that they be
candidly recognized and corrective action be taken.

Third, the evaluation team was impressed by the high
professional standards, the honesty, the courage and the tenacity
of the TA team members. We think that everything possible should
be done to retain their services for at least another two years.
The thrust of our criticisms of the project is never directed at
persons. We have chosen to be direct and open because an
evaluation which is not candid, is worthless to all involved.

Fourth, the evaluation team decided to spell out its
standard of evaluation (see Administrative Annex 7.2 Definition
of FSR/E). This was done for the purpose of clarity and fairness.
This standard was adopted following considerable debate and
knowing well that both the Project Agreement and the
Arkansas/Winrock proposal can and must be seen as legally binding
documents. The trouble with these documents is that they
represent very different approaches to rural development. We felt
that since the University of Arkansas had obtained the contract
to implement ADS-II, it was the contract between this University
and AID which should constitute the standard for evaluating the
performance of the team fielded by the University.













2.2 General Recommendations

Recommendation 1

As it is currently being implemented, the farming systems
component of ADS-II is not performing as an effective
interdisciplinary farming systems unit. The project should be
encouraged to continue its operations while at the same time
restructuring its farming systems component.


Recommendation 2

A truly interdisciplinary approach to FSR/E should be
adopted by the farming systems component. Each local team should
consist of agronomists as well as specialists in
socio-anthropology and rural economics who will jointly plan,
organize and evaluate project activities.


Recommendation 3

To facilitate this interdisciplinary approach to FSR/E, the
rural economics and statistics component (ERS) within the project
should be abolished. The rural economics section should be
disbanded and its staff and budget fully integrated into the
farming systems component. The national agricultural information
component (the national survey and CRIES) should remain within
the ADS-II project but be sufficiently autonomous in terms of
personnel, space and budget so as not to impinge upon FSR/E
activities.

Recommendation 4

Any significant geographical expansion of FSR/E activities
should not be considered until a model and methodology acceptable
to an external evaluation team has been developed.

Recommendation 5

Given the scope of the change recommended and the large
number of new personnel, ADS-II should schedule a 2-3 day retreat
to discuss FSR/E, the approach they will take and how to
establish effective interdisciplinary units.

Recommendation 6

Detailed descriptive and analytical work on economic and
social aspects of farming systems, including relevant rural
institutions, should be incorporated into the interdisciplinary
research activities. This will often require in depth studies
















which cannot be conducted using questionnaires. Case studies,
anthropological techniques, and informal but guided diagnostic
and monitoring surveys are appropriate tools for collecting this
type of information.

Recommendation 7

FSR/E should include research on extension techniques and
develop methods of working with and through appropriate farmer
organizations. Structures which promote farmers' capacity for
change should be identified.

Recommendation 8

Project FSR operations should continue in both the Cayes and
Jacmel regions. The subdivisions into mountain and plains zones
should be maintained and, where feasible, should be linked so
that activities are concentrated into managing a single
watershed. While irrigation is important for increasing
production in the plains, direct involvement of ADS-II personnel
in major construction and maintenance of irrigation systems is
discouraged. Project activities in Jacmel should continue. The
feasibility of managing one watershed completely, including
erosion control on the hillside and irrigated agriculture on the
plain, should be investigated. The present organizational
advantage and easy communication between zones should become a
model for other zones. Activities in the Cayes area need better
geographical focus and concentration. Work on hillside
environments should increase and preferably be concentrated in a
watershed contiguous to that part of the plains where activities
will take place. Organization in a manner which allows direct
communication between the plain and hillside sites is strongly
encouraged.

Recommendation 9

The project needs to' establish a monitoring and evaluation
program on at least 4 levels in addition to that provided by the
agricultural data collection: (1) technical innovations the
adoption, adaption and impact on agricultural production; (2)
farm/farm family changes in farm/farm family activities which
indicate the impact of a technology or intervention on the
farm/farm family as a system; (3) social organization changes
over time; (4) project-documentation of activities and knowledge
acquired, performance of separate components.
















2.3 Specific Recommendations

2.3.1 Socio-economic analysis

Recommendation 10

The technical assistance position for a farming systems
socio-economist, along with the Haitian counterpart, should be
located in Cayes. Their responsibilities should include working
closely with both the Jacmel and Cayes farming systems teams.

Recommendation 11

A major effort should be made to find Haitian
socio-economists to be included in the farming systems teams at
the zone level. Graduates from the University's new Rural
Economics and Development program will not be available for about
two years. In the interim, several resident agronomists should be
given training in a program like CIMMYT's socio-economic training
program.

Recommendation 12

Given the lack of qualified Haitian socio-economists, the
project should provide degree training for several candidates in
agricultural economics and sociology or anthropology.

2.3.2 Agronomy

Recommendation 13

Continued contact and exchange with existing Haitian
research and development programs as well as the International
Agricultural Research Centers should be maintained and where
appropriate, expanded. ADS-II researchers should also establish
contact with other research organizations in the Caribbean
region, e.g., INRA, Guadeloupe; CARDI in the Anglophone West
Indies; the University of the West Indies, Trinidad; and the
University of Puerto Rico.

Recommendation 14

Agronomic analysis of on-farm trials should be done at the
local level. Field agronomists should receive formal training in
experimental design and analysis. They should be providing
increased input into planning the research agenda based on their
knowledge of the local farming systems.














Recommendation 15


The emphasis on varietal testing within the on-farm research
program should be reduced. More interventions relating to system
interactions, tuber crops and other economically important crops
within the system should be included in the on-farm trials.

Recommendation 16

Site characterization for agronomic field trials needs to
improve. Agronomic field data collection should be modified to
include some simple site characterization information and general
observations, which will be useful analyzing and interpreting
results. Farmer evaluations should be added to the 12-point
write up of trial results.

2.3.3 National level data collection

Recommendation 17

The project should use its influence to achieve a
substantial improvement in the design and implementation of the
national agricultural survey. The most urgent changes pertain to
(1) the selection and training of local enumerators, (2) the
design and full analysis of protests of instruments as well as
the survey organization and (3) the planning of small data
quality control surveys on various aspects of land use, land
tenure, animal ownership and cultivation practices.

Recommendation 18

Expansion of the agricultural survey work to a national
level should be postponed until the improvements proposed in
recommendation 13 have been made and the existing data for the
department of the South have been fully analyzed.

Recommendation 19

The national agricultural survey should not be construed as
serving all needs for quantitative information on agriculture.
Standardized, periodic and longitudinal studies should now be
initiated of such critical issues as the volumes of fertilizer
use by region, market prices for basic foodstuffs, and overall
flows of several commodities.








23




Recommendation 20

The implementation of the CRIES system should be accelerated
by the addition of a large training component aimed at
appropriate interpretation of CRIES output. CRIES based
documents should be published and distributed widely, and
specifically aimed at an audience of technicians and policymakers
unfamiliar with specialized quantitative methodologies.










3. Findings


3.1 Major Evaluation Findings

Findings Concerning FSR/E

ADS-II has a number of accomplishments which make it a
worthwhile and, in part, successful project. It has established
a large-scale on-farm testing program which has already scored
two major successes. But ADS-II has been less successful in
establishing an effective interdisciplinary model and methodology
of FSR/E.

ADS-II achieved a major accomplishment by moving from a
reconnaissance survey and its verification to establishing
on-farm trials in about a three month period. The extensive
on-farm testing program covered 293 researcher-managed and 285
farmer-managed trial replications in 1985. Two major successes
have emerged from this testing program, Tamazulapa beans and
Amina rice, as well as the extension opportunity offered by these
many farmer contacts.

The ADS-II project contains a flaw of either design or early
implementation which has seriously impeded the farming systems
component of the project. Socio-economists were physically and
organizationally separated from the farming systems component,
because they were lumped administratively and functionally
together in one unit called the Economie Rurale et Statistique
Section (ERS). This resulted in a situation where the farming
systems teams were not interdisciplinary and had little or no
socio-economic input. They were concentrating entirely on
agronomic testing even though the farming systems within zones
had not been fully characterized and little progress was being
made in the identification of farmer problems and constraints.
Structural changes which fully integrate socio-economists into
the farming systems component must be made. No geographical
expansion or substantive expansion into extension should be
considered until these shortcomings are resolved and a more
complete and effective farming systems research program is
established.

Activities in both Jacmel and Les Cayes show promise of
producing results. Jacmel has devoted a major effort to working
in the hillside environment of Haute Cap Rouge. Erosion control
activities in that area show promise, even though the technique
is different thanthose that will have to be used in many areas of
the country. The possibility of centralizing activities and
attempting to manage a small watershed in a fairly complete
manner is interesting. Les Cayes activities are interesting for
different reasons, but particularly because the region plays an
important role in future AID activities. Although less advanced
in their hillside activities, the rice activity on the plain
promises to have an important impact on agricultural production.












All surveys used by the project have been based on formal
questionnaires. With the exception of the initial reconnaissance
survey, they have all focused on quantitative data. Techniques
like informal surveys and diagnostic surveys, case studies, or
anthropological studies which would provide qualitative
information with the possibility of achieving a deeper
understanding of the farming systems) have been neglected.

The project organization dictates that the basic farming
systems teams are found at the zone level (Haut Cap Rouge, Bas
Cap Rouge, Berault and Maniche). Interdisciplinary integration
should be promoted precisely at the zone level. Basic teams
should be provided with the necessary training and guidance to
design, implement and analyze their own agronomic tests as well
as diagnostic and monitoring surveys. This implies that
technical assistance in particular, and the project in general,
should take on a much stronger training function.

ADS-II has not established an effective training program for
FSR/E. Zone level project staff, who should be the basic farming
systems team, and students primarily receive experience in the
management and execution of on-farm trials. On a personal basis
they may ask questions and receive instruction from the zone
chief or the regional coordinator. They do not receive any
formal training. They are not trained to design, analyze or
interpret test results. They receive no training in FSR/E
methodology and they do not get any experience working in an
interdisciplinary environment. They do not participate in and
are not trained to do reconnaissance or diagnostic surveys.
Except for the study on pigs, they are not involved in farm
monitoring activities. Their understanding of FSR/E is based on
their training at the University and their experience working in
other projects which use the Recherche Developpement approach.

The on-farm research program has made effective use of
research results, technical advice and improved plant materials
from both Haitian activities (Salagnac, La Vallee, ODN, ODVA,
Levy and Damien Farms) and the International Research Centers
(IRRI, CIAT, CIMMYT, ICRISAT). Research results have been
communicated to other researchers through individual contacts and
through the presentation of professional papers (KSU Farming
Systems Symposium, Caribbean Food Crops Society Meeting).

The agronomic testing program needs to do much more analysis
and interpretation of results and pay more attention to isolating
factors which affect performance. This includes improving the
characterization of test sites and target farmers, as well as
designing tests so that the effect of individual factors is
clearly demonstrated. Farmer evaluations should be
systematically solicited on all tests and made an integral part
of the write-up of each test. Test designs do not always
correspond to the stated objective and the analyses of some tests
do not appear to be appropriate for the design actually used.











Research results from previous experiments are not received
rapidly enough to be used effectively in planning future
experiments.

Many of the farmer managed tests are not pre-extension
trials in the sense of approximating farmers conditions.
Fertilizer is used on most trials even though survey results show
that most small farmers do not use fertilizer. These trials
provide little insight into how farmers will adapt a technology
when using it on their own, or what the long term impact of the
innovation might be. Farmer managed trials are often used to
compare a complete technical package to farmers traditional
practices. Farmers typically do not adopt entire packages at one
time, especially if they include an expensive input like
fertilizer. Such tech pack demonstrations do not show the
farmers what individual factors might improve their production
and be within their means.

Haiti lacks national institutions doing effective
agricultural research or effective agricultural extension. This
places a farming systems project like ADS-II in a difficult
situation. Establishing linkages to the decentralized research
and extension systems require even more effort than linking to
centralized institutions. An informal network of research and
extension projects/personnel does exist as an outgrowth of the
Faculty. The project need to improve linkages to the FAculty and
this informal network. Haitians who are trained by the project
and return to teach at the University will provide the greatest
impact on this group. Training activities and project
participation for students, recent graduates, research
specialists and extension personnel can also be used to help
improve this relationship once the project has achieved
credibility. Projects and PVO's are an important part of the
decentralized extension system. Such organizations can provide
extension outreach of technologies tested by the project if more
effort is devoted to establishing the necessary liaison.

ADS-II is presently attached to the MARDNR, but is not
integrated into its functional structure. Politically motivated
changes in Ministry personnel led to a serious conflict in
personalities which required that the project be moved out of the
Faculty. If the project is structurally integrated into the
Ministry, political changes might again threaten its function
capacity. Although not ideal, the practical solution to the
projects' institutional location seems to be to maintain the
present relationship until the new Ministry has proven its
ability to function effectively and the political situation is
stable.

The ADS-II interpretation of FSR/E has been flawed and
incomplete in several respects, particularly with regard to
institutional problems. It has not as yet begun to study rural
institutions and their effects on adoption of innovations.
Innovations have not been monitored after farmers had begun to











use them on their own, to see what changes take place in the
farming system as a whole. It is not clear that the researchers
are sufficiently aware of labor and resource constraints to
include such considerations in their selection of technologies.
The importance of different innovations and the size of the
target group to which they apply have not been specifically
evaluated.

The definition of extension has also been too narrow and
needs to be expanded to include the promotion of rural
institutions (organizations) which facilitate the adoption of
innovations by farmers. Although the scope of this project does
not include the possibility of a large extension program, it can
contribute to a better understanding of rural institutions and
how extension might involve them in the transfer of technology
process. The ability to identify and analyze this potential of
rural institutions needs to be built into extension training
programs.

The technical assistance team leader has three nearly full
time functions in administration, farming systems socio-economic
research, and the national survey and CRIES activities. He
cannot fill all these functions. The addition of an
administrative assistant would allow him to fulfill more of his
technical duties.


Findings Concerning CRIES


CRIES has great promise for many types of planning and
analysis and for many different clients in Haiti. ADS-II is
performing an excellent service and is at the point of starting
to provide exactly what various organizations need. The project
has the resources and the competency necessary to provide this
service rapidly. Comparison with similar efforts elsewhere shows
just how useful CRIES is. The University of Maine Agroforestry
Outreach Research Project has just published a working paper on
"Ecological Zones and Erodibility in Haiti" which represents
about five months of work and includes many statistical tables.
ADS-II using CRIES has just completed a similar research effort
which took only one month to do. More importantly, the CRIES
data is permanently stored in digitized form and can be used in
any future update or research effort while the University of
Maine study is essentially "frozen in time". No matter what the
merits of analysis and interpretation on either side, the CRIES
system is vastly preferable over traditional tools in research of
this type.

The next set of variables to be put into the CRIES data base
will include altitude curves. Considering the enormous amount of
work which this particular variable will constitute it would
appear advisable to concentrate on a few watersheds and/or












administrative divisions of the country at this time. CRIES
products will always be useful to the extent that they are aimed
at specific problems and specific spacial units, and it would
appear a luxury to devote a very major part of the CRIES
resources to the altitude curves alone. CRIES output for the
entire land area of Haiti is much less needed and less useful
than output for specific project areas.

The ultimate hope in the installation of CRIES at ADS-II is
to produce yield models for specific crops. Given the lack of
relevant data for many parts of Haiti, only the most simplistic
and therefore the least useful, yield models can be attempted at
this time. It would be advisable to stay away from this type of
endeavor as long as more precise information on specific yields
in specific areas and under specific conditions cannot be fed
into the data base.

The interpretation of CRIES output is enormously difficult.
There is a built-in contradiction in the system which is hard to
overcome yet must be taken into account at all times. Most of
the data is fed into the base in aggregate form and is hence not
location-specific, while most of the questions one will ask, and
want to see answered by the CRIES system, are precisely
location-specific. Analysts will have to exercise great care if
they want to avoid turning the whole CRIES system into a gigantic
exercise in fiction. Only continuous updating, upgrading and
external as well as internal verification of the variables will
help improve the system in the medium term and make it from a
great promise into a useful planning tool.

Given the present dependence of the CRIES operation upon
technical assistance staff and given the current state of
training of Haitian team members it is inconceivable that CRIES
could be used within the next three years by Haitian government
services without a leading role being given to TA staff members.
The institutional location of CRIES within ADS-II or the Natural
Resources directorate is much less important than the level of TA
support which the system will receive over the next few years.
The fact that CRIES is at the present time well beyond the
capability of what government services can manage does not speak
against it; it highlights the training needs within government
services and Haitian organizations involved in planning
agricultural development.

CRIES is a tool to keep a national resource inventory in a
practical form. At no time should CRIES be taken to be an
expression of research activity or a substitute for
socio-economic and agronomic research. Even if the results of
such research are stored in the CRIES data base, they are in all
likelihood subject to inappropriate generalization and serious
misinterpretation. It is doubtful whether ADS-II is the optimal











location for CRIES but appropriate alternatives do not easily
come to mind. It may be best to leave CRIES in the hands of
ADS-II for the moment while at the same time opening up access to
CRIES output to many organizations and at many different levels.


Findings Concerning The National Survey of Agriculture


The survey of the Departement du Sud used a good sample and
an adequate questionnaire. In these two respects the team has
proven itself to be up to the task at hand and has performed
well.

The fact that the sampling and questionnaire aspects of the
survey were done well is in part due to a very high level of
technical assistance received by the team. In terms of the
organization, supervision and logistics of carrying out the field
research, the team was not fully up to the task. Maybe nobody
could have carried out this kind of a survey while a revolution
was going on at the same time, and this criticism should not be
misunderstood as an accusation of the team. Quite to the
contrary, the team showed a lot of tenacity and courage when it
continued survey operations under adverse conditions.

