0
 
Baseline Survey Report of the Resilience 
Food Security Activity Graduating to 
Resilience in Uganda, Cohort 2
August 2022 | Volume I
IMPEL | Implementer-Led Evaluation & Learning Associate Award
ABOUT IMPEL
The Implementer-Led Evaluation & Learning Associate Award (IMPEL) works to improve the design and implementation of 
Bureau for Humanitarian Assistance (BHA)-funded resilience food security activities (RFSAs) through implementer-led 
evaluations and knowledge sharing. Funded by the United States Agency for International Development (USAID) BHA, IMPEL 
will gather information and knowledge in order to measure performance of RFSAs, strengthen accountability, and improve 
guidance and policy. This information will help the food security community of practice and USAID to design projects and 
modify existing projects in ways that bolster performance, efficiency, and effectiveness. IMPEL is a seven-year activity (2019–
2026) implemented by Save the Children (lead), TANGO International, Tulane University, Causal Design, Innovations for Poverty 
Action, and International Food Policy Research Institute. 
RECOMMENDED CITATION
IMPEL. (2022). Baseline Survey Report of the Resilience Food Security Activity Graduating to Resilience in Uganda, Cohort 2 (Vol. 
I). Washington, DC: The Implementer-Led Evaluation & Learning Associate Award.
PHOTO CREDITS
AVSI Foundation
DISCLAIMER
This report is made possible by the generous support of the American people through the United States Agency for 
International Development (USAID). The contents are the responsibility of the Implementer-Led Evaluation & Learning (IMPEL) 
award and do not necessarily reflect the views of USAID or the United States Government.
CONTACT INFORMATION
IMPEL Activity
c/o Save the Children
899 North Capitol Street NE, Suite #900
Washington, DC 20002
www.fsnnetwork.org/IMPEL
IMPEL@savechildren.org
PREPARED BY:
Baseline Survey Report of the RFSA Graduating to Resilience in Uganda, Cohort 2 (Vol. I)
Acknowledgments i
ACKNOWLEDGMENTS
The authors of this report would like to thank Edmund Emulu and Morris Bwambale for their 
outstanding work on this project as Field Managers for Innovations for Poverty Action (IPA).
We would also like to thank all field staff, enumerators, team leaders, mobilizers, and auditors for all the 
surveys they were able to collect, many times in complicated settings, rainy seasons, bumpy roads, and 
COVID-19. 
We would also like to thank AVSI Foundation for their partnership and their patience and collaboration 
in adapting their program to the context of an RCT. 
We would like to thank the Bureau for Humanitarian Affairs (BHA) for their funding, support and 
guidance in the design and execution of the study.
We would like to thank the Office of the Price Minister (OPM) and Government of Uganda (GoU), for 
which the engagement and commitment of local leaders led to ease in operating in the community.
Finally, we would like to thank all the respondents for their patience and for the hours they spent
answering our questions.
This report was prepared by Antoine Guilhin, Jan Will, Gabriel Olila, and Rachel Strohm from IPA and 
Lasse Brune from Northwestern University.
IMPEL | Implementer-Led Evaluation and Learning
ii Table of Contents
TABLE OF CONTENTS
Acknowledgments...............................................................................................i
List of Tables......................................................................................................iv
List of Figures......................................................................................................v
Acronyms...........................................................................................................vi
Executive Summary ..........................................................................................vii
1. Context and Intervention...............................................................................1
1.1 Background on Graduating to Resilience Activity, Phase 1............................................................1
1.2 Phase 2 Evaluation of Graduating to Resilience.............................................................................2
2. Evaluation of Cohort 2 ...................................................................................5
2.1 Overview ........................................................................................................................................5
2.2 Sampling.........................................................................................................................................7
2.3 Cohort 2 Eligibility Assessment......................................................................................................8
2.4 Trial A: Impact of the refined Graduating to Resilience Activity....................................................9
2.5 Trial B: The Impact of Adding Interpersonal Psychotherapy in Groups.........................................9
3. Baseline Household Survey ..........................................................................11
3.1 Objectives.....................................................................................................................................11
3.2 Questionnaire...............................................................................................................................11
3.3 Field Preparation..........................................................................................................................13
3.4 Field Activities ..............................................................................................................................13
3.4.1 Soft Launch..................................................................................................................13
3.4.2 Data Collection............................................................................................................14
3.4.3 Replacement Strategy.................................................................................................14
3.4.4 Data Quality ................................................................................................................14
3.5 Limitations and Challenges ..........................................................................................................15
3.5.1 Limitations Related to Self-Reported Data .................................................................15
3.5.2 Challenges Related to Field Conditions and COVID-19...............................................15
4. Results.........................................................................................................16
4.1 Characteristics of the Study Sample.............................................................................................16
4.2 Household Demographics............................................................................................................16
4.3 Poverty Indicators........................................................................................................................17
4.3.1 Consumption Per Capita .............................................................................................17
4.3.2 Prevalence of Poverty .................................................................................................18
4.3.3 Depth of Poverty.........................................................................................................18
4.4 Water, Sanitation, and Hygiene ...................................................................................................18
Baseline Survey Report of the RFSA Graduating to Resilience in Uganda, Cohort 2 (Vol. I)
Table of Contents iii
4.4.1 Use of Basic Drinking Water Services..........................................................................18
4.4.2 Percentage of Households in Target Areas Practicing Correct Use of Recommended 
Household Water Treatment Technologies.................................................................................19
4.5 Housing and Assets ......................................................................................................................20
4.6 Land and Agriculture ....................................................................................................................21
4.7 Livestock Holdings........................................................................................................................24
4.8 Employment.................................................................................................................................25
4.9 Off-Farm Business........................................................................................................................26
4.10 Savings and Loans.........................................................................................................................26
4.11 Nutrition and Food Security .........................................................................................................28
4.11.1 Prevalence of Women of Reproductive Age Consuming a Diet of Minimum Diversity..
.....................................................................................................................................30
4.11.2 Prevalence of Children 6–23 Months Consuming a Diet of Minimum Diversity ........30
4.11.3 Prevalence of Exclusive Breastfeeding of Children Under 6 Months .........................30
5. Comparison of Experimental Groups............................................................32
6. Next Steps....................................................................................................34
6.1 Process Evaluation .......................................................................................................................34
6.2 Outcome Monitoring....................................................................................................................34
6.3 Endline Survey for Cohort 2 Evaluation .......................................................................................34
Appendix ..........................................................................................................35
Additional Figures on Cohort 2 Design.................................................................................................35
Additional Figures and Tables ..............................................................................................................37
VOLUME II
Annex A: Field Manual
Annex B: Baseline Survey
Annex C: Pre-Analysis Plan
Annex D: Study Protocol
IMPEL | Implementer-Led Evaluation and Learning
iv List of Tables and Figures
LIST OF TABLES 
Table 1. Sampling of Cohort 2 households as a function of Cohort 1 study status......................................8
Table 2. Number of households in Trial A.....................................................................................................9
Table 3. Number of households in Trial B...................................................................................................10
Table 4. Individual demographic characteristics ........................................................................................16
Table 5. Household demographics..............................................................................................................16
Table 6. Poverty levels................................................................................................................................18
Table 7. Household WASH characteristics..................................................................................................19
Table 8. Household water use and source..................................................................................................20
Table 9. Roof materials...............................................................................................................................20
Table 10. Key assets owned........................................................................................................................21
Table 11. Agricultural land ownership in the host community during the September to December 2021 
season .........................................................................................................................................................21
Table 12. Agricultural land rights of usage in the host community during the September to December 
2021 season ................................................................................................................................................22
Table 13. Agriculture land and inputs during the September to December 2021 season .........................22
Table 14. Main food crops during the September to December 2021 season...........................................22
Table 15. Main cash crops during the September to December 2021 season...........................................23
Table 16. Main food crops during the March to May 2021 season............................................................23
Table 17. Main cash crops during the March to May 2021 season............................................................24
Table 18. Livestock ownership....................................................................................................................24
Table 19. Number of chickens owned ........................................................................................................25
Table 20. Employment................................................................................................................................25
Table 21. Main business types....................................................................................................................26
Table 22. Main Business types among youths (18–35 years old)...............................................................26
Table 23. Outstanding loans .......................................................................................................................27
Table 24. Savings over the last 3 months ...................................................................................................27
Table 25. Food insecurity at the household level.......................................................................................29
Table 26. Nutrition status for women of reproductive age and young children........................................30
Table 27. Subcomponents of the exclusive breastfeeding indicator for children under 6 months...........31
Table 28. Trial A: Comparison of refined graduation and control ..............................................................32
Table 29. Trial B: Comparison of refined graduation with and without IPT-G ...........................................33
Table 30. Future Activities ..........................................................................................................................34
Table 31. Appendix: Phase 1, Cohort 1, households found eligible by AVSI in 2018, prior to Cohort 1 
implementation ..........................................................................................................................................35
Table 32. Appendix: Phase 2, household level randomization...................................................................35
Table 33. Appendix: Trial A .........................................................................................................................36
Table 34. Appendix: Trial B .........................................................................................................................36
Table 35. Appendix: Correlations between HFIAS and FCS components...................................................38
Baseline Survey Report of the RFSA Graduating to Resilience in Uganda, Cohort 2 (Vol. I)
List of Tables and Figures v
LIST OF FIGURES
Figure 1. Map of Kamwenge District, 2018...................................................................................................1
Figure 2. Appendix: Binscatter plot of Negative HFIAS against FCS ...........................................................37
IMPEL | Implementer-Led Evaluation and Learning
vi Acronyms 
ACRONYMS
BHA Bureau for Humanitarian Assistance
FCS Food Consumption Score
FFBS Farmer Field Business School
FFP Office of Food for Peace
HFIAS Household Food Insecurity Access Scale
HH Household
HHS Household Hunger Scale
IMPEL Implementer Led Evaluation and Learning
IPT-G Interpersonal Psychotherapy in Groups
LTFU Long-Term Follow-Up
N Number of Households or Individuals in the Survey Sample
OECD Organization for Economic Co-operation and Development
OPM Office of the Prime Minister
PPP Purchasing Power Parity
RCT Randomized Control Trial
RFSA Resilience Food Security Activity
ROSCA Rotating Savings and Credit Association
SACCO Savings and Credit Cooperative Organization
SD Standard Deviation
UNHCR United Nations High Commissioner for Refugees
USAID United States Agency for International Development
VSLA Village Savings and Loan Association
WASH Water, Sanitation, and Hygiene
Baseline Survey Report of the RFSA Graduating to Resilience in Uganda, Cohort 2 (Vol. I)
Executive Summary vii
EXECUTIVE SUMMARY
This survey reports on data collected as part of the baseline survey of Cohort 2 of the Graduating to 
Resilience activity, which is a Resilience Food Security Activity (RFSA) implemented by the AVSI 
Foundation, an international non-governmental organization, and funded by the United States Agency 
for International Development (USAID) Bureau for Humanitarian Assistance (BHA). The Graduating to 
Resilience activity is implemented in and around the Rwamwanja Refugee Settlement in Kamwenge 
District, Uganda, and is offered to extremely poor households living in both the refugee and host 
communities in this area. The goal of the activity is to graduate extremely poor refugee and Ugandan 
households from conditions of food insecurity and fragile livelihoods to self-reliance and resilience.
The report provides an overview of the RFSA, describes the evaluation design for the activity’s second 
cohort, and provides summary statistics and indicator estimates based on a recently completed round of 
baseline data collection. Innovations for Poverty Action (IPA), with a team of 38 enumerators, 
administered a baseline survey to 2,570 households in November and December 2021, in the 
Kamwenge district and in the Rwamwanja settlement. The survey took place before the kickoff of the 
Graduating to Resilience Cohort 2 and its Interpersonal Psychotherapy in Groups (IPT-G) interventions, 
which will start in February 2022.
