Front Cover
 Front Matter
 Table of Contents
 List of Tables
 List of Figures
 Background and objectives
 Research design
 Data collection procedures
 Description of farming systems...
 Analysis of selected maize and...
 Priorities for adaptive production...
 Evaluation of methods
 Summary and conclusions
 Appendix A. Questionnaire for identifying...
 Appendix B. Informal survey...
 Back Cover

Title: Planning an adaptive production research program for small farmers
Full Citation
Permanent Link: http://ufdc.ufl.edu/UF00086774/00001
 Material Information
Title: Planning an adaptive production research program for small farmers a case study of farming systems research in Kirinyaga district, Kenya
Series Title: Planning an adaptive production research program for small farmers
Physical Description: 3, xiii, 279 i.e. 283 leaves : ill. ; 29 cm.
Language: English
Creator: Franzel, Steven Charles
Publication Date: 1983
Subject: Agriculture -- Research -- Kenya   ( lcsh )
Genre: bibliography   ( marcgt )
non-fiction   ( marcgt )
Spatial Coverage: Kenya
Thesis: Thesis (Ph. D.)--Michigan State University. Dept. of Agricultural Economics, 1983.
Bibliography: Includes bibliographical references (leaves 273-279).
Statement of Responsibility: by Steven Charles Franzel.
 Record Information
Bibliographic ID: UF00086774
Volume ID: VID00001
Source Institution: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
Resource Identifier: oclc - 11402069

Table of Contents
    Front Cover
        Front Cover 1
        Front Cover 2
    Front Matter
        Front Matter
        Page i
        Page i-a
        Page ii
        Page iii
    Table of Contents
        Page iv
        Page v
        Page vi
        Page vii
        Page viii
        Page ix
    List of Tables
        Page x
        Page xi
        Page xii
        Page xiii
    List of Figures
        Page xiv
    Background and objectives
        Page 1
        Page 2
        Page 3
        Page 4
        Page 5
        Page 6
        Page 7
        Page 8
        Page 9
        Page 10
        Page 11
        Page 12
        Page 13
        Page 14
        Page 15
        Page 16
        Page 17
    Research design
        Page 18
        Page 19
        Page 20
        Page 21
        Page 22
        Page 23
        Page 24
        Page 25
        Page 26
        Page 27
        Page 28
        Page 29
        Page 30
        Page 31
        Page 32
        Page 33
        Page 34
        Page 35
        Page 36
        Page 37
        Page 38
        Page 39
        Page 40
        Page 41
        Page 42
        Page 43
        Page 44
        Page 45
        Page 46
        Page 47
        Page 48
        Page 49
        Page 50
        Page 51
        Page 52
        Page 53
        Page 54
    Data collection procedures
        Page 55
        Page 56
        Page 57
        Page 58
        Page 59
        Page 60
        Page 61
        Page 62
        Page 63
        Page 64
        Page 65
        Page 66
        Page 67
        Page 68
        Page 69
        Page 70
        Page 71
        Page 72
    Description of farming systems in the study area
        Page 73
        Page 74
        Page 75
        Page 76
        Page 77
        Page 78
        Page 79
        Page 80
        Page 81
        Page 82
        Page 83
        Page 84
        Page 85
        Page 86
        Page 87
        Page 88
        Page 89
        Page 90
        Page 91
        Page 92
        Page 93
        Page 94
        Page 95
        Page 96
        Page 97
        Page 98
        Page 99
        Page 100
        Page 101
        Page 102
        Page 103
        Page 104
        Page 105
        Page 106
        Page 107
        Page 108
        Page 109
        Page 110
        Page 111
        Page 112
        Page 113
        Page 114
        Page 115
        Page 116
        Page 117
        Page 118
        Page 119
    Analysis of selected maize and bean practices
        Page 120
        Page 121
        Page 122
        Page 123
        Page 124
        Page 125
        Page 126
        Page 127
        Page 128
        Page 129
        Page 130
        Page 131
        Page 132
        Page 133
        Page 134
        Page 135
        Page 136
        Page 137
        Page 138
        Page 139
        Page 140
        Page 141
        Page 142
        Page 143
        Page 144
        Page 145
        Page 146
        Page 147
        Page 148
        Page 149
        Page 150
        Page 151
        Page 152
        Page 153
        Page 154
        Page 155
        Page 156
        Page 157
        Page 158
        Page 159
        Page 160
        Page 161
        Page 162
        Page 163
        Page 164
        Page 165
        Page 166
        Page 167
        Page 168
        Page 169
        Page 170
        Page 171
        Page 172
        Page 173
    Priorities for adaptive production research
        Page 174
        Page 175
        Page 176
        Page 177
        Page 178
        Page 179
        Page 180
        Page 181
        Page 182
        Page 183
        Page 184
        Page 185
        Page 186
        Page 187
        Page 188
        Page 189
        Page 190
        Page 191
        Page 192
        Page 193
        Page 194
        Page 195
        Page 196
        Page 197
    Evaluation of methods
        Page 198
        Page 199
        Page 200
        Page 201
        Page 202
        Page 203
        Page 204
        Page 205
        Page 206
        Page 207
        Page 208
        Page 209
        Page 210
        Page 211
        Page 212
        Page 213
        Page 214
        Page 215
        Page 216
        Page 217
        Page 218
        Page 219
        Page 220
        Page 221
        Page 222
        Page 223
        Page 224
        Page 225
        Page 226
        Page 227
        Page 228
        Page 229
        Page 230
        Page 231
        Page 232
        Page 233
        Page 234
        Page 235
        Page 236
        Page 237
        Page 238
        Page 239
        Page 240
        Page 241
        Page 242
        Page 243
        Page 244
        Page 245
        Page 246
        Page 247
        Page 248
        Page 249
        Page 250
        Page 251
    Summary and conclusions
        Page 252
        Page 253
        Page 254
        Page 255
        Page 256
        Page 257
        Page 258
        Page 259
        Page 260
        Page 261
        Page 262
        Page 263
        Page 264
        Page 265
        Page 266
        Page 267
    Appendix A. Questionnaire for identifying recommendation domains in middle Kirinyaga
        Page 267a
        Page 267b
        Page 268
        Page 269
    Appendix B. Informal survey guidelines
        Page 270
        Page 270a
        Page 271
        Page 272
        Page 273
        Page 273a
        Page 274
        Page 275
        Page 276
        Page 277
        Page 278
        Page 279
    Back Cover
        Page 280
        Page 281
Full Text

Planning an Adaptive Production
Research Program for
Small Farmers: A Case Study
of Farming Systems Research
in Kirinyaga District, Kenya

Steven Charles Franzel

A Dissertation

Submitted to
Michigan State University
in Partial Fulfillment of the Requirements
for the Degree of

Doctor of Philosophy

Department of Agricultural Economics


SCopyright by Steven Charles Franzel
All Rights Reserved




Steven Charles Franzel

This thesis uses the farming systems research (FSR) methods of the

International Maize and Wheat and Improvement Center (CI'MMYT) to plan an

experimental program for farmers in Middle Kirinyaga, Kenya, and to

address several methodological issues concerning FSR. The approach

includes three stages: (1) interviews with extension agents to identify

recommendation domains (RD's), i.e., fairly homogenous groups of

farmers; (2) an informal survey in which researchers interview farmers;

and (3) a formal sample survey. An agronomist collaborated with the

author in mounting the research.

The two RD's identified in Middle Kirinyaga were high income

farmers and low income farmers. Farmers' circumstances are described

and "leverage points" are identified, which represent opportunities for

increasing productivity in ways acceptable to and feasible for farmers.

An experimental program is presented; the two most important research

priorities are:

1. Improving soil fertility and structure through on-farm
experiments to test the effectiveness of readily available
coffee husks as manure.

2. Reducing the draught power bottleneck by selecting bean
cultivars with superior ability to withstand dry planting,
treating seeds against ant damage, and deeper planting.

Two methodological issues are addressed. The first is how to

obtain normative and prescriptive information, i.e., information on

Steven Charles Franzel

farmers' values and decisions. Two techniques, repertory grid (RG) and

hierarchical decision tree models (HDM), are incorporated into the

informal and formal surveys and are evaluated. The techniques were

found useful for assembling data concerning preferences and decisions

in a systematic fashion and for assisting the researcher to develop an

understanding of farmer decision-making.

The second methodological issue concerns the quality of data at

different stages of the investigation. First, data from the RD-

identification exercise are evaluated in comparison to those of the

formal survey. The exercise is found to be reasonably effective for

tentatively classifying farmers into RD's.

Next, the utility of the formal survey is evaluated by comparing

its results with those of the informal survey. The formal survey con-

tributed relatively little to the understanding of farmers' practices

and constraints or to the experimental program developed in the infor-

mal survey. These findings support the hypothesis that the informal

survey can be an effective and sufficient method for planning experi-

mental programs for farmers.


I wish to express my thanks to the individuals and institutions

which provided me with support and counsel throughout my graduate program.

First, I would like to thank Professor Carl Eicher, chairman of my

guidance committee during most of my program, for his invaluable advice

and interest in my development. I would also like to gratefully acknow-

ledge the guidance and help of Professor Eric Crawford, my thesis super-

visor and major professor. Sincere appreciation goes to Dr. Michael

Collinson of the International Maize and Wheat Improvement Centre's

(CIMYT) Eastern Africa Economics Program for his assistance and support

in the design and execution of my dissertation project.

I am grateful for the financial and personal support offered to me

by the Department of Agricultural Economics, Michigan State University,

during my graduate studies. A commendation is due to Professor Lester

Manderscheid for his assistance in making many important decisions

throughout my program and to Drs. Derek Byerlee and Peter Matlon who

supervised my Masters work. Professors Glenn Johnson, Harold Riley,

Russ Freed, and Carl Leidholm offered helpful counsel and comments on

my dissertation.

Appreciation also goes to the CIrMYT Economics Program, which pro-

vided me with generous assistance for carrying out my fieldwork and for

supporting me during part of my write-up phase. During the field phase

of this project, assistance came from several additional sources. First,

I thank all the farmers who willingly sacrificed their time to participate

in this research project. I am also grateful for the cooperation and

assistance of a number of institutions: Scientific Research Division

(SRD) of the Ministry of Agriculture of.Kenya, the Kenya Agricultural

Research Institute, and the District Agricultural and District Admini-

strative Offices of Kirinyaga. I also thank Mr. Njogu Njeru,

formerly Senior Maize Research Officer, SRD, who collaborated with me

in mounting the informal survey and drawing up the experimental program.

Appreciation also goes to my research assistant, Mr. Simon Muthangia,

my four enumerators, and to the sub-chiefs and extension agents of

Middle Kirinyaga, whose cooperation was invaluable. Others who provided

useful assistance include Mr. Fred Chege, Dr. H. A. van Rheenan,

Ms. Juliet Gathee, and Dr. Christina Gladwin.

fly thanks also go to the staff of the Agricultural Economics Com-

puter Service at Michigan State University, particularly Susan Chu and

Paul Wolberg,for their assistance in helping analyze my data. I am also

grateful to Cindy Spiegel for typing the draft and tables for this dis-

sertation and to Lois Pierson for typing the final draft. Finally,

thanks go to my family and friends for their support, and assistance in

completing my graduate program.



List of Tables . .. . . .

List of Figures .........

List of Maps . . . . . . . . . . .

. . xiii




Background to the Methodological Objectives
Statement of the Methodological Objectives
Background to the Problem-Solving Objectives
Statement of the Problem-Solving Objectives
Overview of the Thesis . . . . .

RESEARCH DESIGN . . . . . . .
The Planning Stage of FSR . . . . .
Analytical Approach . . . . . .

. . 1
. . 1

. . 14
. 15

. . 18

S. 19
S. 24

2.3. Theoretical Contributions from Managerial Decision
Theory . . . . . . . . . . .
2.4. Some Practical Considerations in Selecting Methods
for Data Collection . . . . . . . .
2.5. A Review of Approaches to Modeling Farmer
Decision-Making . . . . . . . . .
2.6. Issues Concerning the CIMMYT Diagnostic Survey .
2.6.1. Collecting Information on Farmers'
Values and Decisions . . . . . .
2.6.2. The Utility of an RD-Identification
Exercise and a Formal Survey . . . .
Utility of the RD-Identiffication
Exercise . . . . . . . .
Utility of the Formal Survey . . .
2.6.3. Holism . . . . . . . . .
2.7. The Repertory Grid (RG) Technique and Hierarchical
Decision-Tree Models (HDM) . . . . . .
2.7.1. The Repertory Grid Technique . . . .
2.7.2. Hierarchical Decision-Tree Models . . .
2.8. Summary of Methods . . . . . . . .





S 41



3.1. Selection of Middle Kirinyaga as Study Area . .
3.2. Identifying Recommendation Domains . . . .
3.3. The Informal Survey .. . .. .. ... .
3.3.1. Who to Interview . . . . . . .
3.3.2. Interview Guidelines . . . . . .
3.3.3. Sampling Methods and Reporting of Results
3.4. The Formal Survey . . . . . . . . .
3.4.1. Stratification and Sample Size . . .
3.4.2. Selection of Sample Farmers and Organi-
zation of Fieldwork . . . . . .
Accommodating Logistical Constraints .
Classifying Farmers into Income Strata
Irregularities in the Sample Frame . .
3.4.3. Execution and Analysis . . . . .


4.1. Physical and Socio-Economics Features . . .
4.1.1. Physical Features . . . . . .
4.1.2. Population and Settlement . . . .
4.1.3. Transportation and Marketing . . .
4.1.4. Cooperatives, Credit, and Government
Agricultural Institutions . . . .
4.2. Farmer Objectives and Sources of Income . . .
4.3. Management of the Farming System . . . .
4.3.1. Enterprise Pattern and Land Use . . .
Long Rains Crops . . . . . .
Short Rains Crops . . . . .
4.3.2. Livestock . . . . . ....
4.3.3. Crop Calendar and Management of Food
Supplies . . . . . . . .
4.4. Resource Use and System Constraints . . . .
4.4.1. Land . . . . . .
Land Tenure, the Land Market and Land
Constraints . . . . .
Soil Fertility and Structure . . .

S 55
S 55
S 56
S 62
S 63
S 64
S 66
S 66

. 73
. 73
. 78
. 80

S. 82
. 83
. 88
. 88
. 88
S. 92
. 92

S. 99

S 99
S. 02



4.4.2. Labor . . . . . . . .... 103
Labor Use, Peak Season Labor and the
Labor Market . . . . . . . 103
Labor: A Constraint to Developing the
System? . . . . . . 107
4.4.3. Cash . . . . . . . .... 109
4.4.4. Access to Draught Power . . . . .. 114
4.5. Leverage Points in Middle Kirinyaga . . . .. 117

5.1. Long Rains Management . . . .... ..... .121
5.1.1. Maize Variety Choice . . . . ... 121
Grow Katumani for Early Maize . . .. 127
Grow Katumani for Storage and Sale . .. 130
Grow Hybrid 511/512 for Main Stock of
Maize . . . . . . . ... .135
5.1.2. Bean Variety Choice . . . . .... .139
5.1.3. Method and Timing of Land Preparation and
Planting . . . . . . . ... .143
Plant Population . . . . .. . 156
5.1.4. Maintenance of Soil Fertility and Structure 158
5.1.5. Other Production Practices and Use of
Purchased Inputs . . . . . .... .161
Weeding . . . . . . . ... .161
Pest Control . . . . . .... .162
Use of Purchased Inputs . . . ... 162
5.1.6. Harvesting and Production . . . ... 164
5.2. Short Rains Management . . . . . ... 167
5.2.1. Maize Varieties . . . . . .... .167
5.2.2. Time and Method of Planting . . . ... 167
5.2.3. Use of Purchased Inputs . . . . .. 168
5.3. Farming Systems in Transition: A Summary of
Chapters 4 and 5 . . . . . . . .... 168

6.1. On-Farm and On-Station Experimentation ...... 175
6.2. Outline of Research Priorities . . . . .. 177


6.3. Improve Soil Fertility and Structure . . ... 178
6.3.1. Use of Coffee By-Products as Manure for
Maize and Beans . . . . . . . 179
Proposed Experiment for Low Income
Farmers . . . . . . . . .181
Proposed Experiment for High Income
Farmers . . . . . . . . .182
6.3.2. Training Farmers on Conservation and Use
of Livestock Manure . . . . .... .182
6.3.3. Training Farmers on Correct Method of
Fertilizer Application . . . . .. 183
6.4. Ease the Draught-Power Bottleneck to Permit
Earlier Planting . . . . . . . . . 183
6.4.1. Make Dry Planting More Attractive to
Farmers . . . . . . . . . 183
Proposed Research/Extension Demon-
stration . . . . . . . . .185
6.4.2. Develop Infrequent-Tillage and No-Tillage
Systems . . . . . . . . ... 186
Proposed Experiment for Low Income
Farmers . . . . . . . ... 186
6.4.3. Encourage Farmers to Stagger Plow/Plant
Their Fields .. .. . . .... 187
6.4.4. Increase the Rate of Work and Efficiency
of Ox-Plow Teams . . . . . ... 187
6.5. Improved Maize Varieties . . . . . ... 187
6.5.1. Main Stock Maize Variety for Low Income
Farmers . . . . . . . . ... 188
Proposed Variety Experiments for Low
Income Farmers . . . . . .... .189
6.5.2. Varieties for Early Maize . . . ... 189
Proposed Experiment: Farmer Evaluation
of Dryland Composite . . . . . 190
6.6. Improved Bean Cultivars . . . . . .... 190
6.6.1. Selecting a Cultivar Which Can be Planted
Dry . . . . . . . . . . 191
Proposed Experimentation .. .. .. 192
6.6.2. Selecting a Palatable Variety with High
Returns to Seed Cost--Low Income Farmers . 193


vii I



Improved Storage Management for Maize . . . .
Sunflower Research . . . . . . . .