Many rules of standard survey procedure were violated and
this must have been translated into poor data quality for many of
the interviews. In the absence of comparative studies,
qualitative in-depth exploration of some of the topics in the
survey area and formal verification procedures, it cannot be
assessed which part of the data collection is of high quality and
which part is inadequate. The resulting uncertainty casts doubt
about the usefulness of this data collection for any analytical
purposes.

The team will be ready to take on another department as the
next step in the national survey only after having evaluated both
the data and the survey procedures in great detail. A rush into
new data collection at the present time will only lead to the
collection of bad data.

The institutional capability of the Ministry to conduct such
survey work, even with the help of massive and competent
technical assistance, has not been proven yet.


3.2 Linkage between project and mission strategy objectives

In the most general terms the AID mission strategy of the
recent past has been to strive for (1) greater protection of
fragile lands by (2) promoting integrated agricultural and
infrastructure work in entire watersheds while also aiming at (3)












higher levels of income, nutrition and health of the farm
population. This represents a move away from a strategy aimed
mainly at plains agriculture, without completely abandoning it.

ADS-II generally falls in line with each of these three
mission strategy objectives without overlapping with them
completely. For each case the project makes a contribution but
it remains to be seen how lasting this contribution will turn out
to be since the institutional anchoring of the project is at the
moment still problematic.

The protection of fragile lands has now become a high
priority in the Haut Cap Rouge zone. The project has already
built over 20 km of dry walls and has after initial difficulties
found a way of paying for this infrastructure work without doing
damage to other project objectives. (There is no more community
conflict over this issue since in effect the project is paying
individual farmers to have the walls built using local labor at
going daily wages). In coming years the project will be doing
even better in this respect since the team has decided, based on
past experience, to organize a multiplication plot for the Napier
grass needed to protect the dry walls.

The second part of the mission strategy which aims at the
upgrading of entire watersheds is also well served by ADS-II. In
the Jacmel Zone the project has started to protect an entire
secondary watershed by covering all land from top (Haut Cap
Rouge) to bottom (Bas Cap Rouge) with a combination of drywalls,
grass strips and tree rows. The plan is ambitious but feasable,
and ADS-II intends to replicate it in the Maniche Zone. Logistic
problems will be far greater there but can be overcome. This
kind of clear geographical focus will be good for a team which
has had a tendency to spread itself too thinly over too vast an
area; besides, it falls in line with AID mission objectives.

ADS-II has had significant successes in terms of achieving
potentially much higher levels of farm productivity. The rice
and bean varieties successfully introduced and adapted in Cayes
have now "taken off" to some extent. The third of the Mission
objectives is thus being served. It has to be stressed, however,
that ADS-II is much more a research than an extension project and
that its success in crop experiments can not lead to direct
impact on nutrition and income levels. Such effects are
contingent upon an adequate extension service and the only sure
thing in this respect is that ADS-II cannot itself become an
extension agency. If extension is taken care of in a sensible
way, however, the FSR approach has a much greater chance of
leading to permanent improvements in farming practices than the
traditional "vulgarization", "model farm", and "integrated
development" approaches which have left so many sad reminders all
over the Haitian landscape.








31


In one respect ADS-II departs significantly from the
missions recent strategy: the project has virtually no contact
or collaborative arrangement with PVO's. While AID has had
considerable success channeling trees and swine through PVO's
ADS-II has tried to work through the Ministry and the Faculty of
Agronomy. This aspect of the work has clearly not gone very well
and it will be some time before any significant upgrading of the
government's institutional capability will be observed.














4. MAJOR LESSONS LEARNED

1. Farming systems research is not effective unless
integrated interdisciplinary teams are formed. The attempt to
have a separate, city-based socio-economic survey component
complement on-farm testing by agronomists was fundamentally
flawed and has largely failed. It did not permit the critical
amount of interdisciplinary interaction.

2. The choice of four different zones in two geographically
distant regions made it very difficult to establish
interdisciplinary local farming systems teams. It would have
been more effective to begin work in a single region in order to
establish a methodology and model of FSR/E before expanding
geographically.

3. When a project does not include a full time
administrative position on the technical assistance team, the
team leader cannot devote adequate time to technical tasks
because he is constrained by administrative imperatives.

4. In order to be useful, quantitative survey work has to
follow its own rules, profit from specialized expertise and lead
to a particular mode of analytical work. It is rare to find a
team which combines the ethnographic, socio-economic and
agronomic expertise, and the style of work necessary for
successful FSR/E, with the talent and style typical of good
quantitative work. ADS-II has let the quantitative work dominate
most of its socio-economic analysis, to the neglect of the type
of analysis needed in FSR/E. Questionnaire data of uneven
quality has been substituted for a comprehensive analysis of the
farm system under study. This was the outcome of building two
entirely different tasks into one project. Such a tour de force
should not be repeated.

5. It is always tempting to give a project its "correct"
institutional location. This should be done at the beginning,
and changed only is serious reasons such as major, lasting
personality conflicts, demand it. If a project works out
satisfactorily, an apparently "wrong" institutional anchoring in
the government should not be considered a sufficient reason to
more it around.

6. Just "involving" individuals in short training missions
abroad and similarly attractive activities, will do nothing to
guarantee their cooperation or even interest in a project upon
their return. Training opportunities should be directly linked
to project participation.







33




7. If one intends to introduce a radically new technology,
such as a modern survey or CRIES, one must be prepared to furnish
all the necessary training, resources and management for the
almost indefinite future.

8. A project must have independent control of its budget
and sufficient freedom of action for all critical aspects of its
operation. This is particularly important in bilateral aid
projects working with government ministries. Stated project
goals may well be irrelevant, compared with the hidden survival
and promotion agenda of the civil servant.












5. PROJECT DESCRIPTION

5.1 THE DEVELOPMENT PROBLEM

Like many other development projects, ADS-II started out
without any thorough analysis of the development problem which it
was supposed to solve. None of the four project documents
provides a definition of development or of the development
problem; nor do we find any definitions of under-development, a
concept of which Haiti is a typical expression. The reader has
to reconstruct the underlying development problem from various
parts of the four papers.

One thing on which all project documents agree is the
perception of a crisis which threatens both the Haitian ecosystem
and the population. There is a certain awareness of the vicious
cycle in which increasing poverty leads to more intensive
agriculture on ever steeper hillsides, a practice which then
through soil erosion and depleting soil fertility will in turn
lead to more poverty. This interrelation of an economic, an
agronomic and an environmental problem is addressed in various
degrees of detail, but there is a basic agreement in this respect
throughout the four documents.

Looking at the detail of the problems and solutions listed,
the reader will become aware, however, of two basically different
and almost contradictory approaches to development underlying the
project documents. The first of these approaches is represented
in the Project Paper, the Project Paper Supplement and the
Project Agreement. It is an approach which was very much en
vogue in the 1960's which was presumably the time when the
authors of these papers got their graduate degrees. The key
concept of this approach is modernization. Development is seen
as an evolution of a society from a traditional into a modern
stage, with all the institutional, psychological, political,
economic and social changes involved. Modernization presumes the
existence of a pre-modern status which must be abandoned, and a
modern status which is an ideal worth striving for. The Project
Paper Supplement uses the following quote from the "Five Year
Plan for the Agricultural Sector" of the Government of Haiti:
"(the mandate is) to facilitate the development of the rural
sector from its traditional stage to that of a modernized rural
sector participating fully in national development, and the
promotion of the rural man..." Modernization according to these
papers is seen as the diffusion of modern technology in the
traditional agricultural sector which so far appears to have been
left out of the "national development" happening elsewhere. All
of the goals and purposes fall in line with this approach, and
all of their key arguments make sense only once the underlying
premise of modernization is accepted. Farming Systems Research
is subtly re-interpreted as the introduction of modern crops and
we find much emphasis on the building of institutions capable of











spreading research results and extension services from the top of
the hierarchy to the bottom. The spider-like configuration of
the Ministry of Agriculture and its regional extension branches
lends itself very well to this view of the development problem
and its possible solutions.

The Arkansas/Winrock proposal stands apart from the three
other project documents in terms of its approach to development.
Its thinking is more typical of some of the theories of
underdevelopment of the 1970's. Underdevelopment is not seen
here as a pre-modern status which is to be changed by drawing the
countryside more closely into the urban-dominated national
development. Rather it is understood as the result of factors
which may actually be part of the process of drawing rural areas
into the orbit of powerful cities and urban elites. In this
paper we read of class structure and pronounced economic
stratification, urban middle class values and the many
expressions of poverty of rural Haiti. Even the low level of
agricultural technology is presented in a different light, when
the paper talks of the "persistence of what many regard as
inefficient, traditional cultural practices and resistance to new
technology", and then goes on to talk of the risks involved in
modern technology and the aversion of risk practiced by peasants.

The Arkansas/Winrock proposal not only has a different view
of the problems and processes of underdevelopment, but its
approach to solutions is also different. The focus here is on
detailed conceptualization of constraints in the small farm
environment and on the design of more productive agricultural
techniques responsive to these constraints and developed in
cooperation with farmers. Inherently aware of the pitfalls of a
centrally guided modernization policy, this paper advocates a
bottom-up process in which no modern technologies are a priory
superior to traditional ones unless they are found acceptable
within the constraints of rural society.

The case should not be overstated. The four project
documents have many things in common, not the least because
ADS-II was designed to be a bilateral project from the very
beginning. This implied a heavy emphasis on building up the
institutional capabilities of the Haitian government.












5.2 Project Goal and Purpose

5.2.1. Project Paper (May 1978)

Goal: To improve the nutritional and socio-economic status of
the rural population.

Purpose: Develop in DARNDR the initial institutional capability
to, on a timely and sustained basis within the limits of applied
science, provide Haitian institutions serving farmers and
consumers of farm products with:

a. Reliable statistical data
b. Reliable descriptions of rural economic and social
systems and phenomena
c. Supply of optimal genetic material and appropriate
resources to improve farm technology

5.2.2. Project Paper Supplement (August 1982)

The goal is to develop the institutional capability of
DARNDR to conduct and coordinate a national program of
agricultural research and statistics to support the country's
agricultural development program. The target group remains the
traditional small farmer who will be provided with a package of
appropriate technologies for increasing his farm productivity
levels while simultaneously being made aware of the importance of
environmental preservation.

5.2.3. Project Agreement (July 1983)

The goal of the project is to improve agricultural
production and the productivity level of the small Haitian
farmer. The purpose is to establish within the Ministry of
Agriculture, Natural Resources and Rural Development (DARNDR) the
institutional capacity to conduct a farming systems improvement
program through adaptive research and to develop a program of
agricultural economics and statistical analysis of sufficient
volume and reliability likely to support the country's
agricultural development and increase farm production and income.

5.2.4. Arkansas & Winrock Proposal (Submitted April 1983)

The goal of the project is to assist the Haitian Ministry of
Agriculture, Natural resources, and Rural Development (DARNDR) in
strengthening its operational and administrative capacity to
conduct and coordinate a national program of farming systems
development and agricultural information collection, analysis and
dissemination. The target group is the traditional small farmer
and appropriate technology will be designed to increase his
productivity and assist him in contributing to environmental
preservation.








37


Objectives:

1. Introduce the farming systems approach and develop the
institutional capability to carry out farming systems
research resulting in alternative, adoptable, and
economically viable production technology beneficial to
the small farmer.

2. Develop institutional capability to collect, analyze,
and utilize policy relevant agricultural production,
marketing and socio-economic information about small
farmers.

3. Train Haitian agricultural technicians, statisticians,
data analysts, enumerators, and farmers to continue to
carry out the above objectives after the project
terminates.













5.3. Project Inputs and Outputs

5.3.1. Project Inputs


Project Paper


Long term TA
Ag Research
Statistics

Short term TA
Ag Research
Statistics

Degree Training
Ag Research

Short term Training
Ag Research
Statistics

Commodities


Construction


Total USAID Contribution


252 PM
36 PM


54 PM
12 PM


(Person Months)


360 PM


90 PM
61 PM

$642,000

$158,000

$4,047,000


5.3.1.2. Project Paper Supplement


Long Term TA
Farming Systems
Rural Econ and Ag.

Short Term TA
Farming Systems
Rural Econ and Ag.

Degree Training
Farming Systems
Rural Econ and Ag.

Short Term Training
Farming Systems
Rural Econ and Ag.


Workshops
Farming Systems
Rural Econ and Ag.


Stats


96 PM
90 PM


12 PM
8 PM


96 PM
72 PM


Stats


Stats


(Person Months)


Participants
to


Stats


Stats


5.3.1.1.












Interns
Farming Systems 96 "
Rural Econ and Ag. Stats 48 "

Extension Training 260 "
(1 week per agent)

Farmer Training/Field Days 14,400 "
(1 day per farmer)

Commodities $821,000

Construction $20,000

Total USAID Contribution $3,808,000


5.3.1.3. Project Agreement

Same inputs as the Project Paper Supplement


5.3.1.4. Arkansas & Winrock Proposal

1. Provision of expatriate staff skilled in Farming
Systems Research and Development and in agriculture
information systems development.
Long Term TA 192 PM

2. Provision of funds for program operation, acquisition
of facilities and equipment.
Commodities $821,000

Construction $20,000


3. Training experiences relating to FSR and development
and information systems development.
4 Farming Systems Research Specialists (US training)
12 Agricultural Extension Agents trained in FSR/E
concepts
2 Agricultural Statistician/Information
Administrators (US training)
2 Computer Operators (US training)
4 Survey Analyst/Supervisors
12 Enumerator Supervisors
48 Enumerators
12 Agricultural Student Interns


Total USAID Contribution


$3,808,000












5.3.2. Project Outputs

5.3.2.1. Project Paper

1. Provisional Tech-Packs all crops
1st round improved Tech-Packs 5
2nd round improved Tech-Packs 5
3rd round improved Tech-Packs 1

2. Research Facilities
Hinche Papaye
Damien improvements

3. Trained Personnel
15 ag. researchers with degrees
15 ag. researchers with specific short term training
2 statisticians sampling frame
2 statisticians field sampling
3 statisticians data processing

4. Sample Frame
Aerial photography
Statistically valid sampling frame (year 1)

5. Reorganizational plan for SERA
(Reorganized under CRDA)

6. Research programs organized with coordinators
a. maize, sorghum, grain, legumes, tropical
horticulture, sugar cane, soil conservation,
forestry
b. vegetables
c. livestock, forage management

7. Research Advisory Councils
a. regional
b. national

8. Social systems descriptions for 7 watersheds

9. Cost of production and analysis for tech-packs

10. Baseline and series surveys
a. Area and yields of major crops by land class and
watershed
b. Production, consumption and shipments of major
commodities by watershed
c. Livestock numbers and out-shipment by species and
watershed
d. Prices received by farmers by commodity and point
of sale
e. extent of soil erosion by land class













5.3.2.2. Project Paper Supplement

1. Farming Systems Component
a. Baseline data will have been obtained on the agro-
socio-economics of the target communities.
b. Improved and economically viable alternative farm
production systems will have been developed for
each of the four target communities.
c. Results of 1,200 on-farm co-operative and
verification trials will have been analyzed,
interpreted and reported during the life of
project.
d. Training programs in farm production systems will
have been formulated, tested and implemented for
recent graduates of FAMV who are about to enter
the Research and Extension Services of DARNDR.
e. Viable operative units will have been created to
duplicate the strategy for developing improved
farm production systems in other regions of Haiti.
f. Farmers training programs will have been developed
and executed for facilitating adoption of the
improved technology. It is estimated that 14,400
farmers will have attended farmers field days.
g. Farmers organizations and government services
(credit, storage facilities etc.) will have been
created to institutionalize the improved farm
production systems and environment protection
techniques in the target areas.
h. Approximately 100 FAMV students will have obtained
field training and experience in the areas of farm
management, farming systems research and extension
methods.
i. Approximately 40 technicians of DARNDR will have
received long and/or short-term specific training
overseas in farming systems and rationalized land
use patterns.
j. Additionally, clear evidence of enhanced
productivity and improved land use by the target
communities should be seen by the fourth and final
year of the project.

2. Statistics Component Outputs

a. A cadre of trained DARNDR technicians in the areas
of (i) statistics; (ii) gathering of agricultural
information; and (iii) data processing, analyses
and interpretation.
b. A well-equipped and fully functional data
processing center.












c. Standardized procedures for carrying out nation-
wide surveys and for gathering information
relating to (i) crop production costs; (ii) crop
yields; (iii) variations in farm product prices
and (iv) agro-socio-economic indicators.
d. Standardized procedures for the dissemination of
relevant information (e.g., price elasticity of
demand, disease and pest control practices, newly
released crop varieties, planting dates, etc.) to
rural communities throughout Haiti.
e. A significantly improved DARNDR capacity to: (i)
identify constraints to enhanced agricultural
production and productivity for each agricultural
district of the country; (ii) prioritize and
implement viable solutions to the identified
constraints; and (iii) furnish the necessary
information for the preparation of agricultural
development and marketing projects for the rural
areas.

