KickStart – Washington State University (WSU) Research Study
Milestone 11: Final Impact Evaluation Assessment Report
Mobile Layaway and R2O: Bringing an Innovative Savings Solution to the Rural 
Poor for Small-Scale Irrigation Pump Purchases
23 June 2016
i | P a g e
Mobile Layaway and Rent-to-Own: Bringing an 
Innovative Savings Solution to the Rural Poor for Small￾Scale Irrigation Pump Purchases 
Research Team – Milestone 3 
Project Final Report 
Baseline and End of Project Surveys with Cash, Rent-to-Own and Layaway Savings 
Options for Pump Purchases and a Control Group of Non-Buyers: 
Process and Results Highlights 
Thomas E. Byers, Douglas L. Young, 
Boris Houenou, Di Wu, Tyler Laferriere, 
Patrick Kirui, Teresiah Gitau, Jenniffer Mueni and Andrew Wanyonyi 
Byers is Associate Professor - International Research and Agricultural Development Unit, 
Washington State University and Associate Director for North Africa and Asia Programs, 
Young is Professor - School of Economic Sciences, Washington State University; 
Houenou, Wu and Laferriere are or were Graduate Research Assistants –
School of Economic Sciences, Washington State University 
Kirui, Gitau and Mueni are Graduate Students at the University of Nairobi; Kirui was Senior 
Advisor and Enumeration Team Leader, Gitau was Enumeration Team Leader, and Mueni 
was Senior Enumeration Team Member 
Wanyonyi was Senior Advisor and Enumeration Team Member who now works for the 
Kenya Water Board 
12 June 2016 
The authors are grateful for the invaluable assistance of Mr. John Kihia, KSI Country Director and 
Coordinator - Innovations for his invaluable insights and assistance at KSI Headquarters and in
the field. Administrative and field support was graciously provided by John Kihia, Angela Kere, 
and Esther Banda. Each was instrumental to the success of the End of Project Survey of Buyers 
and Non-Buyers of Treadle and/or Hip Pumps. The authors want to recognize other KSI senior 
and junior staff, who are too numerous to mention here, but who were instrumental in the 
success of all three Survey activities over the last three years. 
ii | P a g e
Table of Contents 
Table of Contents ........................................................................................................................ ii
Tables ........................................................................................................................................... iv
Acronyms...................................................................................................................................... v
Introduction ................................................................................................................................... 6
Comprehensive Description of Project Implementation ............................................................... 7
Project phases ......................................................................................................................................... 7
Phase 1 ................................................................................................................................................. 7
Phase 2 ................................................................................................................................................. 7
Phase 3 ................................................................................................................................................. 8
Clustering and randomization of project participants ..................................................................... 8
Team building, training and coordination with county administration........................................ 9
Development and discussion of descriptive statistics and frequencies .......................................10
Data input and validation...................................................................................................................10
Shortfall in TKT and statistical implications ....................................................................................11
Implementation lessons ......................................................................................................................11
County characteristics and population of interest ..........................................................................12
Application of results ..........................................................................................................................12
Descriptive Statistics for Selected Results .................................................................................. 13
Household, demographic and location results................................................................................13
Characteristics of pump purchaser and children........................................................................13
Discussion of medians for land use and pump utilization ............................................................15
Land use ............................................................................................................................................15
Pump utilization...............................................................................................................................15
Health and nutrition............................................................................................................................18
Micro finance, credit utilization and savings...................................................................................20
Access to MPesa and other finance opportunities.......................................................................20
Attitudes................................................................................................................................................20
Women’s empowerment and small scale irrigation .......................................................................21
Statistical Comparisons of Household Wellbeing of Moneymaker Pump Users versus Pump 
Non-Users: End of Project .......................................................................................................... 23
iii | P a g e
Statistical Comparisons of Household Wellbeing of Moneymaker Pump Purchasers Who Used 
Specified Finance Methods versus Non-Irrigators at End of Project.......................................... 25
Multiple Regression Analyses of Impacts of Moneymaker Pump Use on Selected Household 
Wellbeing Measures (Brief Summary) ....................................................................................... 28
Difference in differences analysis ......................................................................................................28
Regression with panel data.................................................................................................................29
Pathways for Future Scaling-up of the Intervention ................................................................... 30
Important Reasoning for Respondents to Adopt KSI Pumps...................................................... 31
Reasons for Purchasing a Pump.................................................................................................. 32
Special Survey: Shortfall in Utilization of Layaway Finance Option......................................... 33
Implications of results .........................................................................................................................35
Volunteered Comments from the End of Project Survey............................................................ 35
Non-Buyers’ comments .......................................................................................................................35
Buyers’ responses .................................................................................................................................35
Summary, Conclusions and Recommendation............................................................................ 36
Appendix 1: Sub-Counties (previously Districts) in which the study was undertaken............... 42
Appendix 2: Groups contacted to participate in the longitudinal study...................................... 43
Appendix 3: On-Line Appendices (Located in Dropbox Online – request share from Byers or 
Young)......................................................................................................................................... 44
iv | P a g e
Tables
Table 1: Phase 2 and phase 3 projected versus actual participation by individuals.......................11
Table 2: Characteristics of Eastern Kenya Counties............................................................................12
Table 3: Groups and gender of participant pool..................................................................................13
Table 4: Marital status of respondents in BaseS...................................................................................13
Table 5: Frequency of household types in BaseS.................................................................................14
Table 6: Percentages of respondents’ and spouses’ level of education ............................................14
Table 7: Percent for selected variables for buyers and non-buyers of pumps at baseline (Base) at 
April-June 2015 vs. end of project (EOP) at March-May 2016...........................................................14
Table 8: Comparison of medians of selected variables at baseline (BBase) vs. End of Project 
(BEOP) for buyers and non-buyers .......................................................................................................15
Table 9: Median treadle and hip pump usage on an annual basis....................................................16
Table 10: Median rainfall estimates across recent cropping cycles...................................................17
Table 11: Mean rainfall estimates by respondents across recent cropping cycles ...........................17
Table 12: Maize production ....................................................................................................................18
Table 13: Percentages of respondents reporting having experienced food shortages ...................19
Table 14: Percent of respondents reporting hunger............................................................................19
Table 15: Median responses to illness of family members .................................................................19
Table 16: Access to and utilization of MPESA and credit ..................................................................20
Table 17: Exposure to Agricultural Education and Extension...........................................................21
Table 18: Decision making and labor allocation in the household ...................................................22
Table 19: One-tailed t-test comparisons of means of pump users versus means of pump non￾users for 16 wellbeing variables at end of project survey ..................................................................24
Table 20: One-tailed t-test comparisons of wellbeing means of pump non-users to pump users 
who purchased pumps by specified finance methods: End of Project Survey (EOPS) .................27
Table 21: Respondents’ most important reasons for purchasing a Money Maker pump..............32
Table 22: Responses to layaway (EOPTKTS) follow-up survey.........................................................34
Table 23: Other reasons mentioned that impacted ability to continue the layaway program .......34
v | P a g e
Acronyms
Base Baseline Period (February – May 2015)
BBaseS Buyers Baseline Survey
BEOPS Buyers’ End of Project Survey
Cash Cash Pump Purchase Treatment
DID Difference in Differences
ET Enumeration Team
ETLC Enumeration Team Leader and Coordinators
EOP End of Project
EOPS End of Project Survey (March – May 2016)
EOPTKTS End of Project Layaway Survey
IRAD International Research and Agricultural Development Unit
KSI KickStart International
NBBaseS Non-Buyers Baseline Survey
NBEOPS Non-Buyers End of Project Survey
RA-DS Rapid Appraisal – Demographic Survey
RT Research Team
R2O Rent-to-Own Pump Purchase Treatment
SES School of Economic Sciences, Washington State University
TKT Layaway Savings Pump Purchase Treatment*
WSU Washington State University
*TKT is the Kiswahili translation for drip by drip.
6 | P a g e
Mobile Layaway and Rent-to-Own:
Bringing an Innovative Savings Solution to the Rural
Poor for Small-Scale Irrigation Pump Purchases: Baseline and End of Project Surveys of 
Pump Purchases and a Control Group of Non-Buyers:
Process and Results Highlights
WSU Research Team: Final Report, June 2016
Introduction 
Washington State University (WSU) entered an agreement on 15th April, 2014 to provide research 
services to KickStart International (KSI). This report summarizes two of the three project phases 
implemented from April, 2014 through June 2016 while focusing on the “Research” Milestone 3 
activities of 2016.
“Research” Milestone 1 was met by detailing the process and results of a Rapid Appraisal
Demographic Survey (RA-DS) implemented in March-June 2014. “Research” Milestone 2 
reported results on the “Buyers Baseline Survey (BBaseS)” and Non-Buyers Baseline Survey 
(NBBaseS) of Money Maker Max Treadle and Money Make Max Hip pumps. The Baseline 
surveys were implemented in March-June 2015. 
This report satisfies issues addressed by the WSU Research Team. It becomes part of the KSI’s Final
Project Report to USAID- DIV. The latter will cover a more comprehensive list of issues to be 
addressed by KSI’s Marketing Team and others incorporated in the KSI/WSU sub contract and 
presented here.
1. Review of baseline data and end-line data for each indicator in the project as well as 
monitoring and evaluation process.
2. Comprehensive description of project implementation process including development 
impact and relevant implementation lessons.
3. Comprehensive cost-effectiveness analysis of the intervention compared to competing 
alternatives
4. Comprehensive social impact analysis of the intervention.
5. Potential pathways for future scale-up of the intervention
6. Number of direct/indirect beneficiaries impacted by the project – men versus women
Requests 3) and 4) will be responded to by Kick Start International’s marketing in the final 
report for USAID-DIV.
Interpretations of results are referenced, when possible, to explicit objectives for the research 
(KSI, USAID and WSU, 2014):
#1 - How do the two financing schemes (Mobile Layaway and Rent to Own) affect the take-up 
rates of MoneyMaker irrigation pumps (MoneyMaker Max and MoneyMaker Hip). The finance 
treatment comparisons should be made separately by type of pump if both are marketed.
Layaway and rent to own will be compared to each other and to cash sales. What are the key 
7 | P a g e
features of the financing schemes that matter? 
#2 - Do these new financing schemes enable a different segment of farmers to invest in a pump 
compared to cash-only sales? In particular: do the new financing schemes enable higher 
adoption rates by women? Do they enable poorer farmers to adopt the pump? 
#3 - Does the method of financing affect impacts on farm household wellbeing from 
owning/using a KSI treadle or hip pump? Are the impacts higher/lower among farmers who 
use a financing scheme to buy a pump compared to those who buy using cash? Are the impacts 
higher/lower for each financing scheme compared to households who did not take up a pump? 
#4 - Use literature review and survey results to determine viability at scale of each financing 
mechanism and to determine how these mechanisms could break down barriers for asset 
acquisition for African women and poorest of poor farmers generally. 
Appendices are attached electronically to assure access to the bulk of research materials and 
data. The Appendices contain supporting documentation including detailed numerical results 
from the End of Project survey (EOPS) and the End of Project Layaway Survey (EOPTKTS). 
Comprehensive Description of Project Implementation 
Project phases 
Washington State University’s (WSU) International Research and Agricultural Development Unit
(IRAD) collaborated with KickStart International (KSI) to design and implement the project 
entitled “Mobile Layaway and Rent-to-Own: Bringing an Innovative Savings Solution to the 
Rural Poor for Small-Scale Irrigation Pump Purchases”. WSU undertook the three-phase research 
component of a broader KSI project to inform the funding organization, the United States Agency for 
International Development - Development Innovations Ventures, and KSI on the viability of 
three separate small scale irrigation financing methods: 
 Cash purchase,
 Rent to Own (R2O), and
 Layaway (TKT)
Phase 1: The Rapid Appraisal – Demographic Survey (RA-DS) reported in Milestone 1 identified 
farmers in the study area who were part of agriculturally related organizations willing to 
participate with the WSU research team. During the RA-DS, the group leaders of these 
organizations were questioned about their members’ gender, assets, risk of hunger, dominant 
food and cash crops and other pertinent information. Individuals within the organizations chose 
to participate with the WSU research team and were identified as interested candidates for 
August-December 2015 exclusive KSI marketing program to sell pumps by different finance 
schemes. 
Phase 2: The baseline surveys of pump buyers and non-buyers reported in the WSU Research 
Milestone 2 report were implemented in three eastern Kenya counties: Machakos, Makueni and 
8 | P a g e
Kitui during the period February – May 2015. The results from this phase activity was then 
provided to KSI whose job it was to develop a comprehensive marketing activity to these 
individuals without further involvement by WSU. The marketing activity was undertaken to 
assure as many sales as possible to each of the three treatment groups prior to the early rains in 
2015. 