Evaluation of Repertory Grid and Hierarchical
Decision Modeling . . . . . . . .
7.1.1. Repertory Grid . . . . . . .
7.1.2. Hierarchical Decision Modeling . . .
Limitations of HDM . . . .. . .
7.1.3. Use of RG and HDM in the CIMMYT Diagnostic
Survey . . . . . . . . .

7.2. Quality of Information Gathered at Different
Stages of the Investigation . . . . . . .
7.2.1. Effectiveness of the Exercise for Identi-
fying Recommendation Domains ...
Evaluation of Data Describing RD's . .
Accuracy of RD Identification . . .
7.2.2. Evaluating the Utility of Carrying Out the
Formal Survey . . . . . . . .
Evaluation of Data from the Informal
Survey . . . . . . . . .
Contribution of the Formal Survey to
the Experimental Program . . . . .
Principal Constraints . . . . .
Principal Research Priorities . . .
Non-Experimental Variables . . . .
Experimental Program . . . . . .

8.1. Introduction . . . . . . . . . .
8.2. Developing an Adaptive Production Research Program
8.3. Selected Issues in the Evaluation of Methods . .
8.3.1. Repertory Grid and Hierarchical Decision-
Tree Modeling . . . . . . . .
8.3.2. Quality of Data at Different Stages of
the Investigation . . . . . . .
8.4. Suggestions for Further Research . . . . .
8.4.1. Research on Measuring Effectiveness .. ...
8.4.2. Research on Improving Performance . . .



















BIBLIOGRAPHY . . . . . . . . ... ...... .273


Table Page

2.1 Sample Repertory Grid of Important Attributes of
Bean Cultivars . . . . . . . . . . 46
3.1 Comparison of Two Methods for Classifying Farmers
into Income Groups: Classifications Based on Infor-
mation Supplied by Assistant Chief and Classifica-
tions Based on Evaluation of Survey Questionnaires,
Middle Kirinyaga, 1981 . . . . . . .. .70
3.2 Sample Status of Farmers Interviewed in Formal
Survey, Middle Kirinyaga, 1981 . . . . .... .71
4.1 Rainfall Requirements for Maize and Probabilities of
Receiving Lower Amounts of Rainfall, Middle
Kirinyaga, 1953-81 . . . . . . . .... .77
4.2 Cash Income Levels of Households in Middle Kirinyaga,
July, 1980 to June, 1981 . . . . . .... .84
4.3 Sources of Income for Farmers in Middle Kirinyaga,
July, 1980 to June, 1981 . . . . . .... .86
4.4 Land Use and Average Farm Size in Middle Kirinyaga,
1981 . . . . . . . . . . . . 89
4.5 Principal Crop Combinations in the Short Rains and
Long Rains Seasons, Middle Kirinyaga, 1981 . . .. 90
4.6 Livestock Ownership and Uses, Middle Kirinyaga, 1981 93
4.7 Cropping Calendar for Maize and Beans, Middle
Kirinyaga, 1981 . . . . . . . . ... .95
4.8 Food Availability Calendar for Middle Kirinyaga
Farms, 1981 . . . . . . . . ... . 96
4.9 Sources of Cash for Purchasing Maize when Supplies of
Home Stocks Were Last Exhausted, Middle Kirinyaga,
1981 . . . . . . . . . . ... 98
4.10 Farmers Who Have Bought and Sold Land, Middle
Kirinyaga, 1981 . . . . . . . . . 101
4.11 Household Composition and Workers Available for Farm
Work per Household, Middle Kirinyaga, 1981 . . .. 104
4.12 Farmers' Opinions on Their Busiest Months and
Activities During Their Busiest Months, Middle
Kirinyaga, 1981 . . . . .. . . . . 105
4.13 Use and Cost of Hired Labor Among Farmers in Middle
Kirinyaga, 1981 . . . . . . . ... .106


Table Page
4.14 Farm Expenditures of High and Low Income Farmers,
Middle Kirinyaga, September, 1980-August, 1981 . .. 110
4.15 Most Difficult Months of the Year for Cash, Middle
Kirinyaga, 1981 . . . . . . . . . 111
4.16 How Farmers Would Spend an Extra 200 Shs. if They
Received It at Planting Time (March), Middle
Kirinyaga, 1981 . . . . . . . . . 113
4.17 Ownership of Oxen and Ox Plows, Middle Kirinyaga, 1981 116
4.18 Desired Time of Plowing/Planting Compared to Actual g
Time of Plowing/Planting for Farmers not Owning
Oxen, Middle Kirinyaga, Long Rains Season, 1981 . 116
5.1 Maize Varieties Grown in the Long Rains Season,
Middle Kirinyaga, 1981 ...... . . ...... 123
5.2 Repertory Grid Showing Farmers' Ratings of Selected
Maize Varieties Across Criteria They Consider *
Important, Middle Kirinyaga, 1981 . . . . .. 124
5.3 Farmers' Decisions on Whether or Not to Grow Katumani
for Early Maize and for Main Stock: Reasons for
Rejecting and Accepting, Middle Kirinyaga, 1981 . 128
5.4 Farmers' Decision on Whether or Not to Grow Hybrid
511/512 for Main Stock of Maize, Middle I
Kirinyaga, 1981 .. . . . . . . . 137
5.5 Bean Cultivars in Middle Kirinyaga, 1981: Seed
Characteristics, Percentages of farmers Growing,
and Area Grown . . . . . . . . . .. 140
5.6 Repertory Grid Showing Farmer Ratings of Selected g
Bean Cultivars Across Criteria They Consider I
Important, Middle Kirinyaga, 1981 . . . . .. 142
5.7 Farmers' Opinions on the Best Time to Plant Maize i
and Beans, Middle Kirinyaga . . . . .. . 144
5.8 Repertory Grid Showing Perceived Hazards of Planting
Maize and Beans at Various Times, Middle Kirinyaga,
1981 . . . . . . . . . . . . 145
5.9 Timing and Area of Maize and Bean Plantings, Middle
Kirinyaga, Long Rains Season, 1981 ...... . 148
5.10 Farmer's Success at Plowing/Planting at the Time They
Desired, Middle Kirinyaga, Long Rains Season, 1981 . 151
5.11 Method of Plowing/Planting for Farmers Not Owning Oxen I
Who Were Unable to Secure Oxen at the Time They
Needed Them, Long Rains Season, Middle Kirinyaga, 1
1981 . . . . ......... . 153
5.12 Farmers Planting Without Preparing Their Fields, Long
Rains Season, Middle Kirinyaga, 1981 ...... . 155



5.13 Plant Population for Maize and Beans, Long Rains
Season, Middle Kirinyaga, 1981 ...... ..
5.14 Fertilizer and Manure Use on Maize and Beans, Long
Rains Season, Middle Kirinyaga, 1981 . . . .
5.15 Farmer's Use of Purchased Crop Inputs for the Long
Rains Season, Middle Kirinyaga, 1981 . . . .
5.16 Production, Yields per Hectare, and Sales of Maize
and Beans, Middle Kirinyaga, 1980-81 . . . .
5.17 Farmer Practices in Cultivating Maize and Beans,
Middle Kirinyaga, 1981 . . . . . . .
6.1 Estimated Cost.. and Returns to Alternative Measures
for Increasing Soil Fertility, Middle Kirinyaga,
1,981 . . . . . . . . . . . .
6.2 Seed Cost per Hectare of Selected Bean Cultivars,
Middle Kirinyaga, 1981 . . . . . . .
7.1 Comparison of Estimates of Parameters at Different
Stages of Investigation, Middle Kirinyaga, 1981
7.2 Evaluation of Data from RD-Identification Exercise
by Comparing Them with Data from Formal Survey,
Middle Kirinyaga, 1981 . . . . . . .
7.3 Accuracy of Estimates of Parameters in RD-
Identification Exercise and Principal Sources of
Inaccuracy, Middle Kirinyaga, 1981 . . . .
7.4 Evaluation of Data from Informal Survey Report by
Comparing Them with Data from Formal Survey,
Middle Kirinyaga, 1981 . . . . . . .
7.5 Accuracy of Estimates from Informal Survey, Compared
to Formal Survey, and Principal Sources of
Inaccuracy, Middle Kirinyaga, 1981 . . . .
7.6 Researchers' Perceptions of the Most Important Con-
straints Facing Farmers, from the Informal Survey
and from the Formal Survey, Middle Kirinyaga,
1981 . . . . . . . . . . . .
7.7 Researchers' Perceptions of the Most Important
Research Priorities for Farmers in Middle
Kirinyaga: After the Informal Survey and After
the Formal Survey, 1981 . . . . . .
7.8 Levels of Non-Experimental Variables in Maize/Bean
Experiments Based on Informal and Formal Surveys,
Middle Kirinyaga, 1981 . . . . . . .
7.9 List of Proposed Experiments and Demonstrations
Formulated after the Informal and Formal Surveys,
Middle Kirinyaga, 1981 . . . . . . .

S 157

S. 159

. 163

S. 166

S 171

S. 180

. 194


. 221

S. 224

S 233

S. 237

. 241

S. 243

. 246

S. 247



Figure Page

1.1 Overview of an Integrated Research Program . . . 7
2.1 Schematic Representation of Some Determinants of
the Farming System . . . . . . .... 20
2.2 Hierarchical Decision Tree Model: Decision to Grow
Variety "a" (Model is Hypothetical) . . . ... 50

4.1 Average Rainfall in Middle Kirinyaga, 1953-81
(mm. per ten-day period) . . . .. ... . . 75
4.2 Average Monthly Rainfall and Probabilities of
Receiving Lower Amounts of Rainfall, Middle
Kirinyaga, 1953-81 . . . . . . . . . 76
4.3 Maize and Bean Price Trends in Middle Kirinyaga,
1979-81 . . . . . . . .... . . .81
5.1 Decision Tree on Whether to Plant Katumani Variety
for Early Maize . . . . . . . .... .129

5.2 Decision Tree on Whether to Grow Katumani to Store
for Later Sales and Consumption . . . .... .131
5.3 Decision Tree on Whether or Not to Grow Hybrid 511/512 136
5.4 Decision Tree on the Time and Method of Land Prepara-
tion and Planting . . . . . . . .... .149


Map Page

1 Middle Kirinyaga . . . . . . . . ... 59




This study has two types of objectives: methodological and problem-

solving. The methodological objectives concern the use of three tech-

niques, the CIMMYT diagnostic survey, repertory grid, and hierarchial

decision-tree modeling, for analyzing a farm system. The problem-

solving objective involves using farming systems research (FSR) methods

to plan an adaptive production research program for small farmers. In

the first section of this chapter, we present the background to the

methodological objectives, highlighting the emergence of FSR as a tech-

nique for planning farmer technologies. A statement of the methodolo-

gical objective follows. Next, background to the problem-solving

objective, planning an adaptive production research program in Middle

Kirinyaga, Kenya, is presented. Finally, the specific problem-solving

objectives are stated.

1.1. Background to the Methodological Objectives

Rural development planners in the Third World are becoming increas-

ingly aware that information about small farmers is crucial for planning

rural development programs. The widespread failure of large scale,

capital intensive agricultural projects and the increasing concern for

a more egalitarian distribution of benefits have led to increasing

emphasis on small farmer oriented programs (Lele, 1975). The development

and extension of new technologies for small farmers is often seen as an

important and effective measure for enhancing rural welfare.

Nonetheless, there is a growing perception among policymakers and

research administrators that national agricultural research institutions

are not contributing as well as they could be to the development and

diffusion of new technologies for small farmers. Often there is a wide

gap between what researchers recommend and what farmers practice. When

planners seek to impose on the farmer new production systems and pack-

ages which have proven to be effective on research stations, the result

is failure. The usual scapegoats for these failures are: the farmers

themselves, who are alleged to be lazy or irrational; the extension ser-

vice, which is blamed for not transmitting the research station's message

to the farmers; the delivery system, which fails to deliver inputs to

farmers when required; or policy distortions, which make unprofitable

the use of purchased inputs and the cultivation of crops for the market.

However, many studies point to another cause of low adoption rates--

station recommendations which are irrelevant to the farm family's pri-

orities, resource constraints, and physical, cultural, and economic

environment (Winkelmann, 1977). In many countries, extension recommen-

dations are developed by researchers on experiment stations whose work

is aimed at maximizing yields per unit of land area. This yield-

oriented approach often brings forth recommendations which are irrele-

vant to farmer circumstances, for two main reasons. First, the

recommendations are developed under physical conditions different than

those of the farmers, since they are generally formulated based on the

results of experiments mounted on research station plots. These plots

are usually plowed by tractors, kept weed-free, sprayed, and fertilized

so as to ensure a significant response from the experimental variable.

Hence, (1) responses to experimental variables are generally much higher

than could be expected on farmers' fields and (2) it is unlikely that

the actual response functions would have the same shape as the experi-

mental ones.

Second, researchers' criteria for evaluating new technologies are

much different than those of farmers (Collinson, 1980). As long as

inputs have costs, it is never in the best interests of farmers to adopt

those levels of inputs which maximize yields. Many researchers recog-

nize this fact and use the point of maximum profit for making their

recommendations. Unfortunately, the problem of what is best for farmers

to adopt is much more complicated than this; farmers seek to maximize

utility as well as profit. Thus, economic optimality for a given farm-

ing situation depends on the objectives, resources, and priorities of

the farmers concerned. Farm management studies in Africa have shown

that small farmers operate their farms so as to (1) provide a reliable

supply of food for their families, and (2) provide cash for what they

regard as essential purchases (Eicher and Baker, 1982). They take into

account both natural (temperature, rainfall, soils, etc.) and socioecon-

omic circumstances (prices, riskiness, sociol acceptability, etc.) in

deciding what enterprises and management practices to adopt. Farmers

operate a system made up of many enterprises, and are often forced to

diverge'from the ideal management of one enterprise in order to devote
resources to another enterprise, in the interests of overall system
Farming systems research offers an alternative approach in adaptive
research to the conventional yield-oriented perspective, common among
physical and biological scientists. FSR begins with the farmer's own

situation, rather than a pre-defined package of high yielding

technologies. FSR researchers purport to be holistic, that is, they

view the whole farm as a system of interdependent components and focus

on how these components interact in their physical, biological and

socio-economic setting (Shaner, Philipp and Schmehl, 1982). The

approach is most effective if it is multi-disciplinary; hence, FSR is

generally mounted by teams of scientists from both the agricultural

and social sciences.