5.3.2.3 Project Agreement

1. Farming systems component

a. Development of an improved and economically viable
farm production wywtem for the target communities.
b. Data collection and interpretation of a total of
1,200 on-farm co-operative and verification
trials.
c. Preparation of a training manual in farm
production systems for recent graduates of FAMV.
d. Establishment of functional operative units having
the capability of duplicating the strategy for
developing improved farm production.
e. Development and implementation of a farmer
training program for technology transfer. In this
context, an estimated 14,400 farmers will have
participated in field-day exercises.
f. Creation of appropriate farmer organizations and
government services (credit, farms inputs,
marketing etc.) to institutionalize the improved
production systems and environmental protection
techniques.
g. Implementation of a training program for one
hundred agricultural residents in farm management,
farming systems research, and technology transfer.
h. Implementation of appropriate short and long term
overseas training programs for forty (40)
technicians in farming systems and rationalized
land use patterns.
i. Enhanced production and productivity levels
through the use of crop mixes and soil management.












2. Rural economics and agricultural statistics component

Using the sample frame, surveys will be conduced to
obtain inter alia answers to the following:

a. acreage/hectarage and yield of major crops by
target communities and regions;
b. production, productivity, consumption and shipment
of major commodities by region.
c. Livestock numbers by species and region;
d. Prices received by farmers by commodity by season;
e. Volume of pre- and post-harvest losses, (condition
of product storage) and post-natal losses (for
livestock);
f. Wood consumption for fuel, construction and its
nature and source;
g. Quantities of sea foods sold, consumed, stored, or
lost;
h. Farm credit needs and problems associated with
obtaining credit;
i. Total number of days/year worked on someone else's
farm and other sources of employment and revenue.
J. Total number of days/year worked on someone else's
farm and other sources of employment and revenue;
k. Price elasticity through offer and demand;
1. Types, quantities and costs of farm production
inputs used (fertilizers, seeds, pesticides,
implements, etc.);and
m. The Role of cooperatives in promoting production
and income levels; and the types of farmer groups
or organizations;

As a result of the successful implementation of this component
DARNDR will have improved significantly its capacity to:

I. identify constraints to enhanced agricultural
production and productivity;
II. prioritize and implement viable solutions to the
identified constraints, and
III. furnish the necessary information for the
preparation of agricultural development and
marketing projects for the rural areas.

5.3.2.4 Arkansas & Winrock Proposal
1. Documentation
a. Characterization of agricultural information
systems
b. Characterization of existing farming systems
c. Descriptive studies of regional and community
systems (social, economic, physical)
d. Detailed description of selected farming systems
and cropping systems












2. Agricultural Information Collection Instruments and
Procedures
a. Survey instruments for use in routine collection
of agricultural production and marketing data
b. Detailed procedures for carrying out farming
systems research
c. Sampling frames and sampling procedures for the
collection of agricultural production and
marketing information

3. Farming Systems Research Instruments and Procedures
a. Farming systems characterization, including:
1) Initial rapid appraisal
2) Constraints surveys
3) On-farm testing of technology

4. Agricultural Technology
a. Alternative farm technology appropriate for the
Haitian small farmer will be designed tested and
evaluated, and, if appropriate, made available for
dissemination.
b. The appropriate agricultural technology developed
by the project will be adopted by small farmers,
with consequent improvements in productivity.
5. Trained Agricultural Technicians, Survey Analysts,
Enumerators, and Farmers
a. Participating agricultural graduates will be
thoroughly trained in farming systems research.
b. Survey analysts will be trained in the design,
implementation, analysis, and reporting of
agricultural and socioeconomic surveys.
c. A cadre of field enumerators will be trained for
taking agricultural and socioeconomic surveys.
d. Farmers will be introduced to scientific problem-
solving techniques appropriate to farm production
and management.
6. Agricultural Production and Marketing Information
a. Periodic information about marketing of selected
agricultural products, including information about
quantities and prices.
b. Periodic information about selected production
enterprises, e.g. extent of plantings of selected
cash crops, yields, planting intentions, cultural
practices used, inventories of livestock, etc.
c. Periodic information about major land uses.
d. Periodic information about selected socio-economic
characteristics of farmers and farm families (net
income by sources, tenure status, indebtedness and
credit sources used, labor availability and use,
etc.








45


7. Institutions and Institutional Capability
a. An organization capable of carrying out on-going
farming systems research and extension will be
established.
b. An organization which can collect, analyze, and
disseminate selected agricultural information on a
regular and timely basis will be established.












5.4 Project Assumptions

5.4.1. Project Paper
1. Annual population growth can be held to 2.2%.
2. Improved agricultural research capability will result
in increased agricultural output.
3. Long-term U.S. involvement.
4. Adequacy in number of contract technical service
staff.
5. Timely availability of PL 480 counterpart.
6. Successful implementation of project 078 (Integrated
Agricultural Development)


5.4.2. Project Paper Supplement
Assumptions were not specified.

5.4.3. Project Agreement
Assumptions not specified.

5.4.4. Arkansas and Winrock Proposal
1. Design of a national information gathering, processing and
dissemination system will be an output of the project.
2. Continued cooperation of Haitian agriculture agencies.
3. Funding of FSR and agriculture information system is
continued after project temination.
4. Potential availability of superior agriculture technology.
5. No serious disruption of the Haitian agriculture sector or
relevant agriculture agencies.
6. Continued FSR funding by GOH.
7. Continued GOH funding for agriculture information system.
8. Agricultural information is actually desired by potential
users (e.g. GOH agencies).
9. For U.S. training, logistical arrangements not disrupted by
visa problems, etc.
10. Persons can be identified with requisite background and
willingness to participate.
11. Potential availability of alternative small farm technology.
12. Appropriate location for FSR established within DARDNR.
13. Appropriate location for agricultural information agency
must be found.













5.5 ANALYSIS OF PROBLEM AREAS

The two basic problems underlying and justifying all of ADS-
II's activities are the low productivity of agriculture and the
land erosion deriving from current farming practices. In this
the four project documents and the implementation of the project
all agree. They also agree in the three basic strategies to
combat these two interrelated basic problems, namely (1) FSR/E,
(2) the establishment of an agricultural information base, and
(3) the strengthening of institutions capable of carrying out
FSR/E, keeping the data base up to date and analyzing its
contents.

We find considerable differences when we assess the
priorities assigned to the three strategies. The following table
summarizes these differences by assigning rankings (1= highest,
6= the lowest) to the three strategies, as far as could be
determined by looking at documents.


Rank order of priorities
implicitly assigned to three strategies, by project
document and current implementation

Documents Current
Implementation

PP,Supplement, Arkansas/
Amendment Winrock

1. FSR/E
a)crop research 4 2 2
b)rural economy 5 1 5

2. Ag. info. base
a)project specific 3 3 4
b)national 2 5 1

3. Institutions
a)governmental 1 4 3
b)peasant organize. 6 6 6




The numbers in this table refer to a rank order. A high
number does not necessarily mean that no significance is given to
a given strategy, but only that other items are considered as
ranking higher on the list of priorities. The numbers cannot be
more than the impressions and estimates of the evaluator.











Nowhere does a document or the team state explicitly which items
are more important than which other items. Some documents
contain long lists of expected outputs without realistically
assigning priorities.

Looking at the table we find basic agreement on two line
items. It is clear that ADS-II assigns very low priority to
peasant organizations. Even though the Arkansas/Winrock document
appears to realize their importance, they are not in any way
considered critical vehicles for bringing change to Haitian
agriculture. ADS-II is just not that kind of project.

There is also virtual agreement concerning the importance of
establishing a project-specific base of agricultural information.
While other things appear more important in all three columns of
the table, this item is always given a medium priority as
something not to be neglected at any rate.

On all other items there is significant disagreement in the
table. The ADS-II team clearly puts its energy and resources
into the national data base (both CRIES and the national survey)
and the crop research (mainly at Cayes). The Arkansas/Winrock
document is totally oriented towards farming systems research
with all other strategic items serving that cause. The
collaborative documents of AID and the Ministry privilege the
building of institutional capability and show neither
understanding of, nor great interest in what is considered FSR/E
elsewhere. All documents will pay lip service to most of the
items on list but if one has to determine which ones supersede
the others, the rank order summarized in this table is what one
will find.

The technical annexes provide considerable detail on how the
three strategies are used in approaching the basic problems of
Haitian development. Their results are summarized in the
"Findings" section of this report. In the most general terms-
and clearly without doing it justice- one can say that what the
ADS-II staff is implementing resembles the AID/DARNDR documents
much more than the Arkansas/Winrock proposal. For better or for
worse, the farming systems research program does not appear
highest on the ADS-II team's list of priorities, at least not in
terms of what they are currently implementing.

Finally, it deserves mentioning that the larger issues in
development which dominate the theoretical literature on the
topic have never surfaced in either the analytical or the
practical work of the project. The larger issues raised by
Lundahl, Murray, Locher, Anglade and many others are never
discussed. This is particularly unfortunate in terms of issues
concerning state and government. All of the ADS-II documents and
implementation assume that the Haitian government can and will
provide services when given the necessary resources. There is
virtually no support in the recent major works on Haiti for such
an assumption. The only point where project literature shows








49


awareness of a basic aspect of underdevelopment- rather than
partial description of some of its empirical expressions- is when
the Arkansas/Winrock proposal discusses risks and constraints in
peasant agriculture. In this point at least the ADS-II team also
has shown a sensitivity and awareness which can be translated
into appropriate action.












5.6 CURRENT STATE OF IMPLEMENTATION AND OUTPUTS

5.6.1 Documentation

5.6.1. Characterization of agricultural information systems

5.6.1.2. Characterization of existing farming systems

In April/May 1984 project researchers did a reconnaissance
survey in each of the four zones using a 51 page questionnaire.
This survey was conducted by interviewing small groups of farmers
and community leaders in each zone. The survey provides
information about some farm/farm family characteristics, major
crops and crop associations, livestock production and service
infrastructure within each zone. A formal quantitative survey
based on a two page questionnaire was conducted concurrently with
the reconnaissance survey. Since information from the two
surveys became available about the same time, priority seems to
have been given to the quantitative information. Only a limited
description and analysis of information from the reconnaissance
seems to be documented. The documentation which is provided
describes the zones but does not identify or other wise specify
what constitutes a farming system within the zone. It does
identify some of the factors which contribute to differing
farmers' circumstances within the zone, but without trying to
group farmers with similar constraints.

5.6.1.3. Descriptive studies of regional and community systems

The reconnaissance survey identifies elements of social and
service infrastructures in the communities investigated. These
include listing the schools, health facilities, women's groups,
cooperatives, etc. A general description of the agro-climatic
environment is also provided. Since the reconnaissance survey,
data has been collected on rainfall and some soil samples have
been analyzed. A formal survey has also quantified the portion
of farmers involved in different land tenure arrangements and the
amount of land cultivated by those farmers. As yet there is very
little documentation of social systems in a community, economic
systems in a community, or what constitutes a physical system.
No case studies of communities have been done and case studies of
farm/farm families have only begun in the last two months.

5.6.1.4. Detailed description of farming and cropping systems

Some documentation is available in the write up of the
reconnaissance survey on the cultural practices and crop calender
for different crops and crop combinations. Otherwise the whole
of each zone has been treated as a single farming system even,
though there are strong indications that farmer circumstances
vary greatly within a zone.











5.6.2 Agricultural Information Collection Instruments and
Procedures

5.6.2.1 Survey instruments for routine collection of
production and marketing data


Survey instruments have been prepared and used in at least
six different studies by ADS-II. Some of them are obviously the
results of previous experience (e.g. the questionnaire of the
Rapid Reconnaissance Survey) and models found elsewhere (e.g. the
forms of the National Agricultural Survey). Two instruments only
have been designed for "routine" use in the sense of repeated
data collection over time: the sheets concerning the swine
repopulation campaign and the prices of agricultural products in
market.

Generally speaking, problems with data collection at ADS-II
have not been the consequences of inadequate survey instruments,
with one exception. The sheets collecting data on crop trials
omit several variables critical for the evaluation of trial
success and failure, such as soil type, slope, aspect, etc..

5.6.2.2 Detailed procedures for out farming systems research

Technical Report #2 describes procedures for implementing a
farming systems research program. It describes how the
reconnaissance and verification surveys were done and continues
with a discussion of different types of trials and how the
testing process should lead to extension. The rest of the
document describes the data produced by the verification survey
and some of the analysis that has been done on it. The report
does not address the establishment of local farming systems
teams, interdisciplinary planning, implementation and evaluation
of activities, the need for farm monitoring and diagnostic
studies, the types of socio-economic information necessary, or
how to address the farm/farm family as a system.

5.6.2.3. Sampling frames and procedures for production and
marketing information

The area frame sample used in the national survey
departmentt du Sud) is generally excellent but costly. It
combines a random selection procedure with a one-stage
stratification and will produce, given the use of appropriate
weights, representative results. A good sample, however, will
not necessarily lead to good results if nonsampling error is
high.

Beyond the national survey, the team has established
sampling frames by way of a household census in the zones at
Jacmel and Cayes. They have been used as general background
rather than for systematic sampling in the collection of
production data. Much of the more detailed information on











production comes from the hundreds of farmer-managed crop trials
where the sample by definition is limited to cooperating farmers.
Therefore, much of the production data is based on a self-
selected sample.

Marketing information has been collected more or less
systematically for over a year, but the ERS unit considers only
the data for March and April 1986, as reliable. There is no
evidence that a sampling frame of sellers has been established in
the markets nor that markets have been selected for research in a
systematic way, but this will not make much of a difference
since, here again, nonsampling error will be much greater than
sampling error.

5.6.3. Farming Systems Research Instruments and Procedures

5.6.3.1. Farming systems characterization

5.6.3.1.1. Initial rapid appraisal

The reconnaissance and the verification surveys address
farming in each of the four zones but do not specify or
characterize what are the individual farming systems in the zone.
The reconnaissance survey was based on a 51 page formal
questionnaire rather than being an informal guided survey.
Interviews were conducted in small groups with no attempt to get
information from individual farm families. Each small group in a
zone was generally asked questions from different parts of the
questionnaire so there was little verification that even
different communities, not to mention individual farmers, might
give different responses. No attempt has been made to use
informal surveys since the original rapid appraisal, to perfect
procedures or design instruments (topics of inquiry) for use in
such surveys, or to train staff in their use.

5.6.3.1.2 Constraints surveys

Following the reconnaissance survey two formal surveys were
conducted which address certain constraints in the farming
systemss. One was a study of land area cultivated which
included an investigation of land tenure status and the relative
importance of different crops and crop combinations. The second
combined an inventory of livestock ownership with a study of
certain animal husbandry practices and constraints. Both were
done using formal questionnaires. It has been assumed that
farmers answered questions relating to land holdings and
livestock ownership truthfully, even though there are good
reasons to hypothesize that they would not. These surveys,
along with the reconnaissance survey, identify: constraints on
which more information is needed, a number of gaps in the
information available, and areas where the validity of
information available is unclear. They raise at least as many
questions as they answer, but little attempt has been made to
investigate these topics any further. Farm household case











studies have only recently begun to look at income, expenses,
enterprise budgets, etc. Specific studies to address questions
raised have not been done and procedures for doing specific in-
depth studies have not been established.

5.6.3.1.3. Farm Monitoring

The project has avoided becoming involved in farm monitoring
in the classic sense of attempting to collect a large amount of
data periodically using a cost-route or input-output methods.
But it has not identified an alternative either. Case studies
have not been used to provide in-depth information. In effect no
farm monitoring is being done except as applies directly to on-
farm trials or the reintroduction of hogs. Without some form of
monitoring the project has very little information about
farm/farm family characteristics and activities or about farmers'
circumstances and how they differ. It is important to avoid the
data trap of the large periodic surveys, but the project needs
much more information about what farmers do and how they live.
Some form of farm monitoring is required and effective procedures
which combine very limited periodic surveys with informal studies
need to be tested and adapted.

5.6.3.2 Screening alternative technology

Although some work is being done on a research farm, the
project is primarily using researcher managed trials on farmers
fields as the means to screen alternative technologies. They are
evaluated on the basis of statistical analyses of yield results
and some economic analysis is performed. The results from many
sites are pooled, demonstrating a range of yields but with little
attempt to determine the general characteristics about farmers or
about the sites that are associated with a particular yield
potential. Such procedures do not identify for whom and under
what conditions the technology is appropriate.

5.6.3.3 On-farm testing of technology

The on farm testing program in ADS-II consists of both
researcher managed and farmer managed (pre-extension) trials.
More complex trials (split-plot, randomized complete block) and
varietal screening are kept under researcher management, while
farmers are given more control over trials of simpler design.
Some of these, particularly the simpler trials, are not designed
to isolate the effect of individual factors and their
interactions. Better characterization of sites including:
irrigation status, degree of slope (steep, moderate, slight),
basic soil types recognized by farmers, depth of soil, slope
orientation, type of soil preparation, number of weedings, and
crop cultivated and fertilizer used the previous season are
necessary to permit disaggregation of analysis to isolate
individual factor effects. Analysis also needs to be done at the
local level by the staff managing the tests who have the
knowledge of sites necessary to disaggregate. Analysis is based











on pooled data across all repetition cites in a zone and reported
as a series of statistics (F tests, LV, least significant
difference). There is little attempt to interpret the results
and their implications for future adaptive research. Much more
analysis and interpretation of agronomic results is necessary.
Much more effort to incorporate non-agronomic factors including
economic analysis and farmers' evaluation criteria also need to
be included in the process.

Farmer managed trials often compare an improved technical
package with traditional practices. This does not allow
isolating individual factors to get research results nor
demonstrate to farmers the effect of adopting any element of the
package will produce. They are also done using fertilizer which
most farmers do not use and cannot afford. They are not pre-
extension trials in the sense of approximating the manner in
which farmers will use and adopt the technology, and provide
little indication of the technologies long term impact.