Key findings from the baseline survey include the following: 
● Demographics & poverty
o A total of 16,560 people live in the 2,570 households interviewed at baseline.
o The share of adults over 18 years was 41% in the host community and 35% in the 
refugee community.
o The share of children under 5 years was 15% in the host community and 21% in the 
refugee community.
o The share of women of reproductive age (15–49 years) was 21% in the host community 
and 20% in the refugee community.
o In the host community, 38% of households lived on less than Purchasing Power Parity 
(PPP) $1.90 per day and per adult, while 31% of those in the refugee community lived 
under this threshold.
● Water, sanitation, and hygiene (WASH) & housing
o Similar numbers of people in the host and refugee communities had access to basic 
drinking water services (12% and 15%, respectively), but host community members 
were much more likely to boil their water than refugee community members (75% and 
49% respectively).
o Members of the host community lived in higher quality housing, with 91% having iron 
roofs compared to 21% in the refugee community.
● Land & agriculture
o In the host community, 86% of households own land. In the settlement, the Office of the 
Prime Minister (OPM) allocates plots to refugee households when they join the camp 
and grant them temporary usage right on these plots. Most refugee households (93%) 
reported having land for which they had usage rights.
IMPEL | Implementer-Led Evaluation and Learning
viii Executive Summary
o Almost all households in both communities cultivated at least one crop (98% host and 
99% refugee), and the cultivation of at least one cash crop was common (62% of all 
hosts and 54% of all refugees).
● Employment & off-farm business
o Most people engaged in employment activities, mostly daily agricultural wage 
employment. The rate of employment was similar for both communities (71% for host 
households and 74% of refugee households).
o Refugees are somewhat more likely to have their own off-farm businesses, with 43% of 
refugees and 30% of host community members operating at least one. In the host 
community, the most common types of businesses were petty trade (9% across all 
surveyed host community households) and retail shops (4%). In the settlement, the 
most common types of business were petty trade (27% across all surveyed refugee 
households) and produce trading (5%).
● Access to finance
o Most households reported having outstanding loans—77% in the host community and 
68% in the settlement.
o Almost all households reported having at least some savings (92% in host community 
and 94% in refugee community), which were mostly held informally.
● Food security
o Most households reported being either moderately or severely food insecure using the 
Household Food Insecurity Access Scale (HFIAS). The rate of household deemed as 
“severely food insecure” was 39% in the host community and 29% in the settlement. 
o Households in both communities also reported eating diverse diets, with 95% of host 
community members and 86% of refugee community households reporting an 
“acceptable” Food Consumption Score (FCS).
o Dietary diversity is a particular concern for women of reproductive age and children 
ages 6–23 months, in the same proportions for refugee and host community. For 
women of reproductive age, 30% (host) and 29% (refugee) were meeting minimum 
standards of dietary diversity. For children aged 6–23 months, 14% (host) and 15% 
(refugee) were receiving a minimum acceptable diet, which measures dietary diversity 
and frequency of feeding.
o Exclusive breastfeeding rates for children under 6 months varied between communities, 
with 69% of 61 children in this age group exclusively breastfed in the host community, 
and 83% of 146 such children in the refugee community. 
Baseline Survey Report of the RFSA Graduating to Resilience in Uganda, Cohort 2 (Vol. I)
Context and Intervention 1
1. CONTEXT AND INTERVENTION
1.1 Background on Graduating to Resilience Activity, Phase 1
In October 2017, United States Agency for International Development (USAID)’s Bureau for 
Humanitarian Assistance (BHA) awarded the implementation of the Resilience Food Security Activity 
(RFSA) Graduating to Resilience in Kamwenge District, Uganda to AVSI Foundation (AVSI), an 
international non-governmental organization operating in the country since the early 1980s, together 
with a consortium including TrickleUp and IMPAQ International (now American Institutes for Research–
AIR). The Graduating to Resilience RFSA is part of a portfolio of RFSAs that BHA funds in low-income 
countries across Africa. The goal of Graduating to Resilience is to improve food and nutrition security 
and self-reliance among extremely poor households in refugee settlements and host community.
As shown in Figure 1, Kamwenge District (in green) is in the southwest of Uganda and is home to 
approximately 78,5431 mostly Congolese refugees, as well as a non-refugee population facing chronic 
food insecurity.
Figure 1. Map of Kamwenge District, 20182
Graduating to Resilience is based on the “graduation from extreme poverty” approach, which is a 
holistic set of services for targeted “ultra-poor” households,3 designed to help recipients build new 
 1 UNHCR Uganda comprehensive refugee response portal. Accessed March 9, 2022. https://data2.unhcr.org/en/country/uga 2 Graduating to Resilience is located in the northern part of Kamwenge district in this 2018 map of Uganda’s political districts. 
Kamwenge district boundaries were redrawn in 2019, but the activity is still located within the new boundaries of the district. 3 Ultra-poor households are households who have typically been identified through a three-stage targeting process: 1) Target an 
area considered as poor according to socio-economic indicators; 2) Identify the ultra-poor households through a community 
approach such as participatory wealth ranking to identify the poorest and most vulnerable in the community; 3) Administer a 
poverty assessment to households to verify the results and screen out any households who are too well off.
Kampala
IMPEL | Implementer-Led Evaluation and Learning
2 Context and Intervention
livelihoods while building skills and confidence, along with an asset base to diversify income, protect 
themselves from shocks, and sustain well-being. 
BHA awarded Innovations for Poverty Action (IPA) a Cooperative Agreement to design a randomized 
controlled trial (RCT) on the 6-year Graduating to Resilience RFSA to measure the impact and cost￾effectiveness of different variations of graduation activities for refugees and host community. Informed 
by the results from the RCT in year four, AVSI is now implementing a refined version of the Graduation 
Approach.
4 This second version is based on findings from the Cohort 1 endline, and implemented with 
households that were not previously participating in interventions of the activity. As such, the 
Graduating to Resilience RFSA has two cohorts of participants: Cohort 1 (2018–2021) and Cohort 2 
(2022–2024). Each cohort includes extremely poor households from both the refugee and host 
communities in Kamwenge district.
1.2 Phase 2 Evaluation of Graduating to Resilience
In 2021, the BHA-funded Implementer-Led Evaluation and Learning (IMPEL) mechanism awarded IPA the 
evaluation of Phase 2, which consists of both a round of follow-up surveying of Cohort 1 study 
participants to measure longer-term impacts, as well as an evaluation designed specifically around 
Cohort 2 interventions. The latter aims to provide new evidence on how to amplify the impact of the 
Refined Graduation Approach by incorporating low-cost mental health treatments using Interpersonal 
Psychotherapy in Groups (IPT-G) methodology. This report focuses on a recent round of baseline 
surveying that was conducted as part of the design to evaluate Cohort 2.
The study area of the Cohort 2 Graduating to Resilience activity is equivalent to that of Cohort 1. For the 
refugee settlement this includes the entire refugee community. In the host community, the activity is 
implemented and evaluated in six sub-counties contiguous to the settlement: Nkoma, Biguli, Bihanga, 
Bwizi, Nkoma Katalyeba Town Council, and Lyakahungu Town Council. Cohort 2 offers a package of 
interventions to about 7,200 households, split more or less evenly between the refugee and host 
community. The interventions are informed by the knowledge acquired during the implementation of 
the first cohort and its accompanying multi-arm RCT. To distinguish the package of interventions 
administered in Cohort 2 from those of Cohort 1, the approach of this second cohort will be referenced 
in this document as “Refined Graduation.”
The key interventions of the Refined Graduation Approach are: 
1. Consumption support: a small, regular cash transfer provided over 12 months to stabilize 
incomes and enable households to focus on new livelihoods, as well as prevent consumption of 
productive assets. This cash transfer is about $4–5 per household member and per month. The 
total transfer per household over the course of the cash transfer periods sums to about $300.
2. Productive asset transfer: a lump sum cash transfer for any small-scale income-generating 
activity of about $300 that takes place about 6 months after the beginning of the activity. 
 4 The refined variation of the Graduation Approach is mostly based on the “group coaching and cash transfer” approach since 
that proved to be the most cost-effective approach in the Phase 1 evaluation.
Baseline Survey Report of the RFSA Graduating to Resilience in Uganda, Cohort 2 (Vol. I)
Context and Intervention 3
3. Training on technical skills: including financial literacy, enterprise selection, planning and 
management, improved agricultural skills (crop and livestock), and bank linkages. These 
trainings happen in the first 6 months on the activity, before the asset is given out. 
4. Coaching on various themes: including health, nutrition, gender, life skills, and sanitation. Each 
participant will attend 48 group coaching sessions over 24 months and will have 8 quarterly 
individual touch points over the same period.5
5. Access to Savings: creation of Village Savings and Loan Associations (VSLAs) and meetings to 
create a secure place to save income and access low interest loans.
6. Referrals and Linkages: referral for domestic violence, protection, health, nutrition, and linkages 
to services within the public and private sector like extension services, energy, finance, markets, 
and agricultural inputs.
One difference between the Cohort 1 and Cohort 2 activities is the intensity of coaching. Cohort 1 
participants received two different coaching approaches—individual coaching (every 2 weeks) and 
group coaching (weekly). AVSI is employing a hybrid approach for Cohort 2 activities. This hybrid 
approach is mostly based on group coaching sessions happening every two weeks, but also includes 
quarterly individual touch points to maintain individual contact between participants and coaches.
For Cohort 2 activities, AVSI decided to add a group-based psychotherapy intervention component—
Interpersonal Therapy in Groups (IPT-G). In an AVSI study in Palabek refugee settlement in Northern 
Uganda,6 35% of respondents indicated that they were experiencing depression at the time, and 64% of 
these individuals reported that their depression was interfering with their daily lives. Up to 23% of all 
respondents said that they had had suicidal thoughts within the last two weeks. In addition, during 
Cohort 2 sensitization sessions, AVSI collected data that indicated that up to 80% of refugees had 
experienced depression at some point in their lives, as well as 60% of host community members.
Against this backdrop of substantial mental health concerns, AVSI will implement IPT-G with half of the 
Cohort 2 participants. This therapeutic approach is participatory and group-based, empowering isolated 
and vulnerable women7 to improve relationships, develop communication and conflict resolution skills, 
and foster lasting support networks. Within these groups, participants share their own challenges, 
discuss actions they have taken to manage these challenges, and provide support to one another. There 
are eight sessions across a 2-month period (weekly sessions over 2 months) and each session will last 60 
minutes. 
 5 Cohort 1 participants received two different coaching approaches—individual coaching (every two weeks) and group coaching 
(weekly). For Cohort 2 activities, AVSI Foundation is employing a hybrid approach with group coaching sessions and quarterly 
touch points
6 Mental health rapid assessment of refugee and host community in Palabek refugee settlement conducted by AVSI Foundation 
and Strong Mind in 2018.
7 Graduating to Resilience activity primarily focuses on women, with 89% of the primary participants being female.
IMPEL | Implementer-Led Evaluation and Learning
4 Context and Intervention
IPT-G has three phases, each with distinct objectives:
● Initial Phase: Group Sessions 1–2. This phase focuses on creating initial bonds between group 
members and building rapport with one another, so women feel comfortable sharing personal 
information and discussing the reasons for their depression. 
● Middle Phase: Group Sessions 3–6. This phase ensures that all members are actively engaged 
and helping each other by making suggestions regarding one another’s problems. This is also the 
phase where important progress is made for members to fully understand all the symptoms and 
triggers of depression. 
● Termination Phase: Group Sessions 7–8. These sessions prepare members to end formal 
sessions. Members are reminded to continually identify their own triggers of depression in the 
future, and what they should do to respond. Individual action plans are created and reviewed. 