Phase 3: Research during this phase addressed differences which may exist among three 
financing methods to purchase small scale irrigation systems; cash purchase (Cash), mobile 
layaway savings (TKT) and Rent to Own (R2O).
An additional component addressed differences between pump buyers and non-buyers. 
Milestone 3 reports on the individuals, both non-buyers and buyers during 2015 and through 
January 2016. 
Unlike the RA-DS in Phase 1, which addressed farm group leaders, the Phase 2 BBaseS (2015)
and Phase 3 EOPS (2016) and Phase 3 EOPTKTS (2016) questioned the same individual 
responders. These were families who had either purchased a KSI pump via Cash, registered to 
purchase via R2O, or were to be contacted by the KSI Marketing Team to register and purchase a 
pump using (Layaway-TKT). In addition, a non-buyers group was interviewed who did not 
benefit from either a treadle or hip pump purchase during the period of the surveys.
The “Baseline” survey (BBaseS and NBBaseS) and the End of Project Survey (BEOPS and 
NBEOPS) contain a wealth of data including family health, nutrition, income, wealth and assets, 
and children’s educational , costs and returns for different enterprises, production on irrigated 
and rainfed cropland, and related farm information. These data allow longitudinal analysis 
spanning a 12 to 14 month period. 
The “End of Project Layaway” survey (EOPTKTS), a one-time spot survey, was undertaken in 
February/March 2016. This was designed to help identify why so few of sixty-six potential
“Layaway” (TKT) buyers actually took up this financing option prior to January 2016. This result 
was inconsistent with expectations generated during the Research Team’s work earlier in 2015. 
The EOPTKTS seeks to explain the low levels of participation by respondents who had 
previously committed to the purchase of equipment using the “TKT” finance treatment option.
The End of Project Survey (EOPS) undertaken during the period February - May 2016,
interviewed the same individuals interviewed during a similar period one year earlier at Base.
The primary objective of the EOPS was to track impacts of up to 11 months of pump use, or 
through one production period, depending on time of access to equipment.
Clustering and randomization of project participants
Phase 2 and Phase 3 efforts were predicated upon the RT’s ability to identify and randomize 
groups, treatments and subsequently identify individuals. The following process was 
implemented to lay the foundation for meaningful results. 
 Two hundred and fifty-six groups were contacted during the Phase 1 Rapid Appraisal￾Demographic Survey in 2014. These groups consisted of 3,917 individuals, 67% of 
9 | P a g e
whom were female and 33% male. From the 256 groups, 71 clusters were generated 
(Clustering Process and Randomization Procedures {CPRP} document Dropbox.com –
Request sharing rights for access). 
 Clusters were dispersed by topography and geography. Highlands, rivers and plains areas 
offered topographical barriers to information transfer between clusters which was 
minimized. Spatially, a distance > than 15 kilometers between treatments has been 
achieved in all but a few cases. 
 Derivation of clusters, allocation of groups to clusters, treatments, breakout by county of 
clusters, identification of treatments groups and group leaders occurred. This was a multi￾step process; a summary is provided here.
 Each case (represented by a Group Leader), from the Rapid Appraisal-Demographic 
Survey (RA-DS) Dropbox was introduced to Google Earth utilizing EarthPoint 
software (http://www.earthpoint.us/Default.aspx).
 Locations associated with the EarthPoint data (latitude and longitude demarcations) 
were visually reviewed in Google Earth and prepared for clustering.
 Each group leader (case) and a subset of data associated with each leader, was 
introduced into a new spreadsheet for cluster generation utilizing Google Earth as a 
visual foundation tool.
 Each case was allocated to one of seventy-one clusters across the three counties. The 
qualitative allocation is based upon physical topography and spatial dispersion of 
cases across the three counties. (Example from the CPRP, Cases 102 93, 103, 96 fall 
into Cluster 1). 
 Each cluster was assigned a treatment number, 1, 2 or 3, sequentially across all 
clusters ([1, 2, 3][1, 2, 3] [1, 2, 3][1, 2, 3], etc - See column A of the CPRP).
 Each Case was allocated a randomly generated value to allow sorting by Cluster and 
subsequently by Treatment and/or County.
 Cases were sorted according to random values from low to high identifying 
associated Cluster and Treatment affiliations. 
 Treatments were color coded for visual identification
 Sorting by Treatment provided information necessary to select the “sales opportunity 
set” associated with each Treatment. Sales were to be generated by the KSI sales 
force based on this opportunity set. The sales which were subsequently made became
the framework for the sample population or in this case a census of buyers in the 
Buyers Baseline Survey (BBaseS) and the Buyers Baseline End of Project Survey 
(BEOPS).
 Clusters were then selected by choosing for each of the three treatment groups
sequentially. Selection was made after reviewing topographical barrier space or a 
physical distance greater than 15 kilometers from each other. Those which were less 
than 15 kilometers apart from each other were rejected from participation.
From the 3,917 potential buyers’, sales volume was anticipated to exceed statistical minimum 
sample size requirements.
Team building, training and coordination with county administration
The WSU research team was reassembled for the Phase 3 field activity in February, 2016 to
10 | P a g e
implement the EOPS. Three of four Kenyan research associates who had worked on the Phase 1 
and Phase 2 activities returned to implement the EOPS. The team met on Day 1, debriefed and 
translated the EOPS questionnaire into Kiswahili on Day 2, and departed for activities in Machakos 
County on Day 3. This rapid progress was facilitated by successful team building and training during 
the previous two phases. This team continued data collection, field activities, data analysis and input 
of data into MS Excel into the month of May, 2016.
As part of the research agreement with the Kenya National Commission on Science and 
Technology, introductions and authorization to undertake research for the duration of the project 
was received from County Commissioners, County Directors of Education, and County
Directors of Agriculture in Machakos, Makueni and Kitui Counties. In line with the NACOSTI 
agreement, copies of results of the RA-DS had been hand delivered to the administrative units in 
2015 after report completion. The BBaseS results, detailed in the Research Team‘s Milestone 2 
report, were delivered to each administrative unit in April 2016. The final report will be presented 
to the administrative units by KickStart International after submission
Development and discussion of descriptive statistics and frequencies
Three primary surveys were undertaken during the three phases of the project
 Rapid Appraisal-Demographic Survey (RA-DS)
 Baseline Survey (Base)
 End of Project Survey (EOPS)
The EOPS followed closely and was built upon the foundation of earlier works undertaken in 
Phase 1 and Phase 2.
Data input and validation
Data was collected in thirty-nine districts from farmers at locations, 
1. Household sites, 
2. Group meeting locations, and 
3. Markets nearby where they were going to be on the day of the interview.
Data collected on-site was then introduced into a rigorous data input and validation process 
which included:
 Entry into Excel files consistent with questionnaires
 Codes assigned to variables
 Incremental sampling of the data undertaken to identify the prevalence of remaining errors or 
omissions
 Delivered to the WSU partners for a third review utilizing a second incremental sample, 
followed by an aggregate data analysis run for mins/max. 
 Provided final evaluation for cleanliness and validity of the data set by PI and Co-PI.
 Final physical and mechanical checks of data set undertaken 
 Final review for consistency across relationships and variables 
11 | P a g e
 Survey instruments and data sets uploaded for DropBox cloud access by WSU for 
researchers and stakeholders. (Dropbox.com – Request sharing rights for access)
 Data analyzed using Microsoft Excel and IBM Statistical Package for the Social Sciences 
(SPSS)
 Data disaggregated to undertake special analysis projects 
Shortfall in TKT and statistical implications
Despite 67 households expressing an interest in the TKT mobile savings finance option at Base, 
only eight of households purchased pumps by this scheme by EOP. This was the foremost 
challenge to fulfilling research objectives of the project. Although this small subsample 
weakened statistical significance of impacts by finance scheme, many useful conclusions were 
generated. Reasons behind this shortfall are discussed later in this report.
Multiple regression analyses of survey data were weakened by small subsamples and by a 
relatively short time period between the BaseS and EOPS; nonetheless, t-tests and descriptive 
narrative were able to identify important lessons for KSI in design of TKT and for targeting of 
MoneyMaker (MM) pumps in general. The bottom line is that the RT was able to develop 
considerable quantitative and descriptive information in this report at minimal cost.
Table 1: Phase 2 and phase 3 projected versus actual participation by individuals
Action Taken
Cash
Treatment 
Group
Rent to Own 
(R2O) Treatment 
Group
Layaway (TKT) 
Treatment 
Group
Identified, purchased or 
committed to purchase
KSI Projected
90
KSI Projected
> 100
RT Identified
67
Participants who purchased 59 68 8
Implementation lessons 
 Geographic and topographic selection for this type of effort is an integral part of this 
genre of research project and must be provided adequate time and resources to be 
completed during Phase 1.
 Physical location of respondents needed to participate in a longitudinal study of this 
nature, involving expected purchase of technology, may not be known to the marketing 
group. 
 Meetings held at locations away from the buyer’s household require less effort to 
generate greater participation but are subject to long waits both for the enumeration team 
and the respondents who may be waiting in a que for unsustainable periods of time.
 GPS readings of group meeting locations, nearby villages, and/or homesteads of 
respondents must be developed in the first stages of this type of research activity. This 
may be a separate, distinct activity and may require several weeks to generate, dependent 
upon scope of the research project.
 GPS locations should be mapped via Google Earth and EarthPoint or another similar 
mechanism. Being able to arrive within meters of an anticipated meeting location is value 
added to the RT’s efforts and to the Marketing Team as well 
12 | P a g e
 The institutional infrastructure to undertake these activities may not be available but 
must be created to ensure timely completion of surveys.
County characteristics and population of interest
Table 2 displays basic county information from both secondary and primary data sources on
the target counties of Machakos, Makueni and Kitui. Kitui is the most remote in terms of 
distance from Nairobi and also is the most arid. Kitui lagged Makueni and Machakos in
household wealth and displayed more hunger incidence as reported in the Phase 1 RA-DS 
survey.
Table 2: Characteristics of Eastern Kenya Counties
County 
Name Population Rural Poor Literate Estimated Farm
Households
Groups
Interviewed
2009 % % % No.
Machakos 1,098,584 48 59.6 69.7 79,098 132
Makueni 884,527 88.2 64.1 72.7 117,023 63
Kitui 1,012,709 86.2 63.5 74.8 130,943 61
Total 2,995,820 327,064 256
Sources:
Phase 1 Rapid Appraisal Demographic Survey (RA-DS)
Populations from (http://www.scribd.com/doc/36672705/Kenya-Census-2009 )
% Rural from (http://fs12.formsite.com/ICTAuthorityKE/OpenData2014/index.html )
% Poor from O p e n D a t a K e n y a (https://www.opendata.go.ke/Poverty/Poverty-Rate-by-District/i5bp-z9aq )
World Bank (http://data.worldbank.org/country/Kenya/ )
Kenya Open Data https://www.opendata.go.ke/Counties/County-Urbanization-Kitui/747f-z33c
Maize, beans and cassava are the most important staple food crops in all three counties. Cash
crops, including cotton, coffee, and fruit, are found in favorable agro-climatic niches. Farmers
keep cattle, goats and donkeys in all three counties. Machakos County is adjacent to Nairobi and 
offers residents greater opportunities for seasonal off-farm employment.
Application of results
The conclusions of the BaseB, BaseN and EOPS apply to the farm households in each County.
These include 130,943 farm households in Kitui, 117,023 in Makueni, and 79,098 in Machakos.
With an average of six persons per household, the conclusions will apply to a population base 
of about two million people in the three Counties. Of these two million, those individuals or 
families who have access to water, and who purchase, access or rent and then utilize the small 
scale irrigation tools provided, should be considered direct beneficiaries. Secondary and tertiary 
beneficiaries would number in the hundreds of thousands. These begin from the base of almost 
4,000 possible participants identified during the research project who subsequently had 
knowledge of the technology and access to the KSI marketing team members in the Counties. 
Cautiously, one could generalize conclusions to other African locations which share
characteristics similar to the surveyed counties.
13 | P a g e
Descriptive Statistics for Selected Results
Household, demographic and location results 
Table 3 introduces the general background of potential respondents who were members of the 
256 groups contacted. The total membership was 3,917 of which 67% were female members and 
33% male. From this number, sales opportunities existed with 116 female group members and 
78 male group members. This suggests about a sixty/forty female to male ratio. 
Table 3: Groups and gender of participant pool
Gender Potential Buyers
RA-DS Identified groups* 76
Population of members in groups contacted 3,917
Number %
Female Group Members Identified 2589 67.0
Male Group Members Identified 1265 33.0
Female Sales Opportunities 116 59.8
Male Sales Opportunities 78 40.2
* Rapid Appraisal-Demographic Survey – Phase 1
Interestingly, the married couples, on a percentage basis were more likely to be treadle and/or 
hip pump purchasers (see Table 4). 