In FSR, insights into how current practices fit into the farm

system are used as a basis for proposing improvements. Researchers

focus on identifying and overcoming critical constraints which the farm

family faces so that they can better meet their priorities (Norman,

1976). The immediate goal is to offer a few technological improvements

which are compatible with the farm household's activities and circum-

stances, not to replace the current farm system with a radically differ-

ent one. Emphasis is given to identifying and developing new techno-

logies which (1) have potential for increasing productivity and, hence,

enhancing the farm family's ability to meet its own objectives, (2) will

be acceptable to farmers and feasible for them to adopt, and (3) will

promote developments which are consistent with national policy objectives.

Generally, FSR also includes the mounting of experiments on farmers'

1It is incorrect to consider FSR as a completely new approach with
no antecedents. Johnson (1981) and Ramaratnam (1981) discuss the fore-
runners of FSR in farm management and farm home development programs in
the United States in post-World War II years. In fact, some of these
programs were considerably more holistic than is FSR, as currently prac-
ticed. For example, the Kentucky Farm and Home Development Program gave
considerable attention to institutional and human change, the production
of household goods and services, and firm/household interrelationships.
These topics are frequently neglected in most current FSR. A review of
previous holistic approaches to farm research and extension would there-
fore be valuable for current FSR practitioners.

fields, in conjunction with on-station research. In on-farm experiments,

the farmer and extension staff join the researcher in managing the exper-

iments and evaluating the results.

In summary, FSR is an important tool for agricultural research

institutions to use for increasing the effectiveness of their programs.

FSR is important both for identifying the broad, long-term research

priorities of disciplinary and commodity programs as well as for plan-

ning and executing adaptive research to formulate recommendations for


Farming systems research programs are being undertaken by a variety

of institutions throughout the world. Guatemala (Gostyla and Whyte,

1980), Honduras (Whyte, 1981) and Zambia (Collinson, 1982) are among

several national programs utilizing the FSR approach. Moreover, many

international agricultural research centers and bilateral donors are

mounting FSR programs and promoting the use of FSR by national agricul-

tural research institutions.

The International Maize and Wheat Improvement Centre (CIMMYT) has

been active in developing FSR concepts and procedures for planning tech-

nologies for farmers.2 CIMMYT promotes "on-farm research" with a

farming systems perspective (OFR-FSP) which aims to generate higher-

productivity technology for specific groups of farmers, especially in

the short term. The program uses on-farm research methods, such as

1Researchers inmarketing management would view FSRas an example of
the application of "the marketing concept" to a non-profit organization.
The marketing concept states that the function of an organization is to
match the production and services it offers to the needs and wants of
the target consumers it serves (Kotler, 1980).

CIMMYT's procedures receive special emphasis in this thesis since
CIMMYT procedures were used in carrying out the research.

farm surveys and on-farm experiments, and is conceptually based on a

farming systems perspective (Byerlee, Harrington, and Winkelmann,

1982).1 Figure 1.1 presents CIMMYT's view of an integrated research pro-

gram and the role of on-farm research in the program. The on-farm

research team plans new technologies, mounts experiments on farmers'

fields, formulates recommendations, assesses farmers' experiences with

the recommendations and promotes improved technologies through demon-

strations. Moreover, the on-farm research team identifies problems for

station researchers and policy-making bodies to address.

The planning stage, in which researchers obtain information for

planning experiments to test new technologies, is the stage addressed by

the research reported in this thesis. FSR practitioners tend to agree

that the overall objective of FSR is to develop an understanding of the

farm system in order to plan improvements in the system. However, there

is much disagreement concerning methods of data collection and analysis,

and the use of the results for planning experiments. CIMMYT economists

have developed FSR methods and procedures which differ considerably from

those used in farm management research carried out in most Third World

countries (Byerlee, Collinson, et al., 1980). The approach,called the

CIMMYT Diagnostic Survey in this thesis,includes the following steps:

1. Identification of recommendation domains (RD's) or farmer

target groups. An RD is a group of farmers with a similar

1The authors claim that a new acronym is required because of the
increasing amount of confusion over the use of the term FSR. FSR is a
very general term, they argue, and FSR programs have objectives, ranging
from solving a specific problem to increasing the body of knowledge.
They define OFR/FSP as a subset of FSR which emphasizes on-farm research
methods to generate technology to increase productivity for a specified
farmer group. In this study FSR and OFR/FSP will be used interchange-
ably to refer to the subset of FSR described above.

Figure 1.1

Choice of target
farmers and research

National goals. Input
'..-.. .'.'' .' ; .'.'.> .:f

supply credit markets etc.
1C . '


of policy issues

Overview of an Integrated Research Program

New components
_Incorporated into on-farm

Obtain knowledge
and understanding of.
farmer circumstances
and problems to plan

2 Experiment
Conduct experiments
in farmers' fields to

formulate Improved
technologies under
farmers' conditions.

3. Recommend
Analyze experimental
results in light of farmer
circumstances to
formulate farmer

4. Assess
Determine farmers'
experience with

5. Promote
to farmers.

Identification of problems
for station research

Source: International Maize and Wheat Improvement Center, Economics
Staff, 1981.

Developing and
screening new tech-
nological components
(e.g. varieties, new
herbicides, pesticides]


farming system; that is, farmers in the group operate their

farms in a similar manner under fairly homogenous eco-

climatic and socioeconomic conditions. Thus, they have

similar problems and similar development opportunities and

it is likely that a given recommendation will be more or

less applicable to all farmers in the RD. An exercise to

identify RD's in a given area is useful for establishing

the numbers and main characteristics of the different farmer

groups. Policy makers can then decide which groups merit

the attention of the research services.

2. Informal Survey. In an informal survey, a multi-disciplinary

team of researchers using detailed but essentially un-

structured guidelines holds informal interviews with

farmers. The objectives are to develop an understanding

of the farming systems, identify priority research topics,

pre-screen possible new technologies for their introduc-

tion into the system, and plan the formal survey which


3. Formal Survey. In a formal survey, a questionnaire is ad-

ministered by enumerators to a random sample of farmers to

verify the findings obtained in the informal survey and to

measure important parameters in the system. Generally,

the information i.s collected in a single visit to each

sample farm.

4. Planning on-farm experiments. The purpose of the survey

work is to plan on-farm experiments for farmers in the RD.

Experiments are mounted under farmers' conditions and farmers

participate in the planning, monitoring and evaluation of

the results.

The CIMMYT Diagnostic Survey differs from most other FSR procedures1

in five principal ways. First, objectives are focused on field experi-

mentation. The purpose is to provide production researchers with infor-

mation to help them plan technologies which will be acceptable to

farmers and feasible for them to adopt in their existing environment.

Thus, many institutional and policy variables are treated as fixed in

the short run.

Second, the primary data collection instrument in the CIMMYT

approach is the informal survey whereas the formal survey is the prin-

cipal instrument in most other approaches.

Third, the CIMMYT approach gives more weight than many other FSR

approaches to the collection of normative and prescriptive data,2 such

as dataon farmers' attitudes, opinions, and reasons for employing par-

ticular practices. Less emphasis, relative to other FSR approaches,

is placed on collecting quantitative, non-normative data such as levels

of inputs and outputs.

Fourth, the CIMMYT approach does not use quantitative methods to

model the farm system. Rather, an informal technique based on

Other FSR approaches include those of American universities
administering FSR projects in Africa, e.g., Michigan State University
in Senegal, Purdue in Upper Volta; Winrock Foundation in Kenya; West
African Rice Development Association; and the International Center for
Research in the Semi-Arid Tropics.

2Normative information is information about the goodness or badness
of a condition, situation or thing. Prescriptive information concerns
the rightness or wrongness of goals or acts and are always based on both
normative and non-normative (or positive) information (Johnson and
Zerby, 1973).

developing an understanding of the farmer's perspective through intensive

interviewing and observations provides the "model" which is used to

analyze the farm system and test the impact of proposed changes.

The fifth distinguishing feature of the CIMMYT procedures is that

turnaround time is more rapid than in most other FSR approaches. A

"complete" inventory of resources and input and output flows, based on

data collection of one year or more, is not regarded as essential for

planning experimentation for farmers. Rather, the CIMMYT procedures are

sequential--at each stage the understanding attained is used to guide

and focus data collection during the subsequent stage. Turnaround time

ranges from three to six months, measured from the time researchers

enter an area to the time detailed experimental plans are formulated

and a survey report is issued.

CIMMYT's approach to FSR is considerably less holistic than certain

other FSR and farm management approaches, because of several practical

considerations. First, CIIMMYT's mandate is limited to maize and wheat;

thus, proposed improvements in CIMMYT surveys are generally limited to

these two crops and related operations. Second, CIMMYT seeks to work

through national agricultural research organizations; therefore, they

require an approach which is flexible enough to accommodate the manpower

and financial constraints which these organizations face. Unfortunately,

these constraints often limit the holistic nature of the diagnostic

survey, as when multi-disciplinary teams are reduced to only two or

three members. Third, the approach generally avoids proposing changes

in government policy, e.g., credit, land tenure, prices. There are two

10f course, several members of the household may be involved in
making decisions on the farm. Thus, it is often incorrect to refer to
the farmer in the singular, or as "he".

reasons for this: (1) agricultural researchers generally have little

influence over such policies and (2) most governments are more inter-

ested in obtaining assistance to promote the development of agricultural

technologies than assistance to promote changes in their agricultural


The procedures developed by the CIMMYT Economics Program are by no

means a rigid recipe for mounting FSR. For example, the preface to the

CIMMYT manual, "Planning Technologies Appropriate to Farmers: Method-

ologies and Procedures", states that CIMMYT's approach "has evolved from

our experiences with farmers and researchers in many countries. We

fully expect that these guidelines will be improved through the experi-

ences of other researchers" (Byerlee, Collinson, et al., 1980).

1.2. Statement of the Methodological Objectives

The primary methodological objective of this study is to use and

evaluate two methods, repertory grids and hierarchical decision-tree

models, as supplements to the CIMMYT diagnostic survey for collecting

and analyzing normative and prescriptive information about farmer deci-

sions. The field of agricultural economics in general, and farm manage-

ment in particular, has relatively little experience in collecting

non-monetary normative and prescriptive data, compared to other disci-

plines in the social sciences. Collinson (1981a) argues that the CIMMYT

informal survey is "almost an anthropological approach to understanding

the farming system", referring to the need for examining farmers' values

and for probing into the complex reasons underlying farmer practices.

One possible area of improvement in the CIMMYT approach is the incor-

poration of tools from other disciplines for more systematic collection

and analysis of normative and prescriptive information. Repertory grid,

first used by clinical psychologists, is a method for (1) eliciting

from respondents the criteria they use in differentiating among items,

e.g., alternative technologies, and (2) recording respondents' evalua-

tions of how each item performs on each criterion. Hierarchical

decision-tree models, used primarily by anthropologists, depict the

decision process people use when considering alternative actions. Thus

the two approaches assist the researcher in understanding the preferences

and reasoning underlying the decisions farmers make.

Two secondary methodological objectives concern the quality of

information gathered at different stages of the diagnostic survey

sequence.1 First, we test the effectiveness of the exercise for iden-
tifying recommendation domains. This exercise has two purposes:

(1) to delimit RD's and (2) to provide some preliminary data on RD's

in the study area. The method, a single-page questionnaire administered

to local leaders, is admittedly open to a broad range of error.

Collinson (1982) states that "it is important to emphasize that these

[the RD's specified by the exercise] are preliminary groupings. As the

diagnostic sequence is implemented within these identified target groups,

the characteristics of each domain will be more fully understood." If

the errors are significant, the exercise can be wasteful and misleading.

This thesis will evaluate the quality of the information obtained in

the exercise by comparing the information with that of the formal survey.

1This is an important issue because national research institutions
seek to lower the cost of diagnostic studies in terms of cash, time, and
use of scarce skilled manpower. Evaluating the quality of data at dif-
ferent stages of the investigation and the impact of the differences on
an experimental program can guide us in assessing the importance of each

Possible biases leading to incorrect information at the preliminary

stage will therefore be identified and ex-ante corrective measures will

be proposed.

Second, we evaluate the utility of the formal survey, by comparing

its results with the informal survey results. As stated above, the pri-

mary purpose of the formal survey is to verify the findings of the

informal survey. However, the formal survey is too expensive and time-

consuming an exercise if it serves only to confirm informal survey

findings. This thesis will examine the changes in the understanding of

the farm systems and in experimental content which occur as a result of

mounting the formal survey. We will then conclude whether a formal

survey was indeed necessary and identify characteristics of a farm

system which make the formal survey more or less necessary under dif-

fering circumstances.

1.3. Background to the Problem-Solving Objectives

Kenya's agricultural research system is relatively strong, compared

to most African countries. In 1978, there were 255 Kenyan staff above

the B.S. level and 96 expatriates on research stations throughout the

country (Jamieson, 1979).

Since independence, there has been a shift of resources from cash

crops to food crops, and from high potential areas to marginal areas,

reflecting the government's efforts to change certain policies inherited

from the colonial era. The research system has made some significant

achievements in developing technologies for small farmers, most notably

concerning pyrethrum and maize development (Heyer and Waweru, 1976).

However, non-adoption and partial adoption of recommended

technologies in Kenya is widespread and has been well-documented

(Kariungi, 1977; Gerhart, 1975). The 1979-83 five-year development

plan notes the lack of new technologies available for immediate adop-

tion and the irrelevance of many research packages. The plan calls for

reorienting research and extension towards alleviating production con-

straints in smallholder farming systems (Government of Kenya, 1979a).

With these policy objectives in mind, the Committee on Maize and

Pasture Research of the Ministry of Agriculture recognized the potential

usefulness of the farming systems approach and recommended that:

KARI/SRD [Kenya Agricultural Research Institute/
Scientific Research Division] should lay more emphasis on
problem identification of the zonal and farm level through
diagnostic studies which take account of the physical
environment and social and cultural attributes of the
target population (Government of Kenya, 1980).

1.4. Statement of the Problem-Solving Objectives

This thesis holds that FSR is a more effective and efficient

approach for developing small-farmer technologies than is the conven-

tional yield-oriented approach. The general problem-solving objective

of this thesis is to plan an adaptive production research program for

farmers in Middle Kirinyaga. More specifically, the problem-solving

objectives are to:

1. Describe farming systems in the study area.

2. Identify priority research areas for planning technolo-

gies appropriate for farmers. These are points of the

system where changes can increase productivity and

where such changes can be readily acceptable and feasible

for farmers.

3. Examine selected farmer decisions related to the priority

research topics. These decisions are selected because an

understanding of the issues involved will assist research-

ers in proposing new technologies and formulating recom-

mendations for farmers.

4. Plan an adaptive production research program to develop

technologies for farmers in the study area. Recommenda-

tions will also be made for planning technical research

on research stations and for mounting extension programs

in the area.

As stated above, this thesis addresses only one stage in the tech-

nology development process, the planning stage. Therefore, the thesis

does not purport to solve any farmer problems, per se. Rather, our

problem-solving objective is to help researchers solve their problem of

planning biological and physical research to aid Middle Kirinyaga

farmers. The researchers' problem is seen as one of making experimen-

tation more effective in developing technologies for small farmers.

1.5. Overview of the Thesis

Chapter 1 presents the objectives of the thesis--both methodologi-

cal and problem-solving--and the background to these objectives. Chapter

2 discusses the research design of this thesis. The chapter begins

with a discussion of the objectives and attributes of the planning stage

of FSR, highlighting the importance of understanding farmer decisions

for planning new technologies. Theoretical contributions from manager-

ial decision theory are presented which provide a framework for

exploring farmer decisions. Some practical considerations in selecting

an approach and methods are also examined. Next, approaches to under-

standing farmer decision-making are reviewed and their relevance to

meeting the objectives of the planning stage of FSR are assessed.

Finally the approach and specific methods used in this thesis are pre-

sented and discussed.

Chapter 3 presents the procedures for collecting data in this

thesis. First the selection of Middle Kirinyaga as the study area and

the delineation of recommendation domains in and around the study area

are discussed. Next sampling methods, fieldwork procedures, and speci-

fic problems in mounting the informal survey and the formal survey are


The survey results, a description of farming systems in Middle

Kirinyaga, are presented in Chapters 4 and 5. In Chapter 4, natural

and socio-economic features are described and farmer objectives and

sources of income are examined. This is followed by discussion of the

management of the farm system, resource use, and system constraints.