The timeline for Cohort 2 of the RFSA is as follows:
● Beneficiary registration and forming groups: February 2022
● Intervention kickoff: February 2022
● Consumption support cash transfer: February 2022 to February 2023
● Coaching sessions: February 2022 to February 2024
● IPT-G sessions: April to July 2022
● Farmer Field Business School (FFBS), VSLA, and other intervention elements: February 2022 to 
February 2024
Baseline Survey Report of the RFSA Graduating to Resilience in Uganda, Cohort 2 (Vol. I)
Evaluation of Cohort 2 5
2. EVALUATION OF COHORT 2
2.1 Overview
The evaluation of Cohort 2 of the Graduation to Resilience activity has two objectives. The first is to 
assess the impact of the Refined Graduation Approach in the context of the Phase 2 evaluation. The 
second objective is to provide new evidence on how to amplify the impact of the Graduation Approach 
by incorporating low-cost mental health treatments through the IPT-G intervention.
The test of the addition of an IPT-G component is motivated by early-stage evidence from IPA’s research 
in Ghana8 and Niger9 that suggests that the integration of mental health interventions into graduation 
interventions could improve participants’ economic and health outcomes for minimal additional cost. 
However, other research from Kenya suggests that participants in a cash transfer program who were 
also enrolled in a 5-week psychotherapy program did not see any benefits from the mental health 
treatment. There is a clear need for additional research on this topic. 
While there is evidence that IPT-G works in Uganda outside the context of a graduation activity, the 
present study will help to understand whether addressing mental health will cost-effectively improve 
outcomes of a refined version of the Graduation Approach. This is especially important in a refugee 
context where mental health issues may pose a particular challenge for individuals who have 
experienced conflict and trauma. The evaluation of this intervention will examine not only the effect on 
the mental health, but also test for effects on economic outcomes for the primary participants. We 
hypothesize that better mental health allows participants to engage more effectively with their 
livelihoods. Beyond mental health outcomes, potential additional downstream effects on consumption 
can be compared against the additional costs, to evaluate cost-effectiveness. 
The Cohort 2 evaluation is designed as two separate two-arm trials. In both trials, households were 
randomly assigned to experimental arms at the household level. Households participating in both trials 
were recruited from households who were not invited to join Cohort 1. Trial A seeks to evaluate the 
overall effect of the Cohort 2 interventions, while Trial B seeks to measure the additional effects of IPT-G 
programming.
In Trial A, we compare households randomized into offers of participation in Cohort 2 interventions 
(“Refined Graduation”) to households who were randomized out (“control”). The goal of Trial A is to 
evaluate the overall impact of the Cohort 2 interventions. Cohort 2 interventions evaluated as part of
Trial A include IPT-G programming for some households, but the trial is not designed to test for 
differential impacts of this programming element. Instead, the trial will deliver estimates of the impact 
of Cohort 2 interventions averaged across households receiving and not receiving IPT-G programming 
 8 Nathan Barker, Gharad Bryan, Dean Karlan, Angela Ofori-Atta, and Christopher Udry. 2021. “Mental Health Therapy as a Core 
Strategy for Increasing Human Capital: Evidence from Ghana.” Working paper. 
https://www.povertyactionlab.org/sites/default/files/research-paper/working-paper_2754_Cognitive-Behavioral￾Therapy_Ghana_April2021.pdf
9 Thomas Bossuroy, Markus Goldstein, Dean Karlan, Harounan Kazianga, William Parienté, Patrick Premand, Catherine Thomas, 
Christopher Udry, Julia Vaillant, and Kelsey Wright. 2021. “Pathways out of Extreme Poverty Tackling Psychosocial and Capital
Constraints with a Multi-faceted Social Protection Program in Niger.” World Bank Policy Research Working Paper 9562.
IMPEL | Implementer-Led Evaluation and Learning
6 Evaluation of Cohort 2
(see the description of Trial B below, which delivers estimates that can be used to assess the impact of 
the Cohort 2 graduation activity without IPT-G). Households participating in Trial A were selected from 
households that were part of the control group of Cohort 1 and were previously interviewed in survey 
data collection as part of the Cohort 1 evaluation.
In “Trial B,” we compare households who were randomly assigned to offers of participation in Cohort 2
with IPT-G with those who were assigned to offers of participation in Cohort 2 without IPT-G. The 
purpose of Trial B is to measure the additional effect of IPT-G programming. Households participating in 
Trial B were largely recruited from households that were part of the holdout control group of Cohort 1 
and who were not previously surveyed, but the sample also includes households who were part of the 
Cohort 1 “regular,” non-holdout control group.10
Trial A and Trial B have similar but not identical sampling frames. Trial A recruits from Cohort 1 non￾holdout controls. Trial B recruits from a mix of former Cohort 1 non-holdout controls and holdout 
controls (across all villages). Non-holdout controls underwent an additional screening step (mainly 
availability during baseline surveying) during the recruitment for Cohort 1 participants but are otherwise 
comparable to holdout controls. As a result, the two trials are internally valid and since the sampling 
frames are similar, the results from one trial can be used to inform the interpretation of the results from 
the other. But since experimental conditions were not randomly assigned across both trials but within 
each, regression analysis for the two trials will be done separately. We discuss additional details on 
sampling for the two-trial design for Cohort 2 further below. 
The two-trial design is a response to implementation constraints stemming from the desire to integrate 
a study of Cohort 2 with a long-term follow-up of Cohort 1 and the large number of Cohort 2
participants that AVSI is aiming to serve. Ideally, the Cohort 2 evaluation would have been designed as a 
more straightforward, single three-arm design with a control group and two treatment group variations 
instead of the two two-arm trials summarized above. However, to maintain the integrity of the design of 
Cohort 1 for long-term follow-up, any implementation of Cohort 2 interventions needed to be balanced 
across any Cohort 1 treatment and control villages in which Cohort 1 study households will be followed 
up with in future rounds of surveying. If, for example, control villages were to have a relatively higher 
share of Cohort 2 participant households, this would mean that a comparison of Cohort 1 participant 
households in treatment with households in control villages would no longer isolate the effect of the 
Cohort 1 participation. Treatment effects estimates would be confounded either indirectly due to the 
additional economic activity injected by Cohort 2 interventions or directly (if, for example, control village 
households participate in Cohort 2 but Cohort 1 participant households in treatment villages do not). 
The need for balance of implementation of Cohort 2 across Cohort 1 treatment and control villages 
drastically limits the total number of households that could have been recruited for a three-arm Cohort 
2 study since in Cohort 1 treatment villages only a relatively small fraction of households eligible to 
participate in Cohort 1 were not offered participation. In addition, the AVSI organizes its Graduation 
Approach interventions—coaching, savings groups, FFBS—around groups of about 20–30 participants. 
This means that there are “small” villages where there were not enough eligible Cohort 1 non-
 10 Control group households of the RCT around Cohort 1 can be separated into two distinct groups: those recruited for the 
study including for surveying by IPA and those who were initially identified by AVSI as eligible but were randomized out of 
baseline and other follow-up surveying. We refer to the latter group as the holdout control. 
Baseline Survey Report of the RFSA Graduating to Resilience in Uganda, Cohort 2 (Vol. I)
Evaluation of Cohort 2 7
participants to form at least two groups of 20–30 people necessary for a three-arm trial that would 
accommodate two treatment variations (Refined Graduation with and without IPT-G) and would contain 
enough additional such households so that Cohort 2 control condition could be formed. The specific 
design with two two-arm trials that IPA and AVSI are implementing was chosen to accommodate the 
above constraints while maintaining its ability to achieve the two study goals of measuring the impact of 
Cohort 2 relative to a control group and measuring the additional impact of adding IPT-G to the 
Graduation Approach. 
There are an expected 7,200 Cohort 2 participants and not all will be part of the Cohort 2 evaluation’s 
surveys data collection. Specifically, due to constraints on the total sample size in the Phase 2 study, 
while all households participating in Cohort 2 activities were randomized into either receiving IPT-G 
programming or not, only a subset of households will be interviewed in follow-up surveys as part of the 
Cohort 2 evaluation.
The evaluation of Cohort 2 includes the following data collection rounds: 
● Baseline survey: November to December 2021
● Process evaluation: May to November 2022
● Outcome monitoring: October 2022
o Trial A: September–October 2022
o Trial B: December 2022 to February 2023.
● Endline: Mid-2024
2.2 Sampling
To understand the sampling strategy for Phase 2 of the evaluation, it is important to understand the 
original study design employed for Cohort 1. At the start of Cohort 1, AVSI had conducted a social 
mapping exercise as well as a household scorecard exercise yielding a total of 25,104 households 
considered eligible to benefit from the Graduating to Resilience activity. Among those, 11,145 were 
interviewed at baseline. Of those, 6,631 were assigned to treatment and 4,514 to control. 
The assignment of the 25,104 households identified as eligible by AVSI to experimental conditions in 
Cohort 1 was done in three stages: in the first stage, half of the 114 villages (42 in the refugee 
settlement and 72 in the host community) were randomly selected as treatment villages and half as 
control villages.11 In the second stage, households were selected to be targeted for interviews to 
participate in the Cohort 1 study. Households that were ultimately not selected for interviews became 
"holdout" controls. In the third stage, households who were recruited into the study in treatment 
villages were randomly assigned to one of the three treatments arms or to a control group. The Cohort 2
evaluation that is part of Phase 2 takes places in all villages that were part of Phase 1.
Cohort 2 participants sampled for the Phase 2 evaluation are drawn from the 25,104 initially eligible 
households and consists of either a) former Cohort 1 control households some of whom were offered 
participation in interventions as part of Cohort 2 (while other were not and remain in the control group 
 11 Some “villages” were the result of combining or splitting up administratively assigned villages for the purpose of organizing 
implementation and the intervention. In this report, we refer to the villages as they are used in the study rather than referring 
to the local administrative units. 
IMPEL | Implementer-Led Evaluation and Learning
8 Evaluation of Cohort 2
or b) Cohort 1 holdout controls (which both come from treatment and from control villages) but are 
offered participation in Cohort 2 interventions. In addition, Phase 2 includes the long-term follow-up of 
Cohort 1 households. Since the sample for the Cohort 2 study in Phase 2 is made up in part of former 
Cohort 1 control households that had been interviewed in the 2021 Cohort 1 endline, the Cohort 2
baseline survey was only administered to holdout controls. The total sample size of the Cohort 2 study is 
5,872 households but only 2,612 were interviewed in the Cohort 2 baseline survey data collection. 
Table 1 summarizes the sample size of Cohort 2 separately by Cohort 1 (regular, non-holdout) control 
and holdout control status. The total sample size of the Cohort 2 evaluation is 5,872 households.
Table 1. Sampling of Cohort 2 households as a function of Cohort 1 study status
Trial A Trial B Total
Selected from among Cohort 1 controls (previously 
interviewed as part of Cohort 1) 2,662 598 3,260
Selected from among Cohort 1 holdout controls (not 
previously interviewed; in Cohort 2 baseline sample) 0 2,612 2,612
Total number of households 2,662 3,210 5,872
2.3 Cohort 2 Eligibility Assessment
The Graduating to Resilience activity targets poor and extremely poor households that have an 
economically active woman or a male youth.
12 To identify households satisfying these criteria and 
eligible for participation in Cohort 2, AVSI conducted an eligibility assessment to all Cohort 1 households 
that had not received any intervention (holdout and non-holdout controls).
AVSI administered a scorecard questionnaire to calculate a poverty score, from 0 (poorest) to 8 (richest).
In addition, AVSI completed a bucketing exercise with community members to ask them to categorize
households into three categories in terms of poverty: “Extremely poor,” “Poor,” and “Moderate.” AVSI 
considered households as poor or extremely poor if a) their scorecard score was less than 6 or b) their 
scorecard was equal to 6 but the household had been classified by the community as poor or extremely 
poor.