5.6.4. Agricultural Technology

5.6.4.1 Alternative farm technology appropriate for the Haitian
small farmer will be designed tested and evaluated,
and, if appropriate, made available for dissemination.

ADS-II has been able to take advantage of varietal testing
by other projects in Haiti as well as promising plant materials
from International Agricultural Research Centers. Much of the
project's research activities have focused on varieties as the
most likely way to rapidly increase agricultural production of
small farmers Tamazulapa beans and Amina rice are successful
results of this process. But with many of the promising
varieties now tested, the focus needs to shift towards more
complex systems related problems and constraints.

Most varieties or techniques are being tested with
fertilizer even though most farmers do not use it. Technologies
also need to be designed for those farmers who can not afford
expensive inputs.

Much of the hillside activity to date has been limited by
the need to have farmers construct erosion control measures and
the difficulty in providing vegetative materials to help
stabilize the canals or dry walls. Once the vegetation is
available more work can be done on combinations of crops and use
of the hillsides in a manner which both improves production and
controls erosion. Tree crops may provide a useful contribution
but may not be appropriate if the farmer does not own the land.
Other sources of vegetation need to be explored including use of
sugar cane, coffee, and pineapples and even vetiver.












Rice is an important crop on the irrigated plains and the
project has correctly identified the possibility to have an
important and rapid impact on that area through improved rice
technology. This includes work on a simple thrasher and fanning
mill as well as varietal testing.

5.6.4.2 The appropriate agricultural technology developed by
the project will be adopted by small farmers, with
consequent improvements in productivity.

The project has tested two varieties, Tamazulapa beans and
Amina rice, which are popular with farmers in the communities
where the project operates and which have begun to spread outside
those communities. Thus the project can already claim that these
two varieties are being adopted. Monitoring has not been
conducted to determine the extent of adoption, or whether
adopters have different circumstances or characteristics from
non-adopters. One of the basic tenets of the farming systems
approach is that whether or not a technology is appropriate will
depend on farmers' circumstances. Very few technologies will
apply to all farmers in a zone. But the project has made little
progress determining how circumstances differ between groups of
farmers or how test results vary according to either physical
characteristics of the sites, or characteristics of the farm/farm
family. More work is need to determine to whom a technology is
appropriate.

5.6.5 Trained Agricultural Technicians, Survey Analysts,
Enumerators, and Farmers

5.6.5.1 Participating agricultural graduates will be thoroughly
trained in farming systems research.

Participating agricultural graduates receive experience in
the management and execution of on-farm trials. On a personal
basis they may ask questions and receive instruction from the
zone chief or the regional coordinator. They do not receive any
formal training. They are not trained to design, analyze or
interpret test results. They receive no training in FSR/E
methodology and they do not get any experience working in an
interdisciplinary environment. They do not participate in and
are not trained to do reconnaissance or diagnostic surveys.
Except for the study on pigs, they are not involved in farm
monitoring activities. Their understanding of FSR/E is based on
their training at the University and in other projects in Haiti
using the Recherche Developpement approach.

5.6.5.2 Survey analysts will be trained in the design,
implementation, analysis, and reporting of agricultural
and socioeconomic surveys.

Survey skills are generally acquired by a combination of
formal training and experience. Several ADS-II team members have










a considerable amount of experience but none of them has adequate
formal graduate training in survey research. As a consequence,
some skills concerning the mechanics of data collection are
transmitted but the analysis is generally weak and does not go
beyond the most rudimentary description of frequency
distributions and two variable tables. Above all, there is
little understanding for the importance of data quality rather
than quantity. Under these circumstances, the "training in the
design, implementation, analysis and reporting of agricultural
and socioeconomic surveys" which individuals have received
through ADS-II has been of very limited value. This stands in
stark contrast to the high level of statistical and computer
skills of the Port-au-Prince based expatriate staff.

5.6.5.3 A cadre of field enumerators will be trained for taking
agricultural and socioeconomic surveys.

Field enumerators have been trained mostly in informal
sessions, for data collection in various surveys. Judging from
the problems reported on the national survey (see Annex), their
training has not taken them past some of the mechanics of taking
down responses given by farmers. Their training is more adequate
for recording technical information, e.g., on feeding of pigs,
than for probing into the difficult issues of land tenure and
production figures.

It deserves mentioning that the only piece of survey
research which had analytical ambitions beyond the usual
descriptive level, did not make use of ADS-II own enumerators and
socio-economic monitors at all (Pierre, 1985). A US-trained
Haitian sociologist did all the interviews himself and recorded,
in addition to numbers, qualitative data of far greater
importance than anything published in other ADS-II documents.

5.6.5.4 Farmers will be introduced to scientific problem-
solving techniques appropriate to farm production and
management.

It is very difficult to tell what the farmers are learning
from the project. Numerous farmers have attended field day
activities and participated in trials. A number have adapted the
Tamazulapa bean or the Amina rice. It is not clear what they
might have learned from interacting with the project and its
personnel about solving their own problems.












5.6.6. Agricultural Production and Marketing Information

5.6.6.1 Periodic information about marketing of selected
agricultural products, including information about
quantities and prices.

Prices of agricultural products are systematically collected
in four markets in the Jacmel region every 2 weeks. Such data
has not been systematically collected in the Cayes region due to
various implementation problems. The methodology used to get
this information does not distinguish between buying and selling
prices. It also does not account for the traditional "gift"
which is added to the normal measure. A quantity of the product
is weighed to get a price per kilogram, but it is not certain how
accurate these prices are due to problems like those mentioned.
The data may be a good representation of relative prices between
agricultural products, but some verification should be done. No
attempt is made to measure or estimate the quantity of different
agricultural products flowing through a given market. Project
personnel seem to know the hierarchy of markets and merchants,
but no study of market channels for different commodities has
been attempted.

5.6.6.2 Periodic information about selected production
enterprises, e.g. extent of plantings of selected cash
crops, yields, planting intentions, cultural practices
used, inventories of livestock, etc.

Inventories of livestock ownership and area cultivated per
crop have been completed in each region and as part of the
testing of the national survey, however one must question whether
farmers gave truthful answers. Yields have been measured in a
specific survey as well as in the trials and their control.
Trials provide some information about the crop or association
which is being tested, but not necessarily under typical
circumstances. General cultural practices were revealed in the
reconnaissance survey and a few trials on cultural practices were
started in 1984 but largely discontinued. Rotations have not
been studied and without farm monitoring, little information is
available on the labor requirements and economic aspects of
different enterprises.

5.6.6.3 Periodic information about major land uses

The national survey presently being tested at the regional
level provides information about use of cultivated lands on a
very aggregated basis. Such information is not collected
periodically in the areas with farming systems interventions, nor
is it likely that the national survey with produce it on a
regular and periodic basis.












5.6.6.4 Periodic information about selected socio-economic
characteristics of farmers and farm families (net
income by sources, tenure status, indebtedness and
credit sources used, labor availability and use, etc.

The initial reconnaissance and verification surveys provide
some information on socio-economic characteristics of farmers and
farm families. The two studies on land area cultivated and
livestock provide some additional data. Only in the last two
months has the projected resumed collecting socio-economic data
after a year without doing so. It is not clear whether these
efforts will be established on a periodic basis or not.

5.6.7. Institutions and Institutional Capability

5.6.7.1 An organization capable of carrying out on-going
farming systems research and extension will be
established.

An organization has been established which carries out a
large number of on-field trials. Given the many problems
outlined in this evaluation, one can not yet say that it is doing
effective farming systems research and extension.

5.6.7.2 An organization which can collect, analyze, and
disseminate selected agricultural information on a
regular and timely basis will be established.

Although certain improvements need to be made, much progress
has been made towards the design and implementation of a national
agricultural survey. It still remains to be seen if the project
has the human resources to administer such a large undertaking
and to process the data on a timely basis, without depriving
other project components of much needed human resources. Even if
the collection of data can be assured, the quality of both data
and analysis is not impressive when compared with the quantity of
data collected.










Technical Annex 1

Farming Systems Research and Extension and its Application in
ADS-II

John Lichte

1. What is FSR/E
1.1 The role of FSR/E
1.2 Characteristics of FSR/E
1.3 The problem of technical packages
1.4 The basic steps of FSR/E

2. The application of FSR/E in ADS-II
2.1 Position of ADS-II with regard to external structures
2.2 Internal strucutre of ADS-II
2.3 Survey approach
2.4 Establishing an interdisciplinary FSR/E approach
2.5 On-farm testing
2.6 Other interventions
2.7 Diffusion
2.8 Training
2.9 Monitoring and evaluation
2.10 Replicability, sustainability and recurrent costs



1. What is FSR/E

1.1 The role of FSR/E

Farming systems research and extension is an
interdisciplinary approach to the generation and transfer of
farming technology.

A farming systems team typically fills a particular niche
within the process of agricultural technology development,
adaptation and diffusion. Ideally this process is a continuum.
The US Land Grant University system unites research and extension
within a single institutional structure in order to facilitate
the process by maintaining the institutional continuum.


Research
technology
development


Farming System
Research (on-farm
adaptive research)


Extension
(of technology)











The situation in many developing countries is quite
different. Agricultural research and agricultural extension are
often separate institutions and in some cases are even in
different ministries. Agricultural research is often limited
largely to activities on research stations, and researchers often
have little contact with farmers and the reality within which
they live. Therefore, agricultural researchers often do not
consider the constraints farmers face, or their objectives, and
much of their research is not appropriate to farmers. Extension
personnel typically have more contact with farmers and recognize
that research results are not appropriate. Thus, many of the
research results are not accepted by extension organizations for
dissemination to farmers.

A farming systems team is usually inserted between the
research station and extension and ideally consists of personnel
from both. Its objectives include: learning the real world in
which farmers function; transmitting knowledge of farmers'
situations, constraints and objectives back to researchers on-
station; adapting existing technology/techniques to local
farm/farmer circumstances; and involving extension personnel in
the development and adaptation of new technologies. Such a team
is attached either to the research institution, favoring feedback
to researchers who are the source of new technologies, or to the
extension service, favoring the rapid diffusion of appropriate
technologies, once they are identified.


Farmer





Research FSR FSR Extension


It is hoped, that over time, both the researchers and the
extension personnel will adopt a farming systems approach. But
in the meantime, the farming systems unit is expected to assure
the linkage between research and the farmers, extension and the
farmers (with regard to technology development), as well as the
linkage between research and extension.

1.2 Characteristics of FSR/E

Most farm problems, like the farming system itself, cut
across several disciplines. Since present day educational
systems require specialization by discipline, an
interdisciplinary team is required to deal with the different
aspects of a given problem. Farming systems also views the farm
and the farm family's activities as a whole, ie., as a system.











Understanding the range of these activities requires an
interdisciplinary team. CIMMYT describes a minimum team for its
cropping systems activities as an agronomist and a socio-
economist. A farming systems team should more likely consist of
an agronomist (often the main liaison with the agricultural
research institution), an agricultural economist, an animal
scientist, a sociologist, and an extension specialist (also the
main liaison with the extension service). An individual with the
proper qualifications can sometimes serve more than one of these
roles.

The farming systems approach is farmer oriented and as such
is essentially a bottom-up approach because it treats farmers'
objectives, needs, constraints, and opinions as being of primary
importance. Research essentially attempts to understand the
farm/family farming system in order to help farmers solve their
own problems. Farmers are rational managers of complex systems
and are one of the major sources of information and training for
researchers, as well as the final evaluators of any new
technology and its usefulness in their farming systems.

1.3 The problem of technical packages

Although basic research often deals with technical packages,
these are not readily adoptable by small farmers. Such packages
may consist of 4 or 5 major technical components. Any one of
these may require 1 to 3 years of use to master proper
management. With all the interactions when they are combined, it
may easily take farmers 3 to 5 years of experimentation before
they master the management of the technical package to the point
that results improve significantly from their traditional
practices. In the meantime, results may actually be worse than
farmers can attain with traditional practices. Small farmers
usually cannot afford this risk, particularly is the technical
package is expensive. They also cannot afford a long delay in
receiving a return on their investment. Even credit programs
often do not help because they usually require repayment before
they master the technical package and are receiving much benefit
from it.

One of the prime responsibilities of a farming systems team
is to break down technical packages so that individual components
are more readily adaptable by farmers. Introduction of the
components must then be a stepwise process leading toward the
complete technical package and each individual step must be
profitable to the farmer. Although this may sound easy, it
usually is not and requires considerable creativity and adaptive
research.












1.4 The basic steps in FSR/E

Farming systems research is often said to consist of five
basic steps:

1) Selection of a target zone or target population.

2) Characterization of the farming systems and diagnosis of
problems and constraints as well as opportunities where
rapid improvement is possible.

3) The design of on-farm tests and other interventions.

4) On-farm testing and/or evaluation.

5) Diffusion of successful techniques/technology to the farmers
using a similar farming system.

These steps are not hierarchical, but rather iterative, such
that several and all steps may be in process at the same time.

1.4.1 Selection of a target zone

Target zones or target populations are often a political
decision imposed on the project and team. If the zone or
population is large and covers a number of farming systems, it
may be necessary to choose one or two sub-zones or sub-groups in
which efforts will be concentrated. If the team has to make such
a decision, it should be done with an eye towards research
efficiency and effectiveness. The decision should be based on
which zone or group is most important in terms of agricultural
production and where there appear to be opportunities to achieve
results rapidly (improve agricultural production and farm family
income/well being).

Most farming systems practitioners now accept that such
decisions as well as initial research priorities and an initial
characterization of the farming systems can be based on a rapid
reconnaissance survey. This survey should be done by an
interdisciplinary team, preferably including some individuals
with knowledge of the zone or group. The team learns all it can
about the region from secondary sources and then draws up topics
of inquiry (guide d'enquete) which will be used during the
survey. Although the main focus may be on the farm/farm family
activities, their timing, crop calendar, activity calendar,
methods, etc., care must be taken to include socio-economic and
consumption/nutrition aspects of each activity as well as basic
institutional concerns. (See FSSP Networking Papers 10 and 11)
This process tends to be painful, but extremely useful as
individuals with different backgrounds, training and biases pound
out an agreement of what they as a group consider most important.
It is also an excellent team-building experience. Once the
topics of inquiry are established, the entirety becomes the











domain of each individual. Each individual (often working in
teams of two) is expected to ask farmers about the range of
topics. Since no questionnaire is used and interviews are open-
ended, topics of particular interest may be pursued in greater
depth, if the farmers have the time and interest. Such a survey
might take from 1 to 4 weeks, depending on the size and diversity
of the region, with another 1 to 2 weeks required to write up the
results. To a large extent, the topics of inquiry should serve
as an outline for the written report. Such a document provides a
basis for discussion of findings, research priorities and
proposed interventions with interested organizations and
individuals who did not participate in the survey. The results
of such a survey can only be considered preliminary at best, but
they do provide a starting place for project activities in a
short amount of time. Characterization of the farming systems
and the diagnosis of problems and constraints will be incomplete,
and any findings need to be verified. This leads directly into
the diagnosis step.

1.4.2 Diagnosis of problems and constraints

Through a series of informal and formal surveys, information
gaps will be filled, a more complete characterization of the
farming systems will be made, and specific problems and
constraints will be studies/diagnosed in depth. Often this will
include specific focused surveys which follow the same basic
process as the reconnaissance survey, and for which disciplinary
specialists in the area of the specific problem are added to the
basic farming systems team. Each such survey should be written
up in detail as a joint team effort in which all findings and
recommendations are fully discussed and agreed upon. Such
reports document the activities and accomplishments of the
project and contribute to the continuing characterization of the
farming systems, as well as recommending a course of action to
resolve or research a specific problem. A series of such study
documents can provide an important contribution to monitoring the
project.

To study some problems, the diagnostic survey may have to be
repeated to evaluate the performance of a given variable through
time. Where precise quantitative data is needed, formal surveys
are required. Wherever possible, these should be one time
surveys with a limited scope and a specific focus. Some
information, like market prices through the year, will require
using formal periodic surveys.

Many past projects have collected much more data than they
could manage or analyze, particularly those using a cost-route or
input-output approach to data collection. In many cases, the
computer has served to exacerbate this tendency to collect too
much data. One rule of thumb is to collect only as much data as
can be processed and analyzed on hand calculators, but use a
computer to do it faster and better. The time saved should
usually be spent doing additional analysis. In many research












projects, 90 percent of the time and resources are spent
collecting data and only 10 percent in analysis. This is usually
a poor allocation of resources. Less data with better analysis
would often provide more results. Use of the informal diagnostic
surveys which can be written up quickly, is one way to help avoid
this data trap, which has been the downfall of many research
projects.

1.4.3 Design of on-farm trials and interventions

The third step is the design of on-farm trials and other
interventions. The reconnaissance survey should provide an idea
of where to start with such activities, as well as priorities
based on: opportunities for which technologies exist and seem
relatively well adapted so that they have the potential to have a
rapid impact, the importance of the potential impact of a given
technology, and the size of the group to which it could
potentially apply. Diagnostic surveys should provide even more
specific recommendations about what tests and interventions are
needed, and how they might be done. Usually 4 or 5 important
topics are evident in the reconnaissance survey, and work can
begin on these while more information is being gathered.