Table 4: Marital status of respondents in BaseS
Marital Status BBaseS (%) NBBuyers (%) BEOPS (%) NBEOPS (%)
 Married 84.5 73.5 81.7 72.3
 Unmarried 7.2 2.9 8.6 6.4
 Divorced/Separated 1.0 4.4 8.6 17
 Widowed 7.1 19.1 1.1 4.3
N 97 68 117 47
Characteristics of pump purchaser and children 
Table 5 reports that there were few families who did not have male and female household 
members. Few male or female single parent households appeared to be interested in the 
irrigation technology. The result implies that the KSI marketing team should not target 
single-parent households. The results of Table 5 indicate that households with two adults 
and younger children were more likely to buy KSI Money Maker pumps. 
14 | P a g e
Table 5: Frequency of household types in BaseS
Household Status BBaseS NBBuyers
Male and Female Adults 38.7 35.3
Female adult only 2.6 2.9
Male adult only 1.5 0*
Male adult and children < 14 yrs old in the household 0.5 0*
Female adult and children < 14 yrs old in the household .5 2.9
Male and female adults and children < 14 yrs old in the household 56.2 58.8
N 97 68
*/ None reported
Table 6 shows that in each category of education the spouse of the respondent had a similar 
education level, except for University. 
Table 6: Percentages of respondents’ and spouses’ level of education
Education level Respondent Spouse
Primary 58% 53.5%
Secondary 34% 36.6%
College 6.2% 9%
University 1% 9%
There was an approximate 60/40 split between female and male respondents in the 
Baseline who were buyers compared to an 80/20 split for non-buyers. In each case, perhaps 
due to the poor rainfall during the project, there were increases from Base to EOP in those 
who had small savings. These differences moved from 36% to 40% for buyers and from 
43% to 22% for non-buyers.
Table 7: Percent for selected variables for buyers and non-buyers of pumps at baseline (Base) 
at April-June 2015 vs. end of project (EOP) at March-May 2016
Survey 
Activity
Gender Access to 
adequate water 
to irrigate
Household 
Savings
Female 
Respondents %
Male 
Respondents %
Yes
%
<
15,000
KS (%)
>
15,000
KS
(%)
BBaseS 59.8 39.8 80 36 31
BEOPS 60.0 40.0 65.8 40 29
NBBaseS 80.9 19.1 76 43 16
NBEOPS 74.5 25.5 48.9 21 15
15 | P a g e
Discussion of medians for land use and pump utilization 
Land use 
Table 8 presents medians for several key variables related to crop production and to use of 
MoneyMaker irrigation pumps. Medians are preferred over means because they are less 
vulnerable to being skewed upward by a handful of very high values. Readers’ attention is 
directed to the horizontally adjacent columns showing Base and EOP. This presents the desired 
“before and after” impact of pump ownership and use. Irrigation should permit producing 
more crops and more valuable crops. Pump buyers expanded Money Maker Pump (MMP) 
irrigated ac/household of food crops from a median of 0.13 to 0.25. Marketed irrigated 
ac/household expanded similarly. The zero Base gross revenue (GR) for MM irrigated food 
crops is an artifact of medians. It means the bottom 50% of households’ GR from these crops 
was reported as zero. As expected, the quantities are smaller because semi-arid eastern Kenya 
produces relatively little non-food crops like cotton, coffee and tea. Sisal was the exception as it 
tolerates arid conditions. 
The entries of N/A (non-applicable) for Non-Buyers (NB) are expected. This group lacked 
pumps and would not be expected to produce crops under treadle or hip pump irrigation. 
The data show that eastern Kenya farmers in the sample area continued to produce rainfed food 
crops at similar levels at Base and at EOP. Producing food more efficiently on irrigated cropland 
possibly released some labor to continue producing dryland food. 
As expected, NB’s acreages of rainfed food crops changed little from Base to EOP. A slight 
increase in this group’s production of non-food crops was probably associated with improved 
precipitation at EOP. 
Pump utilization 
Among buyers at EOP, 63% of households chose hip pumps and the remainder treadle pumps. 
Table 8 reports median time spent by households was zero at Base with both types of pumps. 
This is inconsistent with the nonzero median acres of MMP irrigated land at Base reported 
below. However, our data show there are means for each of these cases which are positive and 
negate the concern associated with the 0 median in this case. If one extends this rationale to 
other BBaseS and NBBaseS medians one can discern benefits which are apparent over period 
between Base and EOP. 
Table 8: Comparison of medians of selected variables at baseline (BBase) vs. End of Project 
(BEOP) for buyers and non-buyers 
Money Maker Pump (MMP) irrigated food crops BBaseS BEOPS NBBaseS NBEOPS
Total acres of MMP irrigated food crops 0.13 0.25 NA NA
Marketed acres MMP irrigated food crop 0.12 0.25 NA NA
Gross revenue, marketed MMP irrigated food crops 
(KSh/yr.)
0.00 20,000 NA NA
Rainfed food crops BBaseS BEOPS NBBaseS NBEOPS
Total acres, rainfed food crops 2.50 3.00 2.75 2.50
16 | P a g e
(Table 8 Cont.)
Total acres marketed rainfed food crops 2.50 3.00 2.00 2.50
Total gross revenue rainfed food crops (KSh/yr.) 10,000 20,000 2,330 10,500
Rainfed non-food crops BBaseS BEOPS NBBaseS NBEOPS
Total acres rainfed non-food crops(ac) 0.00 0.50 0.00 0.125
Total acres marketed rainfed non-food crops 0.00 0.25 0.00 0.125
Total gross revenue rainfed non-food crops (KSh/yr.) 0.00 8,000 0.00 2,000
Marketed livestock and off-farm earnings BBaseS BEOPS NBBaseS NBEOPS
Gross revenue marketed livestock and livestock products 
(KSh/yr.)
0.00 0.50 0.00 0.125
Household off-farm earnings (KSh/yr.) 36,000 96,000 30,000 60,000
An increase in time spent by households using a pump on 3.00 acres of total rainfed food crops 
at EOP compared to 2.50 acres at Base is expected. An increase is normal as families become 
more familiar with the equipment and better appreciate its benefits. Table 9 shows EOP sample 
households’ median hours/year of pump use increased to 144 and 124 for treadle and hip 
pumps, respectively. This is a substantial investment by families in strenuous activity in 
frequently hot weather. It is expected that non-buyers would not be irrigating and that data for 
pump use would be not applicable (N/A).
Table 9: Median treadle and hip pump usage on an annual basis
Utilization Survey
Treadle Pump BBaseS BEOPS NBBaseS NBEOPS
Ave. hrs./day 0.00 3.00 NA NA
Ave. hrs./day 0.00 3.00 NA NA
No. months/year 0.00 4.5 NA NA
Total hrs. and days/year 0.00 144/18 NA NA
Hip Pump BBaseS BEOPS NBBaseS NBEOPS
Ave. hrs./day 0.00 2.00 NA NA
Ave. hrs./day 0.00 3.00 NA NA
No. months/year 0.00 4.00 NA NA
Total hrs. and days/year 0.00 124/15.5 NA NA
During the period when the project was being implemented, buyers and non-buyers were 
experiencing difficult rainfall conditions with over 70% of non-buyers responding that rainfall 
was below normal until the short rains of 2016 when normal or above normal rainfall patterns 
predominated. This, without question, resulted in making the marketing team’s efforts more 
difficult and may have been the determining factor in its inability to meet the sales needed 
reach minimum sample size levels. In addition, the poor rainfall conditions and short period for 
the famers to use their newly acquired irrigation equipment had an impact on data generated 
during the survey period. In addition, it is likely that safety-first requirements for minimum 
food consumption may have restricted the households’ willingness to market crops which were 
produced during the short-rains which ended in January 2016 in the project area.
17 | P a g e
Table 10 presents the median rainfall for 2014 (recalled in 2015 at Base), rainfall for 2015 at Base, 
and rainfall for 2016 at EOP. The latter result captures perceptions of the short rains which 
ended in January of 2016.
Table 10: Median rainfall estimates across recent cropping cycles
Rainfall 2014–Prior yr. recall
% of Respondents
2015 Current at Base
% of Respondents
2016 – Short Rains 
Current at EOP
% of Respondents
 Normal 34.0 23.7 43.2
 Above Normal 8.4 3.4 35.3
 Below Normal 57.6 72.9 21.6
N 262 262 142
Estimates of mean rainfall by respondents by year are presented in Table 11. This data supports 
the aggregated medians presented in Table 10. Mean rainfall in 2014, among both buyers and 
non- buyers was perceived to have been below normal. This was the same case in the 2015 long 
rains but more non-buyers were affected by the below normal rains during this period. In 
October, 2015 the short rains provided relief to 78% of the buyer respondents and 78 % of the 
non-buyers. The October 2015-January 2016 short rains mark a transition to above normal 
rainfall. Had the project been able to assess the impact of pump utilization after the end of the 
2016 long rains in the June – August period, the RT would have expected to see increased 
income and higher wealth indications. Safety first concerns would have diminished from 
improved food inventories allowing sales to expand. This, of course, assumes that the 2016 long 
rains were successful in making the maize and/or other staple crop.
Table 11: Mean rainfall estimates by respondents across recent cropping cycles
Rainfall 2014
(Prior Year Recall)
2015 2016 – Short Rains
BBaseS % NBBaseS % BBaseS % NBBaseS % BEOPS % NBEOPS %
Normal 23 25 37.6 23.5 35.9 57.4
Above Normal 4.1 1.5 10.3 2.9 42.4 21.3
Below Normal 71.9 73.5 52.1 73.5 21.7 21.3
N 194 68 194 90 92 47
Comparing Tables 10, 11 and 12 illustrates that poor 2014 and 2015 rainfall led to below 
average maize production among those interviewed. This created a barrier for the marketing 
team trying to sell pumps. It also created a challenge for the RT in interpreting irrigation area 
and pump utilization data. 
Conditions during the period when individuals could have utilized their treadle or hip pumps 
to their greatest advantage did not occur until the short rains in the final quarter of 2015 into 
January 2016. There was little access to water for irrigation during 2014 and 2015 from any of 
the sources normally available. Measureable, positive change to income, from increasing treadle 
or hip pump irrigation, would likely not have occurred until the short rains were over in 
18 | P a g e
January 2016. Emerging from a two-year drought, households may not have been willing to sell 
recently produced crops due to fear of future shortage threatening family food supplies. The 
necessity to secure the households’ consumption needs before entering the market to sell any 
surplus would need to be assessed carefully. This could only have been monitored by extending 
data collection beyond the EOPS. This was impossible due to the effective final report deadline 
by early-June 2016. 
The results presented in Table 12 show that maize production increased by EOP in line with the 
earlier discussion on rainfall presented in Tables 10 and 11. Table 12 reports that a cumulative 
72.8% of respondents reported “Average”, “Slightly above average”, and “Much above 
average” maize production in the EOPS. This represents maize production during the 2015-
January 2016 short rains.
Table 12: Maize production
Maize Production Baseline End of Project
Much below average 62.1 13.1
Slightly below average 22.2 14.6
Average 10.3 21.9
Slightly above average 1.5 25.5
Much above average 1.5 24.8
N 261 137
Health and nutrition
Drought continued to affect families into the second half of 2015 with households subject to 
shortages as reported in Table 13 and Table 14. The relationship is evident among individuals 
who experienced food shortages. While the number of respondents was small they are shown to 
have experienced a deteriorating condition between Base and EOP. Own farm production 
shortfalls increased during the research period leading to more respondents experiencing lack 
of funds to purchase food as production shortfalls continued to exacerbate the challenge to food 
security. 
Table 13 reports the combination of production shortfalls and lack of money to buy food 
decreased over the period (Condition 3: 93.8% respondents at Base declining to 77.3% at EOP) 
while at the same time the challenge became more apparent in both shortfalls and lack of 
money classifications. Both of these conditions were reported to have increased (deteriorated) 
from Base - Condition 1: 6.3% to 18.2% at EOP and Condition 2: zero% to 4.5%. This explains 
why Condition 3: seemingly improved; there were both more respondents who were 
experiencing food shortages and the impact had led to greater shortfalls and more respondents 
no longer having money to buy food but the combination of both was no longer as important to 
the family as the “shortage” of production or the “shortage” on money to buy food. This 
continued up until the short rains broke the drought in the fourth quarter of 2015. 
19 | P a g e
Table 13: Percentages of respondents reporting having experienced food shortages
Reason for Experiencing Food Shortage BBaseS BEOPS
1. “Own Farm” production shortfall 6.30 18.2
2. Lack of money to buy food 0.00 4.50
3. “Own Farm” production shortfall and lack of money 93.8 77.3
N 16 22
Sixteen respondents surveyed during the 2015 BaseS and 22 during the 2016 EOPS reported that 
they were affected by hunger on an annual basis across years. The proportions presented in 
Table 14 suggest that almost two-thirds of families “must” reduce meals or “miss” preferred 
meals at times. It seems reasonable to suggest that hunger occurs during periods when poor 
harvests occur and crop inventories are depleted. 