Eligible households were those who satisfied the above poverty criteria and had an economically active 
woman or a male youth to be the designateed primary participants in the household. The primary 
participant is the individual who receives cash and participates in the various Cohort 2 interventions, 
VSLAs, coaching sessions, IPT-G, and trainings. Among the individuals that AVSI selected as primary 
participant, 89% were female.
 12 In this activity, a youth is someone between 18 and 35 years old.
Baseline Survey Report of the RFSA Graduating to Resilience in Uganda, Cohort 2 (Vol. I)
Evaluation of Cohort 2 9
2.4 Trial A: Impact of the refined Graduating to Resilience 
Activity
The purpose of this trial is to evaluate the impact of the “refined” Graduating to Resilience Activity by 
comparing households who were offered Cohort 2 participation (“treatment”) to those assigned to a 
control group (“control”) that is not offered service. Because AVSI Foundation required that half of all 
Cohort 2 participants receive IPT-G, the treatment group in Trial A is a combination of households 
receiving IPT-G and those who do not. While assignment to treatment vs. control was randomized at the 
individual level, assignment of treatment as part of Trial A to IPT-G (vs no IPT-G) was randomized at the 
village level. This was done at the village level because the per-village number of Trial A participants was 
typically too small, especially in the refugee community, to form at least two implementation groups 
(IPT-G versus not) into which households could be separately assigned--and by the nature of the 
intervention, IPT-G gets delivered at the group level and not individually. Because the sample size for 
Trial A is limited and because IPT-G was assigned at the village level in this trial, the trial is not designed 
to detect meaningful differences in the effects of Refined Graduation by comparing between IPT-G 
participants and IPT-G non-participants. As a result, in this trial we focus on the overall effect of Refined 
Graduation relative to the control, averaging across all participants irrespectively of their IPT-G status. 
The total number of households in Trial is 2,662. All were initially assessed for activity eligibility in 2018 
and participated in the Cohort 1 study as members of the control group. The households that form the 
Trial A sample were those households that were additionally found eligible as part of the Cohort 2
eligibility assessment (described in detail in the previous section). Table 2 shows the sample sizes in Trial 
A and their division between refugee and host community.
Table 2. Number of households in Trial A
Control Refined Graduation* 
Refugee 591 595
Host 782 694
Total 1,373 1,289
Notes: *Refined Graduation in Trial A includes both households who are receiving IPT-G and households who are not. See text 
for details.
2.5 Trial B: The Impact of Adding Interpersonal Psychotherapy 
in Groups
The purpose of this trial is to evaluate the impact of adding IPT-G to the ‘refined’ Graduating to 
Resilience RFSA. Trial B compares households assigned to Refined Graduation without IPT-G to those 
assigned to Refined Graduation with IPT-G. 
The total sample size for Trial B is 3,210. As shown in Table 1, 598 of the 3,210 households participating 
in this trial were Cohort 1 non-holdout control group members who have been reassessed to be eligible 
during the Cohort 2 eligibility assessment and have been offered participation in Cohort 2. Another 
IMPEL | Implementer-Led Evaluation and Learning
10 Evaluation of Cohort 2
2,612 households were selected from the Cohort 1 holdout control group and were enrolled into the 
study as part of the Cohort 2 baseline survey.
Table 3 shows the breakdown of the Trial B sample by the refugee and host community.
Table 3. Number of households in Trial B
Control Refined Graduation* 
Refugee 915 915
Host 690 690
Total 1,605 1,605
Baseline Survey Report of the RFSA Graduating to Resilience in Uganda, Cohort 2 (Vol. I)
Baseline Household Survey 11
3. BASELINE HOUSEHOLD SURVEY
3.1 Objectives
In November and December 2021, IPA conducted a baseline survey as part of the Cohort 2 evaluation. 
The survey was only administered to a subset of households that form the sample for Trial B because 
some households were previously interviewed as part of the recent endline of Cohort 1. The objective of 
the Cohort 2 baseline survey is to collect pre-intervention data that can be used to verify that the 
different arms of the study are comparable, will be used to improve statistical precision in analysis of 
follow-up data, and will allow for valid sub-group analysis of follow-up data. In addition, the data from 
this baseline can be used by AVSI to inform activities.
3.2 Pilot Test Survey Training
Before training the enumerators, the survey instruments were bench tested and piloted to fine-tune 
them, to ensure that the questions flowed well, to ensure logic patterns were well implemented and to 
estimate the duration of the interviews. Two further pilot tests were conducted during the training that 
took place between August 17 and September 1, 2021. The training introduced enumerators to the 
survey instruments and explained the survey questions and procedures using the training manual, 
anthropometric protocol, and survey replacement guidelines. 
During training, the enumerators practiced administering the survey instruments through role-playing 
by interviewing each other. A pilot test was conducted in one village, not part of the study sample, to 
allow enumerators to practice administering the baseline and taking anthropometric measures in a real 
field setting and trouble-shoot any bugs in the programmed survey versions. Subsequently, a debriefing 
session was held in which enumerators shared their experiences and clarified issues that emerged 
during the pilot test. A few changes were made to the logical patterns and other sections of the survey 
based on the enumerators’ observations and recommendations.
3.3 Questionnaire
The survey instrument for the baseline of Cohort 2 is a shortened version of the questionnaire used for 
the Cohort 1 evaluation, developed in consultation with BHA, and in accordance with the BHA 
baseline/endline indicator handbook. One additional module was added to capture respondent’s 
wellbeing and aspiration with three open-ended questions that were recorded.
The questionnaire was divided into three parts. The first part, administered to the head of household of 
the target participant, asked about household characteristics, expenses, income, finance, assets, and 
other information about the household. The second part asked the target respondent, who was an 
economically active female (in 95% of the households) or a youth, about his/her mental health, well￾being and social network. The target respondent is the individual in the household who is going to be 
the primary participant if randomized into a treatment group. Those individuals in the treatment arm 
will participate in coaching sessions, IPT-G, and VSLAs. The third part of the questionnaire, administered 
to the caregiver, was about women’s and children’s nutrition.
IMPEL | Implementer-Led Evaluation and Learning
12 Baseline Household Survey
The modules of the questionnaire are as follows: 
Part I: Household survey (administered to head of household)
1. Polygamy
2. PPI. Poverty Measurement13
3. Housing Characteristics
4. Livestock: ownership, inputs, production, and assets
5. Agriculture over the past two seasons: land, inputs, practices, and production, including perennial 
crops
6. Employment
7. Business
8. Assets
9. Organizational support
10. Remittances and Transfers
11. Consumption
12. Finance
13. Food Consumption Score
14. Household Food Insecurity Access Scale
15. Social Cohesion and Trust
16. WASH
17. Deceased and Roster before Displacement
Part II: Target respondent or participant
1. Subjective well-being
2. Open-ended questions about respondent’s well-being and aspiration
3. General wellbeing
4. Social capital
5. Social networks
6. Grit
7. Self-efficacy scale
8. Mental health
9. Financial health
Part III: Women and children’s nutrition (administered to caregiver)
1. Child nutrition and feeding practices
2. Woman's nutritional status and feeding practices
 13 The Poverty Probability Index (PPI®) is a poverty measurement tool that is used to compute the likelihood that the household 
is living below the poverty line. The module used in this survey had 5 questions and was specifically tailored for our context. The 
PPI was also used in the baseline of Cohort 1. See https://www.povertyindex.org/about-ppi
Baseline Survey Report of the RFSA Graduating to Resilience in Uganda, Cohort 2 (Vol. I)
Baseline Household Survey 13
3.4 Field Preparation
As all questionnaire modules but one was administered during Cohort 1 evaluation, need for pre-testing 
was minimal. This module needed bench testing and field testing to test the quality of the recordings 
and the flow of the questions. 
IPA conducted three rounds of pre-testing prior to the launch of data collection. In the first two rounds 
of piloting, the enumerators only administered the new open-ended questions module. These pre-tests 
were conducted by a team of two Field Officers on October 11 and October 13, 2021, in Akempunu 
village in Kahungye sub-county. This village is in the host community around a 30-minute motorcycle 
ride to the closest baseline village. A total of 6 surveys were carried out on both days, to identify any 
potential challenges with the new module, including the quality of recordings. Based on the findings, IPA 
refined the questionnaire after each initial round of pre-testing. A third pre-test was administered on 
October 27 and 28, 2021 when four Field Officers went to Kaswa village in Kabuuga sub-county in the 
host community, located near the two previous villages used for pre-testing. The Field Officers surveyed 
one person each and tested relevant questions from the Cohort 1 endline plus the additional questions 
added for the Cohort 2 survey.
IPA recruited the field team for the data collection for Cohort 2 among enumerators who had worked on 
the Cohort 1 endline. Training of the team took place from October 25 to October 29, 2021. A total of 38 
people participated in the training. The training consisted of presentations by IPA field supervisors and 
support staff, classroom practice, group discussions, and role play, and covered the responsibilities of 
the enumerators, objectives of the study, completion of field forms, and a question-by-question review 
of the household questionnaire. Enumerators were then put into teams of two to practice the 
questionnaire in both English and the languages spoken in the enumeration areas—Swahili and 
Kinyabwisha in the refugee community and Runyankole in the host community. IPA refined the 
translation and complexity of the questionnaires based on feedback from this exercise. IPA also selected 
several enumerator teams to role-play the questionnaire to identify methods to resolve plausible field 
challenges. During the training, enumerators practiced the new open-ended questions extensively to 
ensure that they were administering it well and recording properly.
3.5 Field Activities
This section describes the procedures for the soft launch, fieldwork, replacement strategy, and data 
quality assurance.
3.5.1 Soft Launch
IPA conducted a full-scale soft launch of the baseline survey in the villages of Kakinga and Kanoyonza II in 
the host community and Base Camp 4 in the settlement on November 20, 2021.14 That day, all 38 
enumerators conducted one survey each. The Field Manager and Senior Research Associate observed the 
preparedness of the enumerator teams, contact strategies, familiarity with the questionnaires, and 
 14 The launch of the data collection was slightly delayed while details of the study design, including randomization, sampling 
frame, and sampling were finalized with AVSI.
IMPEL | Implementer-Led Evaluation and Learning
14 Baseline Household Survey
understanding of the objective of the survey. The enumerator teams received feedback on their 
performance, discussed challenges, and clarified any final questions, especially on the module of the new 
open-ended questions. 
3.5.2 Data Collection
IPA conducted data collection in both the refugee settlement and host community between November 
20 and December 18. The field teams included the following personnel:
● In the refugee settlement, IPA deployed a total of 26 field staff, including 19 enumerators, four 
mobilizers, two team leaders and one auditor.
● In the host community, IPA deployed a total of 26 field staff, with an identical team structure as 
the settlement field team.
The Research Associate transmitted the list of households to survey to the field team. Mobilizers visited 
the villages before the rest of the team and worked with the village chairperson or local council to locate 
the selected household respondents and make appointments. Mobilizers conducted a mobilization 
survey with the households to determine the household member who would be the primary participant. 
The participant had to be an economically active woman or a youth. Then, a few days after or the 
following day, an enumerator visited the household to conduct the survey. 
3.5.3 Replacement Strategy
IPA successfully reached and interviewed 1,317 (99%) of the 1,336 households sampled in the settlement, 
and 1,253 (98%) of the 1,277 households sampled in the host community.
3.5.4 Data Quality
IPA ensured data quality through two processes: survey audits and high-frequency checks of recorded 
data. Audit surveys repeated approximately 40 questions from the original surveys. 
Every day, auditors administered a shortened version of the survey to a subsample of respondents 
surveyed the previous day. They asked them questions and probed whenever the answer they received 
differed from the one originally recorded by the enumerator, to assess whether there were any issues in 
survey administration, comprehension, or completion. The audit survey revisited the household module. 