In Chapter 5, we analyze selected farmer practices, focusing on the

area's two principal crops, maize and beans. The chapter highlights

strategies farmers use for meeting their objectives in the circumstances

they face.

Chapter 6, which addresses the problem-solving objective of this

thesis, presents an adaptive production research program appropriate

for Middle Kirinyaga farmers. Criteria for selecting research priori-

ties are discussed in the light of farmer practices, the context in

which experiments will be mounted. Detailed discussions of two priority

research areas and three areas of lesser priority are discussed for

maize and beans. Comments are also offered on other crops. This

section also presents some problem areas for the extension service to

address in Middle Kirinyaga.

Chapter 7 addresses the methodological objectives of this thesis.

Two methods from outside the field of Agricultural Economics--hier-

archical decision modeling and repertory grid from Psychology--are

evaluated as supplements to the CIMMYT diagnostic survey approach. This

chapter also examines how the quality of data changes as one proceeds

through the stages of the diagnostic survey sequence. The chapter

includes proposals on the use of hierarchial decision-making and

repertory grid in the CIMMYT approach, survey methods for identifying

recommendation domains,and the use of the formal survey for verifying

the results of the informal survey.

Chapter 8 summarizes the principal findings of the thesis. Sug-

gestions for further research are also presented.



This chapter presents the approach taken and the methods used in

this thesis for planning an adaptive production research program for

small farmers. First, the objectives and main attributes of the plan-

ning stage of FSR are presented. Next, we discuss the analytical

approach used to diagnose farmer problems, highlighting the importance

of understanding farmer management strategies and decisions. The theo-

retical contributions of managerial decision theory for understanding

farmer decision-making are examined; these serve as a conceptual frame-

work for empirical studies examining farmer decisions. Also, some

practical considerations in selecting an approach and methods are out-

lined, based on principles of the economics of information.

The methods selected for planning an adaptive agricultural research

program are presented in the last section of this chapter. Three

approaches to understanding farmer decision-making are reviewed. The

cognitive, anthropological approach is identified as best suited to the

objectives of this study. Specific methods include the CIMMYT diagnos-

tic survey for obtaining an overall understanding of the farming system

and repertory grid and hierarchial decision-tree modeling for examining

selected farmer decisions in greater depth. In this last section, we

also highlight the principal methodological issues to be addressed in

this thesis.

2.1. The Planning Stage of FSR

As stated in the previous chapter, the objectives of the planning

stage of an integrated on-farm research program with a farming systems

perspective are to develop an understanding of the farming system and to

use that understanding to plan improvements appropriate for farmers.

Understanding how the farming system functions--what farmers do and why--

serves as a basis for identifying problems of farmers and evaluating the

potential success of possible solutions.

There is a broad consensus among FSR practitioners on the attri-

butes of FSR and in particular, the characteristics of the planning

stage. While many of these attributes may also be found in other

research programs, their combination distinguishes FSR from other

approaches. These attributes, summarized below, are discussed in the

principal works on FSR: Technical Advisory Committee, 1978; Norman,

1980; Gilbert, Norman and Winch, 1981; and Shaner, Philipp and Schmehl,


1. Farmer-Based. One of the principal functions of FSR is to

strengthen the links between farmers and researchers. A presumption of

the FSR approach is that it is difficult to introduce improvements to a

system which is not well understood. Therefore, as a basis for develop-

ing new technologies, it is necessary to develop an understanding of the

farmer's aspirations, environment, resources, constraints, and practices.

2. Holistic. Farming systems researchers begin by considering the

farm family's activities--including crop, livestock, household, and off-

farm processes--as a whole and analyzing the various elements which

influence the farming system. Figure 2.1 presents a schematic represen-

tation of some possible determinants of a farming system. The figure

= = = -m m M m m M-

Figure 2.1 Schematic Representation of Some Determinants of the Farming System




Norms. and

I., s';;ulons



-- Other


Farming Consumption.-1
Decision I
rP. Makers) Savings -
J (Farm)








- - --ncome"




Livestock ) <

Broken lines represent results of farmn-ig system

Farming System

-I I
r ,, -- ~-- - -. - -





Source: Gilbert, Norman and Winch, 1981

___*f - -


I ( L I

I .

highlights the importance of the human element, including both exogenous

factors such as community structures, and endogenous factors such as the

farming household's decisions. Moreover, the figure depicts the house-

hold as both a production and a consumption unit with interactions

between the two.

Researchers use a holistic approach to focus on system interactions--

the impact that particular elements of a system have on each other. For

example, maize stover may be an important input into livestock produc-

tion, and farmer practices concerning maize production may be influenced

by the objective to produce stover as well as grain for human consump-

tion. Another important type of system interaction is the production

compromise; farmers are often prevented from managing a particular enter-

prise in what would be an ideal manner for it alone because they wish to

allocate scarce resources over a wide range of enterprises or activities.

In actual practice, many FSR programs are not very holistic due to

practical considerations. For example, in countries where skilled

researchers are scarce, FSR teams may include only two persons. The

team is likely to neglect many factors outside of its expertise which

influence the farming system.

In using a holistic approach, FSR practitioners consider a fairly

substantial array of possible improvements for any given group of

farmers. This does not imply that all parameters are variables, or that

all variables 'require the same degree of attention. Rather, it is likely

that FSR researchers will quickly focus on particular commodities or

operations because making improvements in that area will benefit the

farmer. However, they must consider explicitly how the facet under

study relates to other system components.

Many FSR programs fail to consider a broad range of improvements

for the same reasons they lack holism. For example, a shortage of

skilled manpower and the compartmentalization of research departments

often limit the scope of improvements considered in FSR programs. More-

over, some claim that it is more efficient to limit FSR's focus in the

short run to production improvements; policy, institutional and off-farm

improvements may be incorporated into the approach at a later date when

FSR has gained more experience and credibility.

3. Multi-disciplinary and Inter-disciplinary. Because of its

holistic approach, FSR is multi-disciplinary. Specialists provide exper-

tise from their respective disciplines on: (1) how the farm system

functions, (2) what its key problems are, and (3) how to solve them.

Most FSR researchers highlight the interaction between technical scien-

tists and social scientists as critical to success. The technical

scientist examines the biological and physical environment and evaluates

farmer management of an enterprise in light of what ideal management

should be. The social scientist examines the socio-economic environment,

endogenous factors, and system interactions to explain why the farmer

makes the decisions he does. Working together on a commonly defined

research agenda, they prescreen possible solutions to farmer problems

which will raise productivity and be both feasible and acceptable to

the farmer. Thus, FSR requires "a multi-disciplinary team working in

an inter-disciplinary manner", that is, specialists from several dif-

ferent disciplines working together using mutually understandable

language to solve a particular problem (Gilbert, Norman and Winch,


4. Farmer-group Specific. Because of prohibitive costs, it is

impossible to plan research for individual farmers in developing

countries. Planning research for large regions, or even nationwide,

is also impractical since such research glosses over important vari-

ations between and within areas. Farmers with similar farming systems

have similar problems and opportunities; thus they require a common

experimental program and set of recommendations. Therefore, a prelim-

inary step in an FSR program should be to identify fairly homogenous

farmer groups, or recommendation domains (RD's). The criterion for

including farmers in a single group is whether a single set of recom-

mendations would be generally appropriate for all members of the group.

Farmer-group specificity should not be confused with area-specificity;

it is farmers and not fields which make decisions on technologies. Thus,

farmers of different RD's may be interspersed in a given area.

5. Flexible in Accommodating Technical and Non-Technical

Improvements. FSR has been applied almost exclusively for identifying

improved techniques in crop production. However, the approach can and

is being used to identify other kinds of improvements--in infrastructure,

policy, land tenure, etc.--needed for rural development.

6. Consistent with Societal Objectives. FSR is used to enhance

the welfare of farmers within the guidelines of national policies and

the long-term interests of society. On occasion, short-run measures to

enhance farmer welfare may conflict with these policies and interests.

For example, hillside farmers may be guided by profit considera-

tions to plant annual cash crops, causing severe soil erosion, at a time

when national policy is seeking to minimize soil erosion. Because

farming system researchers draw upon the objectives and considerations

of both farmers and policy makers, they are able to identify solutions

which will serve the interests of both groups. For example, in the

above case, FSR researchers may seek to introduce relay and mixed crop-

ping patterns which are profitable to farmers and conserve the nation's

soil resources as well.

2.2. Analytical Approach

In the planning stage of FSR, researchers examine how the farmer

allocates his scarce resources of land, labor and capital among compet-

ing enterprises to best meet his objectives. Key farmer problems are

identified and technological alternatives are proposed. The following

analytical approach was used in this thesis to develop an understanding

of the farming system and to identify system improvements. The steps

overlap significantly and are not necessarily undertaken in sequence.

1. Describe the Farmer's Environment and Environmental Constraints.

An understanding of the environment and how it shapes the farmer's

activities is essential for understanding the system and evaluating the

appropriateness of proposed improvements. The aspects of the farmer's

environment are: natural (e.g., rainfall, soils, topography), insti-

tutional (e.g., transportation, government programs), and social (e.g.,

ethnic group, societal values, family structure). In most cases, FSR

does not seek changes in the environment but rather changes which the

farmer can make given his environment.

2. Examine Farmer Objectives and Priorities. Collinson (1981)

lists the objectives of small farmers as:

(a) meeting the social and cultural obligations of the community;

(b) providing a stable, reliable supply of food for the family;


(c) providing cash for basic needs;
(d) providing extra cash.
An understanding of the farmer's objectives and the relative weighting
of each objective is required so that researchers can propose improve-
ments which help him meet his objectives or, at least, are not incon-
sistent with them.
Specifically, researchers must assess farmer preferences concerning
aspects of production and consumption alternatives (e.g., risk-yield
tradeoffs, differing maturity length of varieties, storage decisions,
and tradeoffs between farm investment and consumption). Researchers
obtain information about farmers' preferences through interviews and
inferences concerning enterprises farmers pursue, the choice of period
to perform operations in enterprises, and when and how products from I
the enterprise are used.
3. Evaluate Resource Use and Constraints. The farmer has limited

amounts of land, labor and capital resources at his disposal; how he allo-
cates and uses them are key elements to understanding how the system
functions. Comparing resource use to resource availability helps the
researcher identify farmers' resource constraints and priorities. The
researchers understand how a single enterprise should ideally be managed
in order to maximize returns to that enterprise. They then examine how
enterprises and consumption opportunities compete for resources in order
to explain the compromises made by farmers combining enterprises.
4. Examine Farmer Management Strategies. The analysis of farmer
management strategies pulls together the information thus far collected--
how farmers seek to allocate their resources to best achieve their I



objectives given the environmental constraints they face. Collinson

(1981b) states that farmer management strategies are

. devices for reconciling the satisfaction of a variety
of priorities with resource limitations and uncertain
production circumstances. Identifying farmers' manage-
ment strategies, understanding how they satisfy farmers'
priorities and how they compromise.production methods is
a prerequisite for evaluating new techniques proposed for
the system.

5. Identify Leverage Points and Propose System Improvements

When proposing improvements, the researcher must adopt the farmer's

perspective on the problem and help the farmer improve on the strategy

he is already following. Leverage points are points in the system where

there is scope for increasing productivity in ways that are likely to

be acceptable and feasible for the farmer. Possible leverage points in

the crop sub-system include the method and timing of a cultivation

operation and the type of inputs, varieties and enterprises. At

leverage points, a problem exists and there are potential solutions to

the problem, e.g., by using a new input, changing the timing or method

of an operation, or introducing a new enterprise.

This analytical approach is used to examine farming systems in

Middle Kirinyaga in this study. The results are presented in Chapters

4 and 5. Proposals for system improvements are discussed in Chapter 6.

2.3. Theoretical Contributions from Managerial Decision Theory

FSR and the analytical approach used in this thesis both highlight

an understanding of farmers' decisions. Several sets of concepts from

managerial decision theory (Johnson, 1961) provide a useful perspective

for examining farmer decision-making processes. These sets of concepts

are (1) information classifications for decision-making, (2) steps in

the decision process, and (3) knowledge situations.

The first set of concepts presented here is useful for classifying

information people use in making decisions. Three information areas may

each be broken down into two information types: (1) normative

information--information on values, i.e., goodness or badness and (2)

positive information--non-normative information, i.e., information on

what is or what will be. Further, each information type may be broken

down by tense--information on the past, present or future. Thus,

information used in making decisions can be categorized in a 3 x 2 x 3


The decision maker draws upon information from this matrix to make

prescriptions on the rightness or wrongness of an action or goal. A

right action or goal is the "best" action or goal, best meaning "that

indicated by the value beliefs involved in view of what the factual

beliefs involved indicate is possible" (Johnson, 1961). Prescriptions

are always a function of both normative and positive information.

The set of concepts on how decision makers perceive reality implies

a rational model of decision-making but does not rule out irrational or

inconsistent concepts, goals and actions. Moreover, no restrictions are

placed on how beliefs about normative and positive information are

formed. The model is useful because it instructs us that we must col-

lect both normative and positive information if we are to develop an

understanding of farmer decisions. It also shows us that information

on farmer prescriptions is useful for understanding how the farmer uses

normative and positive information to formulate opinions on the right-

ness or wrongness of an action. Of course, whether or not the farmer

takes the action that he prescribes is a function of his intentions and

many other factors in the farmer's environment--constraints he faces,

his own personal initiative, and chance circumstances.

The breakdown of the decision process into steps is a second

development in managerial decision theory for examining farmer decisions.

The steps include problem definition, observation, analysis, decision,

action, and acceptance of responsibility. Indeed, the very process of

generating recommendations for farmers is an attempt to reduce the cost

to the farmer of several of these steps, especially problem identifica-

tion, observation and analysis (Harrington, 1980).

A third useful concept in managerial decision theory is that of

knowledge situations. Managers acquire information to arrive at deci-

sions, as stated previously. Increments of information have increasing

marginal cost and decreasing marginal value; thus some point exists at

which marginal costs and returns are equal. The following five states

of knowledge explain alternative situations managers face in making


1. Certainty: The manager considers his present knowledge adequate

for making a decision. He has no interest in acquiring further informa-


2. Risk: The manager regards present knowledge as adequate for

making a decision and the cost of additional knowledge is equal to its


3. Learning: The manager feels that he does not yet have enough

information to make a decision. The value of learning additional infor-

mation exceeds the cost of obtaining that information.

4. Inaction: The manager feels that he does not have enough

information to make a decision and that the cost of additional informa-

tion exceeds the value of that information. Therefore, no action is


5. Forced Action: The manager would be in some other knowledge

situation if it were not for some outside force which made it necessary

for him to reach a decision.

The recognition of knowledge states requires us to include the

acquisition of information in our model of decision-making. For example,

in a case where farmers know about a maize variety and 20 percent of the

farmers are using the variety, we might conclude that 20 percent of the

farmers have decided to use the variety and 80 percent have rejected it.

This analysis may be correct, but some users may be in the certainty

state, others may be in the risk state, and still others in the forced

action state. Those not using the variety could be in any of the five

states. Which states farmers are in will have important bearing on our

technology planning. For example, if most non-users are in the cer-

tainty or risk states, the variety is not appropriate for these farmers.

However, if most non-users are in the learning or inaction stage, the

problem may be an extension problem. Thus, the concept of knowledge

states enriches our understanding of the reasons behind farmer decisions

and strategies.

2.4. Some Practical Considerations in Selecting Methods
for Data Collection

Research institutions generally face a number of constraints in

planning experimentation--most notably, lack of trained manpower, budget

limitations, and the time period available to report survey results.

Therefore, it is necessary to design a program of data collection to

accommodate these constraints. Clearly, the planning team must seek to

collect the minimum amount of data needed to plan experimentation which

will likely provide benefits to the client population. Further, the

data must be collected and analyzed in the most efficient manner pos-

sible, minimizing the use of trained manpower, budget costs, and time.

Ideally, researchers continue collecting data until the marginal returns

of collecting additional data are equal to the marginal cost.