Over three quarters reported that the family regularly experiences hunger. The data collected at 
baseline represented responses that were given during a period of short rainfall, see Table 10 
and Table 11, corresponding too much below average maize production (see Table 12). 
Whether perception of the hunger over longer periods of time is correlated with rainfall and 
maize production cannot be determined definitively from the data set, but this seems 
reasonable from the results in Tables 10-14.
Table 14: Percent of respondents reporting hunger
Hunger Indicator Baseline Survey 
Family regularly experiences hunger during any week of year 78.7 %
Household must reduce meals/day during the year 62.0 %
Preferred meals are missed during the year 63.1 %
The median number of times that the respondent is ill each year was two at Base for buyers and 
three for non-buyers. The non-buyers appeared to be ill more times per years but for shorter 
periods of time and less severe.
Table 15: Median responses to illness of family members
Illness Indicators BBaseS NBBaseS
Number of times each year that respondent is ill 2 3
Family member is too sick to work during each sickness period (Days) 7 4
Child is unable to work or attend school due to illness each year (Days) 14 12
Days children 5 < are sick each year 5 4
Days children, => 10 yrs <=14 yrs, are unable to work or attend school 
due to illness each year 3.5 3
Estimated total days children are unable to work or attend school due to 
illness each year. 17.5 12
20 | P a g e
Micro finance, credit utilization and savings
Micro finance across Kenya is growing through the use of MPESA but can also be observed in 
the level of participation in savings clubs. In just a few years, MPesa has penetrated the most 
remote markets in Kenya. Furthermore, 87.6% of the respondents reported membership in a 
“Merry-Go-Round” savings club (see Table 16). 
The Kenya cell phone market is moving toward saturation even in remote locations. Anyone 
can borrow a phone to use MPesa but over time will create the environment necessary where 
cell phones are in almost every pocket. As this process evolves so will the expansion in smart 
phone usage and the benefits which flow from conducting economic activities online. The use of 
treadle and hip pumps should provide part of the revenue necessary to benefit from this 
landmark technology. 
Access to MPesa and other finance opportunities
Familiarity with MPesa and experience with Merry-go-round savings groups will work 
together over time to provide the resources necessary to take more loans while being assured to 
be able to pay-back the loan. Surprisingly, 65.8% of the respondents answering this question 
said they used MPesa to purchase their pump. Given contract requirements, this would include 
all R2O or TKT sales through either the vendor or KSI marketing agents. Informal discussions 
with respondents revealed that many of them used Merry-Go-Round” savings clubs to 
purchase their most expensive assets. 
Table 16: Access to and utilization of MPESA and credit
Response
Yes (%) No (%)
Familiar with MPesa 98.5 1.5
M-Pesa used to purchase your pump 65.8 34.2
Could have purchased w/o access to MPesa 68.4 31.6
Household members took loans, borrowed cash or received in-kind transfers 56.7 43.3
Member of “Merry-Go-Round” savings club 87.6 12.4
Attitudes
Attitudes regarding the environment, education and information sources influence technology 
adoption and can modify the state of knowledge and access to technology. Table 17 reports 81% 
of the respondents said they had attended educational sessions to improve their ability to 
produce crops. The most popular source of information on production and marketing was 
government extension services at 53.9%. The total of 15.6% of respondents who received 
information via cell phone is encouraging considering the short period of time that this 
technology has been available compared to radio and extension.
In the context of the small scale irrigation tools, by far the most important source for 
information is the vendor, 65% (see Table 17). This result identifies a major challenge to 
acceptance R2O and TKT. Cash sales are made directly from a vendor but this is not the case for 
21 | P a g e
R2O or TKT which are the result of interaction between KSI marketing and the potential buyer. 
65% of the owners rely on the vendor as their most useful information source but neither R2O 
nor TKT buyers have a direct seller/buyer relationship with a vendor at purchase. It may be 
difficult for the latter to access important information on product improvement, operation, 
maintenance. New technology information which could improve utilization may not be 
available to the TKT and R2O buyer if there is no formal link with the vendor until a challenge 
arises.
Table 17 confirms that a diverse set of media sources and organizations are used by farm 
households to access information. The transition to cell phones and now toward smart phones 
will transform access to information. These new information communication technologies 
should be utilized to link vendors with participants who have purchased via R2O or TKT.
Attitudes toward accessing information this way are changing rapidly empowering 
participants, who are already proud of their new mobile phone use, to gather and share 
information.
Table 17: Exposure to Agricultural Education and Extension
Agricultural education sessions and reception of information
Yes 
(%)
No
(%)
% of respondents who attended educational sessions on better methods for 
growing crops 81.2 18.8
Received the most useful information on agricultural production and 
marketing from (Type of communication Technology Tool): %1 N
Cell Phones 14.6 19
Government Extension Services 53.9 70
Radio 22.3 29
Pump Vendor 3.8 5
TV 2.3 3
Internet 3.1 4
Received the most useful information on small scale irrigation from (Type of 
Communication Technology or Median): %1 N
Cell Phones 1.3 2
Government Extension Services 20.6 32
Radio 10.3 16
Pump Vendor 65.2 101
TV 1.9 3
Internet .6 1
1 Percentage of those responding yes to receiving information through a type of communication media/organization 
or individual.
Women’s empowerment and small scale irrigation
Women make up over 67% of the 3,900 members of the 256 groups that were contacted during 
Phase 1. Among potential buyers, the breakdown was approximately 60% female and 40% 
22 | P a g e
male. Marketing to women and low resource producers was listed as KSI’s second explicit 
objective for the research (KSI, USAID and WSU; 2014). Consequently the RT devoted 
considerable attention to understanding the role of women in the household decision making 
structure. Table 18 presents the results of this inquiry. 
What is striking about the responses associated with this section are that in each, except one, 
women are making the decisions that are not made jointly at rates many percentage points 
higher than their male counterparts. With the exception of “primary responsibility to pay for inputs, 
bought in the market, which are for use on any irrigated plot” this result holds true. In the latter case, 
males were identified as the main decision maker by 22.8% of the respondents and 22.2% of the time 
female respondents were making the decisions; the remainder, 55%, reported decisions were made jointly. 
Strong pluralities of decisions were made jointly or “Both Equally” in 12 of the 14 decisions listed in 
Table 18. Women were reported to be making over 30% decisions in response to 7 of the 14 questions. 
These are italicized in bold print in Table 18. In the “Leadership” category, 40% or more of the decisions 
are made by women. These findings should inform KSI’s attempts to target women producers in its 
attempt to provide treadle and/or hip pumps to the population. 
Table 18: Decision making and labor allocation in the household 
Function
By whom are decisions in the household made and labor 
allocated?
Mainly 
husband or 
adult male
Mainly wife 
or adult 
female
Both
Equally
A. Production (Any type of Irrigation–pump/buckets/other N (%) N (%) N (%)
Who makes decisions about what non-vegetable crops 
are to be grown under irrigation? 
13(12.6) 26(25.2) 64(62.1)
Who makes decisions about what vegetables crops are 
to be grown under irrigation?
24(14.0) 52(30.2) 96(55.8)
B. Resource Control
Who has primary decision-making power over the use 
of productive resources and assets such as land, 
agricultural inputs and/or credit?
40(17.4) 69(26.5) 151(58.1)
C. Income Control
1.1. Who has primary control over revenue generated from 
irrigated crops which were subsequently sold in the 
formal market? 
12(7.3) 36(21.8) 117(70.9)
Who has primary control over revenue generated from 
irrigated crops which were subsequently sold to 
neighbors? 
13(7.8) 47(28.3) 106(63.9)
Who has primary control over revenue generated from 
MM pump produced crops which were subsequently 
sold in to individuals from close-by villages? 
11(11.1) 27(27.2) 61(61.6)
Who has primary responsibility to pay for inputs, 
bought in the market, which are for use on any irrigated 
plot(s)? 
41(22.8) 40(22.2) 99(55.0)
D. Leadership
In your household, who takes leadership in 
organizations that are related to irrigation? 
40(16.3) 101(41.1) 105(42.7)
In your household, who takes leadership in other 
organizations, open to participation by both women and 
men, which are designed to generate monetary income?
28(11.0) 104(41.6) 122(24.4)
23 | P a g e
(Table 18, Continued)
In your household, who takes leadership in other 
organizations, open to participation by both women and 
men, which are designed to generate/create social 
benefits but not monetary benefits? 
28(11.0) 109(42.8) 118(46.3)
E. Time Expenditure or Savings
Who in the family allocates the most time, on an annual 
basis, to the production of irrigated crops? 
35(19.3) 73(40.3) 73(40.3)
Who in the family allocates the most time, on an annual 
basis, to the production of irrigated cash crops? 
38(21.6) 67(38.1) 71(40.3)
Who in the family allocates the most time, on an annual 
basis, to the production of irrigated staple food crops?
21(18.3) 42(36.5) 52(45.2)
Who in the family controls the money generated from 
sales of crops grown and sold which required use of the 
treadle/hip pump for irrigation?
16(15.5) 27(26.2) 60(58.3)
Statistical Comparisons of Household Wellbeing of Moneymaker Pump 
Users1 versus Pump Non-Users: End of Project
The third research objective asks: “Are the impacts higher/lower for each financing scheme 
compared to households who did not take up a pump?” That question is addressed in the 
following section. This section statistically examines the precursor: Were household impacts 
higher or lower for all pump users, regardless of finance method used, versus those who did not 
irrigate. 
Table 19 presents 1-tailed t-test comparisons of means of pump users versus pump non-users 
for 16 wellbeing variables at EOP. One-tailed tests are used so that the alternative hypothesis 
implies irrigation improves wellbeing. As conventional, the null hypothesis (H0) is specified so 
that it rejection implies tentative acceptance of the theoretically desired alternative. For the 
hunger example, H0: Proportion hungry for irrigators at EOP > Proportion hungry for non￾irrigators at EOP. Table 19 reports this null is rejected at the 0.05 level so one may tentatively 
accept the desirable alternative that irrigators’ family hunger is less at EOP. This corresponds 
with the reported proportions hungry of 0.104 (about 10%) for irrigators (pump users) versus 
0.243 (about 24%) for dryland farmers (pump non-users) (see data columns 2 and 3, row 4 in 
Table 19). 
The nine results where the accepted alternative hypotheses imply that irrigation improved 
household wellbeing, and are explained by normal economic theory, are superscripted “+” in the 
Altern (Alternate) Hypothesis column. These include rows 3, 4, 5, 6, 7, 8, 9, 10 and 15; however, 
rows 7, 8 and 10 fail statistical significance. Rows 7 and 8 both relate to net revenue for rainfed
food crops where improvement in net revenue would not be expected. 
Greater net income from irrigation suggested student enrollment increases might be permitted 
by making tuition and school uniforms more affordable, but the results were not significant. 
1
 Users and non-users of pumps are not identical to buyers and non-buyers because 14 buyers had not 
used their pumps by EOP and two non-buyers had used a pump that had been given to them by World 
Vision. 
24 | P a g e
This is to be expected due to the very short period during two very poor rainfall years (see 
Tables 10 and 11) when most production would be retained for home consumption. Non-food
crops like coffee, tea, and cotton are not common in semi-arid eastern Kenya. Consequently, the 
weakly significant results in rows 13 and 14 should not be given great importance. No definitive 
theory was identified for the influence of irrigation on off-farm earnings. On one hand, by using 
labor more efficiently to produce food, some household members might be liberated to work off 
the farm. But theory could equally predict that with irrigation family members could produce 
more value on the farm rather than by working in the weak off-farm unskilled labor market in 
Kenya. Available data did not permit the RT to favor one theory over the other. 
Table 19: One-tailed t-test comparisons of means of pump users versus means of pump non￾users for 16 wellbeing variables at end of project survey 
Variable Mean or
proporti
on users
Mean or 
proportion 
non-users
Alternative. 
Hypothesis (HA), 1-
tail test
Accept 
HA
1-tail test
p - level 1-
tail (reject 
HO) & S.
2
Theory for HA
1. Primary students
(PS)
2.2 2.3 Mean PS: pump users 
> mean nonusers
No 0.400, NS Lower budget 
constraint for 
utility max.
2. Secondary
students (SS)
1.6 1.5 Mean PS: pump users 
> mean nonusers
No 0.264, NS Lower budget 
constraint for 
utility max.