The respondent re-interviewed was the same as the one interviewed the first time for this section (head 
of household or primary participant). IPA conducted audit surveys with 10% of the household 
respondents. Whenever discrepancies arose, enumerators and auditors met to understand the origin of 
the discrepancy and verify the correct response, sometimes verifying answers with the household 
respondent. IPA research staff then corrected the verified answer in the database. The goal of the audit 
survey was also to verify that enumerator did the survey with the right respondent and that the 
enumerator did not commit fraud. No case of fraud was detected.
IPA also performed high-frequency checks daily on incoming data using Stata. This could be done every 
day since the team uploaded data to the server every day after coming back from the field. IPA research 
staff wrote code to look for data outliers, logical inconsistencies, key variables, missing data, and 
differences among enumerators on key variables. If IPA research staff identified any issues during this 
Baseline Survey Report of the RFSA Graduating to Resilience in Uganda, Cohort 2 (Vol. I)
Baseline Household Survey 15
process, they consulted with enumerators to clarify or correct the answers. IPA discovered no 
intentional data manipulation by enumerators. 
3.6 Limitations and Challenges
3.6.1 Limitations Related to Self-Reported Data
Self-reported data have limitations, such as the possibility of exaggeration or omission of information, 
inaccurate recollection of experiences or events, reporting of untruthful information, and reduced 
validity when respondents do not fully understand a question. This is especially true in areas such as a 
refugee settlement where a high density of assistance programming may encourage respondents to 
believe they may receive a benefit for providing one answer over another. As outlined above in data 
quality, IPA deployed audit surveys to check the validity of answers.
3.6.2 Challenges Related to Field Conditions and COVID-19
The field team encountered infrastructural challenges due to the poor state of roads in the survey area, 
heavy rains, large distance between households in some host community villages, and restrictions 
introduced in light of COVID-19. The survey period fell during the rainy season which further 
exacerbated the low quality of roads and slowed the data collection process by complicating access to 
some areas. In addition, the population density in the host communities is low and households were 
separated by large areas of farmland or accessible only through small, steep paths where not even 
motorcycle taxis were able to pass. For this reason, enumerators had to walk up to 30 minutes to reach 
some respondents. Restrictions introduced to curb the COVID-19 pandemic resulted in further 
challenges for the field team. A curfew at 7:00 PM combined with long ways between Kamwenge where 
the field team was based and the survey sites led to the team having to leave the field early to make it 
back in time before the curfew. The distances and the curfew did not add extra days of data collection 
since the IPA accounted for them when planning for the data collection.
IMPEL | Implementer-Led Evaluation and Learning
16 Results
4. RESULTS
4.1 Characteristics of the Study Sample
The tables present summary statistics from data of the households interviewed as part of the Cohort 2
baseline survey. Given the differences between communities, the tables below provide summary 
statistics disaggregated by community. The key variables show USAID indicators and indicators on 
household demographics, assets, livestock, farming, employment, business and finance, to provide an 
outlook of the situation of the population before AVSI began implementing the package of interventions 
for Cohort 2.
In total, 2,570 households were interviewed at Cohort 2 baseline, 1,317 households were from the host 
community and 1,253 were from the settlement. In the tables below, the number of observations 
corresponds to the numbers of households, unless specified otherwise.
4.2 Household Demographics
Table 4 shows demographic characteristics of both samples at the individual level (not the household 
level). These numbers are based on 8,098 individuals in the host sample and 8,462 individuals in the 
refugee sample.
Table 4. Individual demographic characteristics
Host Refugee
Number of individuals in sampled households 8,098 8,462
Share of adults aged 18 or over 3,310 (41%) 2,920 (35%)
Share of children under 5 years of age 1,208 (15%) 1,762 (21%)
Share of women of reproductive age (15–49) 1,699 (21%) 1,658 (20%)
Youth (18–35) 1,847 (23%) 1,738 (21%)
Notes: The number of observations corresponds to the number of individual of each category.
Table 5. Household demographics
Host 
(N=1,253)
Refugee 
(N=1,317)
Mean Standard Deviation (SD) Mean SD
Average age of household head 45.90 14.61 39.54 12.32
Average number of children per household
Under 18 3.81 2.02 4.20 2.17
Under 5 0.96 0.90 1.33 1.02
Under 3 0.52 0.63 0.78 0.71
Household head is under 30 years old 14% 35% 21% 41%
Female household head 44% 50% 50% 50%
Baseline Survey Report of the RFSA Graduating to Resilience in Uganda, Cohort 2 (Vol. I)
Results 17
Host 
(N=1,253)
Refugee 
(N=1,317)
Mean Standard Deviation (SD) Mean SD
Widow(er) household head 21% 40% 12% 33%
HH has adult female (18+) and no adult male 15% 36% 18% 38%
HH has adult male (18+) and no adult female 15% 36% 18% 38%
HH has adult male (18+) and adult female 98% 13% 98% 13%
Household head education
No formal education 28% 45% 38% 48%
Completed at least 1 year of primary 71% 45% 62% 49%
Completed At least 1 year of secondary 9% 28% 14% 34%
4.3 Poverty Indicators
4.3.1 Consumption Per Capita
To calculate whether a household falls under the poverty threshold, IPA first calculated households’ 
daily consumption. In Cohort 2 baseline, IPA used a short module questionnaire based on a 
comprehensive consumption module from Cohort 1 endline.15 To compute the total consumption value, 
IPA predicted the consumption per capita using the consumption value of a subset of types of food as 
well as the size of the household. The subset of types of food comprised the types of food with the 
highest consumption value at endline: millet, ground nuts, onions, cassava, Irish potatoes, matooke, 
sweet banana, beef, chicken, milk, maize, rice, and beans. These were the types of food that 
represented the biggest shares of the households’ food expenditures at endline. During baseline 
analysis, Irish potatoes and matoke were excluded from the variables used in the prediction; there was a 
harvest of Irish potatoes during most of the baseline, making its consumption value less comparable to 
the endline that was conducted over many months, and the price of matooke during the baseline was 
much higher than what was observed during the endline (even though quantities remained similar). To 
consider household economy of scale and the fact that children (under 18 years old) consume less than 
adults, we used the concept of adult equivalence and specifically the Organization for Economic Co￾operation and Development (OECD) scale. The number of adult equivalents in the household 
corresponds to the following formula:
𝑁𝑁𝑁𝑁. 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝑒𝑒𝑒𝑒 𝑒𝑒𝑒𝑒 𝑒𝑒𝑒𝑒 𝑒𝑒 = 1 + 0.7 ∗ ( 𝑛𝑛 𝑛𝑛 𝑜𝑜𝑜𝑜 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 − 1) + 0.5 ∗ ( 𝑛𝑛 𝑛𝑛 𝑜𝑜𝑜𝑜 𝑐𝑐ℎ )
 15 In the Cohort 1 endline, IPA had calculated it from a consumption module that listed all the types of expenses of the 
household in the past 7 days, 30 days, 12 months. Some expenses were asked about the past 7 days (food consumption for 
instance), some others were asked about the past 30 days or past 12 months. IPA converted these expenses to a daily 
household total expenditure and divided it by the number of household members (in terms of adult equivalents, see 
explanation below).
IMPEL | Implementer-Led Evaluation and Learning
18 Results
4.3.2 Prevalence of Poverty
IPA compared household average daily per capita consumption to the international poverty line of 2011 
Purchasing Power Parity (PPP) $1.90 to determine if consumption falls below or above the poverty line. 
To do this comparison, IPA converted the international poverty line to Ugandan shillings using the 2011 
USD PPP exchange rate and adjusting for inflation. All the amounts in U.S. dollars (USD) mentioned in 
this section are in 2011 PPP USD.
Approximately 31% of households in the refugee community fall below the poverty line, as compared to 
38% of households in the host community. These results are summarized below in Table 6.
4.3.3 Depth of Poverty
Depth of poverty measures, on average, how far below the PPP $1.90 consumption poverty threshold 
are the poor in the study sample. IPA calculated the mean depth of poverty by converting the PPP $1.90 
per person per day poverty threshold into local currency units and then inflating it using Uganda’s 
Consumer Price Index. In the refugee community, the mean depth of poverty is 51%, meaning it would 
cost about USD $0.97 (51% * $1.90) per household member per day to lift the average household in the 
refugee sample out of poverty. In the host community, the mean depth of poverty is 48%, meaning it 
would cost about $0.91 per household member per day to lift the average household out of poverty. 
These results are summarized in Table 6.
Table 6. Poverty levels
Host Refugee
N Mean SD N Mean SD
BL01. Prevalence of Poverty: Percentage of 
people living on less than $1.90/day 2011 PPP 1,252 38% 48% 1,317 31% 46%
BL02. Depth of Poverty of the Poor: Mean
percentage shortfall of the poor relative to the
$1.90/day 2011 PPP poverty line
470 48% 52% 402 51% 50%
Notes: For the indicator BL02, the number of observation corresponds to the number of households below the $1.90/day 2011 
PPP poverty line.
4.4 Water, Sanitation, and Hygiene
4.4.1 Use of Basic Drinking Water Services 
The BHA indicator handbook describes basic drinking water services as “improved sources or delivery 
points that meet three technical criteria: (1) are protected from outside contamination by nature of their 
construction or through active intervention, in particular from outside contamination with fecal matter, 
(2) where collection time is no more than 30 minutes for a round-trip including queuing, and (3) that 
consistently produce (i.e., year-round) basic drinking water.” The rate of households accessing such 
infrastructure in the survey area is low, only 12% in the host and 13% in the refugee community.
Baseline Survey Report of the RFSA Graduating to Resilience in Uganda, Cohort 2 (Vol. I)
Results 19
4.4.2 Percentage of Households in Target Areas Practicing Correct Use 
of Recommended Household Water Treatment Technologies
Households are considered practicing correct use of recommended household water treatment 
technologies if they use one of the evidence-based household water treatments. These treatments are:
● Chlorination (chemical disinfection)
● Flocculants/disinfectant (physio-chemical disinfection)
● Filtration (physical removal)
● Solar disinfection (UV/heat disinfection)
● Boiling (disinfection by heat)
Most households reported the use of at least one of these techniques. The most common is boiling. In 
the host community, 75% of the households boil their water, and 49% of the households in the 
settlement do it. There are a number of potential reasons that could explain this difference. Drawing on 
information about the context and the baseline data, we note that, first, households in the refugee 
community generally have worse access to firewood to boil water. In the host community, firewood 
available for collection from the surrounding forested areas. In contrast, inside the settlement, 
households are not allowed to cut trees and so they would need to buy firewood or walk to the host 
community to collect it. Second, households in the settlement seem to have access to cleaner water: 
80% of the households in the settlement used boreholes compared to only 30% in the host community, 
where 39% used unprotected wells. Those using boreholes may not feel the need to boil water (even if it 
is still recommended). 
We tested the statistical significance of the differences in terms of WASH behavior between the two 
communities. All these differences are statistically significant at the 10% level, some are statistically 
significant at the 5% level.
Table 7. Household WASH characteristics
Host Refugee
N Mean SD N Mean SD
BL16. Percentage of households using basic 
drinking water services 1,253 12% 32% 1,317 15% 36%
BL17. Percentage of households with soap and 
water at a handwashing station on premises 1,253 8% 27% 1,317 3% 18%
BL18. Percentage of households in target areas 
practicing correct use of recommended 
household water treatment technologies 1,253 75% 43% 1,317 58% 49%
BL19. Percentage of population in target areas 
practicing open defecation 1,253 2% 14% 1,317 11% 31%
IMPEL | Implementer-Led Evaluation and Learning
20 Results
Table 8. Household water use and source
Host Refugee
N Mean SD N Mean SD
Reported the use of the following water 
treatment technique
Boiling 1,253 74% 44% 1,317 49% 50%
Filtration 1,253 3% 18% 1,317 5% 21%
Chlorination 1,253 0% 7% 1,317 6% 23%
Flocculant/disinfectant 1,253 0% 5% 1,317 3% 18%
Solar disinfection 1,253 0% 0% 1,317 0% 05%
Source of water
Tube well/borehole 1,253 29% 46% 1,317 80% 40%
Unprotected well 1,253 39% 49% 1,317 2% 13%
Public tap/standpipe 1,253 12% 33% 1,317 5% 22%
Surface water (river, pond, etc.) 1,253 5% 23% 1,317 5% 22%
Protected well 1,253 5% 23% 1,317 3% 18%
Protected spring 1,253 2% 16% 1,317 3% 17%
Piped to yard/plot 1,253 3% 18% 1,317 0% 3%
Unprotected spring 1,253 2% 14% 1,317 1% 9%
Piped into dwelling 1,253 1% 11% 1,317 1% 8%
4.5 Housing and Assets
Table 9. Roof materials
Host Refugee
N Mean SD N Mean SD
Iron sheet 1,253 91% 28% 1,317 21% 41%
Tarpaulin 1,253 0% 6% 1,317 78% 42%
Straw, thatch, sod 1,253 8% 27% 1,317 0% 5%
Notes: In the settlement, 15 individuals have other types of materials on their roofs.