A holistic, multidisciplinary FSR approach appears to be incon-

sistent with efforts to minimize costs and accommodate manpower and

financial constraints. However, empirical findings on how farmers adopt

changes limits the scope of the objectives of an FSR program, and thus

its costs (Byerlee, Harrington and Winkelmann, 1982). Small farmers,

characterized by capital scarcity, risk aversion and possessing a healthy

degree of skepticism, generally change in relatively small steps, adopt-

ing one or a few relatively minor changes at a time, rather than seeking

to transform their system of farming all in one step (Mann, 1977; C.

Gladwin, 1979; Gerhart, 1975).1 Therefore, the objective of the FSR

approach in the short run should be to develop a few system improvements

which require relatively small degrees of change on the part of the

farmer. To accomplish this, researchers should use data collection pro-

cedures which help them to efficiently identify a few worthwhile changes

and develop a sufficient understanding about the farming system and

This is not to denigrate the "package" concept in technology dif-
fusion. Indeed, the issue of package recommendations versus the single
recommendation approach is a red herring; the issue is the degree of
change implied by a single recommendation or set of recommendations, not
the number of changes (Collinson, Personal Communication).

farmer attitudes concerning the changes to plan appropriate


A.sequential approach to collecting data, in which data collection

at each stage is increasingly focused on priority problems, is more

efficient than setting out from the start to collect a complete data

set on stocks and flows of inputs and outputs. Researchers using a

sequential procedure begin by developing a broad overview of the farm-

ing system and then focus on a few variables at leverage points. At

each stage, an iterative procedure is followed: researchers collect

and evaluate data to decide what additional data are required. Thus,

at each stage, data are used to refine one's understanding and estab-

lish hypotheses which in turn are used to further focus data collection

and eliminate issues and topics which are not relevant. Throughout the

exercise, researchers are weighing the value of additional information

against the costs of obtaining that information (Byerlee, Harrington

and Winkelmann, 1982).

Two other practical considerations are also important in selecting

FSR planning methods. First, the methods should be conducive to multi-

disciplinary team work, that is, they should be readily comprehensible

to all team members so that each can play a role in contributing to the

objectives of the FSR exercise. Second, the methods should be readily

comprehensible to policy makers; if they can understand the process by

which results were obtained they will have greater confidence in those


* 2.5. A Review of Approaches to Modeling Farmer Decision-Making

I Thus far, we have described the principal aspects of FSR, presented
an analytical approach for identifying system improvements, and discussed
I some theoretical and practical considerations for developing an under-
standing of farmer decisions and the farming system. In this section,
we review broad approaches to modeling farmer decisions and select an
I approach suitable to meet our objectives. In the following sections, we
present specific issues we will address in using this approach for
I developing our understanding of the farming systems and identifying
system improvements.
I Broadly, there are three different approaches to understanding
I farmer decisions which are found in farm management literature on devel-
oping countries. The three approaches and examples from each are pre-
I sented below.
1. Multivariate analysis approach. Adoption and diffusion studies
I commonly use multivariate analysis to examine farmers' decisions. In
I these studies researchers correlate the outcome of a decision to char-
acteristics of individual or groups of adopters (e.g., age, zone, access
Sto extension, characteristics of village leaders), usually using multiple
regression. Exploring the association between specific characteristics
I and adoption is useful for identifying the factors which promote and
I inhibit adoption. Examples of multivariate analysis studies include
Gerhart's study of hybrid maize adoption in Western Kenya (Gerhart,
i 1975) and comparative diffusion research by Rogers, et al. (1970).
Scientists from a variety of disciplines--geography, economics, commu-
I nications--use multivariate analysis for explaining farmer decisions.



2. Behavioral Approach. Researchers using this approach establish

a choice criterion, such as maximization of profit subject to a number

of constraints, and test the hypothesis that farmers make decisions

which are consistent with the outcomes of these models. Further, beha-

vioral models are also used to identify constraints, evaluate the impact

of new technologies and policy changes on the farming system, and identi-

fy and evaluate farmer management strategies (Eicher and Baker, 1982).

Behavioral models are used most often by economists and particular

methods include response functions (Wolgin, 1975), utility function

analysis (Walker, 1980), linear programming (Heyer, 1972) and systems

simulation (Crawford, 1982).

3. Cognitive Anthropological Approach. Advocates of this method

assume that the farmer's own perspective of the world around him is the

best starting point for model building (H. Gladwin, 1979). Intensive,

informal interviews are used to develop an understanding of the criteria

farmers use in making decisions and the management strategies they

pursue. Greater emphasis is given to collecting normative and prescrip-

tive information than in the other two approaches. Examples of the

cognitive anthropological approach include participant observation (Hill,

1972), hierarchial decision modeling (C. Gladwin, 1976) and repertory

grid (Baldwin, 1977). The CIMMYT diagnostic survey procedures also rely

heavily on this approach.

For the purpose of developing an understanding of farmer decisions

and management strategies and using this information for planning pro-

duction research, the third approach, the cognitive anthropological

approach, is'the most useful.

Multivariate analysis studies are important for gaining a broad

overall view of adoption patterns. Furthermore, they often are able to

pinpoint strong associations which may be used to understand why adoption

does or does not take place. For example, Gerhart found a strong cor-

relation between climatic zone and adoption of hybrid maize in Western

Kenya. Using supplementary rainfall and experimental data he showed

that available hybrids were simply not suitable for certain areas. But

regression studies investigate association of farm and farmer character-

istics with adoption, not causation of adoption. Therefore, they

cannot generally be used to explain the reasons why farmers take parti-

cular decisions, reasons which may be fairly complex. For example,

education is a common variable in regression equations to "explain"

adoption. But lack of education is rarely ever a reason for non-

adoption; rather it is a proxy for other possible causes of non-adoption:

inability to read instructions, less willingness to experiment, less

exposure to media, etc. In fact, there is usually a very weak link

between independent variables and dependent variables in regression

equations which "explain" adoption. Certainly, we are in the dark about

which policy measures would be most effective if we do not know the

actual causes of non-adoption.

Regression equations are especially poor in explaining adoption

when independent variables are used as proxies for farmer attitudes,

such as risk. For example, Gerhart sought to associate the area planted

to some well known insurance crops (cassava, groundnuts, sorghum and

millet) with the decision to adopt hybrid maize, in order to test the

importance of risk in inhibiting adoption. The proxy is questionable,

as Gerhart acknowledges, since the insurance crops may be grown for a

variety of other reasons, many yield differently in different areas,

etc. Furthermore, Gerhart points out that even if the proxy is accept-

able, the association tells us nothing about the nature of the risk

farmers perceive--losing their seed investment in a drought year, expos-

ing themselves to more variable returns, etc. Thus, information from

multivariate studies is not sufficient for developing an understanding

of farmer decisions and management strategies necessary for identifying

potential system improvements.

Behavioral models, seeking to confirm whether farmer behavior is

consistent with certain choice criteria, also oversimplify the actual

decision criteria farmers consider when making a decision. Expected

utility modelers, for example, derive utility functions for farmers

which establish the farmers' preferences of tradeoffs between risky

high-paying options and more stable, low-paying ones (O'Mara, 1971;

Walker, 1980). Disregarding the host of measurement problems encoun-

tered (Petit and Dijon, 1980; Young, et al., 1979), the model tells us

little about sources of risk and other factors besides risk which affect


Other behavioral model approaches have similar pitfalls. Some

researchers use production functions to compare existing resource allo-

cation with optimal allocation, and infer that the difference is due to

risk aversion (de Janvry, 1972). The problem in this case is that a

multitude of other factors may be involved (Young, et al., 1979).

Linear programming and systems simulation may be useful, if costly,

methods for modeling the system and testing the effects of including

new technologies or enterprises in existing farm systems. However,

these models are based on, and are not substitutes for, an understanding

of the farming system. How this understanding is obtained is not often

made clear; "understanding" generally appears to come through ex-post

analysis based on farm modeling (Byerlee, Harrington and Winkelmann,

1982). In any case, behavioral models tend to sidestep examining the

actual reasons which farmers have for making decisions and instead, test

whether the outcome of the decision is consistent with certain general

behavioral choice criteria.

Researchers using behavioral models are generally guided in their

data collection efforts by the input-output framework. As comprehensive

as this approach may appear it is often found lacking. Informal methods

of intensive interviewing and observation are needed to penetrate the

relationships and patterns of activities which lie beneath the surface

of input-output data (Haugurud, 1979). Researchers using the cognitive

anthropological approach emphasize understanding the logic behind pro-

duction decisions rather than gathering facts about production. "Know-

ledge of farmers' reasoning is as necessary an input to a successful

rural development project as is agronomists' or economists' reasoning

from a distance" (C. Gladwin, 1979).

An understanding of "farmers' reasoning", the key element in the

analytical approach adopted for this study, can best be developed using

the cognitive anthropological approach. For example, farmers in Middle

Kirinyaga are divided over the issue whether the best time to plant

maize and beans is before or after the rains begin. The agronomist may

have his own opinion, but he needs to understand the farmer's perspective

However, we do not contend that the cognitive anthropological
approach is always useful for explaining farmers' reasoning. Indeed
many actions and decisions may not be subject to analysis using any
approach. For example, farmers may not be able to explain why they
plant in a particular manner or why they use a particular variety.

and the flexibility farmers have in order to decide whether time of

planting is a leverage point and what research opportunities are asso-

ciated with it. Farmers in Middle Kirinyaga point to many advantages

and disadvantages to planting at each of the two times--before or after

the rains begin. Some of the disadvantages can be characterized as

risks, whereas others are fairly certain results of planting at a par-

ticular time. When the farmer actually does plant is a function of

when he intends to plant and any external forces which interfere with

his intentions. Sifting through the various criteria he regards as

important and understanding the reasoning he uses, cannot be accomplished

using a multivariate or behavioral model. These models can indirectly

provide important clues of association or possible outcomes of alterna-

tive decision paths, but they are not substitutes for the informal,

analytical approach for developing an understanding of the farmer's

perspective and using this understanding to identify research opportuni-


2.6. Issues Concerning the CIMMYT Diagnostic Survey

The principal method selected for this study is the CIMMYT diagnos-

tic survey, outlined in Chapter 1. The diagnostic survey has several

practical advantages:

1. It is relatively inexpensive to implement in terms of time,

finances, and manpower. A team of researchers, two at a minimum, can

identify recommendation domains and complete an informal survey in one

to two months. A formal survey can be completed and data analyzed in

an additional two to four months.

2. The method uses a sequential data collection procedure to focus

on important issues as described in Section 2.4. During the informal

survey, researchers evaluate the data collected and reformulate data

needs on a daily basis. By the end of the informal survey,leverage

points are identified and system improvements are proposed.

3. Interview procedures are informal and data collection and

analysis techniques are easy to learn. Thus, the method is conducive

to multi-disciplinary teamwork and can be easily understood by policy


Examples of diagnostic surveys being used for planning experimen-

tation include: International Maize and Wheat Improvement Centre

(1979), and Shumba (1981). Successful results, measured in terms of

farmer adoption of technologies proposed in diagnostic surveys, are

reported in Moscardi (1982).

Three important issues concerning the CIMMYT diagnostic survey are

presented below. The first two issues concern the methodological objec-

tives of this thesis: (1) the collection and analysis of normative and

prescriptive information in the CIMMYT diagnostic survey and (2) the

advantages and disadvantages of including an RD-identification exercise

and a formal survey in the CIMMYT approach. The third issue, that the

CIMMYT approach is relatively less holistic than many other FSR

approaches, is presented in order to qualify the problem-solving results

of this thesis, that is, the adaptive research program developed for

Middle Kirinyaga.

2.6.1. Collecting Information on Farmers' Values and Decisions

A key issue in mounting a CIMMYT-style diagnostic survey is how to

collect the relatively large amount of normative and descriptive infor-

mation--information on farmers' values and decisions--which is required.

The call for more emphasis on these kinds of information for understand-

ing decision processes in farm management is by no means a new one.

However, farm management and farming systems research in LDC's still

emphasize collecting positive data on inputs and outputs, from which

normative and prescriptive information are extrapolated. The reasons

behind this tendency are not difficult to ascertain. First, farm man-

agement work in the U.S. and Europe tends to be positivistic and this

approach has been transferred to LDC's. Second, rewards in terms of

salaries and prestige among peers are earned on the basis of using the

most sophisticated quantitative methods, often with little consideration

as to their role in solving problems. Third, communication is diffi-

cult between researchers (whether expatriate or local) and farmers,

and it is easier to "measure" than to communicate. Fourth, conceptual

and semantics problems, which seriously affect the collection of posi-

tive data, are even more troublesome in the collection of normative

and prescriptive data. Fifth, many agricultural economists seek to

mimic the positivistic approach to research which they observe among

other agricultural scientists. Thus, they shy away from the study of

values, objectives, decision processes, and prescriptions, labeling

such studies as "unscientific".

1See, for example, Johnson, 1961.

The CIMMYT approach and other similar approaches, e.g., Hildebrand

(1981) and Bartlett and Umearokwu (undated), use informal, direct

interview methods for developing an understanding of farmer values,

prescription and the reasons behind farmer decisions. Farmer responses

are corroborated by direct observation, and intensive reasoning is used

to piece together why farmers do what they do. Unfortunately, the

approaches provide little systematic direction on how to identify and

evaluate the criteria which farmers use in making decisions. At pre-

liminary stages of the assessment of the farming system, supplementary

tools for investigating farmer decisions are probably not necessary.

But once leverage points are proposed, researchers need a fairly detailed

understanding of farmers' reasoning behind the decisions they make

concerning the leverage point. For example, maize variety is a pro-

posed leverage point in this study, as discussed above. Using the

CIMMYT approach, researchers would ask the farmer why he plants the

varieties he plants as opposed to other alternative varieties. They

may also ask about the farmer's past experiences, and test hypotheses

on relationships between varieties planted and other variables of the

system, e.g., farm size, income level, etc. However, a host of problems

may confront the researcher at this point. A farmer may give more than

one reason for why he plants or does not plant a particular variety and

if a list of possible reasons is presented to him to consider, the num-

ber of reasons would certainly increase. Second, two farmers may give

the same reasons for planting different varieties; that is, they dis-

agree about some positive characteristic of the varieties. Moreover,

constraints and chance occurrences may prevent farmers from planting
the variety they want to plant. Unless a fairly uniform set of

responses is obtained, the situation can become quite confusing. It is

in these cases that it would be useful to have easy-to-understand,

systematic tools for rapidly assembling the information on values and

prescriptions and for modeling the decision so as to better understand

the farmer's reasoning.

This thesis will use and evaluate two methods, Repertory Grid and

Hierarchical Decision Models, as supplements to the CIMMYT Diagnostic

Survey, for eliciting and evaluating normative and prescriptive informa-

tion on farmer decisions. These methods come from the disciplines of

psychology and anthropology, respectively, which have relatively more

experience in collecting and evaluating such information than does

agricultural economics. The theoretical underpinnings and proposed

contributions of the two methods are discussed below, in Section 2.7.

2.6.2. The Utility of an RD-Identification
Exercise and a Formal Survey

Utility of the RD-Identification Exercise

During the initial stages of an investigation, researchers are

interested in gathering preliminary information about farmers in order

to demarcate RD's, to develop a preliminary understanding of farmer

circumstances in each RD, and to identify appropriate data collection

methods to use in ensuing stages. Few farming systems studies use

primary data collection methods to meet these objectives. Rather, most

researchers rely on secondary information and reconnaissancee surveys",

i.e., informal discussions with farmers and persons knowledgeable about

the area. Formal data collection exercises at this stage are shunned

because they are expensive and time consuming, and because in initial

stages of an inquiry the researcher does not know enough about the area

he is interested in to develop a suitable questionnaire.

However, Collinson used a single-page questionnaire survey of

extension agents to identify RD's in Central Province, Zambia (Inter-

national Maize and Wheat Improvement Centre, 1979). He concluded that

the method is a low-cost and effective, albeit preliminary, means of

identifying RD's and providing information about them. However, no

formal technique was used to arrive at this conclusion. In this thesis,

we compare the information obtained in the exercise identifying RD's

with the information obtained in the formal survey which followed. We

assume that if the information and RD's based on the formal survey are

similar to those developed in the RD-identification exercise, then the

exercise is an effective means of gathering preliminary information

about RD's.