3. Ac cropland 2015
(AcCr)
2.12 2.26 Mean AcCr pump: 
users > mean 
nonusers+
Yes 0.001, S at 
0.01 ***
Profit max
4. Proportion.
Hungry
0.104 0.243 Mean Proportion 
hungry: users < 
nonusers+
Yes 0.012; S at 
0.05 **
Utility max
5. Net Rev/Ac irrig.
crops, KSh
46,933 4,039 Net Rev/Ac irrig. 
Crops: users 
>nonusers+
Yes 0.0001; S at 
0.01 ***
Profit max
6. Net Rev/Farm
irrig. crops, KSh
15, 849 762 Net Rev/Farm irrig. 
Crops: users > 
nonusers+
Yes 0.00000009, 
S at 0.01 ***
Profit max
7. Net
Revenue/ACRE for
Rainfed Food Crops,
KSh
359 1,854 Net Revenue/ACRE 
for Rainfed Food 
Crops: users < 
nonusers+
No 0.264, NS High 
opportunity 
cost of labor 
for users
8. Net
Revenue/FARM for
Rainfed Food Crops,
KSh
6,957 7,528 Net Revenue/FARM 
for Rainfed Food 
Crops: users < 
nonusers+
No 0.450; NS High 
opportunity 
cost of labor 
for users
9. Wealth/FARM,
KSh
452,408 243,946 Wealth: users > 
nonusers+
Yes 0.004; S at 
0.01 ***
Reinvestment 
of profit in 
farm assets
10. Acres irrigated
using Bucket
0.234 0.243 Acres irrigated using 
Bucket: users < 
nonusers+
No 0.462, NS High 
opportunity 
cost labor
2 For the 1-tailed test, the null hypothesis is that one mean > or < the other mean. Rejecting this null implies that 
means are ranked with a reversed inequality for the alternative hypothesis. Significance (S) declines progressively 
from 0.01, 0.05, and 0.10 levels. These levels are denoted by ***, ** and *, respectively. NS denotes not significant at 
the lowest 0.10 level. 
25 | P a g e
(Table 19 Continued)
11. Total acres 
rainfed food crops
4.417 2.740 Total acres of rainfed 
food crops: users < 
nonusers
No 0.0007, S at 
0.01 ***
Uncertain
12. Total gross 
revenue rainfed food 
crops (KSh/yr)
34,959 25,346 Total gross revenue 
rainfed food crops: 
users < nonusers
No 0.067, S at 
0.10
Uncertain
13. Total acres 
rainfed non-food 
crops
0.55 0.198 Total acres rainfed 
non-food crops: users 
> nonusers
Yes 0.06, S at 
0.10
Uncertain
14. Total gross 
revenue rainfed non￾food crops (KSh/yr)
28,700 4,745 Total gross revenue 
rainfed non-food 
crops: users > 
nonusers
Yes 0.07; 
Significant 
at 0.10
Uncertain
15.Gross revenue 
marketed livestock 
and livestock 
products (KSh/yr)
62,801 19,148 Gross revenue 
marketed livestock 
and livestock 
products: users > 
nonusers+
Yes 0.068, S at 
0.10 
Profit max
16. Household off￾farm earnings 
(KSh/yr)
151,840 137,389 Household off-farm 
earnings: users > 
nonusers
No 0.306, NS Uncertain
Statistical Comparisons of Household Wellbeing of Moneymaker Pump 
Purchasers Who Used Specified Finance Methods versus Non￾Irrigators at End of Project
This section statistically tests results pertaining to the third research objective: “Are the impacts 
higher or lower for each financing scheme compared to households who did not take up a 
pump?” Table 20 presents statistical answers to this question for the same 16 household 
wellbeing variables included in Table 19. The explanation of and justification for 1-tailed t-tests 
remains the same. No theories are listed for alternatives to the null hypotheses as in Table 19. 
The prevailing theory, although not listed in Table 19, is that pump use and irrigation, 
regardless of finance method, should improve wellbeing relative to dryland farming. Wellbeing 
means or proportions are presented for:
a) Non-users 
b) Cash pump purchasers
c) R20 purchasers and
d) TKT purchasers. 
For acres of cropland (Ac cropland) 2015, Proportion Hungry, Net Revenue per acre (Net Rev) 
and per farm for Irrigated crops (Irrig. crops), all three finance options show better wellbeing 
than non-irrigators at the strong 0.01 level. Cash has better point estimates for Ac cropland and 
TKT for Proportion Hungry, but R2O is slightly stronger over Cash for Net Rev per acre and per 
farm for Irrig. crops. Results for TKT are weaker because fewer than ten households purchased 
equipment using this finance/treatment method. Irrigation would be expected to have less 
effect on rainfed crops and rows 7, 8, 11 and 12 fail to show poor significance across finance 
26 | P a g e
methods. As introduced earlier, non-food crops are not common in semi-arid eastern Kenya. 
Results for rows 13 and 14 fail significance for Cash and R20; it is not applicable (N/A) for TKT 
due to a small sample size. Wealth for cash purchasers at KSh 675,378 easily has highest 
superiority over non-irrigators and is significant at the 0.01 level. R2O and TKT lack statistically 
significant superiority over non-irrigators. The significance across finance methods is erratic.
Gross revenue from livestock is weakly superior for cash relative to non-irrigators and fails 
significance for the other two finance methods. Off-farm earnings are statistically superior at the 
0.01 level for cash compared to non-irrigators, but for R2O and TKT lacked statistical 
significance. A plausible explanation may be that higher off-farm earnings, at a mean of KSh 
204,267, permitted cash households to pay cash for their irrigation pump. An alternative 
explanation may be that cash buyers had more opportunities to generate wealth due to more of 
them having access to the equipment during the long rainy season in 2015. On the other hand, 
R20 and TKT purchasers showed statistically weaker off-farm earnings when compared to non￾irrigators.
Taken collectively, the results in Table 20 show that cash purchasers had the highest household 
wellbeing advantages over non-irrigators. Cash pump purchasers had the highest and most 
significant net revenue advantage from irrigated food crops. Moreover, cash pump purchasers
experienced significantly less hunger, were significantly wealthier, had significantly greater 
nonfarm earnings and irrigated less cropland with a bucket. The results show R2O purchasers 
were in second place for wellbeing compared to Cash. 
Layaway results should be regarded cautiously since this subsample contained only eight 
households. As a result, TKT (layaway) results were not expected to be substantially revealing 
given the few sales that occurred among treatment participants. All t-tests for TKT households
lack statistical power as a result of small sample size. TKT households earned more net revenue 
from rainfed crops and irrigated more acres with a bucket than other types of households. TKT 
pump purchasers were less wealthy and earned the least from non-farm sources compared to 
the other households. Only a small number of households used the layaway micro-credit 
scheme to purchase MoneyMaker pumps. This issue was addressed in a follow-up Layaway 
survey undertaken to assess why so few of potential buyers in this treatment purchased 
equipment. Lack of money and shortness of payback period were the most common responses 
but others were perceived as equally important by other potential buyers. The results of this 
data analysis are introduced in Table 20.
27 | P a g e
Table 20: One-tailed t-test comparisons of wellbeing means of pump non-users to pump 
users3 who purchased pumps by specified finance methods: End of Project Survey (EOPS)
Variable
Non-users
Mean or 
proportion
Cash
Mean 
or
propor.
R2O4
Mean 
or
propor.
TKT5
Mean 
or
propor.
Cash
p-level,
1-tail 
(reject 
HO) & S6
R2O
p-level 1-tail 
(reject HO) & S
Layaway(TKT)
p-level 1-tail
(reject HO) & S
1. Primary students (PS) 2.3 2.25 2.26 1.5 0.400, NS 0.047, S at 0.05 0.037, S at 0.05
2. Secondary students (SS) 1.5 1.9 1.56 1.2 0.062, S at 
0.10
0.04, S at 0.05 0.114, NS
3. Ac cropland 2015 (AcCr) 2.27 4.92 4.036 3.07 0.007, S at 
0.01
0.02, S at 0.05 0.318, NS
4. Proportion. Hungry 0.243 0.05 0.17 0 0.0013; S
at 0.01
0.170, NS 0.000006, S at 
0.01
5. Net Rev/Ac irrig. crops, 
KSh
4,039 44,850 54,524 19,800 0.001; S at 
0.01
0.0000454, S at 
0.01
0.117, NS
6. Net Rev/Farm irrig. 
crops, KSh
762 16,216 17,603 6,100 0.0006, S 
at 0.01
0.00005, S at 0.01 0.054, S at 0.10
7. Net Revenue/ACRE for 
rainfed food crops, KSh
1,854 -2,460 259 12,391 0.088, S at 
0.10
0.278, NS 0.157,NS
8. Net Revenue/FARM for 
rainfed food crops, KSh
7,528 1,720 8,293 22,144 0.145; NS 0.440, NS 0.111, NS
9. Wealth/FARM, KSh 243,946 675,378 316,984 174,725 0.004; S 
at 0.01
0.98, NS 0.117, NS
10. Acres irrigated using 
Bucket
0.243 0.0374 0.1875 0.458 0.00007, S 
at 0.01
0.176, NS 0.251, NS
11. Total acres rainfed food 
crops
2.740 4.33 4.69 3.286 0.014, S at 
0.05
0.007, S at 0.01 0.284, NS
12. Total gross revenue 
rainfed food crops (KSh/yr)
25,346 33,402 34,490 42,643 0.142, NS 0.123, NS 0.104, NS
13. Total acres rainfed non￾food crops
0.198 0.75 0.5 0.5 0.127, NS 0.166, NS N/A
14. Total gross revenue 
rainfed non-food crops 
(KSh/yr)
4,745 88,000 15,957 0 0.204; NS 0.103, NS N/A
15. Gross revenue marketed 
livestock and livestock 
products (KSh/yr)
19,148 116,777 26,830 18,778 0.08, S at 
0.10
0.142, NS 0.484, NS
16. Household off-farm 
earnings (KSh/yr)
137,389 204,267 122,757 89,750 0.076, S at
0.01
0.308, NS 0.054, S at 0.10
 
3 Users and Nonusers of pumps are not the same groups as Buyers and Non-buyers of Pumps because 14 buyers had 
not used their pumps by EOP and two Non-buyers had used a pump that had been given to them by World Vision. 
4 R2O is rent to own. The farmer takes possession of the pump immediately after a down payment and may keep it as 
long as his/her rental payments are current.
5 TKT is laying away savings by MPesa. When the full price of the pump is laid away, the farmer is entitled to collect 
his/her pump.
6 For a one-tailed test, the null hypothesis is that one mean > or < the other mean. Significance (S) means this null is 
rejected and that the alternative hypothesis, namely that means are ranked with a reversed inequality, may be 
tentatively accepted. Significance is reduced progressively from 0.01 to 0.05 to 0.10 levels. NS denotes failure to reject 
the null at the 0.10 level. 
28 | P a g e
Multiple Regression Analyses of Impacts of Moneymaker Pump Use on 
Selected Household Wellbeing Measures (Brief Summary)
The previous t-tests of impacts of irrigation on household wellbeing are limited by not 
controlling for confounding factors which could have changed household wellbeing between 
Baseline (B) and End of Project (EOP). These include weather, physical location, farm size and 
other factors. Multiple regression computes marginal changes in the dependent wellbeing 
variable associated with one explanatory variable while other explanatory variables are held 
constant.
Two multiple regression approaches were attempted: Differences in differences (DID) and 
panel data regression (Abadie, 2005; Angrist & Pische, 2008; Bertrand et al., 2004; Card and 
Krueger, 1994). The unit of observation was the household interviewed at both baseline and 
EOP. Matching of households, in both non-irrigators and irrigators groups, over the two 
surveys was a key design of this research.
Difference in differences analysis 
Difference in Differences (DID) analysis, developed in the 1990’s, is the most modern technique 
to show “before and after differences” in a dependent variable like household income or hunger 
after some “intervention” such as adoption of irrigation, passage of new legislation, or political 
upheaval. It does so by utilizing a control group, in our case non-irrigators, who went through 
the Base to EOP period experiencing the same weather and other external forces. Including the 
control group requires merging the buyers and non-buyers data sets. DID also enters a set of 
“control variables” relevant to the wellbeing measure. These variables control for possible 
confounding. In our case, these included gender of respondent; low, normal or high rainfall; 
Kitui, Makueni, or Machakos county of residence; traditional family of male & female adults 
with children under 14, education of respondent secondary or higher, main occupation of 
household is farming, possession of adequate water to irrigate cropland, use of Hip pump, 
household savings greater than KSH 15,000, total annual off-farm earnings of household 
members and household wealth. Separate DID equations were estimated for nine household 
wellbeing factors: (1) number of primary students, (2) number of secondary students, (3) 
number of college and university students, (4) cropland ac planted last year, (5) hungry any 
week in year (yes/no; binary), (6) reduce meals any day in year (binary), (7) consume less 
preferred meals in year (binary), (8) net revenue/ac of food crops and (9) net revenue/farm of 
food crops. Six of these variables, (1), (2), (4), (5), (8) and (9) were also included in the t-tests 
reported in Tables 19 and 20. Unfortunately, the DID regression was not feasible due to 
technical statistical reasons when finance scheme was added as a control variable. 