In the host community, houses typically have iron sheets on the roof while in the refugee settlement 
United Nations High Commissioner for Refugees (UNHCR) tarpaulins are the dominant type of roofing. In 
the refugee settlement, iron sheet roofs used to be prohibited to keep housing temporary. They were 
allowed 2 years ago and those who could afford started installing them. 
Baseline Survey Report of the RFSA Graduating to Resilience in Uganda, Cohort 2 (Vol. I)
Results 21
Table 10. Key assets owned
Host Refugee
N Mean SD N Mean SD
Chairs, bench, or stool 1,253 96% 19% 1,317 92% 26%
Beds, cot—wooden or metal 1,253 90% 30% 1,317 09% 28%
Dining table 1,253 47% 50% 1,317 30% 46%
Radio/transistor/stereo 1,253 40% 49% 1,317 27% 44%
Television 1,253 5% 21% 1,317 6% 24%
Mobile phone 1,253 80% 40% 1,317 70% 46%
Solar power/electric inverters 1,253 63% 48% 1,317 63% 48%
Motorcycle 1,253 4% 20% 1,317 3% 17%
During the Cohort 2 baseline survey, IPA asked about a subset of the most valuable assets that were 
identified during the Cohort 1 endline. Asset ownership rates are higher in the host community than in 
the refugee community (or equal in the case of solar panels). 
4.6 Land and Agriculture
In the baseline survey, IPA collected data about the two previous agricultural seasons, the March to May 
2021 season and the September to December 2021 season. 
IPA collected data on land use, input use, practices used, and crops cultivated and sold. IPA also 
collected data on perennial crops like matoke. Given that the season was still ongoing while IPA was 
administering the survey, the questionnaire also had questions about expectations on input use, 
harvest, and income (e.g., whether they expected to sell their harvest and if so, the quantity they 
expected to sell).
In the host community, most households (86%) own land. Host community households own 0.87 acres
on average (one acre equals 4,047 m2
). 
Table 11. Agricultural land ownership in the host community during the September to December 2021 
season
N Mean SD
The household owns any plot 1,253 86% 0.35
Surface of plot owned by the household (acre) 1,253 0.87 1.17
In the settlement, the Office of Prime Minister (OPM) in charge of the settlement allocates plot to the 
refugees when they arrive in the camp. The refugees have temporary rights to use the plot. They cannot 
sell the land. In the settlement, 93% of the households have such plots. On average, they can use a 
surface of 0.69 acres.
IMPEL | Implementer-Led Evaluation and Learning
22 Results
Table 12. Agricultural land rights of usage in the host community during the September to December 
2021 season
N Mean SD
The household has temporary usage rights on any plot 1,317 93% 26%
Surface of plot that household can use for free (acre) 1,317 0.69 0.59
Farming provides food to the household and can also be a source of income when the harvest is sold. 
Tables 13 through 17 below show the input use during the September to December 2021 seasons and 
the main cash and food crops during the March to May and the September to December 2021 seasons.
Table 13. Agriculture land and inputs during the September to December 2021 season
Host Refugee
N Mean SD N Mean SD
Used organic fertilizer 1,253 36% 48% 1,317 61% 49%
Used inorganic fertilizer 1,253 13% 34% 1,317 13% 33%
Used pesticides 1,253 15% 35% 1,317 33% 47%
Used herbicides 1,253 29% 45% 1,317 44% 50%
Hired agricultural labor 1,253 37% 48% 1,317 51% 50%
Used irrigation 1,253 2% 13% 1,317 4% 20%
Used mechanical plow 1,253 6% 23% 1,317 0% 7%
Used knapsack chemical sprayer 1,253 34% 47% 1,317 56% 50%
Used motorized grain mill 1,253 52% 50% 1,317 35% 48%
Table 14. Main food crops during the September to December 2021 season16
Host Refugee
N Mean SD N Mean SD
Household is cultivating any crop 1,253 98% 0.13 1,317 99% 0.11
Crops considered as top three food crops
Maize 1,253 50% 50% 1,317 70% 46%
Kanyobwa beans 1,253 23% 42% 1,317 46% 50%
Irish potatoes 1,253 21% 41% 1,317 22% 41%
Mixed beans 1,253 30% 46% 1,317 7% 26%
Other beans 1,253 14% 35% 1,317 18% 38%
Cassava 1,253 14% 35% 1,317 7% 26%
Millet 1,253 15% 35% 1,317 1% 7%
Nambale beans 1,253 7% 26% 1,317 4% 19%
 16 Top three food crops are sorted by overall sense across sample. We used the same type of ordering in other tables.
Baseline Survey Report of the RFSA Graduating to Resilience in Uganda, Cohort 2 (Vol. I)
Results 23
Host Refugee
N Mean SD N Mean SD
Kabulangiti beans 1,253 2% 15% 1,317 9% 28%
Sweet potato 1,253 9% 28% 1,317 0% 7%
Table 15. Main cash crops during the September to December 2021 season
Host Refugee
N Mean SD N Mean SD
Any cash crop 1,253 61% 0.49 1,317 53% 0.50
Crops considered as top two cash crops ...
Maize 1,253 43% 49% 1,317 36% 48%
Irish potatoes 1,253 9% 29% 1,317 11% 31%
Kanyobwa beans 1,253 10% 30% 1,317 7% 25%
Mixed beans 1,253 11% 32% 1,317 2% 13%
Other beans 1,253 5% 22% 1,317 3% 16%
Nambale beans 1,253 4% 20% 1,317 1% 10%
Millet 1,253 5% 22% 1,317 0% 6%
Ground nuts 1,253 4% 19% 1,317 1% 9%
Table 16. Main food crops during the March to May 2021 season
Host Refugee
N Mean SD N Mean SD
Household was cultivating any crop 1253 95% 22% 1317 97% 18%
Crops considered as top three food crops …
Maize 1253 52% 50% 1317 83% 38%
Kanyobwa beans 1253 24% 43% 1317 51% 50%
Mixed beans 1253 32% 47% 1317 5% 23%
Other beans 1253 14% 35% 1317 15% 35%
Cassava 1253 17% 38% 1317 7% 25%
Irish potatoes 1253 17% 37% 1317 6% 25%
Kabulangiti beans 1253 2% 14% 1317 1% 30%
Nambale beans 1253 7% 25% 1317 5% 21%
Sweet potato 1253 8% 27% 1317 1% 9%
Ground nuts 1253 6% 23% 1317 1% 12%
IMPEL | Implementer-Led Evaluation and Learning
24 Results
Table 17. Main cash crops during the March to May 2021 season
Host Refugee
N Mean SD N Mean SD
Any cash crop 1,253 44% 50% 1,317 42% 49%
Crops considered as top two cash crops …
Maize 1,253 22% 41% 1,317 32% 47%
Kanyobwa beans 1,253 8% 27% 1,317 8% 27%
Irish potatoes 1,253 6% 24% 1,317 4% 20%
Mixed beans 1,253 8% 28% 1,317 1% 10%
Other beans 1,253 5% 22% 1,317 3% 16%
Nambale beans 1,253 2% 14% 1,317 1% 11%
Cassava 1,253 1% 12% 1,317 1% 11%
Kabulangiti beans 1,253 0% 7% 1,317 2% 14%
Yellow beans 1,253 1% 9% 1,317 0% 6%
4.7 Livestock Holdings
In terms of livestock holdings, most households have owned at least one type of livestock in the past 6 
months. The most common type of livestock owned is chicken, owned by 68% by households in the host 
community and 76% of the households in the settlement. Livestock can be a source of income through 
livestock trading or livestock product trading like eggs and milk. It can also be a source of food, from 
either slaughtering or livestock products consumption.
Table 18. Livestock ownership
Host Refugee
N Mean SD N Mean SD
Livestock owned in the past 6 months
Chicken 1,253 68% 47% 1,317 76% 43%
Pig 1,253 53% 50% 1,317 40% 49%
Goat 1,253 39% 49% 1,317 26% 44%
Duck 1,253 11% 31% 1,317 22% 41%
Cow 1,253 13% 34% 1,317 5% 22%
Sheep 1,253 10% 31% 1,317 6% 24%
Rabbit 1,253 3% 18% 1,317 1% 11%
Other 1,253 1% 4% 1,317 5% 19%
No livestock 1,253 13% 34% 1,317 12% 32%
Livestock products: produced
Baseline Survey Report of the RFSA Graduating to Resilience in Uganda, Cohort 2 (Vol. I)
Results 25
Host Refugee
N Mean SD N Mean SD
Eggs 1,253 33% 47% 1,317 36% 48%
Milk 1,253 5% 22% 1,317 3% 16%
On average, households own 3.83 chickens in the host community and 3.87 chickens in the refugee 
community, as shown in Table 19.
Table 19. Number of chickens owned
Host Refugee
N Mean SD N Mean SD
Number of chickens owned 1,253 3.83 5.11 1,317 3.87 5.23
Number of chickens owned among households 
owning at least one chicken 851 5.63 5.32 1,003 5.08 5.45
4.8 Employment
Most households engaged in some form of employment, mostly through agricultural daily labor. They 
offer their services to other households to dig, weed, or harvest. Some households also do other daily 
labor, for instance in construction or brick laying. The rate of employment is similar among 
communities.
Table 20. Employment
Host Refugee
N Mean SD N Mean SD
Any employment 1,253 71% 45% 1,317 74% 44%
Employed in
Agricultural daily labor 1,253 64% 48% 1,317 62% 49%
Tending animals for other households 1,253 3% 17% 1,317 2% 15%
Other daily labor 1,253 9% 28% 1,317 13% 34%
Household work for another household 1,253 2% 13% 1,317 2% 15%
Salaried/formal employment 1,253 4% 19% 1,317 3% 18%
In terms of agricultural daily labor, 41% of the adult women were employed in this activity versus 37% of 
the adult men. Among youths (18 to 35 years old), 41% were employed in agricultural daily labor versus 
only 2% among people above 35 years old.
In terms of other daily labor like construction or brick laying, less than 1% of the adult women had been 
employed in this activity versus 9% in of the adult men. Among youths, 6% were employed in this 
activity versus less than 1% for people above 35.
IMPEL | Implementer-Led Evaluation and Learning
26 Results
4.9 Off-Farm Business
This section covers off-farm businesses. These businesses were typically petty trade (selling small 
quantities of food or goods at the market for instance), produce trading (trading maize, or other types of 
food in large quantities), retail shops, or boda boda (moto taxi). Other off-farm businesses included 
selling alcohol, cooked food, groceries, and second-hand clothes. Engagement in off-farm business was 
higher in the settlement (43%) than in the host community (30%).