Utility of the Formal Survey

The role of the informal survey in farm management investigations

in developing countries has increased in importance in recent years,

relative to the formal survey. Shaner, Philipp and Schmehl (1982),

comment at length on the advantages and disadvantages of informal

methods. The principal advantages are that it promotes free and in-

depth discussion of problems and issues and that it helps the researcher

to get acquainted with local words, concepts and ideas. On the other

hand, its principal disadvantages are that it is non-random, and that

because questions are not standardized, quantification is difficult and

results are less reliable. The authors conclude that researchers must

be "cautious in generalizing from informally collected data." Thus,


they seem negative about using informal surveys as a basis for planning
experimentation, except in special cases when problems and opportunities
are so apparent that formal methods of data collection are unnecessary.
Other farming systems researchers view the informal survey as being
a generally effective and sufficient measure for gathering information
about farmers to plan agricultural experiments. They argue that, whereas
a formal survey may increase the accuracy and precision of information,
the increased costs, measured in terms of time as well as resources,
outweigh the value of the increased benefits. Hildebrand's Sondeo
(Hildebrand, 1981) and Gathee (1979) provide examples of using an infor-
mal survey to plan experimentation.
Collinson occupies a middle ground between those arguing that only
the informal approach is required and proponents of the formal survey.
On the one hand, the informal survey is the "pivotal" step in the diag-
nostic approach, but on the other hand, he generally advocates mounting a
single-visit formal survey to verify the information gathered in the
informal survey (Collinson, 1982).
Little work has been done to formally compare the information and
implications for research from informal surveys with those of the ensu-
ing formal survey for the same group of farmers. Indeed, if the formal
survey exercise does not lead to significant improvements in the accuracy
of information and the design of experiments appropriate for farmers,
one can argue that it is superfluous. In this thesis we compare the
data and the proposed experimental program developed in the informal
survey with those developed from the formal survey in order to examine
the utility of mounting a formal survey.



2.6.3. Holism

A principal weakness of the CIMMYT approach, as practiced in

Eastern Africa, is that it tends to be less holistic than certain other

FSR and farm management approaches. Typically, multi-disciplinary teams

mounting CIMMYT diagnostic surveys include only agricultural scientists

and economists, excluding sociologists, rural non-farm enterprise

specialists and others who have expertise on rural development. Thus,

it is not surprising that leverage points in these surveys almost exclu-

sively involve agricultural production inputs or operations. Moreover,

relatively little attention is given to such topics as household-firm

interactions, marketing and credit networks, and other factors influenc-

ing the farming system.

The approach used in this thesis is subject to the above weaknesses.

The team was composed of an economist and an agronomist, with only

limited input from other agricultural scientists. Leverage points were

heavily weighted towards experimentation on maize and beans, the area's

two principal crops. The researchers sought to analyze how household

processes influenced production processes but certainly lacked the

expertise of rural sociologists or human ecologists in addressing such


In our case, a more holistic approach was an ideal towards which

we strove but which we were prevented from fully attaining by three

principal considerations. First, skilled researchers are in scarce

supply in Kenya, making it difficult to recruit researchers to partici-

pate in survey exercises. Second, logistics and financial constraints

limited the size of the team. Third, we had no formal access to policy

makers outside of production research so we chose to limit our proposals

to those concerning production research. However, we believe that there

were sufficient incremental benefits to mounting an FSR exercise with

only two team members and focusing only on agricultural production

research to warrant the costs of the exercise. Ideally, we recognize

that the make-up and focus of the team should be much broader than it

was in our case.

2.7. The Repertory Grid (RG) Technique and Hierarchical
Decision-Tree Models (HDM)

In this thesis, we evaluate the incorporation of two techniques,

RG and HDM, into the CIMMYT diagnostic survey for assembling and analyz-

ing data about farmer values and decisions. These techniques are

described in detail below.

2.7.1. The Repertory Grid Technique

Repertory grid is a method from cognitive psychology which seeks

to elicit and measure people's perceptions of their environment. The

method was developed by clinical psychologist G. A. Kelly in the 1950's

as a therapeutic procedure, based on his human "personal constructs"

theory.1 This theory holds that individuals build conceptual models,

based on their own experiences, which are used to guide their future

actions. In these models, an individual arranges features of his per-

ceived environment, called "elements", by discriminating on the basis

of attributes into bi-polar scales which express meaningful contrasts.

These scales are called "personal constructs".

1A detailed presentation of Personal Construct Theory is beyond the
scope of this paper. A brief sketch is given here only to acquaint the
reader with its basic precepts. A summary of the theory is found in
Bannister and Mair (1968).

For example, an individual may perceive a set of elements called

"bean cultivars" because they have a set of similar attributes which

separate bean cultivars from other sets of phenomena. He evaluates

them through his own personal constructs, with bi-polar scales for each

attribute, e.g., high yielding-low yielding, good tasting-bad tasting,

etc. How these constructs are built into mental models is the subject

of further work (Bannister and Mair, 1968).

Repertory grids are matrices of scores for a set of elements across

a set of constructs. Elements and constructs are generally elicited

from individuals themselves to minimize interviewer bias. Elements are

rated on each construct using a consistent procedure. The resulting

matrix describes an individual's repertory of feelings about a set of

elements. An example of a repertory grid for evaluating farmer opinions

of alternative bean cultivars is shown in Table 2.1.

Table 2.1
Sample Repertory Grid of Important
Attributes of Bean Cultivarsa

Cultivar 1. Cultivar 2. Cultivar 3.

Yield in Season of
Sufficient Rainfall 4 4 3
Yield in Season of
Low Rainfall 1 1 2
Storing (Without
Insecticide) 3 3 4
Taste 5 3 5
Price 2 2 4
Disease Susceptibility 1 3 1

aRatings: 5 = excellent, i.e., variety performs very well.
1 = poor, i.e., variety performs poorly.
Source: Data are hypothetical.

Repertory grids have been used by researchers in a number of fields

other than clinical psychology, e.g., market research (Hudson, 1974) and

urban geography (Harrison and Sarre, 1975). This author knows of two

applications of this technique to agricultural development in less

developed countries. Townsend (1975) examined how farmers in a Colom-

bian settlement project perceived of their own farms in contrast to

other farms, such as the "best" farm they knew, a particular neighbor's

farm, etc. Floyd (1977) used a similar approach in Trinidad. Both

researchers reported that the method is useful for developing an under-

standing of how farmers view the circumstances they find themselves in

and the priorities they have for improving their situations. Further,

Baldwin (1977) examined the perceptions of English tomato growers towards

past technological improvements in tomato production. Solving farmers'

problems was not the objective of any of these three applications of

repertory grid to agriculture. Rather, the researchers were concerned

with using the technique to understand how farmers perceive their world

around them.

In this thesis, repertory grids will be constructed to identify the

criteria farmers use in deciding among selected technological alterna-

tives concerning leverage points. In the early stage of the field work,

four potential leverage points were selected where it appeared that

repertory grids could be useful in obtaining farmers' appraisals of

available technological alternatives at these leverage points. Reper-

tory grids were constructed using samples of 5 to 15 farmers and the

important aspects concerning the set of alternatives were elicited from

the farmers themselves. During the formal survey the grids were

constructed for randomly selected farmers, using the aspects identified

in the formal survey.

2.7.2. Hierarchical Decision-Tree Models

The hierarchical decision-tree model (HDM) is the second method to

be tested in this study as a supplement to the CIMMYT diagnostic survey.

HDM's present the decision process in a tree form, with decision criteria

at the nodes, or branching points, of the tree. HDM's are used by

researchers in a number of disciplines, including anthropology, psy-

chology, and economics, to model decisions made by individuals.

The underlying theory behind HDM is that "people in choosing alter-

natives, do not make complex calculations of the overall utility of each

alternative. Rather, people tend to use procedures which simplify their

decision-making calculations," due to their inability and/or unwilling-

ness to process all information available to them (C. Gladwin, 1979).

Indeed, social scientists are becoming increasingly concerned with the

simplifying procedures which individuals use in making decisions. For

example, in a review of behavioral decision research across a number of

disciplines, Slovic, Fischoff and Lichtenstein (1977) claim that

. whereas past descriptive studies consisted mainly of
rather superficial comparisons between actual behavior and
normative models, research now focuses on the psychological
underpinnings of observed behavior.

Researchers examining simplifying procedures in decision-making include

Tversky (1972), Simon (1969), and Abelson (1976).

HDM is an important approach to modeling the simplified decision

process which individuals use. The model is built in the following man-

ner. First, intensive interviewing is used to elicit decision criteria

from respondents. Farmers who have not yet made a decision, i.e., are

in the "learning" or "inaction" knowledge state, are excluded from the

analysis. The criteria are then grouped into three categories:

1. Orderings of alternatives on some attribute (e.g., "in a

season of sufficient rainfall, is the yield of variety 'a' greater than

that of variety 'b'?");

2. Explicit choices that performance on one attribute is more

important than performance on another attribute (e.g., "you say that

variety 'a' yields better than 'b' when rainfall is sufficient but that

'b' yields better when rainfall is low. So is it better to plant a

variety expecting low rainfall or one expecting sufficient rainfall?");

3. Constraints which must be passed or satisfied (e.g., "did you

have cash available for buying variety 'a' seed?").

Once a set of criteria are identified, they are arranged in a flow

chart in a logical manner. For example, the hypothetical tree, shown

in Figure 2.2, summarizes the decision criteria which an assumed group

of 32 farmers consider in deciding whether or not to plant variety "a".

Yield is the first criterion because it is the most important criterion

to farmers in the group interviewed. Since yield in sufficient rainfall

was a more important criterion than yield in low rainfall, the former

precedes the latter. The criterion involving the trade off between the

two yield criteria follows directly after the two yield criteria. The

storage issue follows the yield criteria because farmers who test a

variety for the first time and find that it has very low yields will not

know, or care, much about storage characteristics anyway. "Not having

cash for seed" is a constraint and is therefore listed in the tree after

the intention to plant variety "a" has already been established.

Figure 2.2
Hierarchical Decision Tree Model: Decision to Grow
Variety "a" (Model is Hypothetical)

Variety "a" Gives Highest Yield Of
All Varieties When Rainfall is Sufficient

-Yes No

Variety "a" Gives Highest Do Not
Yield of.All Varieties Plant
When Rainfall is Low 3 Cases

Yes No

Better To.Plant a
Plant Variety Expecting Sufficient
2 Ca se Rainfall or One Expecting
2 Cases Low Rainfall

Rainfall Both Low Rainfall

Enough Plant
Yes 0o 2 Cases

Had Cash
Available For Do Not Plant
purchasing Seed
18 Cases

In any case, the ordering of the criteria on the tree does not

affect a farmer's outcome, as long as the researcher has obtained the

principal reason why the farmer does not plant the variety. For

example, if a farmer answers no to the first question, "does variety

'a' give the highest yield of all known varieties when the rainfall is

sufficient?" but later says that his reason for not using "a" is

because it doesn't store well, it would be incorrect for the researcher

to let the farmer "get off" the tree at the top box, "do not plant",

because variety "a" does not give highest yield when rainfall is suf-

ficient. Rather, one of two problems has occurred. Either the infor-

mation obtained from the farmer does not represent the farmer's opinions

on the issue or, more likely, the tree needs further development to

accommodate a person who (1) feels that variety "a" does not yield higher

than other varieties when rainfall is sufficient and (2) claims that

poor storage is the reason he does not use the variety.

The tree shows the relative importance of criteria in a particular

farmer decision. For example, in the hypothetical example of Figure

2.2, 60.percent of the farmers do not plant Variety "a" because it

stores poorly. Thus, it is implicit that storage is the most important

criterion and that yield and cash availability are less important con-

siderations. Furthermore, the tree can be used to show how farmers

explicitly weigh the relative importance of any two particular criteria.

For example, Figure 2.2 shows that farmers feel that it is more import-

ant that a variety performs well when rainfall is sufficient than that

it performs well when rainfall is low.

The use of decision tree models to explain farmer decisions is

relatively new. Roumasset (1974) used decision trees to estimate the

risk of alternative fertilization techniques on rice in the Philippines.

C. Gladwin (1977, 1980) has used HDM in Mexico, Guatemala, and Alabama

to model farmer decisions on the adoption of new methods and inputs and

for decisions on which crops to plant. Her earlier work involved using

HDM for ex-post evaluation of recommendations to provide feedback to

project planning. More recently, she has used HDM for ex-ante research

planning. For example, her work on cropping choices in the Altiplano

of Guatemala indicated that the least cost method of promoting cash crop

production was to improve yields of the subsistence crop, corn

(C. Gladwin, 1980). Further, she has reported using HDM's in two

"sondeos", multi-disciplinary survey exercises similar to CIMMYT infor-

mal surveys, for helping to plan experimentation. However, the models

used were not developed during the sondeos; they were refinements of a

previously developed model on cropping decisions in a neighboring area.

This thesis builds on Gladwin's use of HDM in FSR for guiding

technology planning. However, one departure from her use of the method

was made. Gladwin develops an HDM from interviews with 20 to 30

farmers and then tests the model for its ability to explain the deci-

sion among another 20 to 30 farmers. Further, she attaches considerable

importance to the ability of the model to "predict the decisions of a

new, different (if possible, random) sample of decision makers." Thus,

a high "success rate" (rate of prediction) is an important objective of

this exercise. In this thesis, HDM's are developed by researchers inter-

viewing a small number of farmers, 10 to 15, during an informal survey.

The questions underlying the model are then included in a formal survey

to a random sample of farmers. The objective is not to test, per se,

the model developed in the informal survey by examining the prediction

rate. Rather, the objective is to use the model to develop appropriate

questions for the formal survey so that decision trees can be recon-

Sstructed, based on formal survey responses, which represent the deci-

sions of randomly selected farmers. Thus, we do not presume that the

model developed in the informal survey is accurate enough to predict

decision outcomes; nor do we have any qualms about adding to or chang-

ing the model developed in the informal survey; in order to incorporate

individuals from the formal survey who do not "fit" into the original


2.8. Summary of Methods

The cognitive anthropological approach involves intensive inter-

viewing to develop an understanding of the logic behind farmer manage-

ment strategies and practices in order to develop an understanding of

the farming system. Thus, the approach requires normative, positive,

and prescriptive information about farmer decisions. Three particular

methods will be used to develop this understanding: (1) the CIMMYT

diagnostic survey, which is the overall approach for developing the

understanding of the system and (2) and (3) the repertory grid tech-

nique and hierarchical decision-tree model, which are used for examin-

ing particular farmer decisions.

The first step was to select a study area and identify recommenda-

tion domains. Next, an informal survey was undertaken together with an

agronomist. During the informal survey, we identified system leverage

points and used RG and HDM to obtain more information about farmer

decisions concerning the leverage points.

A formal survey was then mounted to verify the information from the

informal survey and quantify selected parameters useful for planning

experimentation. The questions used for developing RG's and HDM's in

the informal survey were incorporated into the formal survey. Thus, the

RG's and HDM's constructed from formal survey data may be compared with

those developed during the informal survey.

RG and HDM will be evaluated on the contributions they make as

supplements to the CIMMYT diagnostic survey for developing an understand-

ing of particular farmer decisions. Unfortunately the exercises could

not be carried out completely independently of each other. However, in

general, the results from a particular exercise can be compared with

results from another exercise.

Two secondary methodological objectives are to compare the quality

of information gathered at different stages of the survey sequence. The

effectiveness of the exercise to identify recommendation domains is

evaluated by comparing the data obtained with data obtained in the for-

mal survey. The utility of carrying out a formal survey, in addition to

an informal survey, is evaluated by (1) comparing the data obtained with

those obtained in the informal survey and (2) assessing the implications

which the formal survey results have on changing or refining the pro-

posed research and extension program planned following the informal




This chapter examines the survey procedures--sampling methods,

interviewing techniques and fieldwork procedures--employed for develop-

ing an understanding of the farming systems in the study area. First,

the reasons for selecting Middle Kirinyaga as the study area are pre-

sented. Next, the exercise for identifying recommendation domains is

examined. Finally, methods and procedures for the informal survey and

formal survey are discussed.