It should be stated at the outset that the DID literature emphasizes that it works best when (a) 
the intervention shows a pronounced break between before and after and (b) the sample size is 
relatively large. Neither of these conditions applied to our data. Regarding (a), only one year 
elapsed between Base and EOP due to project term limitations. This limited the amount of time 
for experience with the pumps to accumulate and for their full effects to be realized. Indeed for 
most buyers, they would have only seen large improvement in their wellbeing after the final set 
of data had been collected as discussed earlier in the report. Also, different households used the 
29 | P a g e
pumps to varying degrees during this short period. Regarding (b) our sample with complete 
data included only 96 pump users (“irrigators”) and 70 non-users (“non-irrigators”). In contrast, 
some DID analyses contain more than 1,000 observations. Our small sample was constrained by 
the ability of KSI to market equipment as effectively as had been hoped. As discussed above, 
this was likely due to drought conditions that affected household’s ability to buy pumps, and 
also their ability to access water for irrigation from 2014 up to the short rains in the four quarter 
of 2015.
In light of these limitations, it was not surprising that a limited number of statistically 
significant relationships emerged at 0.01, 0.05, or 0.10 levels. R2’s, showing “goodness of fit” 
were low ranging from 0.037 for number of primary students to 0.435 for acres of cropland. Net 
revenue per farm from food crops was the only wellbeing variable with a statistically significant 
difference from baseline to EOP. Net revenue per farm from food crops was also positively and 
significantly correlated with farming as principal occupation and savings exceeding KSh 15,000. 
Attention to farming on a full-time basis and adequate savings to purchase inputs improves 
farm profit. 
Wellbeing variable (6), hungry any week of the year, was significantly related with education 
greater than secondary and savings greater than KSh 15,000. Both these coefficients were 
negative as predicted by theory; in other words, greater education and savings reduced hunger. 
Number of primary students was positively and significantly related to savings exceeding KSh 
15,000. Greater savings plausibly made school related costs more affordable. No other wellbeing 
variables displayed significance with the control variables. The category, “greater than KSh 
15,000 savings” was most frequently significant. Education equal to or greater than secondary 
school ranked second. Savings and education are frequently significant in the technology 
adoption literature.
A complete matrix of DID results with coefficient estimates, standard errors of coefficients, 
significance levels, and R2’s is included in the online appendix. The appendix also includes a 
table of theoretically expected coefficient signs for several explanatory variables for several 
possible dependent variables. The range of variables differs slightly from the lists above.
Regression with panel data 
Panel data regression is an older technique where observations on the same entities, in our case 
farm households, are observed over two or more time periods
(http://www.indiana.edu/~wim/docs/10_7_2011_slides.pdf). The data set also includes a set 
of explanatory variables. There is no control group.
Results from the panel data regression are similar to those for DID. Cropland acreage, hungry 
any week of the year (and its duplicates) and net revenue/farm from food crops again show the 
most statistically significant relationships. The category, “greater than KSh 15,000 savings” was
again the strongest explanatory variable. The DID and panel data approaches are 
methodologically different, and there were minor differences in the included variables, so 
comparisons should be made with caution. Readers with greater interest in the panel data 
results are referred to the online “Appendix 3” showing estimated coefficients, standard errors, 
and significance levels for each explanatory variable for each wellbeing regression.
30 | P a g e
Pathways for Future Scaling-up of the Intervention
The authors consulted extensive literature on agricultural technology adoption in other regions 
of Kenya and other African countries to attempt to ascertain how different financing 
mechanism could break down barriers to asset acquisition (Gitau et al., 2015). This literature 
review summary was included in the Milestone II report and is repeated here. Two of the three 
eastern Kenyan counties in this research were populated by dominantly rural households, all 
three possessed mainly poor households solely dependent on farming. On the positive side, 
from 70 to 75 percent of the counties’ populations were literate. This shows the potential that 
farmers could understand how to operate and care for new agricultural equipment like KSI’s 
small scale irrigation equipment. The authors believe this KSI-WSU research might be 
generalized to Kenya’s Kwale and Tharaka agro-ecological zones because they are similar to the 
eastern counties (Kenya Food Security Group, 2015). Truly arid regions such as Turkana, 
Laikipia and Isiolo are mainly inhabited by pastoralists rather than farmers. This makes them 
poor candidates for the study’s results.
Some factors might discourage layaway savings elsewhere in Africa. Mobile banking has not 
been as successful everywhere in Africa as in Kenya. For instance, Somalia’s agricultural 
ventures are discouraged by the political instability of the country. Islamic rules against paying 
any form of interest on loans could inhibit use of R2O in Arabic countries in North Africa 
(Savin, 2012). The eastern Kenya results will be difficult or impossible to duplicate in most
Saharan and southern African countries that experience arid and/or desert conditions. The 
mobile layaway and R2O financing schemes will not overcome a genuine lack of water (Kirui et 
al., 2013).
Mobile layaway and R2O financing could be beneficial to women by permitting them to expand 
their asset base. Women were very active in the farm groups surveyed in the RA-DS. The 
percent of women ranged from 63 in Makueni to 82 in Kitui (Young et al., 2014). 
Based on a popular brochure, KickStart International (2015) asserts 93% of women feel 
empowered after buying the pump. Women empowerment by the Kick Start project has tripled 
the number of students attending schools, 28% of children in the communities using pumps can 
attend private schools, an increase from 9%. These exceptional benefits are also attributed to the 
fact that women control 36% of the household income generated through irrigation. 
Communities embracing the technology experience a claimed doubling in the income per 
household KickStart International (2015). 
Literature from other countries reveals that certain variables have a consistent positive effect on 
financial innovation and agricultural technology adoption. These include higher education, 
exposure to extension programs which teach farmers how to use the new technologies, and 
adoption of other financial innovations. These relationships are shown by studies in Kenya and 
Nigeria (Owuor, 2009; Omonoma, 2010). The age range 21-49 years has higher technology 
adoption rates. KSI’s Rent-to-Own and Layaway financial options, as well as Cash Purchase, 
may facilitate adoption by a broader range of households. In Ethiopia, which shares agro-
31 | P a g e
climatic conditions with Kenya, microfinance credit, access to extension services, and shorter 
distance to the nearest road and/or market positively influenced adoption of irrigation 
(Liverpool, 2010; Goshu, 2012). Ethiopia possesses extensive agricultural lands. Malawi has 
greater availability of water and land which may contribute to its higher adoption of human￾powered pumps. Higher share of off-farm income, participation in farmers’ self-help groups, 
and contact with extension services were shown to promote adoption of pumps and similar 
technology in Madagascar, Tanzania, and Malawi (Moser, 2006; Kabunga, 2011; Mohamed, 
2008; Amudavi, 2009). 
From the literature review, the authors believe Kick Start pump technology has strong potential 
to alleviate poverty and enhance food insecurity in Africa, especially due to an increased 
empowerment of African rural women. However, households must be able to afford the pump 
and hoses as a lump sum purchase or over time through layaway or rent-to-own. The authors 
believe there are niches in the agricultural population throughout Africa under current 
economic growth trends that can afford this investment.
Important Reasoning for Respondents to Adopt KSI Pumps
One hundred-ninety four of the surveyed respondents in the BaseS were potential buyers of 
human-powered irrigation pumps (Young et al., 2015). Of these, 59 indicated an interest in cash 
purchase, 68 in rent to own, and 67 in TKT-layaway savings. As a control group, another 68 
households in the same locations as the respondents, who had neither purchased a pump nor 
who were interested in doing so, were surveyed. The baseline surveys asked for considerable 
information about access to water, production of food, non-food crops, livestock, and household 
wellbeing. The wellbeing measures included family nutrition, family health, children’s 
enrollment in education, farm income from various enterprises, and household wealth. General 
farm management information elicited use of credit, land allocation to different crops, irrigation 
intentions, revenues and costs of rainfed or irrigated crops, labor utilization and exposure to 
agricultural education and remoteness from public services was also elicited. The bulk of this 
information is included in the on-line appendices.
During March-April 2016 the RT team conducted End of Project (EOP) surveys of the same farm 
households among both pump buyers and non-buyers. The EOP surveys used a subset of 
identical questions from the Baseline. By the EOP, among 168 respondents, only eight (8) had 
used the layaway finance treatment to acquire pumps, with the remainder using cash and rent 
to own to acquire pumps. This implies an attrition of 59 households who intended to use 
layaway to purchase pumps to those who actually did so. Finally, in May 2016, 43 households 
who had expressed an interest in layaway at baseline, but had dropped from the program or 
had yet to receive their pump by the time the EOPS was undertaken, were surveyed to learn 
more about their inability to complete the purchase.
The Base, EOP, and Layaway surveys all contained a number of self-classification questions. We 
believe the answers to these can assist KSI in targeting future pump promotion. Qualitative 
responses from participants also aid in the interpretation of quantitative results. 
32 | P a g e
Reasons for Purchasing a Pump 
Table 21 reveals that economic reasons dominated plans for purchasing a pump in the Baseline 
survey. These included “Increase income,” Enable to irrigate larger parcels,” “Inexpensive,” 
“No operational costs,” “No costs involved,” and “Easy to pay.” These six reasons accounted 
for 53% and 57% of the first and second most important reasons, respectively. 
A second grouping of “important” reasons related to ease of use and efficiency in use of water 
and labor. The EOP results showed 49% of buyers considered the ability to irrigate larger 
parcels as the most important reason for purchasing their irrigation pump. When compared 
with the Base result of 16% one might conclude that use of the equipment allowed respondents 
and their family members to more clearly recognize the benefits of bringing more land of 
irrigation. At Base increasing income was seen to be the major motivating factor for purchase. 
This may be related to marketing efforts which emphasize profit through intensification that 
can generate surplus production which can be sold after food security concerns have been met.
This compares to only 4 percent responding that increased income was the most important 
reason for purchasing the pump, down from 17 percent in the Baseline. The expectations thus 
become intertwined and related as more knowledge is gained on the use of the equipment. Both 
security issues and efficiency issues tend to promote the interest in irrigating more land, 
especially when poor rainfall is putting the home at risk for shortages. Some 18% indicated 
reasons such as planting vegetables or caring for livestock. Home gardens and water near the 
homestead for livestock were important reasons for using the irrigation equipment when water 
was available. But there were many respondents who used their pumps for multiple purposes
such as watering their chickens and making bricks for their homes along with producing the 
mortar for sealing their homes against weather. For those using the pumps there were very 
important additional benefits which fell into this grouping. The multiple purposes category was 
the second “Most Important” reason for purchasing a pump. This can be interpreted to mean 
that if the first most important reason for a respondent to purchase a pump was to irrigate more 
land than his/her second most important reason for the purchase may fall in the multiple￾purpose category while the third may be income related. The multiple-purpose category should 
be an important category to consider when developing sales messages. 
Table 21: Respondents’ most important reasons for purchasing a Money Maker pump 
1st and 2nd Most Reasons for Purchase
Important Reason 1st Most Important
Reason 
(Respondents)
(%)
2nd Most Important 
Reason 
(Respondents)
(%)
Baseline: Increase income 33 17% 38 20%
EOP: Increase income 4 4% 12 11%
Baseline: Enable to irrigate larger 
parcels
35 18% 29 15%
EOP: Enable to irrigate larger 
parcels
54 49% 12 11%
Baseline: Ease to use 32 16% 24 12%
EOP: Ease to use 14 13% 3 3%
33 | P a g e
(Table 21, Continued)
Baseline: Inexpensive 14 7% 21 11%
EOP: Inexpensive 3 3% 2 2%
Baseline: Multiple purposes
(farming, brick making, etc.)
13 7% 18 9%
EOP: Multiple purposes 20 18% 15 14%
Baseline: No operational costs 17 9% 17 9%
EOP: No operational costs 6 5% 13 12%
Baseline: Reduction of 
workforce/workload 
16 8% 14 7%
EOP: Reduction of 
workforce/workload
10 9% 7 6%
Baseline: More efficient water 
use 
15 8% 3 2%
EOP: More efficient water use 0 0% 0 0%
Baseline: Nonresponses 13 7% 25 13%
EOP: Nonresponses 0 0% 44 40%
Baseline: Other responses 6 3% 4 2%
EOP: Other responses 0 0% 3 3%
Baseline: Total 194 100% 194 100%
EOP: Total 111 100% 111 100%
Compared to the Baseline responses, 27% less buyers provided a second reason for purchasing 
a pump. As in the Baseline, economics again predominated at EOP, accounting for 60% of 
primary reasons for purchase. 