Table 21. Main business types
Host Refugee
N Mean SD N Mean SD
Any business 1,253 30% 46% 1,317 43% 50%
Number of businesses 1,253 0.33 0.55 1,317 0.52 0.68
Has any business of 
Petty trade 1,253 9% 29% 1,317 27% 44%
Produce trading 1,253 2% 14% 1,317 5% 22%
Retail shop 1,253 4% 20% 1,317 2% 13%
Boda boda driver 1,253 2% 15% 1,317 3% 16%
Notes: Other off-farm businesses include selling alcohol, selling cooked food, groceries, second-hand clothes etc. 70% of the 
households in the host and 57% of the households in the settlement don’t have any off-farm business.
Table 22. Main Business types among youths (18–35 years old)
Host Refugee
N Mean SD N Mean SD
Any business 1,847 12% 32% 1,738 20% 40%
Number of businesses 1,847 0.12 0.33 1,738 0.21 0.43
Has any business of
Petty trade 1,847 3% 17% 1,738 12% 32%
Retail shop 1,847 2% 13% 1,738 1% 10%
Produce trading 1,847 0% 7% 1,738 2% 14%
Tailoring 1,847 1% 7% 1,738 1% 9%
Boda boda driver 1,847 1% 11% 1,738 1% 12%
4.10 Savings and Loans
Most households reported having some savings while at the same time many also reported having 
outstanding loans. Part of it can be explained by the participation in a VSLA or other savings groups that 
Baseline Survey Report of the RFSA Graduating to Resilience in Uganda, Cohort 2 (Vol. I)
Results 27
can allow participants to save and take loans. However, many respondents also reported taking out 
loans from friends and neighbors, as well as shopkeepers.17
Table 23. Outstanding loans
Host Refugee
N Mean SD N Mean SD
Has any outstanding loan 1,253 77% 42% 1,317 68% 47%
Number of outstanding loans 1,253 1.52 1.3 1,317 0.99 0.96
Main source of loans
VSLA 1,253 20% 40% 1,317 9% 29%
Friend 1,253 11% 31% 1,317 14% 35%
Savings and credit cooperative organization 
(SACCO) 1,253 13% 33% 1,317 7% 25%
Shopkeeper 1,253 5% 23% 1,317 14% 34%
Neighbor 1,253 2% 12% 1,317 15% 35%
Other savings groups 1,253 13% 34% 1,317 0% 3%
Other sources 1,253 5% 23% 1,317 3% 17%
Moneylender 1,253 1% 8% 1,317 4% 21%
Rotating savings and credit association (ROSCA) 1,253 4% 21% 1,317 1% 10%
In terms of savings, a large majority of respondents reported having saved some money in the past 3 
months. Most of them did so in savings groups, 64% in the host and 82% in the settlement. Other 
popular saving methods include carrying the money or storing it at home despite the risks. Households 
in our sample do not use commercial banks. The closest commercial bank is in Kamwenge town, which is 
far away for most respondents (one hour away by public transport from the biggest trading center of 
the study area). 
Table 24. Savings over the last 3 months
Host Refugee
N Mean SD N Mean SD
Saved in any place 1,253 92% 27% 1,317 94% 23%
Saved in ...
VSLA, SACCO, or another savings group 1,253 64% 48% 1,317 82% 39%
Pocket, clothes, or bag that is carried 1,253 55% 50% 1,317 34% 48%
 17 For further information about loan use, we can draw on data from previous data collection rounds, specifically data from 
control village households collected as part of the Cohort 1 endline survey. In that data, the average amount of outstanding 
loans was PPP $177 in the host communities and PPP $77 in the settlement. The median amount of outstanding loans was PPP 
$67 in the host communities and PPP $38 in the settlement. In the Cohort 1 endline, in the control villages, 46% of the 
households were part of a VSLA. Among those households, 21% had no outstanding loans, 14% had outstanding loans solely 
with the VSLA and only 3% had outstanding loans from both VSLA and neighbors, or friends of family.
IMPEL | Implementer-Led Evaluation and Learning
28 Results
Host Refugee
N Mean SD N Mean SD
A secret place at home 1,253 21% 41% 1,317 39% 49%
Mobile money 1,253 15% 36% 1,317 24% 43%
Box in the household 1,253 4% 18% 1,317 8% 27%
With family member outside the household 1,253 2% 15% 1,317 0% 7%
With a friend 1,253 1% 12% 1,317 1% 10%
Commercial bank 1,253 1% 10% 1,317 1% 7%
Other 1,253 1% 11% 1,317 1% 10%
With a neighbor 1,253 1% 9% 1,317 0% 6%
With a shopkeeper 1,253 0% 5% 1,317 0% 4%
4.11 Nutrition and Food Security
Food security is essential to well-being, and elimination of hunger is the second goal of the United 
Nations Sustainable Development Goals. Food diversity is important to ensure sufficient intake of 
micronutrients that are key for individuals’ development to their full potential, in particular in the first 
years of life. 
In terms of food insecurity at the household level, IPA computed two indicators, the Food Consumption 
Score (FCS) and the Household Food Insecurity Access Scale (HFIAS). IPA also computed the Household 
Hunger Scale (HHS), using a subset of the HFIAS questions.
The FCS is a measure of food diversity and frequency of the consumption of the different food groups.18
The FCS is calculated with the weighted average of the number of days the households consumed 
different categories of items: staples, pulses, vegetables, fruits, meat/fish/eggs, milk, sugar, oil, and 
condiments. 
The HFIAS is a measure of the food insecurity experienced by the household.19 It is calculated from a 
series of questions about the food insecurity and hunger experienced by the household over the past 4 
weeks and the frequency of those episodes.
The HHS is a measure of hunger and food deprivation experienced by the household.20 It uses a subset 
of the HFIAS questions that are focused on food deprivation.
There may seem to be a mismatch between FCS and HFIAS when comparing the discrete categories that 
are based on the respective scores, but further analysis shows that these two indicators are consistent 
with each other and strongly correlated. In terms of FCS, most households are categorized as 
“acceptable” (respectively 95% and 86% in the host and refugee community) while in terms of HFIAS, 
many of them are categorized as “moderately food insecure” (respectively 48% and 54% in the host and 
 18 https://inddex.nutrition.tufts.edu/data4diets/indicator/food-consumption-score-fcs. Accessed on March 16, 2022. 19 https://www.fantaproject.org/sites/default/files/resources/HFIAS_ENG_v3_Aug07.pdf
20 https://www.fantaproject.org/sites/default/files/resources/HHS-Indicator-Guide-Aug2011.pdf
Baseline Survey Report of the RFSA Graduating to Resilience in Uganda, Cohort 2 (Vol. I)
Results 29
refugee community). The apparent difference in assessments by indicator is mainly related to the pre￾defined score cut-offs for each category. The HFIAS index is constructed such as lower means better. To 
be consistent with the FCS index, we used the negative of the HFIAS so that higher means better. The 
negative of the HFIAS and the FCS are strongly positively correlated, see Appendix Table 31. In other 
words, households that are doing relatively better according to the HFIAS also are also doing relatively 
better according to the FCS. In addition, correlations of the individual components of each score have 
the expected sign: more frequent consumption in the different food categories as elicited in the FCS is 
negatively correlated with reported experiences of food insecurity, see Appendix Table B1. 
To further the analysis on food insecurity and the apparent mismatch between FCS and HFIAS category 
labels (but not between the scores, which are strongly correlated as described above), Table 25 also 
shows the categories based on the Household Hunger Scale (HHS). On this scale, most households are 
categorized as having, ”little to no hunger in the household.” That is the case for 85% of the households 
in the host community and 91% of the households in the settlement, broadly in line with the FCS-based 
categorizations.
Table 25. Food insecurity at the household level
Host Refugee
N Mean SD N Mean SD
BL10. Percentage of households with poor, 
borderline, and acceptable food consumption 
score (FCS)
Acceptable (> 35) 1,253 95% 22% 1,317 86% 34%
Borderline (21.5–35) 1,253 5% 22% 1,317 13% 34%
Poor (0–21) 1,253 0% 3% 1,317 0% 6%
HFIAS
Food Secure 1,253 6% 24% 1,317 11% 31%
Mildly Food Insecure 1,253 7% 26% 1,317 7% 25%
Moderately Food Insecure 1,253 48% 50% 1,317 54% 50%
Severely Food Insecure 1,253 39% 49% 1,317 29% 45%
HHS
Little to no hunger in the household (0–1) 1,253 85% 36% 1,317 91% 28%
Moderate hunger in the household (2–3) 1,253 13% 34% 1,317 8% 27%
Severe hunger in the household (4–6) 1,253 2% 15% 1,317 0% 0.07
In terms of HFIAS, host community households are more “severely food insecure” (39%) than refugee 
community households (29%). That could be explained by the fact that refugee households have been 
receiving food or cash from the World Food Program (WFP) since 2018. That support was equivalent to 
PPP $15 per household member in November and December 2021. But note that according to the FCS, 
hosts have a higher rate of, ”acceptable” scores. 
IMPEL | Implementer-Led Evaluation and Learning
30 Results
IPA also collected data on nutrition about women of reproductive age (15–49 years), children 6–23 
months, and children 0–5 months. Related summary statistics are shown in the following subsections. 
4.11.1 Prevalence of Women of Reproductive Age Consuming a Diet of 
Minimum Diversity
This measures the food diversity of women of reproductive age to ensure they receive a sufficient intake 
of micronutrients. Women ages 15–49 years were asked whether they consumed at least five of 10 
specific food groups during the previous day and night to evaluate the diversity of their consumption. As 
shown in Table 26, the share of women of reproductive age who had a diet of minimum diversity was 
30% in the host community and 29% in the settlement.
4.11.2 Prevalence of Children 6–23 Months Consuming a Diet of 
Minimum Diversity
This indicator measures the percentage of children 6–23 months who receive a minimum feeding 
frequency and minimum dietary diversity. IPA asked caregivers the liquid and solid types of food 
consumed by these children in the day and night preceding the survey. As shown in Table 26, the share 
of children having a minimum acceptable diet was 14% in the host community and 15% in the 
settlement.
4.11.3 Prevalence of Exclusive Breastfeeding of Children Under 6 Months
According to the BHA indicators handbook, “Exclusive breastfeeding means that the infant received 
breast milk (including milk expressed or from a wet nurse) and might have received oral rehydration 
solution (ORS), vitamins, minerals, and/or medicines, but did not receive any other food or liquid, 
including water.” Exclusive breastfeeding reduces the risk of mortality from infectious diseases that 
could come from food or water and is recommended for children until 6 months. 
Exclusive breastfeeding is higher in the refugee community (83% of children) than in the host 
community (69% of the children). The difference between the two communities seems to relate to 
caregivers in the host communities being somewhat more likely to give food or liquids to their children 
under 6 months than their counterparts in the settlement, see information in Table 27. 
Table 26. Nutrition status for women of reproductive age and young children
Host Refugee
N Mean SD N Mean SD
BL11. Prevalence of women of reproductive age 
consuming a diet of minimum diversity 947 30% 45% 1,174 29% 45%
BL39. Prevalence of children 6–23 months 
consuming a diet of minimum diversity 357 14% 34% 579 15% 35%
BL13. Prevalence of exclusive breastfeeding of 
children under 6 months 61 69% 47% 146 88% 33%
Notes: For the indicator BL11, the number of observations corresponds to the number of households with at least one woman in 
reproductive age. For the indicator BL39, it corresponds to the number of households with at least one child 6–23 months. For 
the indicator BL13, it corresponds to the number of households with at least one child under 6 months.