3.1. Selection of Middle Kirinyaga as Study Area

Middle Kirinyaga was selected for this project because it was felt

that an adaptive production research program could offer substantial

benefits to local farmers, as well as contribute to several of the Kenya

government's policy objectives. The specific reasons for selecting

Middle Kirinyaga are discussed below.

First, Middle Kirinyaga is made up of mostly low-income, small

scale farmers, and does not have any important cash crops, such as

coffee, tea, or cotton. As stated in Chapter 1, the Kenya government

places a high priority on increasing incomes in low-income, small

farmer areas (Government of Kenya, 1979a).

Second, supplies of maize and beans, the area's two most important

foods, are frequently exhausted. There is potential for stabilizing

farmers' food supplies and turning Middle Kirinyaga into a surplus food

area, since the physical and climatic conditions of Middle Kirinyaga are

suitable for producing both maize and beans. In the aftermath of food

shortages of 1980-81, Kenya has placed high priority on achieving self-

sufficiency in maize and bean production (Government of Kenya, 1982).

Third, maize and bean production levels are very low and adoption

rates for recommended inputs and practices are also low. The reasons

for non-adoption are not clearly understood. A diagnostic survey could

help researchers understand the reasons behind non-adoption and make

appropriate policy recommendations. Revised research recommendations

based on on-farm experiments could contribute significantly to increas-

ing production.

Fourth, the institutional environment appears to be favorable,

relative to other areas in Kenya. For example, transportation and access

to input and output markets are adequate and research and extension ser-

vices in the area are fairly well developed.

3.2. Identifying Recommendation Domains

The next task was to identify recommendation domains (RD's) in the

study area and to delineate the farmer groups to be studied. Two sepa-

rate tasks were required. One was to identify geographical, or across-

area, differences among farming systems. Across-area differences are

generally caused by physical factors--e.g., climate and altitude--but

may also be a result of historical or socio-economic differences--e.g.,

ethnic group or government settlement schemes.

The second task was to examine within-area differences in farming

systems, that is, to check whether two or more farming systems may be

interspersed in a particular ecological zone. The principal causes of

within-area differences in farming systems are often socioeconomic and

historical factors such as ethnic group, income, or participation in a

government credit program.

A critical issue in identifying RD's concerns whether a particular

difference between two groups of farmers is an important enough differ-

ence to justify separating the groups into different RD's. If the

differences among two groups of farmers are significant enough that we

will likely require different sets of experiments to meet their needs

and circumstances, then the two groups of farmers belong in separate

RD's. Thus the objectives of the survey become the benchmark for evalu-

ating the importance of differences among farmers.

The exercise to identify RD's in Middle Kirinyaga took approximately

two weeks and involved three methods. First, secondary data on the area

were assembled. Unfortunately, these proved to be of limited use

because they were averaged across several dissimilar areas within

Kirinyaga. Second, researchers, government officials, and local leaders

familiar with the area were informally interviewed about across-area and

within-area differences. Next, a short, two-page questionnaire was

administered to extension workers and local officials in each subloca-

tion, the smallest administrative unit in Middle Kirinyaga. The

questionnaire covered principal characteristics of the farming systems

which were likely to vary between farms and across areas. Individual

The questionnaire is presented in Appendix A. Sublocations in
Middle Kirinyaga range from 50 to 150 square kilometers and have 500 to
4,000 inhabitants.

questions concerned such aspects as physical environment, cropping

pattern and practices, livestock, and sources of income.

The interviews were conducted by the researcher; each interview

lasted about 40 minutes. Twelve questionnaires covering 15 sublocations

in Middle Kirinyaga and adjoining areas were completed in one week.

Data tabulation and analysis took another three days. The results of

the exercise are reported in Franzel (1981). The methods delivered a

rapid, low-cost, preliminary identification and description of RD's in

the area.

Map 1 presents the boundaries of Middle Kirinyaga, based on the

exercise identifying RD's. The boundaries represent fairly distinct

changes in the eco-climatic and socio-economic environment. For

example, as one moves north, out of Middle Kirinyaga, altitudes and

rainfall increase and temperature and solar radiation decrease. Coffee

and dairy are important enterprises and there are higher cash incomes

and higher population densities. Maize takes much longer to mature

and only one crop is grown per year whereas in Middle Kirinyaga, two

crops are grown.

The southern boundary of Middle Kirinyaga is marked by a change in

soils, from light, red loam to heavy, black clay. The black soils area

is characterized by somewhat different crops, significant differences in

the cropping calendar, larger farms, and greater numbers of cattle. To

1The boundaries were revised slightly following the informal survey
exercise. Boundaries shown are those from the final assessment of RD's
in Middle Kirinyaga.

21n fact the northern boundary of Middle Kirinyaga is the 4,300
meter countour, the line below which farmers are forbidden from growing
coffee. However, this edict has not been enforced since the coffee boom
of 1978.

Sm m m m m m m m m m m m

Y ,KIAN JAN VA \ ~*I,(
411IE 11 ItPp, b~iyp~l I
I~s~b11 It 0 4 oi!kuo
Kicrnbwe Ktu
1pKibingoU Kirimunge Sch.
ii II Sch

sKAIT4RUtI 1r Ndombo a
isti l Nyongali, '
I KNon0,i B A

Map 1. Midl K a

,, nial It OM13UINIun
Sch 00\glt
Ka Kngai (. im

'I ; ~\ I cut

rr `ec hh KEY

rlBoundaries of Middle Kirinyaga

5 AG A A it Area included in Middle Kirinyaga
!iboaqan Sch "dbfollowing RD-identification
f ~~~a exercise but excluded later ,)
,app- c9Q g $ Kndon,,

Map 1. Middle Kirinyaga

the east and west of Middle Kirinyaga are dry, hilly areas with

relatively poor road access. These areas have less fertile soils,

lower population density and much uncultivated land.

The task of assessing within-area differences among farmers in

Middle Kirinyaga was more difficult than identifying across-area dif-

ferences. Two factors were considered in the investigation:

1. Characteristics of the Farming System: We sought to identify

whether there were important differences in the way farmers operated

their farms--their priorities, the resources they used, their con-

straints, and the strategies and practices they employed to use avail-

able resources to best meet their priorities.

2. Potential for Change: Second, we were concerned with the

relative potential for change among farmers in the area. Two farmers

may be operating their farms in the same manner, but have different

potential for change because of different resource availabilities.

Researchers may divide a homogenous, dryland cropping system, adjacent

to an unused flooded swamp area, into two RD's: one with the potential

for growing irrigated rice and the other without this potential.

In Middle Kirinyaga, it appeared that access to cash income was an

important determinant of the farming system. For example, access to

cash influenced whether the farmer undertook certain enterprises, such

as owning exotic-breed cattle. Managing these cattle requires substan-

tial cash and other resources for purchasing feed, transporting water to

the home, and protecting the animals against disease. Further, it

appeared that low income farmers made much less use of purchased inputs

such as hybrid maize seed and maize insecticide than higher income

farmers. Also, high income farmers tended to own oxen or were able to

hire oxen as soon as the rain started, thus taking full advantage of

the brief rainy periods for growing their crops. On the other hand,

fewer low-income farmers own oxen and those without oxen tended to

plant late, relying on social contacts rather than cash payments to

secure an ox-plow team.

We also hypothesized that income level was associated with many

other aspects of the farming system aside from enterprise choice and

crop husbandry. For example, food security appeared to be an entirely

different problem for each of the two groups. Nearly all high income

farmers appeared to obtain a regular flow of cash from a non-farm enter-

prise or a farm enterprise such as dairy; therefore cash was always

available for purchasing food when required. However, low income

farmers were forced to hire out their labor when their food supplies

ran short.

These differences in the way farmers operate affect the type of

experiments to be planned for the two groups. For example, maize experi-

ments should incorporate differences in non-experimental variables,

e.g., time of planting, variety, and plant population, for the two

groups. Further, the number of and range in experimental variables

can generally be greater for high income farmers, since low income

farmers lack cash for purchasing improved inputs.

Two further issues concerned the number of income groups to estab-

lish and how to draw lines between them. We decided to classify

farmers into two groups--those who could afford modest investment in

their farms and those who could not, because it was thought that this

division would cover most of the variation among farmers in the

area. A set of proxies for income were drawn up to differentiate high

income farmers from low income farmers for the informal survey. The

proxies for high income farmers included grade cattle ownership, house

type, past land purchases, and type of off-farm income. A subjective

weighting of these variables was used to allocate farmers between the

two groups. In only a few cases, was there any uncertainty as to which

group a farmer belonged.

3.3. The Informal Survey

The informal survey lasted about five weeks and was carried out by
the author and an agronomist who spent two weeks in the study area.

The procedures followed correspond closely to those outlined in

Byerlee, Collinson, et al., 1980. Two to three farmers were interviewed

each day at their farms and about 60 farmers were interviewed, overall.

Researchers spent about the same amount of time evaluating the informa-

tion as in visiting the farms.

Wealth is often used as a means of stratifying farmers in farm
management studies. However, this approach is not appropriate for our
purposes for two reasons. First, high income and low income farmers
did not appear to have appreciable differences in the more common
measures of physical wealth, such as land or number of cattle. Second,
one of the most important questions concerning the planning of experi-
mentation is who has money for purchasing inputs and who does not.
Certainly, cash income is more closely associated with this distinction
than is the fairly lumpy resource base.

The agronomist needed substantially less familizarization time
than the economist since he was of the same ethnic group as study area
farmers, and he had lived and worked in an area just outside the study
area. A sunflower agronomist, maize breeder and bean agronomist each
spent 1 to 2 days in the field and were consulted on a number of issues
during the course of the survey.

3.3.1. Who to Interview

Before proceeding too far with the informal survey, it was

necessary to define "farmer" and "farm" in the context of local circum-

stances. Middle Kirinyaga was settled by families who were given title

deeds for pieces of land in the late 1950's. Since that time, many

farmers have subdivided their farms among their wives, children, or

other relatives. In this study, a farmer is defined as a person or

group of persons who manages a farm, that is, makes the decisions con-

cerning the allocation of resources on the farm and controls the output

from the farm. In the typical case of a husband, wife, and children,

the husband and wife share the decision-making function, with the hus-

band making the more general cash-related decisions, e.g., when to plow,

which seed to use, when to sell, and the wife making the more detailed

day-to-day decisions, e.g., how to thresh, how to space the crops, and

when to harvest. Many other systems of management and family organiza-

tion were found in Middle Kirinyaga and it was not always clear whether

one was dealing with an individual farm operation. For example, it was

common for a man to divide his land among one or more children, parti-

cularly his sons. In general, if the son was not married, the division

was in name only and decisions were made in common. However, if the

son was married, it was likely that he was operating separately from

his father. Polygamous marriages also presented complex situations. If

the husband was living on the farm, it was likely, but not always true,

that the farm was managed as a single unit. However, if the husband

was working away from home, it was likely that there was little if any

coordination between the wives; they represented different farms though

each may have been receiving assistance from the same spouse. Another

common arrangement was for a farmer to give an acre of land to his

mother to manage for herself.

We decided that any persons) managing a unit of land and its out-

put independently from others was suitable to be interviewed, with one

exception. We did not interview elders who were living with their

children and had been allocated a small piece (usually one acre) to farm

because (1) they represent a very small fraction of total production and

(2) they are not a target group for change agents or research services

proposing system improvements. We also made an effort to interview the

husband and wife together when it was apparent that management was

shared between them.

3.3.2. Interview Guidelines

We began farmer interviews by asking a short set of screening

questions as shown in Appendix B, Part 1. These questions were used to

(1) identify a particular household and the farm area associated with

it, and (2) determine which RD this household belonged to.

Two types of interview guidelines were used in the informal survey.

The first set, a general overview of important subject areas for develop-

ing an understanding of the farming system, is outlined in Appendix B,

Part 2. The guidelines are adopted from Collinson, 1980 and 1982, and

are divided into seven broad topics. Generally, each topic was covered

in a 1 to 1-1/2 hour visit with a farmer. Researchers interviewed one

to four farmers in each RD on each topic. The more complex a subject

area and the greater the variation in responses, the more farm interviews

were required to cover that area.

The second set of guidelines, developed after the informal survey

had begun, is presented in Appendix B, Part 3. As researchers develop

an understanding of the farming system, they identify leverage points

and draw up guidelines for gathering more information about them. These

guidelines help the researcher to assess farmer preferences using

repertory grids, to model selected decisions using hierarchical decision-

tree models, or to simply gather more information about a particular

topic. RG's received first priority because a detailed understanding of

farmer preferences among alternatives is useful for identifying parti-

cular farmer decisions to model and for constructing the HDM models

themselves. Repertory grids were constructed to evaluate farmer pre-

ferences among:

1. Alternative maize varieties
2. Alternative bean varieties
3. Alternative times for planting maize
4. Alternative times for planting beans

Hierarchical decision models were constructed to explain the following

farmer decisions,based on an evaluation of farmer preferences from RG's:

1. Decision to plant Katumani variety for early maize
2. Decision to plant Katumani variety for main stock of maize
3. Decision to plant hybrid 511/512 for main stock of maize
4. Decision on the time of planting for maize and beans.

Each RG and HDM was constructed from a sample of seven to twelve farmers.

3.3.3. Sampling Methods and Reporting of Results

Sampling in the informal survey was purposive and efforts were made

to interview farmers at different income levels, farmers at different

locations within the survey area, and farmers living along roads as well

as those living at a distance from roads. Simple methods to reduce

sampling bias, such as to interview the "nth" farmer on the left along

a particular path, were also used.

While the survey was still continuing, results were written up

corresponding to the topics listed in each of the guidelines. Repertory

grids and hierarchical decision tree models were also developed. The

survey results and proposed areas for experimentation were summarized

in Franzel and Njeru (1981), issued about two weeks after the survey was


3.4. The Formal Survey

3.4.1. Stratification and Sample Size

The farmer sample was stratified on the basis of income level, the

variable which differentiated the two recommendation domains. Determin-

ing the sample size was a difficult problem, since no data were available

for the target groups on the standard deviations of critical variables,

e.g., farm size, enterprise choice, income, farmer practices, etc.

However, there appears to be a consensus among many farm management

practitioners that 20 to 30 farms in an independent stratus are adequate

for producing reliable estimates for each stratum and for making compari-

sons between strata (Upton, 1972; Lynch, 1976; Bernsten, 1979). Since

there was no basis for claiming that the standard deviation for critical

variables was greater for one target group than for the other, it was

proposed that the sample size be the same number for each strata.

Forty-five farmers were selected from each stratum, to allow for the

possibility of having to exclude the questionnaires of non-cooperating

farmers and farmers who did not fall in either RD. The target sample

size was thus 90 farmers.

Fortunately, a relatively accurate sample frame exists for Middle

Kirinyaga. In the mid-1950's, all land in the area was demarcated by

the colonial government and accurate records are kept on all title deed

holders. Further, the government stipulates that all land transactions

purchases be officially approved and this law is strictly adhered to.

3.4.2. Selection of Sample Farmers and Organization of Fieldwork

Several problems remained in selecting a random sample: logistical

constraints, the fact that the stratum of individual farmers was not

known, and irregularities in the sample frame. The following three

sections describe each of these problems and outline the manner in which

they were resolved.

Accommodating Logistical Constraints

Logistical problems had to be taken into account, since our

resources for mounting the survey were limited. Our workforce was com-

posed of four enumerators,l two supervisors and one vehicle. Further,

locating sample farmers would require a local contact. Therefore,

clustering seemed to be the most appropriate method.

Determining the size and number of clusters for each group was the

next step. The questionnaire was to be administered in two visits, and

it was decided that visits were best scheduled on consecutive days at

the same time each day. This way, the likelihood of the farmer forgetting

We selected two female interviewers, since women play such an
important role in farming in Middle Kirinyaga.

the second appointment was minimized and enumerators were able to

remember details from the first interview for use in checking the con-

sistency of responses in the second-interview. Four enumerators,

interviewing two farmers per day, could complete eight farms per two

days. We decided to add a full extra day to each cluster to allow for

missed interviews, questionnaire checking, and reinterviewing.

We selected a multi-stage sampling method as the most efficient way

to meet the requirements of randomness given the logistical constraints.

Sampling was carried out in three stages:

1. Selection of sublocations to study. Map 1 shows that parts

of sixteen sublocations are included in the study area. Sublo-

cations were randomly selected giving greater weight to those

sample units with higher population.