Special Survey: Shortfall in Utilization of Layaway Finance Option
The answers reported in Table 22 address the specific reasons why respondents, who had 
committed to purchasing a pump using the TKT treatment, subsequently did not buy, bought 
but could not complete payments, or were not able to buy either the treadle or hip pump. The 
respondents included in the End of Project TKT-Layaway Survey (EOPTKTS), had hoped to 
purchase and had verbally committed to purchase a KSI pump. This was the major focus of the 
study; KSI and USAID-DIV were particularly interested in the viability of the layaway savings 
financing scheme. Unfortunately, there were not nearly enough sales (8) to expect useful results 
from comparisons. At EOP the research team had no insight into the reasons why this had
occurred and therefore designed and implemented the unplanned follow-up survey to learn, 
directly from the participants, why they had not completed the purchase process. 
Forty-three of the original 61 who had not completed the layaway process were successfully
contacted. The other 29% were unable to be contacted since this survey had to be done by cell 
phone or they preferred not to interview on this issue. Women made up 58.1 percent and men 
41.9 percent of those who responded to the follow-up survey. Twenty-seven percent stated they 
did not understand the program. This was very likely due to the changing of KSI field 
representatives during the project period, inability to easily explain how and why MPesa would 
be used to carry out payments and, in at least three cases, because the potential buyer did not 
receive a follow-up visit by a KSI representative 
34 | P a g e
Table 22: Responses to layaway (EOPTKTS) follow-up survey
EOPTKTS Responses Yes 
(%)
No
(%)
N
Received a “free” pump from another group 0 100 43
If opting out, was first TKT payment made 48.8 51.2 43
Treadle or Hip pump received after 2014 2.3 97.7 43
Respondent was asked to agree or disagree with following statements: 4.7 95.3 43
The TKT program was confusing 4.7 95.3 43
Did not understand the MPesa system when we signed up 14.0 86.0 43
Time limit for laying away the purchase amount was too short 23.3 76.7 43
Did not make enough money to pay off the pump in the period allowed 46.5 53.5 43
Heard about another finance option, R2O, and preferred that method 4.7 95.3 43
Felt misinformed by the KSI representative about the TKT program 0 100 43
The Money Maker people came to us in August 2015 and wanted to change 
the terms of our agreement; I decided to drop out of the program 
0 100 43
The respondents were asked if there were other reasons for having stopped participating in the 
layaway program. Their responses are listed in Table 23. One issue that should be noted is that 
three individual reported that there had been no follow-up by KSI marketing personnel to 
arranged for participation in the program. This represents a minimum of 4.9% of the total 
potential TKT buyer sample that was not contacted again by KSI until the RT arrived to 
interview them at EOP. 
Table 23: Other reasons mentioned that impacted ability to continue the layaway program
1rst Other Reason Reported 2
nd Other Reason Reported
1= Financial constraints 1=Financial constraints
2= Other priorities such as fees and food 2=Other priorities such as fees and food
3= Sickness 3=Sickness
4= Has other type of pump such as gas pump 4=Has other type of pump such as gas pump
5= Had other loans such as from KWFT 5=Had other loans such as from KWFT
6= Preferred R2O 6=Preferred R2O
7= Wanted to deepen the water source first 7=Wanted to deepen the water source first
8= Group disintegrated; it's hard to mobilize 8=Group disintegrated; it's hard to mobilize
9= Husband refused 9=Husband refused
10= Has group pump 10= Has group pump
11= No follow up by KSI people 11= No follow up by KSI people
12= Lost the number to pay cash through 12= Lost the number to pay cash through
13= Group member was conned hence feared to pay 13= Group member was conned hence feared to pay
14= I prefer cash to be picked at home since 
transactions costs are high
14= I prefer cash to be picked at home since 
transactions costs are high
15= Water source is so deep the pump cannot raise 
the water
15= Water source is so deep the pump cannot raise the 
water
35 | P a g e
Implications of results
As of May 2016 at the EOP, only eight households had completed layaway payments and 
received their pump through the layaway scheme. Over 46% of respondents in the Layaway 
survey stated they “Did not make enough money to pay off the pump in the period allowed”. 
Forty-eight percent of the respondents reported they had made their initial layaway payment. 
Perhaps the other 52% were reluctant to layaway precious cash without an immediate return. 
However, there are many reasons provided in Tables 22 and 23 that suggest there may have 
been other reasons that were not wealth related or MPesa related. It appears that the potential 
layaway participants were willing to test the finance treatment but did not have the time 
sufficient to purchase or did not have the cash necessary to continue in the program. This was in 
the second year of poor rainfall and poor production of their staple food.
A study in neighboring Tanzania proved a significant positive relationship between farming 
technology adoption and a household’s size, access to credit, a household’s productive assets 
and extension contact (Mohamed and Temu, 2008). Higher wealth favors adoption of new 
technology and that may have allowed Cash and R2O finance treatments to move forward more 
efficiently than use of the TKT treatment. However, there is little evidence found in the responses 
of the “expected” TKT buyers to suggest that the 48% who had made initial payments would not 
eventually completed the payment process to acquire their pump. The transactions costs for KSI 
to allow this prolonged purchase process would have been larger than planned. This appeared to 
be derailed by the inability of the respondents to make more money for payment and to do so 
within the time frame allowed for repayment. When combined these two categories account for 
70% of those who dropped from the program which represents 30 of the 67 original respondents. 
Volunteered Comments from the End of Project Survey
Insights are often provided by permitting farmers to speak for themselves. Enumerators 
recorded the following unsolicited comments by respondents during the EOP surveys.
Non-Buyers’ comments
 She was given the pump by World Vision. She uses the pump, but the seedlings were 
given by World Vision so she incurred no cost of the irrigated crops
 Needs pump using rent to own
 Can we get a reservoir liner?
Buyers’ responses
 Find a way to increase pressure to be able to till a larger area
 Farmer needs spare parts for an oil press
 Do you offer tanks
 Lost a pump
 Hasn't used the pump
 Farmer uses the pump to water his chickens
36 | P a g e
 Water dries up after three days, and
“I thank the money maker people for the initiative of making a manual 
pump available to the common man and for making it available”
Summary, Conclusions and Recommendation
The responses summarized in this report revealed that farmers in three counties of eastern 
Kenya (Machakos, Mkueni, and Kitui) were interested in KSI’s small scale irrigation equipment. 
The program introduced, in the counties, presented individual participants the opportunity to 
either purchase equipment via cash or via Cash and one of the treatments, Rent to Own or 
Layaway. Treatments were separated topographically and spatially to reduce any spillover of 
information from one treatment to another. The design layout was successful and focused on 
individuals who had been identified from 256 groups, located across the three counties. Each
respondent spent several hours with the enumerators in the interview sessions across two years. 
Two of the three financing schemes (Cash and Rent to Own) provided immediate economic and 
food security reasons for being demanded by respondents since the equipment was provided at 
time of transaction. Layaway had been identified as a mechanism for introducing pumps to 
women and the base of the pyramid producers over time with the equipment available after full 
payment had been made. Layaway respondents may have found it more difficult to acquire the 
pumps otherwise. Layaway was tied directly to the use of MPesa to make payments. The vast 
majority of the participants had knowledge of and access to the mobile payment tool. 
Unfortunately, all three of the treatments attracted lower levels of sales than required to meet 
the RT’s identified minimum sample size numbers. Layaway, in particular was drastically 
under-represented, with fewer than ten individuals receiving pumps via this method at EOP, 
out of 68 who had committed to purchase, via layaway, at base. 
Statistical interpretation of results was hindered due to the small number of layaway sales. Cash 
appeared to be the preferred method for purchase of MoneyMaker Max and MoneyMaker Hip 
irrigation pumps, with rent to own a close second.
The pumps were viewed as a means of increasing farm income and intensifying or expanding 
irrigated cropland. The farmers also reported that the pumps offered a more convenient and less 
demanding form of irrigation.
A diverse set of media sources and organizations were used by farm households to access 
information and this marked an important opportunity for KSI’s marketing in the future. Cell 
phones had become prevalent in the most remote locations. A transition toward smart phones 
will transform access to information and improved access to mobile banking. New information 
communication technologies should be utilized to link vendors with participants who have 
purchased via R2O or TKT. Attitudes toward accessing information this way are changing 
rapidly empowering participants.
Women are becoming empowered to make many of the important household and production 
37 | P a g e
decisions. Strong pluralities of decisions were made jointly or “Both Equally” in 12 of the 14 
decisions reported. However, women were reported to be making over 30% of the decisions in 
response to 7 of the 14 questions and out of the total number of questions asked, men were 
identified as the main decision maker only once. In the “Leadership” category, 40% or more of 
the decisions are made by women. These findings should inform KSI when attempting to target 
women producers with the opportunity to purchase treadle or hip pumps. 
Transactions costs for KSI need to be minimized but over 40% of potential layaway buyers 
defaulted after their first payments and withdrew from actively trying to secure their pump.
Lack of money and too short a repayment period were identified as the two main constraints for 
successful participation in the layaway program. KSI should try to determine how best to keep 
the base of the pyramid and women clients engaged and active in their layaway programs 
during periods of climatic stress.
In the End of Project (EOP) results farmers reported a preference for increasing area of irrigated 
cropland to increasing profit. Expanding irrigated cropland area may reflect a greater concern for
food security than for profit from marketed crops. The lesser importance of profits might relate to 
the lower popularity of the layaway financing method because it imposes a delay in boosting 
food production. 
Statistical t- tests also evaluated changes in selected household wellbeing measures for non￾irrigators versus irrigators who used each of the three finance methods at EOP. With rare 
exceptions, irrigators using all three finance options versus non-irrigators showed significant 
increases in acres of total cropland, net revenue per acre and per farm from irrigated food crops, 
household wealth and reductions in household hunger. Wealth for cash purchasers at KSh 
675,378 easily achieved highest superiority over non-irrigators and is significant at the 0.01 
level. Rent-to-own and layaway irrigators lacked statistically significant superiority in wealth 
over non-irrigators. Off-farm earnings is statistically superior at the 0.01 level for cash compared 
to non-irrigators, but rent-to-own and layaway lacked statistical significance. A plausible 
explanation may be that higher off-farm earnings, at a mean of KSh 204,267, permitted cash 
households to pay cash for their irrigation pump. On the other hand, poorer rent-to-own and 
layaway pump purchasers showed statistically weaker off-farm earnings compared to non￾irrigators. This would support the findings that layaway participants who dropped out either 
had to little money or could not make all payments in the time period allotted.
Taken collectively, these results showed that cash purchasers had the highest household 
wellbeing advantages over non-irrigators. Rent-to-own purchasers were in second place and
layaway a distant third. These results are disappointing relative to the project aspirations that 
the innovative rent-to-own and layaway finance options would sharply improve financial and 
physical wellbeing for poor households.
To better control for confounding variables aside from irrigation that could have boosted
household wellbeing between Base and EOP, two multiple regression approaches were 
attempted: Differences in differences (DID) analysis and regression with panel data. Both 
methods showed that only Household Savings Exceeding KSH 15,000 and Higher Education 
Level consistently improved several household wellbeing measures. Both these variables tend 
38 | P a g e
to be positively correlated with wealth. The results corroborate the t-test results that irrigation 
tended to improve wellbeing primarily in wealthier households.
The results in this report offered insight into selected reasons for the major shortfall in take up of 
pumps by layaway. More investigation is needed on this issue beyond the scope of this project 
which terminates in mid-2016. KSI’s marketing team is strongly advised to explore all possible
remedies when addressing reasons given, by respondents, for dropping from the layaway 
treatment; a response which led to a partial collapse of the layaway financing option in eastern 
Kenya during 2015-2016. 
KickStart International’s equipment is demanded and utilized by farmers in the three counties
where the project was implemented. Demand for the technology exists far beyond this limited 
area of eastern Kenya, as was described in the report. Cash and R20 options are comparatively 
easier to introduce than the layaway option. However, continuing to utilize all three options and 
applying ICTs to disseminate and receive information will help drive down transaction costs for 
both purchase and maintenance functions. These evolving technology drivers will continue to 
improve the value added from the transfer of KSI treadle and hip pump technology. With that 
next step taken, many more can aspire to exit from the poverty trap in which so many 
individuals and families currently exist.
39 | P a g e
References
Angrist, J. D.; Pischke, J. S. (2008). Mostly harmless econometrics: An empiricist's companion. 
Princeton University Press.
Amudavi, D.M. et al. 2009. “Assessment of technical efficiency of farmer teachers in the uptake 
and dissemination of push–pull technology in Western Kenya.” Crop Protection, 28(11): 
987-996.