Care must be taken in designing on-farm trials and other
interventions, to consider all aspects of the problem. This is
one reason why trials should be selected and designed by
interdisciplinary teams, not just agronomists, and why
socio-economic research should be an integral part of all team
activities. The use of fertilizer provides an instructive
example. If fertilizer is to be used, one must consider if it
provides sufficient yield response to be interesting, on what
crop or crops, and under what conditions. Economically, one must
consider whether the increase in production is sufficient to pay
for the fertilizer and the extra costs associated with its use.
These include the cost of application, the cost of increased
weeding if needed, and the cost of increased harvest labor. One
must determine when fertilizer should be applied for greatest
efficiency of response and what form of fertilizer should be
used. One must consider whether the farmers have the labor to
apply it at the proper time, and the cost or change in return if
it is applied at a different time. This cost may not only be in
terms of the crop being fertilized. Increased labor allocation
to that crop may delay or replace activities on other crops or
enterprises, which have a cost or reduce total farm returns. In
addition, whether the farmers can get the fertilizer at the time
it is needed must be considered if the supply is uncertain. The
institution or institutions that could provide it must be
identified. Their relationship to farmers or farmer
organization, and research and extension should be considered.
Finally, whether the farmers have or can get the money to buy
their fertilizer at the time it is needed must be considered. If











they cannot, fertilizer is of no use. If credit is necessary and
feasible, then the institutional arrangements to provide it must
be considered. All of this, at least implicitly, should be part
of designing a trial which uses fertilizer.

Use of fertilizer implies that at least another step should
be taken. Since fertilizer is of no use if farmers cannot afford
it, an effort needs to be made to determine which farmers have
sufficient resources to potentially use fertilizer, and which do
not. Data should be collected which allows the team to stratify
farmers according to their resource base and/or their access to
inputs. It also implies that there should be a different set of
trials for these farmers who have resource or access constraints
which limit the use of inputs and those farmers who have the
potential to use them.

Fertilizer use also needs to be looked at over time.
Frequently, the management of fertilizer use is best viewed on
the basis of an entire rotation, rather than for each crop. To
do this, one needs to know both typical rotations and the recent
crop history of different parcels. It also requires research to
determine the residual effects of fertilizer according to the
form, quantity and time of application. This residual effect
(mostly of P and K) should be determined for different crops as
well as the results possible by applying a complement of
nitrogen. Although total production may be reduced, fertilizing
the rotation rather than the crop may be much more economical.

This example makes it very clear that the research team
needs to know the farming systems) very well to do a good job of
designing, refining and adapting agronomic trials. It also
underlines the need for interaction between disciplines in the
planning of tests and other interventions.

1.4.4 On-farm testing

1.4.4.1 Types of tests

The fourth step is doing on-farm trials and other
interventions on a limited scope with a research intent. On-farm
trials fall somewhere on a continuum between researcher-managed,
researcher-executed (RMRE), and farmer-managed, farmer-executed
(FMFE). Complicated trials similar to those done on stations
would have to be both managed and executed by project staff on
land borrowed or leased from farmers. Such trials should not be
the main focus in an FSR program, but may be necessary for
certain specific objectives. For example, soil scientists may
feel that it is necessary to test responses to N, P and K and all
of their interactions in the soils and conditions of the target
farmers. But, they cannot expect farmers to understand the test
or its results, or to be any help in its execution. Most
projects do not have the manpower and resources to execute many
RMRE on-farm trials, so they use researcher-managed and farmer-
executed trials. If the research staff cannot be there for all











activities, the farmer must understand the test well enough to do
some of the activities without supervision. This constrains
researcher management by requiring tests simple enough that the
farmer can understand and execute some activities on his own.
With some experience, farmers can correctly execute 4-plot tests,
but initially more than 2 or 3 plots with simple treatments will
probably lead to execution problems which compromise the results.
Superimposed trials are often farmer executed with the farmer and
researcher sharing the management. The researcher usually
decides when and how much of an input to apply and may decide
with the farmer on what part of the field to place it, while the
farmer makes other management decisions. Finally, farmer-managed
and farmer-executed trials are those in which the farmer has the
input or technique and can apply it as he wants. These are true
pre-vulgarization trials in the sense that one can observe how
the farmers begin to adapt the technique or input, once they are
no longer constrained by the researchers plot size requirements,
dates, etc. The relationship of trials to the adoption process
will also depend on whether inputs are provided free of charge or
whether there are other incentives for farmers to participate.
Participation is no indication of adoption as long as fertilizer
or other incentives are provided. The fields of farmers in the
adoption process also should be monitored to identify strategies
farmers use to adapt the technology to their system. Only after
this adaptation has taken place can a realistic assessment of the
technology's potential impact be made.

1.4.4.2 Test designs

Test designs, the size of plots and the number of
repetitions, should all be a function of the test's objective.
But determining specific objectives is complicated by the fact
that efficient procedures for isolating factors which influence
performance often conflict with effective procedures to
demonstrate that influence to farmers. Complex designs with
numerous plots can provide information about several factors and
their interactions in a single test, but are likely to be
incomprehensible to farmers especially if the plots are
randomized. Simple 2 or 3-plot tests with only one treatment,
but possibly several levels, are the best for communicating with
farmers. A reasonable compromise is the 4-plot, 2-treatment
trial which also provides information on interactions. Farmers
can probably also understand "with" and "without" treatment on a
group of varieties using a strip-plot design (not randomized).
If more complex designs are deemed necessary to rapidly isolate
factors influencing performance, it may be best to borrow or rent
a piece of land in the community where such tests are completely
managed and executed by research staff.

1.4.4.3 Plot size

Plot size should also be a function of the test's objective.
Large plots are not needed to observe if a variety or fertilizer
is performing well. Cultural practices may require slightly











larger plots (100 m2). Plots used to calculate labor time need
to approach the size of farmers' parcels to provide data of any
accuracy.

1.4.4.4 Number of repetitions

The number of repetitions of a trial should also be related
to the objective. In the initial stages of testing, many tests
correspond to observation trials in which researchers want to
learn what works well or what factor has an effect. Typically,
this does not require statistical results, but 3 to 6 successful
repetitions on which to base an initial judgement about whether a
new variety or technique is better than what is presently used.
In the early years, there should be an abundance of topics the
team wants to test, but it is constrained by the manpower and
resources available for testing. Many fewer topics can be tested
using 30 repetitions than 6 repetitions, and farmer evaluations
will often show a need to redesign the test after the first
attempt. Given the objectives of observing and making a
judgement, using many repetitions at this stage is an inefficient
use of resources.

At the second stage of testing, the team will want to
determine quantitatively how much a new variety or technique is
better than the existing technology, and begin to isolate factors
which identify why or under what conditions there is a
difference.

Much greater precision is needed to allocate portions of the
difference to individual factors. Variance also needs to be
reduced to determine meaningful differences in results.
Therefore, a larger number of repetitions are necessary, but
still may not be sufficient to produce statistically significant
results, unless efforts are also made to isolate and eliminate
major sources of variability. Identifying the sources of
variability should make an important contribution to
understanding the farming systemss. The variability of results
will increase as farmers take over management of a technology and
adopt it to their individual constraints and circumstances. If
these circumstances are not fairly homogeneous, the adaptations
may be quite large, producing a large variability in the results
at this stage. The technology may require adaptation to the
important differences in the farming system identified as factors
affecting performance. At this stage again, it may be very
difficult to get statistically significant results.

1.4.4.6 Use of statistics

Although farming systems tests should be exposed to rigorous
analysis, it is not clear that a .05 or .01 significance test of
yield differences is always an appropriate decision rule. At the
first stage of observation trials, statistical analysis may not
be appropriate at all. At the second stage of more rigorous
testing, statistical analysis may be appropriate, but at what











level? Even a variety or technique which increases yields 50
percent, may not test statistically significant at the .05 or .01
level because of high variability in the on-farm environment.
Decision rules, whether statistical or otherwise, should be
sufficiently flexible to allow use of variety or techniques with
high potential.

1.4.4.7 Economic analysis

Economic analysis should also be done on most of the trials.
If the crop(s) is(are) profitable, and/or the technology is low
cost, economic results may be statistically significant, even
though the yield results are not. In many cases, these economic
results are more important to the farm family than the yield
results. Attention should also be paid to the potential
consequences, for example, the risk involved in a large expense
or investment. The beneficial consequences of even moderate
improvements when people are desperately hungry, should be
considered as well. Judgments concerning a technology should
take economic results into account, whether or not statistical
analyses are performed.

1.4.4.8 Evaluating other interventions

Agronomic testing is a specific form of evaluation for which
a whole body of scientific procedures have been developed.
Procedures for evaluating other interventions are not as well
established. Yet, evaluating other interventions, not only to
determine it they were successful, but also to establish how they
worked and why, is as important as testing agronomic
interventions. Such evaluations are an important part of
learning how the social, institutional and economic systems of
farmers function and interact. As in agronomic testing, such
knowledge is necessary to refine and adapt existing and future
interventions. All interventions need to be monitored in a
manner that provide information on how, why and under what
conditions the intervention works, as well as on its impact.
This monitoring should continue through the diffusion stage to
provide information on the adoption process of farmers, their
adaptation of the technology over time, problems or improvements
which arise from this adaptation, and indications of the
technology's long term impact.

1.4.5 Diffusion

In a farming systems program, diffusion of a new technology
begins during on-farm testing, as those farmers with tests
observe and learn to use the technology. As testing and
adaptation continue, a number of farmers in a small area have an
opportunity to try the technology and many of their neighbors
observe and find out about it from those with tests. That this
process works is demonstrated by those few cases in which a
technique or variety really responds to farmers needs, and within
several years, spreads to the entire community. But farming











systems research is an intensive effort and requires far too many
resources to be undertaken in each and every community. The
farming systems approach attempts to identify large groups of
farmers who are homogeneous, in the sense that they face similar
circumstances and constraints and use a similar farming system.
If these research domains are properly identified, then a
technology developed in one community should be appropriate to
farmers throughout the research domain. But in most cases, the
farming systems unit will not have the resources to diffuse the
technology throughout the research domain. This diffusion beyond
the communities where research is done, is typically the task of
a regional extension/development program. Strong linkages should
exist between the farming systems unit and an active regional
extension program, to assure the dissemination of technologies to
those populations for which they are appropriate. The linkages
are obvious for research development programs in which the
farming systems research unit is attached directly to an
extension program. In other situations, the linkages may be less
obvious and much harder to establish. But, research of any type
serves no purpose unless the research results are appropriate to
and diffused to a target population.


2. The application of FSR/E in ADS-II

2.1 Position of ADS-II with regard to external structures.


The ADS-II project finds itself in an unusual and a rather
uncomfortable position for a farming systems project. It is not
directly attached to either a research or an extension structure.
In fact, it could be said that there are no national institutions
doing effective agricultural research or effective agricultural
extension in Haiti. A national extension structure exists
through the offices of the District Agronomist, but it seems to
function effectively only in isolated cases, where it is
supported by a project and external funding. Research exists in
isolated projects and in a non-systematic manner through the
efforts of the University Faculty (FAMV) and its students. This
leaves ADS-II in the position of being unable to really link to
either research or extension structures in a formal,
institutionalized manner. It also makes it very difficult for
the project to define the boundaries of these activities. On one
hand, it ends up screening varieties and doing the selection of
local varieties because such services are not offered by a
national agronomic research institution. On the other hand, it
is involved in doing seed multiplication and the extension of
certain successful varieties because there is no national
institution effectively serving these functions. It has
effectively sought out varieties from other projects, programs,
and international centers, which have been screened and the
better varieties tested. The more effective linkages to
International Centers have largely been based on personal
relationships, except for CIMMYT, which has a local presence,











i.e., a maize program in Haiti. Given the lack of an effective
agricultural research institute in Haiti, external linkages are
very important and should be strengthened and broadened.
Relationships with other Caribbean agricultural research centers
like INRA of Guadaloupe, the University of Puerto Rico and the
University of the West Indies should be developed, since they
could serve as sources of varieties and technology already
adapted to the general Caribbean agro-climatic zone.
Relationships with IITA's and CIAT's roots and tubers programs
would also be useful. They have developed a mini-setting
technique which would be one approach to help resolve the
constraint of the high cost and scarcity of planting materials
for important roots and tuber crops.

Through the faculty and the Group de Recherche pour le
Developpement (GRD), there is a nascent network of farming
systems research programs. Several of the Haitian project staff
are informally members of this network. This relationship should
be strengthened, but must grow slowly and be handled with some
diplomacy. The Recherche/Developpement orientation in which
Haitians are trained and operate is one form of a farming systems
approach. ADS-II, as the "new kid on the block", does not yet
have any credibility vis-a-vis these older and better established
programs. Until that credibility is established, the project
will be in no position to take the leadership role in Haitian
farming systems research to which it aspires.

This network, which seems to be an outgrowth of the
University Faculty, groups the key research and extension
personnel working in the government. It is doubtful that
expatriots can have any great impact on this group. More likely,
it must be reached through the slower process of gaining project
credibility and training Haitian project staff who return to
teach at the faculty and interact at this level. The continued
relationship of Haitian project staff to the faculty is probably
the key linkage to Haitian research. Most of the relationships
in this decentralized research (and to some extent extension)
system seem to revolve around the faculty.

Another level of influence, also related to the faculty, is
through the training provided to students and the resident
agronomists serving their civilian service obligation. Many of
these individuals will become research and extension personnel.
But so far, the project does not seem to take their training very
seriously. No systematic training is provided in farming systems
methodology, to students or even to project staff. Training of
students and the resident agronomist staff is largely restricted
to providing experience implementing and managing on-farm trials.
It is not clear that any systematic training in the design and
conception of such trials is provided. Training in the design
and use of reconnaissance and diagnostic surveys is not provided.
Students and resident agronomist staff receive little training
related to identifying problems and constraints, and designing
research or interventions to help resolve them. They seem to be











perceived more as extra bodies to help execute the very ambitious
number of trials the project conducts, rather than an important
resource to be trained and prepared for the future.

Projects and PVO's are an important part of the
decentralized extension system in Haiti. ADS-II has established
contact with a number of such organizations in the project zones
and throughout Haiti as a source of research, varieties, planting
materials, etc. But it seems that, to date, little effort has
been made to develop these relationships as a means of
facilitating extension outreach of project tested technologies.
It would seem that developing these relationships should be one
of the priority tasks of the project extension specialist.

ADS-II is presently attached to the Ministry of Agriculture,
Rural Development and Natural Resources, but is not integrated
into its functional structure. For those of us who think that
structural location is an important part of institution-building,
this is a serious problem. But given the recent history and
existing situation in Haiti, it is not in the best interest of
the project to change its location and institutional situation at
this time.

The project was moved out of the Faculty and into this
situation because politically motivated changes in Ministry
personnel led to "serious conflict of personalities". The CIMMYT
project was forced into the same kind of institutional
arrangement. The present political climate in Haiti provides
little assurance that Ministry personnel will not change again in
the near future. If such changes again take place, structural
integration might well threaten the functional capacity of the
project. Although far from ideal, the practical solution seems
to be to leave the "project right where it is until the new
Ministry has proven its ability to function effectively, and the
political situation is stable.

2.2 Internal Structure of ADS-II

The internal structure of ADS-II is also rather unusual for
a farming systems project. Project staff are divided between two
separate hierarchies, the farming systems component (SP) and the
Rural Economics and Statistics Component (ERS). The ERS is
subdivided into National Agricultural Statistics (EN) and Rural
Economics (ER). The EN is primarily responsible for a national
agricultural survey and has an important subgroup responsible for
establishing the computer based Comprehensive Resource Inventory
and Evaluation System (CRIES). The ER section is responsible for
the economic analysis of field trials and the socio-economic
aspects of the farming systems program. Regional staff are also
divided between SP and ERS. This division is inappropriate and
needs to be changed. One of the basic tenets of the farming
systems approach is that the farming systems team should be
interdisciplinary. But so far the farming systems component of
this project is staffed entirely by agronomists. The ER group











has done a few studies concerning some aspects of the farming
system. But there has been very little feedback to the
agronomists (SP) from the rural economics surveys. Even if the
ERS had produced surveys effectively, there still would not have
been joint interdisciplinary planning, implementation and
evaluation of on-farm trials and other project activities.

To achieve proper interdisciplinary integration, the rural
economics (ER) section needs to be abolished, its staff
integrated into the farming systems (SP) component, and its
budget added to that of SP. Since the farming systems activity
has to take place at the field level, the Haitian socio-economist
(Panyon) and the authorized technical assistance position for a
farming systems socio-economist should both be moved to Cayes.
They should have responsibilities to work with the Jacmel team as
well, but need to work together, at least initially, so that
effective training can take place. The national survey and CRIES
activities have a tendency to absorb all of the technical
assistance and other computer-qualified social science manpower.
The pace at which the national survey and CRIES have been rushed
forward in the first 2 years has had a detrimental effect on
farming systems activities. The implicit priorities suggested by
this resource use is that the national survey and CRIES have
replaced the farming systems activities as the main focus of the
project. The evaluation team does not believe that such a change
in priorities is appropriate. The socio-economists working in
the SP should be reserved exclusively for that component, or
other arrangements made to assure that the national survey and
CRIES do not continue to absorb all social science manpower to
the detriment of farming systems activities. Haitian
socio-economists are needed at the sub-regional level to interact
with the chefs des zones (Haute Cap Rouge, Bas Cap Rouge,
Maniche, Berault) and the resident agronomists, because these are
in fact the basic farming systems teams. At the moment, it
appears that BA level socio-economists are very rare in Haiti,
and will be until the Faculty's Rural Economics and Development
program starts graduating students in about 2 years. In the
meantime, it would be useful to send two to four agronomists to a
socio-economic training program like the one provided by CIMMYT.
Given the lack of Haitian socio-economists, degree training for 2
agricultural economists and 2 sociologists/anthropologists should
be included in the project. Master's level training is generally
most appropriate, but several of the best project participants
should be permitted to seek Ph.D. level training so that at some
point they can teach in the University. These are the people who
in the long run will have an impact on the Haitian interpretation
of the farming systems approach. If their duties permit, project
staff should be permitted to teach at the University on a limited
basis.