2. Selection of areas within each sublocation. Each subloca-

tion consists of five to fifteen portions, each of which is

contained on a separate cadastral map. Sublocation-portions,

consisting of 25 to 150 farms, were selected to further cluster

the sample.

3. Selection of sample farmers. Finally, twenty farms were

selected from each map using a random number table. Twenty

was believed to be the minimum number of farmers required to

give at least five high income farms and five low income farms.

Classifying Farmers into Income Strata

Because there were many more low income farmers than high income

farmers in Middle Kirinyaga, a simple random sample would have given us

more low income farmers and fewer high income farms than we required.

Thus, we needed a method for stratifying farmers into income groups

before selecting our farmer sample. We decided to use the assistant

chiefs, who each preside over a sublocation, to obtain this information.

The assistant chiefs were asked to obtain information on the income

proxies listed in Section 3.2 for each farmer: house type and number

of zinc roofs, number of grade cows, number of oxen, off-farm jobs, and

whether the farmer had ever purchased land. No single proxy was suffi-

cient to establish the income group of a farmer. However, when informa-

tion on each proxy was assembled and evaluated it was usually clear

which income group the farmer belonged to. In many cases, the researcher

or a supervisor was able to discuss the grouping procedure with the

assistant chief and verify the classification of each farmer.

This procedure also had secondary benefits. First, it permitted

us to estimate the ratio of high income farmers to low income farmers

in the study area. Second, our sample size, and thus level of precision,

was increased. Fortunately, it was relatively easy to test the quality

of information provided by the assistant chief by comparing his data

with data obtained from the survey. Where important differences existed,

we consulted the chief and/or the farmer to resolve which was correct.

Thus the method was also useful for checking the validity of the data.

Table 3.1 compares the classification of farmers into income groups

based on information obtained from assistant chiefs with classifications

established after examining the completed questionnaires. Seventy-

seven percent of the initial classifications were correct, assuming of

course, that the final assessment based on survey data, was correct.

Errors were not biased in any particular direction. Moreover, many of

the errors were not errors in classification but errors in the sample

Table 3.1

Comparison of Two Methods for Classifying Farmers into
Income Groups: Classifications Based on Information
Supplied by Assistant Chief and Classifications Based
on Evaluation of Survey Questionnaires,
Middle Kirinyaga, 1981

Number of Farmersa

Data from Assistant Chiefs agrees with
data from survey 64 (77%)

Farms classified as high income, based on
information from chiefs, which were
found to be low income in survey 11 (13%)

Farms classified as low income, based on
information from chiefs, which were
found to be high income in survey 8 (10%)

Total 83 (100%)

aFour additional farmers interviewed were not classified by
assistant chiefs.

Source: Survey data.

frame. For example, in several cases, a high income title holder was

selected but a low income relative farming on the title-holder's farm

was interviewed.

Once the groupings were completed, the first five high income

farmers and the first five low income farmers were selected for inter-

viewing. If a farmer was unavailable for interviewing, the next farmer

on the list from his income group was selected. Of the 90 interviews

conducted, 3 were discarded because of suspicion that false information

was given. Of the 87 remaining questionnaires, 49 were for low income

farmers and 38 were for high income farmers. Low income farmers were

greater in number primarily because the quota of five high income

farmers could.not be obtained from sample lists in two of the subloca-

tions and because the three rejected farmers were all high income


Table 3.2 shows that 72% of the high income farmers and 84% of the

low income farmers interviewed were sample farmers, that is, farmers

selected from the sample list. Most of the remaining farmers inter-

viewed were also listed farmers, who replaced sample farmers not avail-

able to be interviewed.

Table 3.2

Sample Status of Farmers Interviewed in Formal Survey,
Middle Kirinyaga, 1981

High Income Farmers Low Income Farmers
Number Percent Number Percent

Sample Farmers 29 72 37 84

Replacement farmers
from sample list 11 27 5 11

Other farmers 0 2 4

Total 40 100 44 100

Source: Survey data.

Irregularities in the Sample Frame

There were two additional problems with the sample frame. First,

some high income farmers have more than one title deed; they appear in

the sample frame more than once and thus have a greater chance of being

selected. However, since only six high income farmers and no low income

farmers in our sample were in this category, no adjustments were made in

sample selection. However, adjustments were made in calculating the

ratio of high income farmers to low income farmers, since the number of

high income farmers on the list was biased upwards.

The second problem with the sample frame, referred to above, was

that some farmers had given a piece of their land to a relative, usually

a son, who operated independently from the title deed holder. These

relatives did not appear on the sample frame list, therefore some names

on the list actually represented two or more farmers. We decided not to

devise a system for randomly choosing a farmer under these circum-

stances; the choice would have had to be made by the enumerator himself

after arriving at the farm and could have created ill feelings between

him and the farmer. Therefore, the enumerator interviewed whoever was

willing to be interviewed on arrival at the farm. This system appeared

to work well; in some cases title deed holders were interviewed and in

other cases, relatives farming independently were interviewed.

3.4.3. Execution and Analysis

The formal survey was carried out during a five-week period coin-

ciding with the long rains maize harvest, August through September, 1981.

The survey was preceded by a three-week period of enumerator training

and questionnaire protesting. The questionnaire was translated into

Kikuyu, the native language of the area. Data tabluation began immedi-

ately after the survey was completed and most farmers were revisited to

check or clarify some of their initial responses. The preliminary

survey results and proposals for experimentation were summarized in

Franzel and Njeru (1982).



The objective of the following two chapters is to describe the farming

systems of Middle Kirinyaga and to examine farmers' management strate-

gies, as a basis for proposing experiments for developing system improve-

ments. In this chapter, natural and socio-economic features of Middle

Kirinyaga are presented. Next, overall system management is described

in a series of sections on farmer objectives and sources of income,

management of the farming system, and resource use and system con-

straints. Finally we outline the system leverage points, those areas

of the system where opportunities for improvement appear brightest. In

Chapter 5, we focus on farmers' practices concerning the leverage

points as a prelude to presenting proposals for a production research

program in Chapter 6.

4.1. Physical and Socio-Economics Features

4.1.1. Physical Features

Kirinyaga District extends from the summit of Mt. Kenya in the

north to low-rainfall lowlands in the south. The area selected for this

study, Middle Kirinyaga, is an area of flat to mildly sloping terrain at

an altitude of 1,200 to 1,350 meters (see Map 1). Soils are red, friable

clays of volcanic origin, high in humic content (3.7 percent carbon in

the A-horizon) and well-drained [Government of Kenya, 1970].

Rain falls in two seasons: the "long rains", March through May,

and the "short rains", October through December, as shown in Figure

4.1. Average rainfall in the long rains is 590 mm whereas the short

rains average 341 mm. The figure also shows the brevity of each season.

In the long rains, rainfall is over 20 mm per ten-day interval for a

period of 80 days; in the short rains the rain lasts only about 50 days.

Figure 4.2 presents average monthly rainfall and the probabilities

of receiving lower amounts in some years. The data highlights the

unreliability of rainfall in Middle Kirinyaga. For example, in four

years out of ten, rainfall is below two-thirds of the average in four

of the six highest-rainfall months, March, May, October and December.

In two years out of ten, rainfall is below one-third of the average in

three of these months. The figure also shows that the starting point

of each rainy season is more variable than is the ending point. For

example, although the long rains season never extends into June, it is

not certain whether the rains will begin in March or April.

Approximately 14 percent of the long rainsseasons over the past

28 years had less than 400 mm and may be characterized as poor seasons.

The corresponding number for the short rains is 56 percent. In the

last six years, two long rains seasons and three short rains seasons have

been poor, according to this definition.1

Table 4.1 shows the monthly moisture requirements of maize, Middle

Kirinyaga's principal crop, relative to the moisture available in

1In fact, the quantity of rainfall is a necessary but not suffici-
ent condition for high crop yields since it ignores the distribution of
the rainfall. Therefore, the measure used here probably underestimates
the number of seasons when rainfall is "poor".

Figure 4.1 Average Rainfall In Illddle Kirlnyaga, 1953-01 (nm. per ten-day period)

Rainfall (m)
per 10-day

1.1 10.4


Jan Feb Mar Apr May June July Aug Sept Oct Nov Dec

Source: Data for 1953-76 from Ndomba Government Farm. Data for 1977-81 from Mwea/Tebere Cotton Research Station

Figure 4.2 Average Hontly Rainfall and Probabilities of Receiving
Lower Amounts of Rainfall, Hiddle Kirinyaga, 1953-81











25.9 23.
9.4 7 20.5 18.5
S-.5 --

6: .






Average rainfall (28 year
average 11

Maximum amount received
in 4 lowest rainfall
years out of 10 > 1....

Maximum amount received
in 2 lowest rainfall
years out of 10 .1*- 1 ---




Jan Feb Mar Apr May June July Aug Sept Oct Nov Dec

Source: Data for 1953-76 from Ndomba Government Farm. Data for 1977-81 from r.1ea/Tebere
Cotton Research Station.






26.6 28.5







m-mm mmm m m = m m M M

Table 4.1

Rainfall Requirements for Maize
Lower Amounts of Rainfall,

and Probabilities
Middle Kirinyaga,

of Receiving

Rainfall Required Less Than Less Than
for Maize This Amount This Amount
is Received is Received
Long Short Potential aina in 2 Years in 4 Years
Month Rains Rains Evaporation Average Current' Out of 10 Out of 10

January 155 192 26 0 0.0 0.0
February 88 198 28 3 0.0 2.5
March 86 191 76 12 23.6 41.6
April 82 164 264 100 180.4 214.7
May 107 139 169 185 82.0 112.7
June 120 120 25 0 2.5 9.4
July 84 112 23 0 2.8 11.0
August 61 135 20 22 3.8 10.4
September 178 18 3 1.8 7.9
October 95 211 109 51 29.6 60.1
November 84 153 192 240 96.7 134.0
December 147 178 39 0 9.6 22.3


year preceding survey, 1980 long rains (March-August) and 1980-81 short rains (September-

Source: Rainfall requirements for maize are based on percentage of potential evaporation required.
These percentages by month are extrapolated from Brown and Cocheme, 1969.
Data for potential evaporation are from Mwea Tebere.
Rainfall data are from Figure 1.

Kirinyaga. The perils involved in the high unreliability of the rainfall

now become apparent. In the long rains, rainfall is quite sufficient in

late March, April and May, but falls below requirements from June to

August, even in normal years. However, the generally deep soils allow

the maize to take advantage of residual moisture during this period.

The situation is much more precarious during the short rains. October

and November rainfall are sufficient in normal years, but October rain-

fall falls far below requirements in four years out of ten. Rainfall

in December, January, and February is normally far below requirements.

Table 4.1 also shows rainfall in Middle Kirinyaga during the 1980-

81 year preceding the survey. Rainfall was considerably below normal;

thus data on production and income may be somewhat unrepresentative.

Nevertheless, the high degree of unreliability in rainfall--when the

rains will start, when they will finish, and how much rain will fall--

coupled with the general insufficiency of rainfall in all but the first

two months of each season cause grave problems to the farmers. Crop

failures occur and late planters, of course, are the most susceptible.

Hence, farmers seek to plant their crops as close as possible to the

start of the rains.

4.1.2. Population and Settlement

Middle Kirinyaga, as defined in this study, covers an area of approx-

imately 170 square kilometers and has a population of about 35,000,

giving a population per square kilometer of about 200. Before the mid-

1950's Middle Kirinyaga was practically uninhabited and was used by

Kikuyu farmers living on the slopes of Mt. Kenya for cattle grazing.

In the late 1950's the area was demarcated by the colonial government

and plots ranging from 2-6 ha. were given to farmers, mostly Kikuyus

from the crowded upper areas of Kirinyaga. These farmers migrated down

to Middle Kirinyaga to settle.

A single family was sometimes able to secure several plots, one for

each son in the family. Since demarcation, many of the farmers have

sub-divided their land among children and other relatives, who manage

their farms separately from that of the original title deed holder.

A continuous flow of farmers from upper Kirinyaga and other dis-

tricts have come to settle in Middle Kirinyaga, purchasing land from

those who obtained land during demarcation. Survey data show that approx-

imately 15 percent of the farmers have lived in the area for less than

ten years.

Household size is approximately 5.5 persons and most households

consist of a man, his wife(s) and children. However, about one-third of

the households are headed by women, because the husband is deceased or

is working away from home.

Farm size is approximately two to four hectares per household1 and

most farmers have only one piece of land, at their homestead. About 30

percent have another piece of land away from the homestead which they

are renting, borrowing or owning. This piece may be in the farmer's own

home area or even outside the area--in lower or upper Kirinyaga.

A household is defined as a group of people who join together to
make decisions about the management of the farm and the disposal of
the produce. Throughout this paper, decisions are attributed to "the
farmer" or "he". In fact, these decisions are more likely to be col-
lective decisions of the household or of the females alone, since they
are more active in farming than their husbands.

4.1.3. Transportation and Marketing

Transportation is excellent throughout the area. A tarmac road

passes east-west linking the zone with Nairobi, 100 km. to the south,

and Embu, 30 km. to the north. Feeder roads are also numerous and well-

kept but some are closed during periods of heavy rainfall.

Market centers for purchasing inputs,and selling produce are numer-

ous and nearly all farms are located less than 10 km. from a market

center. At these centers, a wide range of inputs are available, such

as fertilizer, improved maize seed, chemicals, and tools.

Overall, Middle Kirinyaga is a grain exporting area, but grain

imports are necessary in times of drought. Crops are bought and sold in

local markets and in addition, most grains and legumes may be bought

from or sold to National Produce Board Agents. Board agents are found

in all major market centers and Board prices are officially fixed by the

government. However, the prices at which the agents buy and sell pro-

ducts often fluctuates in accordance with trends in the open market.

Food grains are not permitted to enter or leave a district, except

through official Board channels, at official prices.

Figure 4.3 shows market prices and official buying prices for maize

and beans over 1979-81, and highlights the market price fluctuations

which occurred over the period. Produce price increases range from 50

to 350 percent as measured from harvest time up to the "hungry" season

which precedes the next harvest. In fact, such wide fluctuations are

not atypical. For example, the 1979 long rains crop harvested from July

through October was quite satisfactory and prices of maize and beans

were relatively low. However, the following three seasons were very poor

Figure 4.3 Maize and Bean Price Trends in Middle Kirinyaga, 1979-81



Per Kg.

P = Planting date
H = Harvesting month
..... Official buying price
for beans
----- Official buying price
for corn



1979 1980

S O N D J F M A M J J A S 0

Source: Prices are for Kagio Market, and were collected by the Divisional Agricultural Office, Ndia
Division, Baricho.

for both Middle Kirinyaga and Kenya as a whole and prices fluctuated

severely. Moreover, policy factors exacerbated the national maize sup-

ply situation during this period of poor harvests.1 The result was a

severe shortage of maize in Kenya throughout 1980; maize was not avail-

able in many cities and towns through many months of the year, and

prices increased accordingly. The price vagaries and supply uncertain-

ties demonstrate the importance of providing food from one's own home


4.1.4. Cooperatives, Credit, and Government
Agricultural Institutions

Cooperative activity is very low in the area, with only a few

farmers belonging to cotton or dairy cooperatives. In fact, few farmers

pursue either of these enterprises on a commercial basis.

Credit is available from commercial banks, the Agricultural Finance

Corporation, and from the Ministry of Agriculture's Seasonal Credit

Scheme. However, very few farmers have access to credit. For example,

a farmer is required to have two hectares under pure stand maize in

order to qualify for the Seasonal Credit Scheme. Moreover, the only

collateral most farmers have is their land and they are understandably

not willing to risk losing their land should they default on their loans.

Thus, the only farmers taking advantage of loan facilities are a very

small number of high income farmers who have extensive holdings to

A permanent secretary of the Office of the President blamed the
food shortages on "poor planning" [Daily Nation, March 13, 1982]. For
example, the government continued to export maize, even after it was
clear that a supply shortage was imminent, until strategic reserves
were depleted. Further, credit supplied to farmers for maize produc-
tion was severely curtailed, due to institutional bottlenecks in the
newly formed Seasonal Credit Scheme.

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

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