Bertrand, M.; Duflo, E.; Mullainathan, S. (2004). "How Much Should We Trust Differences-in￾Differences Estimates?" Quarterly Journal of Economics 119 (1): 249–275..
Card, David; Krueger, Alan B. (1994). "Minimum Wages and Employment: A Case Study of the 
Fast-Food Industry in New Jersey and Pennsylvania". American Economic Review 84 (4): 772–
793.
Gitau, Teresiah, Boris Houenou, Douglas Young and Thomas Byers. July 23, 2015.
Can the results of the KSI-WSU research in eastern Kenya during 2014-2015 be 
generalized to other regions in Kenya and to other African countries? KSI/WSU Project 
Research Paper 2015-1.
Goshu, Degye, Belay Kassa and Mengistu Ketema. 2012. “Is food security enhanced by 
agricultural technologies in rural Ethiopia?” African Journal of Agricultural and Resource 
Economics, 8(1): 58 – 68.
Kabunga, Nassul Ssentamu, Thomas Dubois and Matin Qaim, 2011.”Heterogeneous 
information exposure and technology adoption: the case of tissue culture bananas in 
Kenya”. Agricultural Economics 43(2012):473-485.
Kenya Food Security Steering Group. The 2013 Long Rain Seasons Assessment Report. Web. 
Retrieved on 10 July 2015. 
<http://www.humanitarianresponse.info/en/system/files/documents/files/LRA%20
2013_National%20Report_Final.pdf>
KickStart International. Empowering women with technology. February 2015. 
http://www.kickstart.org/about-us/Women's%20Empowerment%20Report%20-
%20KickStart%20International%20-%20February%202015.pdf
Kirui, Oliver., Okello, Julius., Nyikal, Rose & Njriani, Georgina. Impact of mobile phone-based 
money transfer services in agriculture: evidence from Kenya. Quarterly Journal of 
International Agriculture, 52, 2013, (2): 141-162.
Liverpool, Lenis Saweda O., and Alex Winter-Nelson. 2010. “Poverty status and the impact of 
formal credit on technology use and wellbeing among Ethiopian smallholders.” World 
Development, 38(4): 541-554.
40 | P a g e
Mohamed, Khalid S. and Andrewe Temu. 2008. “Access to credit and its effect on the adoption 
of agricultural technologies: the case of Zanzibar”. African Review of Money and Banking￾2008:45-89.
Moser, Christine M., Christopher B. Barrett .2006. “The complex dynamics of smallholder 
technology adoption: the case of SRI in Madagascar”. Agricultural Economics 35(2006): 
373-388.
OpenDataKenya. Retrieved from https://kenya.socrata.com/Poverty-Rate-byDistrict/i5bp￾z9aq/
Retrieved 10 July 2015.
Omonona, B.T., J.O. Lawal; A.O. Oyinlana.2010. “Determinants of credit constraint conditions 
and production efficiency among farming households in southwestern Nigeria.” The 
Social Sciences, (20100401) (v5 n4): 326-331.
Owuor George.2009. “Is micro-finance achieving its goal among smallholder farmers in Africa? 
empirical evidence from Kenya using propensity score matching.” Paper submitted for 
visual presentation at the XXV11 International Conference of Agricultural Economists (August 
2009).
Savid, Osman., Abdelghani, Echchabi. & Abdi, Aziz. Investigating mobile money acceptance in 
Somalia: an empirical study. Pak. J. Commer. Soc. Sci., 6, 2012, (2): 269-281.
WSU and KSI Staff. (March 2015a). Final condensed version of Buyers Baseline Questionnaire. 
WSU and KSI Staff. (March 2015b). Final condensed version of Non-Buyers Baseline 
Questionnaire.
KSI, USAID and WSU. KickStart – Washington State University (WSU) Research Study
Milestone 2: Implementation Plan: Mobile Layaway and Rent to Own: Bringing an 
Innovative Savings Solution to the Rural Poor for Small-Scale Irrigation Pump 
Purchases, KickStart – Washington State University (WSU) Research Study
Wu, Di and Doug Young (July 30, 2014). “Estimation of minimum sample sizes for Kenya
baseline and end-of-project surveys based on data from the May-June 2014 Rapid
Appraisal- Demographic Survey” Unpublished research paper.
Young, Douglas and Thomas Byers. Research proposal by WSU to KSI, 12/08/2013 and 
application by WSU to NCOSTI to do research in Kenya, 3/25/2014.
Young, Douglas and Di Wu (July 30, 2014). “Estimation of minimum sample sizes for Kenya 
baseline and end-of-project surveys based on data from the May-June 2014 Rapid 
Appraisal- Demographic Survey” Unpublished research paper-Dropbox
41 | P a g e
Young, Douglas, Boris Houenou, Wu Di, T.E. Byers, Patrick Kirui, Andrew Wanyonyi,
Teresiah Gitau, and Jennifer Mueni (July 2014). “Summary of results of rapid
appraisal - demographic survey for KSI/WSU eastern Kenya project”. RA-DS Results 
Report.
Young, Douglas, B. Houenou, T. Byers, T. Gitau, P. Kirui, J. Mueni and A. Wanyonyi. 
August 2015. Milestone Two: Baseline Survey of Pump Buyers by Three Finance 
Options and of Non-Buyers: Process and Results Highlights and Appendices.
42 | P a g e
Appendix 1: Sub-Counties (previously Districts) in which the study 
was undertaken
1=Athi River 21=Machakos
2=Ikombe 22=Makueni
3=Kaiti 23=Masinga
4=Kalama 24=Matungulu
5=Kalawa 25=Mbooni East
6=Kangundo 26=Mbooni West
7=Kanyangi 27=Migwani
8=Katangi 28=Mtito Andei
9=Kathiani 29=Mukaa
10=Kathonzweni 30=Mutitu
11=Kibwezi 31=Mutomo
12=Kilome 32=Mwala
13=Kithimani 33=Mwingi Central
14=Kitui 34=Mwingi West
15=Kitui Central 35=Ndalani
16=Kitui East 36=Nguluni
17=Kitumbi 37=Nzaui
18=kwa Vonza 38=Wambiti
19=Kyangwithia East 39=Yatta
20=Lower Yatta
43 | P a g e
Appendix 2: Groups contacted to participate in the longitudinal study 
Group Name
1=Athi River 29=Kyekonge Women group 57=Nzilani
2=Ikombe 30=Kyeni Kya Kyaani 58=Nzukini
3=Kaiti 31=Kyeni Kya Mukameni 59=Pawa Group
4=Farm Field Schools 32=Kyeni Kya Nzueni
60=Persons with Disability 
Group
5=FSA 33=Mango good member 61=Rise and Shine
6=Hand in hand 34=Maroda farmers 62=Sakai Seed Bulkers
7=Ika Wiyike 35=Matendo SHG 63=St.Francis Mwanyani
8=Iuni Pepper Growers 36=Mbee Nthei 64=Tei wa Kyamboa
9=Jikaze Women Group 37=Mbikila Witei 65=Twikumbe SHG
10=Kaka
38=Mbingoni water 
harvesting SHG 66=Umiisyo wa Kitaita
11=Kakuyuni Farmers 39=Mikuyuni 67=Umoja SHG
12=Kamen
40=Mitaboni Kunena na 
kuwika 68=Waekandeka
13=Kanguma Green Growers 41=Mitalani 69=Wendano wa Kyambingu
14=Katutuini Farmers
42=Mithanga horticultural 
Group 70=Wendano wa Maiuni
15=Kavutini Women Group
43=Mituki Iveti Women 
Group 71=Wendano wa Manyatta
16=Kiembeni Utiini Sita 44=Mkombozi 72=Wendano Waaka Kiusini
17=Kikesa Horticulture 45=Movers and shakers 73=Wendo wa Mbuini
18=Kipawa SHG 46=Mwanzo Mpya 74=Wendo wa Musyai
19=Kisiuni 47=Mwooni SHG 75=Wendo wa Omwe
20=Kithia SHG 48=NCCK 76=Wikwatyo
21=Kithiasani 49=Nduluile Wendo 77=Wingu Makanda SHG
22=Kithoni Farmers 50=Ngetha Tu Tethie 78=Wiyisi
23=Kola Horticulture 51=Ngomeni kwa Mwania 79=Woni Wa Mbee
24=Kuleta SHG
52=Ngumbuki 
Shg/Mangumu shg 80=Young Mbilini
25=Kwa Munini 53=Nguno 81=Yumbuni Disabled
26=Kyambele 54=Nguumo Women Group 82=Umiisyo waaka iiani
27=Kyatuni Traders Business 
Ladies 55=Nkolo
28=Kyawema 56=Nzeve Nzau SHG
44 | P a g e
Appendix 3: On-Line Appendices (Located in Dropbox Online –
request share from Byers or Young)
Byers, Thomas. 2015. Clustering of Potential Buyers Across Topographical and Spatial Areas, 
File Name: FINAL Clustering Results.xlsx
Byers, Thomas. (June 2016). Buyers Baseline Survey Data Set, File Name: FINAL BBaseS Data 
Set.xlsx
Byers, Thomas. (June 2016). Non-Buyers Baseline Survey Data Set, File Name: FINAL NBBase 
Data Set.xlsx
Byers, Thomas. (June 2016). Buyers End of Project Survey Data Set, File Name: FINAL BEOPS 
DATA SET.xlsx
Byers, Thomas. (June 2016). Non-Buyers End of Project Survey Data Set, File Name: FINAL 
NBEOPS Data Set.xlsx
Byers, Thomas. (June 2016). Layaway EOPTKT Survey Data Set, File Name: FINAL EOPSTKT 
Data Set.xlsx
Byers, Thomas. (June 2016). Base – EOP Data Set, File Name: FINAL BaseS and EOPS Data 
Set.xlsx
Byers, Thomas. (June 2016). Buyers and Non-Buyers Output, File Name: FINAL BBaseNBBase 
Output Data.xlsx
Byers, Thomas. (June 2016). Buyers End of Project and Non-Buyers End of Project Output, 
FINAL BEOPSNBEOPS Output Data.xlsx
Gitau, Teresiah, Boris Houenou, Douglas Young and Thomas Byers. July 23, 2015.
Can the results of the KSI-WSU research in eastern Kenya during 2014-2015 be generalized to 
other regions in Kenya and to other African countries? KSI/WSU project research paper￾Dropbox. 
WSU Staff (July 2015a). Table 1. Descriptive statistics by three finance options of buyers, and for 
all buyers and all non-buyers, for 71 selected variables, baseline surveys of three eastern Kenya 
counties, March-June 2015. [Table 3 of Milestone Two Appendices 08-10-15] 
WSU Staff (July 2015b). Table 2. Statistical tests of equality of sample means of all buyers versus 
all non-buyers for 42 selected continuous variables, baseline surveys of three eastern Kenya 
counties, March-June 2015. [Table 1 of Milestone Two Appendices 08-10-15]
WSU Staff (July 2015c). Table 3. Selected sample proportions by finance option and statistical 
tests of equality of proportions for all buyers versus all non-buyers, baseline surveys in three 
eastern Kenya counties, March-June 2015. [Table 2 of Milestone Two Appendices 08-10-15]
45 | P a g e
WSU and KSI Staff. (March 2015). Final version of Buyers Baseline Survey Questionnaire, File 
Name: FINAL BBaseS Q'aire.docx
WSU and KSI Staff. (March 2015). Final version of Non-Buyers Baseline Survey Questionnaire, 
File Name:FINAL NBBaseS Q'aire.docx
WSU and KSI Staff. (March 2016). Final version of Buyers End of Project Survey Questionnaire, 
File name: FINAL BEOPs Q’aire.doc
WSU and KSI Staff. (March 2016). Final version of Non-Buyers End of Project Survey 
Questionnaire, File Name: FINAL NBEOPS Q'aire.docx
WSU and KSI Staff. (March 2016). Final version of End of Project TKT Survey Questionnaire, File 
Name: FINAL TKT - Layaway Q'aire.docx
Young, D. Hypothesized coefficient signs (directions of influence) from theory for the 
explanatory variables in DID dependent variable equations, WSU/KSI research project on 
adoption of human-powered irrigation pumps in eastern Kenya, 2014-2016. WSU staff paper. 
May 2016. File Name: Hypothesized Coeff Signs.docx
Young, Douglas and Boris Houenou. June, 2016. Descriptive statistics for EOP survey 
explanatory variables in DID regression. File Name: Descr Stats Regression Models' 
Explanatory Vars.docx
Young, Douglas and Boris Houenou. June, 2016. Complete DID Regression Results
File Name: Results for Panel Data Regression.docx
Young, Douglas and Boris Houenou. June, 2016. Complete DID Regression Results
File Name: DID Regression.docx
Young, Douglas. June 2016. End of Project Retained DID Variables Data Set. File Name: FINAL 
EOP Ret DID Vars.xlsx