2.3 Survey Approach

2.3.1 Reconnaissance survey

During April and May, 1984, the project staff conducted a
series of rapid reconnaissance surveys in the four target
subregions of the project. This survey was based on a 51-page
formal questionnaire and was done by Haitian and technical
assistance researchers. Interviews were conducted with small
groups of farmers and community leaders in different parts of
each of the 4 zones. A number of problems exist with both the
content and the manner in which the survey was implemented, but
the fact that the reconnaissance survey was done, that it allowed
researchers to become acquainted with the target zones, that it
helped establish research priorities, and that it helped provide
a starting place from which to begin on-farm testing, overshadows
the flaws. As is always the case, the researchers undoubtedly
learned much more than was ever documented. But the analysis and
evaluation of that information does not seem to have been carried
through in very much detail. The documentation remains a
description of the zone, with some differentiation by subzone.
Nowhere is there an identification of specific farming systems
within the zone and the factors which characterize them. The
research domains for which different techniques/technologies are
appropriate are also not clearly identified or hypothesized.

2.3.2 Formal surveys

One of the problems may have been that the effort which was
devoted to the design and completion of the formal (2 page,
quantitative) survey distracted researchers from a thorough
evaluation of the reconnaissance information. Information from
the formal survey seems to have been considered more important,
perhaps because it was quantitative and based on formal sampling
procedures. This survey provided basic demographic information
on 300 to 600 households in each of the 4 zones, basic
inventories of land area cultivated and livestock ownership,
input use, cultural and animal husbandry practices, and
information on the importance of different food and cash crops.
Stratification by land area cultivated for the Jacmel region
verified that resource inventories and input use are correlated
with the size of farm. Although a number of these correlations
proved to be statistically significant, additional information
about such relationships does not seem to have been pursued. Nor
does such information seem to be reflected in the continuing
research program. No effort has been made to identify farmers
with different resource endowments as separate
research/recommendation domains. Fertilizer is an integral part
of almost all trials and demonstrations even though the surveys
identify and verify that "small farmers used little or no
fertilizer", and by implication probably cannot afford to do so.












As in any initial studies, numerous questions are raised by
these surveys. What are the effects of land tenure, of
demographic differences, of different resource endowments, of
access to different land types and irrigation, of the dependence
on hired labor and its different forms, of resource poor farmers'
dependence on wage labor, etc.? But in ADS-II, even though
identified, most of these issues and problems have not been
further investigated and have not been used to identify
homogeneous groups of farmers with similar circumstances and
constraints. Almost no diagnostic investigations have been done
to attempt to answer such questions. Instead, the project seems
to have devoted all of its farming systems related effort to a
very ambitious program of on-farm trials, without having
adequately identified and characterized the farming systems
context, or undertaking parallel studies to do so.

In part, this situation is due to the ineffectiveness of the
ERS in contributing to the farming systems aspect of the project.
Following the original verification survey, the ERS also
completed studies of land area cultivated by crop and land tenure
arrangement, and livestock ownership and husbandry. These formal
surveys provided quantified information, but one must use the
results with great care since farmers may well have perceived
that it was not in their best interest to give truthful answers
about land ownership and livestock owned. The project considers
these as constraint studies and they do in fact suggest a number
of constraints concerning land use, land tenure and animal
husbandry. But, little additional effort has been made to
investigate these constraints in detail or to establish the
effect of these constraints on the cropping system and/or farming
system. These constraints were at least generally identified
during the reconnaissance survey. The constraint studies as
implemented, put numbers on these constraints, but did little
more to explain them. They also do not propose research or
interventions to help resolve the constraints. Informally, they
probably did influence topics chosen for testing and the pig
program intervention.

The ERS has also collected price data at 2-week intervals on
agricultural products in four markets in the Jacmel region. The
similar survey in Les Cayes has been hampered by implementation
problems, in part due to conflicting lines of responsibility and
authority. Perhaps due to this problem, the price information
has not been published and it is unclear if it has been analyzed.
The marketing study does not include marketing channels, although
project staff seem to know the hierarchy of markets and different
types of merchants. There also does not appear to be any effort
to identify the quantities of agricultural products flowing into
and through the market system. The price study does not specify
or attempt to identify differences in merchants' buying and
selling prices. Since the product is not physically bought or
sold by the enumerator, the traditional gift of some amount











beyond the normal measure is not included. Although the
precision of this price data may be questioned for the reasons
stated, it probably still provides a fairly accurate measure of
relative commodity prices through time.

The ERS is monitoring the reintroduction of pigs. This
survey covers husbandry practices, health concerns and costs and
revenues. As a third generation of pigs become available, the
study will expand to include a comparison of recommended
facilities and nutrition with traditional farmer practices. The
ERS is also monitoring the watershed erosion control activities
and is just finishing a description and cost analysis. It is not
clear that the ERS or the project as a whole has been involved in
monitoring other technologies, outside of the on-farm trials.
Project staff know that use of Tamazulapa beans and Amina rice
has spread, but it is not evident that any effort has been made
to measure their adoption or monitor adaptations.

2.3.3 Farm monitoring

The ERS and the project as a whole have not been involved in
either periodic or diagnostic monitoring of farm/farm family
activities, until the last two months. As a consequence of doing
neither specific diagnostic studies nor periodic monitoring, the
characterization of the farming systems has really not improved
since the initial reconnaissance and formal survey. Little
progress has been made specifying issues, problems and
constraints so that research could be designed to help resolve
them. Work is just beginning on farm budgets, farm constraints,
farm family revenues and expenses, etc.

The project leadership has made the choice not to become
involved in large scale and intensive farm/farm family monitoring
activities in the farming systems component of the project. This
decision is based on the fact that projects using cost-route or
input-outputs approaches to data collection are often swamped by
huge quantities of data. The surveys require so much supervision
and the data take so long to process, that little or no analysis
is completed until years after the data is collected, if at all.
The project leadership is wise in avoiding this data trap.
Although it is less clear that it is cognizant that the national
survey is the same kind of phenomenon, absorbing all of the
administrative and computer capable human resources available.
But it is not clear that they have developed an effective
alternative. The ERS has not functioned effectively in
performing surveys for the farming systems component. (The ER
began a new series of surveys in April, 1986, with the arrival of
new personnel but has little output to show for the previous 15
months.) But it is doubtful that surveys designed and
implemented from Port-au-Prince can serve the needs of the local
field teams. Feedback from Port-au-Prince to the field teams is
very limited. (Field staff tend not to be aware of the results
of the series of surveys which the ERS did in 1984.) Surveys,
and information gathering in general, need to become the function











of local field teams, reinforced with socio-economists to provide
an interdisciplinary perspective. The personnel in ERS who are
involved in farm monitoring should be moved out of Port-au-Prince
and integrated into the regional teams. Priority should be given
to the Cayes region because of its importance in USAID plans.

The way the project was structured was to have enumerators
and monitors for survey work attached to the farming systems
field staff. Their time was to be divided between helping the
farming systems staff with field trials and doing surveys for the
ERS. They did participate in the series of ERS surveys in 1984,
but have since worked mostly on field trials except for those
periods when they were involved in the regional testing of the
national survey.

Certainly much more information is needed about farms and
family activities than has been provided to date. The question
is what information, how large a sample, and how can it be
collected most effectively. But first, basic farming systems and
strata need to be defined. I expect that several farming systems
will be found in each of the four zones. In the plains, it will
probably be useful to distinguish rice areas from irrigated and
non-irrigated areas growing other crops. Within each zone the
importance of area cultivated, crop combination and cropping
patterns, degree of slope, land tenure status and labor
availability need to be investigated. Some of these factors,
like slope and cropping patterns and perhaps land area cultivated
and labor availability, will probably be correlated. Which
factors are most important need to be determined empirically. It
is evident that economic returns for landowners, renters and
share croppers will all be different, so this should be taken
into account. It would seem necessary to assume on the basis of
what is already known that there are at least 2 or 3 agro-
ecological environments per zone, that 2 or 3 resource strata
should be considered, and that at least the 3 major land tenure
statuses need to be taken into account for each zone. On this
basis, it appears that some kind of 32 or 33 matrix of different
farmer circumstances needs to be considered for each of the four
zones. It is difficult to see how this can be done using formal
surveys and statistical sampling methodology without becoming
involved in a very large data collection and management exercise
of the type that have swamped so many other projects. With the
high variability that must be expected, statistical procedures
will often not allow a distinction between different results on
the basis of small samples. If statistical procedures can
provide only very limited results on a survey of a size the
project can manage, one must question to what extent the formal
survey approach alone is an effective research methodology. It
still cannot provide the high quality qualitative information
that informal surveys and case studies provide. It can be used
to provide information on yields for different crops and crop
combinations and market prices for agricultural products, if
supplemented by qualitative studies which help define and
interpret the price information. Formal surveys may be useful to











collect information on land area cultivated per farm and per crop
and on livestock ownership. But only if farmers gain enough
confidence in project personnel to give truthful responses. It
may be better to develop proxies which provide a qualitative
interpretation of land area and livestock ownership, rather than
what may well be erroneous and biased quantitative information.
(Many of these comments also apply to the national survey.)

Given the many problems involved with formal surveys and the
limitations on the use of the data they can provide, it seems
that it would be more effective to make greater use of informal
surveys. In many cases they are needed to guide the design and
interpretation of the formal surveys. Informal guided surveys of
individual farm/farm families can provide indicators of the
percentage of farmers who have a certain trait or use a specific
practice. But they also have the flexibility to explore how and
why specific practices are used. Informal (guided) diagnostic
studies permit the rapid investigation of a specific topic and
communication of the results directly to researchers so that they
(and not the monitors) have the information to interpret the
results. Even relatively accurate labor profiles can usually be
built using informal surveys, avoiding the need for a massive
collection of labor data.

ADS-II researchers also need to adopt the attitude that
farmers have a great deal to teach them, and they need to go out
and learn what it is. In most zones there will be one community
that grows a specific crop or does some other activity better
than other communities. Within a community there will often be
farmers who grow one crop better than anyone else. Researchers
need to identify these farmers or communities and find out what
they are doing that is different than everyone else. The
difference may be due to soil type, which is important
information even though it is not replicable. But frequently the
difference will be due to a practice or a variety which can be
applied by or adapted to other farmers as well. Project
personnel should monitor the fields of such farmers in the same
manner as agronomic trials to determine what the difference is
and how it can be used by or adapted to other farmers.

Some areas where the project seems to need additional
information include: labor constraints and labor profiles for
crops, crop combinations and other farm/family activities by
gender (based on detailed calendars of crop and other
activities), use of farm labor vs. off-farm labor and the
importance of wage labor, labor associations and exchange labor,
off-farm activities and income, crop budgets and the economic
potential of different crops, the aggregation of crop, livestock,
and other activity budgets into whole farm budgets, the
empirically important stratification of farm/farm families by
resource endowment, land tenure status (or the combination of
land tenure statuses), and other important factors like
significant off-farm income, use of rotations in cropping
patterns, and the costs and benefits of hillside interventions.












2.4 Establishing an interdisciplinary FSR/E approach

The second aspect of the ERS problem is that most of its
resources and effort have gone into CRIES, and designing a
national agricultural survey and implementing it twice a year on
an ever expanding scale. This preoccupation with the national
survey and CRIES has seemingly influenced the perspective of
researchers as well as having a detrimental effect on the
physical and human resources available for farming systems
activities. These seems to be a tendency for some researchers to
believe that only those factors which are important on a national
scale are relevant to and worth investigating in the local
farming systems. Rather than a national agricultural survey
based on an aggregation of knowledge of local farming systems
(which the project has not yet had time to develop), the ERS
seems to have confined investigation of local farming systems to
those variables which a priori were considered important for the
national agricultural survey.

Because of the many problems outlined above, it seems
necessary to abolish the ERS, disband the rural economics section
and integrate its staff and budget into the farming systems
component, and create a National Survey and Statistics component
which does not impinge on the staff, budget and functioning of
the farming systems component.
With the addition of the socio-economists to the farming
systems teams, an effort should be made to insure that they
become fully integrated interdisciplinary units. This does not
mean that the socio-economists continue to do all the survey and
monitoring tasks and the agronomists continue to design and
implement on-farm trials. It means that socio-economists and
agronomists design and implement on-farm trials together, as well
as designing and implementing diagnostic and monitoring
activities together. Whoever is responsible for a given test
should do both the agronomic and economic analysis, with help as
needed from other project staff. Only when the socio-economists
begin to think a little like agronomists and the agronomists
begin to think a little like socio-economists, will a truly
interdisciplinary farming systems effort be possible. An
important part of developing an effective farming systems team is
organizing activities such that an agronomist and a
socio-economist (and animal scientist, entomologist,
anthropologist, etc.) are on the same farm or in the same field
at the same time. They should discuss the different realities,
problems, constraints and opportunities they each perceive until
they agree on a course of action concerning research and
interventions which they believe would be most beneficial to the
farmer. In this process, the farmers will most often be training
them, the so called experts, since no one knows the farming
system as well as the farmers, nor can answer why the farmers do
something in such and such a manner. This is one of the main
reasons for recommending that much more of the research be done
using informal but guided diagnostic survey methods, designed and











conducted jointly by agronomists and socio-economists. This type
of activity requires them to deal with their different
perceptions of reality and places them in contact with the true
experts from whom they need to learn, the farmers. Anything
which hinders this contact between researchers and farmers, like
depending on enumerators to collect most of the information,
impedes farming systems research.

Overtime, members of the basic farming systems field teams
at the zone level need to be trained to design and conduct their
own diagnostic and monitoring surveys, and to do much of their
own analysis. As this development of the zone level field teams
takes place, one, if not the major role of regional and national
project staff and technical assistance, will be to provide that
training and continued guidance. Design and analysis skills do
not develop quickly and will require continued guidance and
training, long after local teams are capable of implementing such
activities on their own. This training should include the use of
informal but guided diagnostic surveys which are written to
provide documentation on activities, analyze a problem or
constraint, and recommend a course of action for research and/or
interventions to help resolve the problem. Formal surveys should
be de-emphasized but not stopped. This implies that local field
teams need access to at least programable calculators and
possibly computers, to do their own analysis. (Such equipment
may have to be located at the regional office, due to limitations
of physical facilities at the zone level). Initial analyses of
data, including agronomic trials, should be done at the zone
level. A second level of analysis should be done to look at
possible aggregations and conclusions across a region. A third
level of analysis should be done at the national office to look
at possible aggregations between regions and conclusions that can
be drawn.

2.5 The On-Farm Testing Program

ADS-II has been extremely successful in moving very rapidly
from initial reconnaissance surveys to a very extensive on-farm
testing program. A total of 293 researcher-managed and 285
farmer-managed trial replications were conducted in 1985. The
team was able to identify technology, particularly varieties,
from the preceding PDAI project and other sources, that responded
to constraints, priorities and opportunities to have a rapid
impact identified during the reconnaissance and verification
surveys. Several of the varieties, particularly Tamazulapa beans
and Amina Rice seem to be an improvement over local varieties and
have been widely accepted by farmers growing these crops in the
research zones. This varietal testing has provided direction to
the on-farm trial program over the first two years as well as
some initial successes. This is fortunate, since the lack of
progress in characterizing the farming systems and identifying
problems and constraints, has meant that there was little
feedback to orient the trials program. But most of the readily
identifiable and interesting varieties of major crops have now











been tested, and it will be necessary to move towards testing
more complex cropping systems techniques. The diagnostic studies
and the testing program need to develop together. Certain
aspects of the testing program should be considered during this
reorientation process.

As the farming systems teams become seriously involved in
monitoring technologies and farm/farm family activities and in
diagnostic studies, they may initially find that they have less
time and manpower to devote to on-farm trials. It will likely
become necessary to make choices between doing fewer trials,
doing fewer repetitions per trial, and doing fewer trials with
complicated designs, such that farmers can execute a larger
portion of the trial repetitions with less supervision. But even
simple tests must clearly isolate one factor as the cause of any
difference between the test and the control to be an effective
demonstration. Although there are exceptions, farmers rarely
adopt whole technical packages at once, but rather adopt them one
component at a time. Farmers want to know the effect of each
component and the components within their resource means.
Therefore, comparing multifactor technical packages against
farmers' traditional practices is not only poor research, but
poor extension as well. Such simple tests are less glamorous and
require longer to achieve any particular research goal, but are
often more effective in the long run. An effective compromise,
for farmers who have attained some understanding of testing, is
the 4-plot, 2-factor trial. This trial design demonstrates the
effect of each factor separately as well as their interaction,
against a control.

The number of repetitions of any trial should also be
reconsidered in light of the objective of the particular stage of
testing. The initial testing stage of most research themes does
not require large numbers of repetitions. One is merely
observing if something works and whether it is worth devoting
more resources to explore that topic. Even if judged as being
worth pursuing, farmer feedback will usually require a redesign
of aspects related to their system or cultural practices.
Therefore, it is often more efficient to treat the first stage as
observation trials with enough repetitions to make a judgement (4
to 8), but with no intention of statistical analysis. At a
second stage of testing the focus should be on isolating factors
which affect performance. It is important to know how a
technology performs over a wide range of conditions, as they are
tested in ADS-II. But it is more important to isolate why and
under what conditions it performs well. With this knowledge, it
is often possible to adapt the technology slightly for different
conditions, or at least specify the conditions under which it
performs well. So far, it does not seem that this is being done
in ADS-II. Statistical analyses should be attempted at this
stage as a tool to help identify if sites or situations with
different conditions do in fact get different results. But
frequently the different conditions will only be determined
during the course of testing, so that the number of repetitions











is not sufficient to obtain statistically significant results.
During analysis, the repetitions should be grouped and analyzed
to compare the effect of different factors. But this, not just
running an F-test, is the purpose of analysis. However, unless
one is working with an immense number of repetitions, grouping
and comparing different sets is also not likely to give
statistically significant results. It is only when you think you
know the answer and set out a number of repetitions under similar
conditions to test the hypothesis, that you are likely to get
statistically significant results. And it is really only then
that statistical significance is a meaningful decision rule.

But, what level of significance is required? Agronomists
and most other researchers are trained that results with
significance levels below .01 or .05 percent are not acceptable
for professional publications. But this has nothing to do with
farmers' decision making processes. Furthermore, when research
is done on-station, a CV of about 15% is expected, not the 40-60%
encountered when working in farmers' fields. Is it reasonable to
use on-station decision rules when off-station variability is so
much higher? Often, it is more appropriate to report the
probability of a given increase in yield rather than testing
whether or not it is significant at a set level. Farmers may
well be interested in a low cost technique which gives them a 70%
chance of increasing production by some known amount.

More important than the statistical significance of the
result is the relationship between cost and result. Farmers will
often accept a low cost technique with mediocre results before
they will accept using fertilizer which produces a much larger
increase in yield. High interest rates (interest rates in rural
Haiti are reported to run as high as 30% per month) may cause the
real cost of such inputs to be several times higher than the
purchase price, and the benefits several times lower than
expected. There is also an important risk factor with high cost
inputs, that resource poor farmers cannot afford. If the crop
fails, will the farmer have to sell his land to repay the
fertilizer or the loan? A development strategy for resource poor
farmers typically requires a series of low cost marginal
improvements in the beginning, which only some years later allows
them to make larger investments. ADS-II should be devoting more
attention to low cost techniques for the majority of resource
poor farmers. Fertilizer should probably not be used in tests,
unless it is specifically being tested, or there is a strong
indication that the specific target farmers can and do use
fertilizer on a regular basis. The use of fertilizer on most
trials, and the lack of any stratification of trials according to
resource endowment are important indications that very little
socio-economic consideration has gone into the testing program.

ADS-II needs to do much more analysis and integration of its
test results than has been done up to now. But this cannot
effectively be done by statisticians in Port-au-Prince who know
nothing about the technology, test sites, factors influencing












performance, etc. This type of analysis must be done by the
field staff who design and run the trial. ADS-II needs to
provide training, guidance, and equipment to agronomists and
other field staff at the zone level so that this type of analysis
can be done. It will also require an effort to characterize
farmers and test sites, so than there is information available
about different conditions.

ADS-II needs to reconsider what it is doing in terms of pre-
vulgarization trials. From a research point of view, the idea of
pre-vulgarization trials is to get one step closer to the manner
in which the farmer will use the innovation. Only the factor
being tested should be different from normal farmer practices.
If fertilizer is not typically used by farmers, it should not be
applied to pre-vulgarization trials. Only in this manner does
one begin to get an idea of how farmers might use and adapt an
innovation, and what its impact might be.

This conflicts with one particular tradition of extension
demonstrations in which a package of a new variety, fertilizer
and improved cultural practices are demonstrated in farmer
fields. This type of demonstration mistakenly assumes that
farmers adopt whole packages. Many farmers in Haiti cannot
afford fertilizer and farmers typically do not adopt whole
packages. To be a good demonstration, farmers need to be able to
observe the effect of the variety alone or individual cultural
practices. Therefore, good extension demonstrations will be like
the pre-vulgarization trials described and will test/demonstrate
the effect of a single factor.

Many of ADS-II pre-vulgarization trials use the package
demonstration approach such that they are neither research nor
good demonstrations. The trials for the selection of the local
maize, Alizene, in particular, fall in this category. The
project needs farmers to multiply new generations of the selected
seed, but is not likely to have anything to test or demonstrate
until after numerous generations of selection. Superimposing
fertilizer, among other factors, on this selection process may
cause higher yielding but less hardy plants to be selected. But,
it is not clear that this result is an objective of the selection
process. There is also a tendency to take the variety that does
the best in a RMRE varietal screening trial directly to a pre-
vulgarization package demonstration, with no intervening testing.
This process may test/demonstrate the genetic potential of a
variety in a given agro-climatic environment, but in no way tests
the appropriateness of that variety to farmer conditions nor the
type of results that might be produced under farmer conditions.
This is multi-locational testing, but not farming systems
research. The introduction of such varieties without more
knowledge of the conditions under which they do or do not perform
well is potentially risky for the farmers.











2.6 Other Interventions

Several other interventions besides agronomic testing
deserve a brief comment. The most important of these is the
whole hillside erosion control activity. The experience until
now is really quite limited. The project is only now getting
some dry walls, canals, grasses and trees in place on the slopes
so that their effect can be monitored and evaluated in the
future. The original approach of constructing a complete system
on a hillside using group labor has been found to be ineffective.
The structures get built, but not maintained because the groups
are interested in the "wages" paid, not the results. When two
different projects in the same region pay different "wages",
whether cash payments, food for work, trees, grasses and other
vegetative material, etc., it causes problems. In Fond de
Freres, Les Cayes, ADS-II was offering only about 3 gourdes a day
plus the vegetative material. Farmers thought they were being
robbed because another UN project nearby was paying 13.75 gourdes
a day. The work has stopped and it is not clear when or if it
will continue. In Haute Cap Rouge, the FAC project was only
providing vegetative materials, so when ADS-II started paying 3
gourdes a day plus providing vegetative materials, it ruined
FAC's erosion control program in that area. The project would
like to consider the 3 gourdes per day as a minor incentive, but
in fact, it seems to be so close to day wages in those regions
that farmers perceive it as such. The approach of working
through individual farmers is proceeding very well in Haute Cap
Rouge. But, it is still not clear if the work is being done
because farmers appreciate the results or because of the 3
gourdes per day payment. The critical factor is whether or not
they will be maintained.

Some people have argued that the dry wall technique is not
economical because it requires a lot of work and takes land out
of production. But in Morija, Haute Cap Rouge, there is so much
stone readily available, that piling them up in dry walls may
actually increase the land available for cultivation.

The techniques used in Fond de Freres, Les Cayes, of digging
small canals to hold moisture and planting the ridge created with
grasses or other vegetative material, still has not been
adequately tested by the project. La Vallee project personnel,
who have used the technique quite successfully, say that the
critical factor is the vegetative strip, not the canal. But
under many circumstances, the vegetation will not grow adequately
unless there is a canal to retain additional moisture. For
whatever reasons, the canals in Ford des Freres had little
surviving vegetation and thus had little chance to be effective.
It makes no sense to have farmers dig such canals, unless the
project can assure the delivery and planting of a sufficient
quantity of grasses or other vegetative material.












The project was fortunate to be able to borrow state-of-the-
art techniques from a project like La Vallee as the starting
point for its research and intervention. One tentative result
seems to be that it takes about 18-24 months of erosion control
to improve a parcel's productivity. Another is that farmers who
are not owners of the land, need to use grasses, sugarcane,
pineapples or other vegetation which produces a rapid return
rather than trees. Trees may provide an incentive for the owner
to want his land back, as well as reducing the production of the
annual crops which provide the rapid return that renters,
sharecroppers, etc. need.

Attention should be devoted to alley cropping techniques as
well as the dispersed planting of trees now being attempted.
Coffee could be an interesting alley crop because of its economic
potential, if it can be grown successfully without additional
shade, once it matures. Bananas, plantain, cassava, pigeon peas
and other crops can be used to provide shade when it is small.
Alleys consisting of trees and shade tolerant grasses may well be
necessary for effective erosion control. Vetiver is also an
intriguing crop because of its possible economic potential and
its ability to grow on soils where almost nothing else will grow.
The need to pull it out by the roots is a serious problem, but
some form of management in strips and/or intercropped with a
cover crop like velvet beans (being tested in La Vallee) may be
worth investigating. In areas where leaving individual strips of
vetiver is acceptable, it makes a good erosion control
vegetation. Some farmers in La Vallee are also using vetiver
straw extensively to mulch hillsides.

Interventions with garden crops may also have good economic
potential. Knowledge of the marketing potential from any given
location seems to be critical and has not yet been studied.
Madian-Salignac has had tremendous success with cabbage, which is
evident from visiting the area. Project staff claim that farmers
gross as much as $3000 per acre. The reintroduction of pigs is
obviously of major economic importance to farmers and should be
pursued. The proposal to build cisterns and have farmers repay
the local credit union, constituting a revolving fund so the
activity can be continued, is an excellent idea. It makes sense
in terms of agricultural production and provides an opportunity
to study, work with, and evaluate an important rural institution.
Testing low cost adapted rice equipment like the thresher and
fanning mill are also useful for the irrigated plain. Although
the economic potential of tuber crops, particularly yams, was
demonstrated in 1984 tests, there has since been only very
limited activity on that topic. Certainly the availability and
cost of planting material is an important constraint. But
perhaps the project can test technological solutions to this
constraint like the mini-setting technique developed at the
International Institutes. According to one source, such
techniques are already being tested in a number of places in the
Caribbean.












2.7 Diffusion

Diffusion of technologies developed is a major problem for
ADS-II. The national extension structure through the 9 District
Agricultural Offices has fallen on hard times due to political
problems and a lack of funding. It is not clear that any
institution has the organizational capacity to extend adapted
technologies effectively. But ADS-II has neither the funding nor
the staff to do the job itself. It was not designed as a
research/development project. Therefore, its extension efforts
must largely be limited to those few communities where it has
research activities, and then primarily undertaken with a
research intent. Even the extension to a geographical area like
Haute Cap Rouge, is not within the scope and funding of the
project. Within the limitations of the project scope and budget,
the extension objective should be to study rural institutions and
extension approaches to determine how the diffusion of
innovations can be facilitated and implemented effectively. Even
the irrigation of the Cyvadier Plain is viewed by the evaluation
team as an attempt to test the concept of working on a whole
watershed, rather than an extension activity per se. It also
explains what may appear to be a contradiction: the evaluation
team supporting limited irrigation activities (if relatively
inexpensive and contracted out) in Cyvadier but opposing project
involvement in irrigation on the vast Cayes Plains. ADS-II needs
to accept that it does not have the means to do extension outside
its research intervention communities. To take on any kind of
regional or zone extension responsibilities would require a major
increase in funding and staff. Even if such funding and staff
were available, the project should not expend its extension
orientation until it has proven that it can do good and effective
farming systems research. It also should not expand
geographically until it has developed an effective model and
methodology for its farming systems component which is found
acceptable to an external evaluation team.

What ADS-II does need to do is expand its liaison activities
with organizations doing extension activities. These include not
only the District Agricultural Offices, but other projects and
PVO's. Project personnel have many informal linkages to such
organizations, but ways to improve these linkages and make them
more concrete should be sought. To help improve these extension
linkages, the role of the extension specialist on the technical
assistance team should probably be redefined. It should be
expanded to provide extension advice to the Cayes region as well
as Jacmel. It should also include establishing effective
linkages and liaison with extension organizations in both
regions.












2.8 Training

In the future, one important linkage to both research and
extension programs should be through providing training programs,
workshops, etc. The project should develop the capacity to train
researchers and extension personnel in the methodology and
implementation of effective FSR/E. This may include workshops in
the methodology and training exercises in reconnaissance and
diagnostic surveys. It may also include working with a local
team for a few weeks to experience how an interdisciplinary team
designs, implements, and evaluates a program together.
Diagnostic surveys will hopefully include representatives of the
local extension services) and research specialists in the area
of the problem being diagnosed. Such training can help develop
very important linkages to research and development programs.
But this can not be done until the project trains its own staff
and develops interdisciplinary farming systems teams which
function effectively. It will also require operating effectively
for some time before ADS-II will acquire the credibility
necessary for other projects and programs to want ADS-II to help
train their personnel.

ADS-II has provided or planned the following training
activities to date:

1 Master's degree Statistics Completed
1 Master's degree Agronomy In progress
1 Ph.D. degree Agronomy To begin in 3 months
1 Master's degree Agronomy Candidate selected
2 Agronomists FSR/E workshop Martinique
3 Agronomists FSR/E workshop Kansas State Univ.
2 Agronomists Watershed mngt wkshp IITA
2 Administrators Project evaltn wkshp Univ. of Minnesota
7 Statisticians CRIES trng 2 weeks Michigan State Univ.
and data processors
1 Data processor Digitizing trng 2 mo Michigan State Univ.

1 Workshop sponsored on on-farm trials
2nd on-farm trial workshop with CIMMYT planned
15 FAMV (1st & 3rd yrs) student summer interns
9 Resident agronomists (6 mos 1 yr) departed
8 Resident agronomists (1 2 yrs) in training
12 Field staff monitors
Survey analyst/supervisors
Enumerator supervisors
Enumerators

But the quality of the farming systems training in the
project is weak. Zone level project staff including the resident
agronomists and students primarily gain experience in managing
and implementing on-farm trials. No training in FSR/E
methodology has been provided. Zone level project staff, who
should be the basic FSR/E teams, have not gained










interdisciplinary experience, do not know the basic steps of
FSR/E, and do not participate in farm/farm family monitoring nor
reconnaissance and diagnostic surveys. They are not trained in
the design and analysis of on-farm trials, even though managing
and implementing on-farm trials is their primary activity. Most
of what zone level project staff, including resident agronomists,
know about a farming systems approach, they learned at the
University and from working in other Recherche-Developpement
projects in Haiti. The zone level staff, including the resident
agronomists, do not understand that the ADS-II project design is
different than the typical Recherche-Developpement project in
Haiti, nor the intent of that difference in design. Therefore,
ADS-II has very little credibility, even with its own staff.

One on the keys to resolving this unfortunate situation is
to establish integrated interdisciplinary FSR/E teams at the zone
level. But, this is easier said than done because there are not
enough Haitian socio-economists in the project to accomplish
this. And, very few socio-economists are available in Haiti.
The University will begin a 2-year option in Rural Economics and
Development in its bachelor's level program for agronomists this
year. But no graduates will be available for 2 years. In the
interim, 2 to 4 project agronomists should be trained in a
program like CIMMYT's economic program as quickly as possible.
Technical assistance in socio-economics also needs to be
reinforced. The additional farming systems socio-economist
already authorized is a first step in establishing this capacity.
However, that position should be located in Les Cayes and
restricted to the farming systems component activities. The
national survey does and will continue to absorb all of the
administrative and computer capable manpower it can get.
Although the team leader is eminently qualified and wants to
participate in the farming systems activities, it is doubtful
that he can escape the administrative imperatives of the national
survey, even with the addition of an administrative assistant.
The TA socio-economist, working with his counterpart and
agronomists who have attended the CIMMYT economics program,
should be able to begin the introduction of socio-economic
aspects into the farming systems component. It ADS-II continues
beyond the remaining two years, a sociologist/anthropologist
should be added to the TA team. In addition, at least 2 degree
programs in Agricultural Economics and 2 degree programs in
Sociology/Anthropology should be added to the training component.

The technical assistance team and program coordinators need
to take on a much stronger training role than has been their
function to date. If basic farming systems teams are to be
established at the zone level, the regional coordinator and TA
personnel must make training those teams their primary function.












2.9 Monitoring and Evaluation

2.9.1 Monitoring Technologies

Numerous elements of technology monitoring are incorporated
in farm monitoring. These include the effect that using the
technology has on other aspects of the farming system and the
impact that this has on the agricultural production of the farm.
Depending on how farm monitoring is done, it may or may not
provide an indication of the proportion of farms in a community
which adopt a technology, and how its affect may vary according
to strata or farm type. This also is necessary to monitor the
impact of the technology. Finally, it is necessary to know the
limits of adoption, both in terms of geography and farm type, to
estimate the total impact on the region. Such information should
be gathered over time to monitor the adoption of Tamazulapa beans
and Amina rice. If these varieties can be distinguished from
others, it may be possible to estimate their impact by monitoring
the these varieties through major markets, relative to other
varieties. There is also a need to learn how farmers adapt or
change their use of a technology over time as they learn to
manage it and integrate it thoroughly into their farming system.


2.9.2 Monitoring Social Relationships

One of the keys to successful extension is to study rural
institutions and their capacity to promote farmers' adoption of
innovations. This assumes that over time there will be changes
in rural institutions and social relationships that help increase
farmer capacity to change. While recognizing that such changes
may take place very slowly, an effort should be made to monitor
this change and evaluate the extent to which it was brought about
by extension activities.


2.9.3 Monitoring the Project

ADS-II has produced more documentation and devoted more
effort to monitoring its own activities than many projects. It
has prepared 18 documents which describe the design and
preparation of the 3 major project components. Much less
information is available on the implementation of and progress in
the three major components. Annual reports provide some of the
information, but often in a form that does not lend itself to
analysis over time. It would be useful, for example, to do a
description of the evolution of maize/bean trials over the
various seasons, and then update it periodically. Certain
research topics like fertilizer and density/spacing trials seem
to have been largely dropped from the testing program with little
indication of why they were dropped. Documenting failures so
that others can avoid the same mistake is just as important as
documenting successes. The physical and human resources and











administration necessary to implement the national survey should
be documented, since it is likely to be a major issue in the
sustainability of that component of the project. The training
program also needs to be carefully monitored since the project
seems to be meeting its specific training objectives in terms of
the number of degree programs, short courses, and workshops
attended, but without making much progress towards the goal of
training teams with the capacity to do effective FSR/E.
Integration of socio-economics into the farming systems
component, in terms of staff and research substance, should also
be monitored since it is a necessary but not sufficient condition
for attaining effective interdisciplinary FSR/E.

2.10 Sustainability, replicability, and recurrent Costs

2.10.1 The farming systems component

As yet, the farming systems component of the project is not
functioning effectively. This is due at least in part to the
failure to establish an effective interdisciplinary approach, the
lack of socio-economic input into the FSR/E process, and the
failure to establish and train integrated interdisciplinary
farming systems teams at the zone level to design, implement and
evaluate local activities. Since what exists to this point is
not something the evaluation team would want to see replicated,
expansion to an additional zone or zones is strongly discouraged
at this time. The evaluation team has proposed an internal
restructuring of the project that should help resolve many of
these problems. With such restructuring, the evaluation team is
confident that a methodology and model of effective FSR/E can be
established within the remaining two years of the project. It is
this model and methodology that one would then hope to replicate.

It is particularly the methodology and model of effective
FSR/E that one would hope to see sustained over time, helping
provide adapted technical innovations to improve Haitian
agriculture as well as innovations concerning the process of
disseminating new techniques to farmers. Until this methodology
and model are actually established and then function effectively
over a period of time to train the staff and prove the
credibility of the FSR/E approach, it can not be sustained.
Training the staff and establishing credibility is often a long
term process and would probably require an additional four years
of project activity. The project must prove that this model and
methodology can accomplish something that the existing
Recherche/Developpement forms of FSR/E do not.

But the financial sustainability of this activity like other
development programs in Haiti, will be in question for the
foreseeable future unless there is continued funding from USAID
or at least Title III. Even the salaries of Haitian staff seem
to be supplemented by projects such that they would be
redistributed to other activities without a continued source of












external funding. For these reasons, the sustainability of the
farming systems component and the project in general may depend
largely on their contribution to USAID's strategy and activities
for helping develop Haitian agriculture.

2.10.2 The National Agricultural Survey

ADS-II has designed and conducted a large scale agricultural
survey at the level of the two southern regions which include
Jacmel and Les Cayes. The project intends to replicate this at
the national level covering all regions within the next year.
Project personnel believe they have the capacity to organize and
administer a survey of national scope, using 180 enumerators.
This task requires tremendous administrative manpower and a very
large data processing and analysis capability. There has been a
tendency to deprive the farming systems component of much needed
socio-economic manpower to conduct the survey in the two regions.
This should not be allowed and the evaluation team has proposed
restricting the new TA socio-economist to the farming systems
component so that it does not continue. It is not yet clear if
the project has the administrative and data processing capability
to conduct a national survey several times larger than the
present two-region activity.

The national agricultural survey requires a high level of
capability in two different areas: logistical organization and
administration and the ability to design, analyze, verify and
modify the survey itself. To develop a Haitian capacity to
undertake these activities without technical assistance will
require degree training for several people as well as
considerable on-the-job experience. ADS-II is providing only one
Master's degree program in statistics. On the logistics and
administration, degree training is necessary for a good
conception of what has to be done, but job experience is even
more critical. People with administrative skills need to be
identified and then trained to handle the very difficult job of
running this kind of activity. Several people also need to be
trained very well in survey and census methodology. These people
need the conceptual ability to improve the design and
verification activities as well as to do the analysis and
interpretation. Such training might combine work in Demography
and Agricultural Economics with regard to large scale surveys,
and might profitably include some form of internship at USDA
working with an Agricultural Census. Until people with such
training and considerable experience are available, the national
survey needs technical assistance to continue.

Training alone does not insure sustainability because of
Haiti's financial constraints. This type of activity is
expensive and will require project or Title III funding for the
foreseeable future.











2.10.3 CRIES

CRIES is a high-tech tool for storing and managing large
amounts of data related to geographical units which can be used
in agricultural sector and project planning. Several agencies
both within and outside the Ministry of Agriculture are seriously
considering purchase of the computer equipment to use the CRIES
system. Basic training can be provided by ADS-II personnel so
that access to and use of the data base is replicable. But the
continued development of the data base and its effective use for
project planning or questions of agricultural sector policy
requires a very high level of conceptual and analytical
capability. Sustainability would require providing Haitians with
high level training and extensive experience to take over this
function. Such training is not presently provided in the
project.

On the financial side, the high-tech equipment is rather
expensive to purchase and maintain. Data entry and digitizing
require maintaining a staff of people with computer skills.
Although not requiring the number of staff or the expense of the
other two components, the Haitian government would have great
difficulty absorbing the CRIES activity on its very limited
budget, without the help of project or Title III funding.

As can be seen from this analysis, most of the ADS-II
project activities will not be sustainable at the end of four
years. Most of the activities require more formal degree
training than is provided in the project and more on-job
experience than will be provided within a four year period.
Given the Haitian government's extremely difficult budget
situation, it is doubtful that any project activity would be
continued without project or Title III funding.












3. Findings Concerning FSR/E

ADS-II has a number of accomplishments which make it a
worthwhile and, in part, successful project. It has established
a large-scale on-farm testing program which has already scored
two major successes. But ADS-II has been less successful in
establishing an effective interdisciplinary model and methodology
of FSR/E.

ADS-II achieved a major accomplishment by moving from a
reconnaissance survey and its verification to establishing
on-farm trials in about a three month period. The extensive
on-farm testing program covered 293 researcher-managed and 285
farmer-managed trial replications in 1985. Two major successes
have emerged from this testing program, Tamazulapa beans and
Amina rice, as well as the extension opportunity offered by these
many farmer contacts.

The ADS-II project contains a flaw of either design or early
implementation which has seriously impeded the farming systems
component of the project. Socio-economists were physically and
organizationally separated from the farming systems component,
because they were lumped administratively and functionally
together in one unit called the Economie Rurale et Statistique
Section (ERS). This resulted in a situation where the farming
systems teams were not interdisciplinary and had little or no
socio-economic input. They were concentrating entirely on
agronomic testing even though the farming systems within zones
had not been fully characterized and little progress was being
made in the identification of farmer problems and constraints.
Structural changes which fully integrate socio-economists into
the farming systems component must be made. No geographical
expansion or substantive expansion into extension should be
considered until these shortcomings are resolved and a more
complete and effective farming systems research program is
established.

Activities in both Jacmel and Les Cayes show promise of
producing results. Jacmel has devoted a major effort to working
in the hillside environment of Haute Cap Rouge. Erosion control
activities in that area show promise, even though the technique
is different thanthose that will have to be used in many areas of
the country. The possibility of centralizing activities and
attempting to manage a small watershed in a fairly complete
manner is interesting. Les Cayes activities are interesting for
different reasons, but particularly because the region plays an
important role in future AID activities. Although less advanced
in their hillside activities, the rice activity on the plain
promises to have an important impact on agricultural production.

All surveys used by the project have been based on formal
questionnaires. With the exception of the initial reconnaissance










survey, they have all focused on quantitative data. Techniques
like informal surveys and diagnostic surveys, case studies, or
anthropological studies which would provide qualitative
information with the possibility of achieving a deeper
understanding of the farming systems) have been neglected.

The project organization dictates that the basic farming
systems teams are found at the zone level (Haut Cap Rouge, Bas
Cap Rouge, Berault and Maniche). Interdisciplinary integration
should be promoted precisely at the zone level. Basic teams
should be provided with the necessary training and guidance to
design, implement and analyze their own agronomic tests as well
as diagnostic and monitoring surveys. This implies that
technical assistance in particular, and the project in general,
should take on a much stronger training function.

ADS-II has not established an effective training program for
FSR/E. Zone level project staff, who should be the basic farming
systems team, and students primarily receive experience in the
management and execution of on-farm trials. On a personal basis
they may ask questions and receive instruction from the zone
chief or the regional coordinator. They do not receive any
formal training. They are not trained to design, analyze or
interpret test results. They receive no training in FSR/E
methodology and they do not get any experience working in an
interdisciplinary environment. They do not participate in and
are not trained to do reconnaissance or diagnostic surveys.
Except for the study on pigs, they are not involved in farm
monitoring activities. Their understanding of FSR/E is based on
their training at the University and their experience working in
other projects which use the Recherche Developpement approach.

Many of the farmer managed tests are not pre-extension
trials in the sense of approximating farmers conditions.
Fertilizer is used on most trials even though survey results show
that most small farmers do not use fertilizer. These trials
provide little insight into how farmers will adapt a technology
when using it on their own, or what the long term impact of the
innovation might be. Farmer managed trials are often used to
compare a complete technical package to farmers traditional
practices. Farmers typically do not adopt entire packages at one
time, especially if they include an expensive input like
fertilizer. Such tech pack demonstrations do not show the
farmers what individual factors might improve their production
and be within their means.

Haiti lacks national institutions doing effective
agricultural research or effective agricultural extension. This
places a farming systems project like ADS-II in a difficult
situation. Establishing linkages to the decentralized research
and extension systems require even more effort than linking to
centralized institutions. An informal network of research and
extension projects/personnel does exist as an outgrowth of the
Faculty. The project need to improve linkages to the FAculty and











this informal network. Haitians who are trained by the project
and return to teach at the University will provide the greatest
impact on this group. Training activities and project
participation for students, recent graduates, research
specialists and extension personnel can also be used to help
improve this relationship once the project has achieved
credibility. Projects and PVO's are an important part of the
decentralized extension system. Such organizations can provide
extension outreach of technologies tested by the project if more
effort is devoted to establishing the necessary liaison.

ADS-II is presently attached to the MARDNR, but is not
integrated into its functional structure. Politically motivated
changes in Ministry personnel led to a serious conflict in
personalities which required that the project be moved out of the
Faculty. If the project is structurally integrated into the
Ministry, political changes might again threaten its function
capacity. Although not ideal, the practical solution to the
projects' institutional location seems to be to maintain the
present relationship until the new Ministry has proven its
ability to function effectively and the political situation is
stable.

The ADS-II interpretation of FSR/E has been flawed and
incomplete in several respects, particularly with regard to
institutional problems. It has not as yet begun to study rural
institutions and their effects on adoption of innovations.
Innovations have not been monitored after farmers had begun to
use them on their own, to see what changes take place in the
farming system as a whole. It is not clear that the researchers
are sufficiently aware of labor and resource constraints to
include such considerations in their selection of technologies.
The importance of different innovations and the size of the
target group to which they apply have not been specifically
evaluated.

The definition of extension has also been too narrow and
needs to be expanded to include the promotion of rural
institutions (organizations) which facilitate the adoption of
innovations by farmers. Although the scope of this project does
not include the possibility of a large extension program, it can
contribute to a better understanding of rural institutions and
how extension might involve them in the transfer of technology
process. The ability to identify and analyze this potential of
rural institutions needs to be built into extension training
programs.








95



The technical assistance team leader has three nearly full
time functions in administration, farming systems socio-economic
research, and the national survey and CRIES activities. He
cannot fill all these functions. The addition of an
administrative assistant would allow him to fulfill more of his
technical duties.

















Technical Annex 2:

Agronomic Considerations Concerning ADS-II

By Mimi Gaudreau


Contents:


Screening Technology
On Farm Testing
Design of Alternative Technology
Findings
Environmental Impact













1. Screening Technology

ADS-II has generally been screening technology directly on
farmers' fields. The teams have taken varieties from sources in
Haiti and from the International Centers and put them in farmers'
fields as researcher-managed trials. Many of the varietal tests,
particularly for corn, beans, and rice, have progressed rapidly
from researcher-managed to pre-extension trials and large scale
seed multiplication within the span of 1 1/2 yrs. While good
technologies can take off and be accepted by farmers in a short
amount of time, there are some issues posed by the ADS-II methods
that need to be addressed.

ADS-II has evaluated the researcher-managed trials primarily
on the basis of yield increase and to a more limited extent
performed economic analyses of the results. By pooling the
results from many sites, the teams have been able to determine
the range of yields one might expect with the technology in
question. They have made little attempt to determine if there
are general characteristics about farmers or about the sites
that are usually associated with a particular yield potential.
For whom is the technology appropriate and under what
conditions?

For some of the interventions, particularly corn and beans,
field tests have been conducted since July, 1984. The first
major trials were put in second season, 1984, with follow-up
research first season 1985. In many cases, by second season,
1985, the trials were at the pre-extension stage and larger scale
seed multiplication trials were put in during first season 1986.

Essentially there are research results for two second season
and one first season plantings. How different are the climatic
and biological (insects, plant disease) conditions of first
season compared to second season? What do farmers do differently
these two seasons? Do farmers plant different local varieties
of the same crop during each of these seasons? Do farmers look
for different plant characteristics e.g. disease resistance,
growth habit etc. of the same crop for each season? This is just
a partial list of some of the questions ADS-II researchers
should be considering in screening their interventions. They
may or not be important in Haiti but it is unclear that they
have been addressed at all.

Putting technologies directly on farmers' fields can pose
problems. Because the technologies are still being researched,
there is a certain element of risk involved. Some farmers are
just not in a position to accept that risk. If something goes
wrong with the trials researchers may lose the confidence of
their cooperating farmers. The ADS-II researchers have had to
deal with this problem several times. In one case a sorghum











variety was being tested that did not head out at the same time
as the local varieties. Farmers got impatient and took the
plants out for cattle feed and demanded compensation for the lost
crop. It is important to build in a compensation provision and
to know as much as possible about the materials being tested
before they go into farmers' fields.

If interventions are too risky or not enough is known about
their potential impacts, they should be tested first in a
controlled environment. The Les Cayes team is currently testing
pearl millet at the Levy Farm. Pearl millet is a new crop for
the Les Cayes area. Depending on the results of this work, the
team may try to introduce it into the area. Levy Farm would be
available for additional experimental work and for seed
multiplication if ADS-II wanted to rent land there. Levy Farm
is not a research station and would need to be reimbursed for
the loss of production on plots used by ADS-II. A formalized
arrangement with Levy for some types of technology screening
might be worth considering.

In Jacmel, ADS-II has been working with Foster Parents Plan
on their "model" farm. This would provide a site for both seed
multiplication and some screening for the plains work in Bas Cap
Rouge. For both regions, screening that is too risky to go onto
farmers' fields on the mountain slopes should be done on rented
land that is totally under researcher control.

When screening technologies, it is important to use farmers'
criteria for evaluating interventions. ADS-II has incorporated
criteria such as plant height, ear height & fullness (for corn),
disease resistance (bean), taste, cooking characteristics and
grain quality (rice), grain color and flour quality for maize as
well as spathe cover. The researchers have talked to women
about the technologies being tested, particularly about those
aspects related to women's domain post-harvest activities.

Much of this discussion has been oriented to varietal
screening since to date most of the ADS-II has been in this area.
Initially there were some fertilizer experiments conducted but
little additional work has been done with form of fertilizer,
dose, and time and method of application to ensure that the most
cost effective method would be found. Fertilizer as well as
pesticide use have been treated as improved cultural practices
and put forward as a package. More work needs to be done to
break the package down into component parts that can be adopted
in a sequential manner by farmers.

Fertilizer use has also been incorporated into the selection
of local variety (chicken corn and Alizene) plots. The
characteristics used in the selection process include plant
height, ear height, covered spathe, ear size and fullness.
Selecting plants that produce well under favorable conditions may
ultimately reduce the ability of the plant to perform well over a
wide range of conditions. Local varieties have adapted to be











able to produce well in a wide range of environments. If high
yield potential materials are selected and farmers are unable to
provide favorable conditions e.g. fertilizer, pesticide, the
results may not be as promising as expected.

A final aspect in screening technologies that needs to be
considered is what happens prior to putting trials in the field.

Usually in selecting technologies that go into farmers'
fields the FSR team goes through a process called ex ante
analysis. It can be done formally or informally. It is an
evaluation of expected results and consideration of system
interactions such that some technologies may be screened out at
this point and never make it to the field. Some factors that
are taken into account at this point include: Are the inputs
available on a regular basis? Does this technology put
additional burdens on the farm family? If so, what members?
How much yield increase is expected? Is it economically
advantageous? Does it generally meet the criteria of
acceptability for farmers taste, milling characteristics, seed
quality?

One tries to compare what is already known about the
technology with what is known about the system to see if they
are compatible. The researcher also try to anticipate problems
and speculate on possible solutions or the resulting impact.

When the rice harvesting and threshing equipment were
introduced in Les Cayes the researchers of ADS-II gathered
information to try to anticipate the farmers' reactions and
problems. They seemed to have done an ex-ante analysis in this
case. For some of the other technologies, it is unclear whether
or not this type of analysis has been performed before going to
farmers' fields.


2. On Farm Testing

The on-farm research program of ADS-II has been divided into
researcher managed trials (essai type chercheur) and
pre-extension or verification trials (essai type
pre-vulgarization or essai de verification). They differ in
complexity and in the level of farmer participation or
management.

The researcher-managed trials tend to be more complex
(split-plot, randomized complete block (RCB)) with several
treatments and no more than two repetitions(blocks) per site.
The verification trials are simpler, usually a comparison of
local variety versus improved variety or farmer practice versus
improved practice. The farmers assume more responsibility for
field activities and the researchers record what happens and
collect the agronomic data. Both types of trials are conducted




Full Text
xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID EZPG0CKDE_YQ7DOS INGEST_TIME 2013-03-21T13:43:17Z PACKAGE AA00011404_00001
AGREEMENT_INFO ACCOUNT UF PROJECT UFDC
FILES