Baseline Survey Report of the RFSA Graduating to Resilience in Uganda, Cohort 2 (Vol. I)
Results 31
Table 27. Subcomponents of the exclusive breastfeeding indicator for children under 6 months
Host Refugee
N Mean SD N Mean SD
Age in months 61 2.93 1.70 146 2.65 1.63
Breastfed yesterday 61 100% 0% 146 99% 8%
Yesterday, the child had
Plain water 61 11% 32% 146 9% 29%
Formula 61 0% 0% 146 1% 8%
Animal milk 61 7% 25% 146 1% 12%
Juice 61 2% 13% 146 1% 8%
Clear broth 61 0% 0% 146 1% 8%
Yogurt 61 0% 0% 146 0% 0%
Thin porridge 61 13% 34% 146 8% 26%
Other liquids 61 0% 0% 146 2% 14%
Solid or semi-solid, food 61 15% 36% 146 6% 24%
Notes: The number of observations in this table corresponds to the number of households with at least one child under 6 
months, excluding about 10% of the children under 6 months for which the caregiver was absent and respondents available did 
not have the knowledge to serve as a proxy.
IMPEL | Implementer-Led Evaluation and Learning
32 Comparison of Experimental Groups
5. COMPARISON OF EXPERIMENTAL GROUPS
This section compares households participating in the different experimental conditions of Trials A and B 
to confirm that the households across conditions, within each trial, are comparable to each other on 
average. Note that this analysis combined data from the Cohort 2 baseline (for newly enrolled study 
households from Cohort 1 holdout controls) as well as from the Cohort 1 endline (from Cohort 1 
controls that were selected to participate in Cohort 2). This is in contrast to the rest of this report which 
discusses the newly collected Cohort 2 baseline data. The data is combined for the baseline balance 
analysis since data from the two types of households will be similarly combined for the analysis of 
endline data, thus, the comparisons presented here are the ones that are relevant to assess balance for 
follow-up analyses. The results confirm that the groups are comparable as the variables are balanced 
among groups on average. Few, if any, variables have statistically significantly different means between 
treatment and control village households, implying that the random assignment was successful, and 
these households are comparable on average."
Table 28. Trial A: Comparison of refined graduation and control
(1) (2) (1)–(2)
Treatment: Refined 
Graduation* Control T vs. C
N Mean SD N Mean SD P-Value
Age of household head 1,272 43.21 13.93 1,357 43.01 13.85 0.75
Number of children under 5 years old in 
the household 1,272 1.17 1.00 1,357 1.11 0.94 0.15
Living on less than $1.90 PPP/day 1,272 39% 49% 1,357 38% 49% 0.86
FCS score 1,272 50.28 14.46 1,357 49.42 14.55 0.14
HFIAS score 1,272 8.98 6.21 1,357 9.11 6.37 0.57
Owns radio 1,272 34% 47% 1,356 35% 48% 0.65
Owns mobile phone 1,272 77% 42% 1,356 74% 44% 0.07
Bought inorganic fertilizer in the first 
season of 2021 1,272 8% 27% 1,357 7% 26% 0.52
Sold any crop in the first season of 2021 1,272 55% 50% 1,357 52% 50% 0.16
Owns chicken 1,272 75% 43% 1,357 72% 45% 0.15
Employed in agricultural daily labor 1,272 60% 49% 1,357 60% 49% 0.92
Household engages in any business 1,272 31% 46% 1,357 30% 46% 0.53
Joint Test P-Value: 0.31
Notes: *Refined Graduation in Trial A includes both households who are receiving IPT-G and households who are not. Trial A is 
not designed to separately analyze the effects of the activity with and without IPT-G and the balance table is aligned with the 
analysis that will be carried out using follow-up data.
For Trial B, Table 29 compares the two treatment groups separately. Recall that there is no control 
group in this trial.
Baseline Survey Report of the RFSA Graduating to Resilience in Uganda, Cohort 2 (Vol. I)
Comparison of Experimental Groups 33
Table 29. Trial B: Comparison of refined graduation with and without IPT-G
(1) (2) (1)–(2)
Refined Graduation + IPT￾G Refined Graduation IPT-G 
vs. Not
N Mean SD N Mean SD P-Value
Age of household head 1,544 42.03 13.6 1,613 42.39 13.61 0.37
Number of children under five in the 
household 1,544 1.21 1.01 1,613 1.17 0.97 0.27
Living on less than $1.90 PPP/day 1,544 34% 48% 1,612 35% 48% 0.64
FCS score 1,544 49.33 13.51 1,613 48.75 13.78 0.32
HFIAS score 1,544 9.19 5.57 1,613 9.61 5.84 0.07
Owns radio 1,544 32% 47% 1,613 31% 46% 0.72
Owns mobile phone 1,544 74% 44% 1,613 75% 44% 0.65
Bought inorganic fertilizer in the first 
season of 2021 1,544 9% 28% 1,613 8% 27% 0.69
Sold any crop in the first season of 2021 1,544 43% 50% 1,613 43% 49% 0.66
Owns chicken 1,544 73% 45% 1,613 72% 45% 0.75
Employed in agricultural daily labor 1,544 62% 49% 1,613 64% 48% 0.22
Household engages in any business 1,544 37% 48% 1,613 36% 48% 0.58
Joint Test P-Value: 0.74
IMPEL | Implementer-Led Evaluation and Learning
34 Next Steps
6. NEXT STEPS
IPA will conduct a process evaluation on the quality of the implementation of Graduating to Resilience 
RFSA interventions and will conduct follow-up surveys to evaluate the impact of the interventions. 
6.1 Process Evaluation
IPA will conduct a process evaluation to understand the extent to which the interventions are 
implemented as planned. Findings from this process evaluation will be critical for interpreting the results 
of the impact evaluation. IPA will monitor RFSA implementation throughout the RFSA delivery period. 
IPA’s methodology for the process evaluation comprises a mix of site visits and observations, face-to￾face interviews, discussion groups, desk-based research, and review of existing reports and secondary 
data. The process evaluation will happen around July–August 2022.
6.2 Outcome Monitoring
IPA will conduct an outcome monitoring survey in two phases. The goal of Trial A outcome monitoring 
will be to evaluate the short-term impact of IPT-G sessions. The Trial A outcome monitoring will happen 
around September–October 2022, about a month after the end of the IPT-G sessions. The goal of Trial B 
outcome monitoring survey will be to evaluate the impact of the activity 6 months after the asset 
transfer. That survey will happen around December 2022.
6.3 Endline Survey for Cohort 2 Evaluation
IPA will conduct a follow-up survey 2 years after the beginning of the Cohort 2 activity to evaluate the 
impact of the Cohort 2 Graduating to Resilience activity. It will happen around mid-2024. It will be 
administered to approximately 4,500 households.
Table 30. Future Activities
Activity 2021 2022 2024
Cohort 2 baseline data collection X
Process monitoring X
Cohort 2 Outcome monitoring data collection X
Cohort 2 endline data collection X
Baseline Survey Report of the RFSA Graduating to Resilience in Uganda, Cohort 2 (Vol. I)
Appendix 35
APPENDIX
Additional Tables on Cohort 2 Design
Table 31. Appendix: Phase 1, Cohort 1, households found eligible by AVSI in 2018, prior to Cohort 1
implementation
Treatment Villages Control Villages
Baseline by IPA in 2018
Spillover control (1,100 /1.100)
T1 (1,100/1,100)
T2 (1,100/1,100)
T3 (1,100/1,100)
Holdout
700 / 1,900
Baseline
Pure control
1,100 /1,100
Holdout
4,100/5,600
Add Cohort 2 
Participants Equally 
Across T and C Villages
Add Cohort 2 
Participants Equally 
Across T and C Villages
In Phase 1 of the evaluation, IPA randomly assigned some eligible households to the holdout control 
group. These households were not surveyed at baseline in 2018 and were not enrolled in Cohort 1 
interventions. Cohort 2 treatment households needed to be added to Cohort 1 such that treatment 
villages and remaining control villages have the same shares of Cohort 2 participants, to preserve the 
Cohort 1 for the long-term follow-up (LTFU) of.
Table 32. Appendix: Phase 2, household level randomization
Cohort 1 Treatment Villages LTFU Cohort 1 Control Villages
Baseline by IPA in 2018
Remaining Spillover-C:
550 (ref.)/550 (host)
Spillover-C→Coh-2 
(assessed and randomly 
selected)
T1 
T2 
T3 
Holdout
→ All Coh-2 assessed and 
selected
Baseline
Remaining Pure-C: 183 
(r)/474 (h)
Pure-C→Coh-2 (assessed 
and randomly selected)
Holdout
→ PARTLY Coh-2 
assessment, calibrated 
to match share of Coh-2 
targets in Cohort 1 
treatment Villages
In Phase 2 of the evaluation, AVSI conducted an eligibility assessment of households from Cohort 1 that 
had not received Cohort 1 interventions, i.e., Cohort 1 control households (Pure Controls, Spillover 
Controls, and Holdout Controls). After the eligibility assessment, some eligible households were offered 
to participate in the activity, while others were not and remained in the control group. All study 
participants who were previously in Holdout Control in Cohort 1 were offered to participate in the 
activity. 
IMPEL | Implementer-Led Evaluation and Learning
36 Appendix
Table 33. Appendix: Trial A
Cohort 1 Treatment Villages LTFU Cohort 1 Control Villages
Remaining Spillover-C IF 
found Cohort 2 eligible
Vs.
SC→Coh-2 assessed and 
selected
Remaining Pure-C: IF 
found Coh-2 eligible
Vs.
PC→Coh-2 assessed and 
selected
IPA selected Trial A participants among former Cohort 1 participants, in spillover control group, and in 
pure control group. In each group (Spillover Control and Pure Control), some households were randomly 
selected to be offered the activity and some remained in control group.
Table 34. Appendix: Trial B
Dropped Cohort 1 Treatment Villages Large Host (Only!) Cohort 1 LTFU
Former Coh-1 Pure 
Control:
Former Coh-1 Hold-Out 50% Coh-2 with IPT-G
Vs.
50% Coh-2 without IPT-G
50% Coh-2 with IPT-G
Vs.
50% Coh-2 without IPT-G
In Phase 2, some Cohort 1 control villages were dropped from the long-term follow-up evaluation of 
Cohort 1. In these dropped villages, there will be no control group and all eligible households were offered 
to participate. Half of these households were offered to participate in the Graduation Approach with IPT￾G, and the other half were offered to participate in the same activity without IPT-G.
In the host community, Trial B also takes place in former treatment villages that had enough holdout 
control households to form at least two interventions groups. 
Baseline Survey Report of the RFSA Graduating to Resilience in Uganda, Cohort 2 (Vol. I)
Appendix 37
Additional Figures and Tables
Figure 2. Appendix: Binscatter plot of Negative HFIAS against FCS
IMPEL | Implementer-Led Evaluation and Learning
38 Appendix
Table 35. Appendix: Correlations between HFIAS and FCS components
FCS components, number of days for which food type was consumed out of past 7 days
HFIAS components, binary indicators: In 
the past four weeks… 
Main 
staples
Nuts 
and 
pulses
Vegetables
Meat, 
fish, and 
animal 
products
Fruits Milk 
products Fats/ Oil Sugar Spices/ 
Condiments
Did you worry that your household would 
not have enough food? -0.02 0.03 -0.01 -0.20 0.01 -0.09 -0.21 -0.15 -0.02
Were you or any household member not 
able to eat the kinds of foods you 
preferred because of a lack of resources?
-0.01 0.05 -0.05 -0.19 -0.03 -0.15 -0.16 -0.15 0.00
Did you or any household member have to 
eat a limited variety of foods due to a lack 
of resources?
-0.03 0.01 -0.06 -0.16 -0.07 -0.12 -0.12 -0.13 -0.01
Did you or any household member have to 
eat some foods that you really did not 
want to eat because of a lack of resources 
to obtain other types of food?
-0.01 0.07 -0.04 -0.22 -0.03 -0.13 -0.17 -0.16 -0.01
Did you or any household member have to 
eat a smaller meal than you felt you 
needed because there was not enough 
food?
-0.02 -0.02 -0.06 -0.11 -0.12 -0.13 -0.07 -0.14 -0.02
Did you or any household member have to 
eat fewer meals in a day because there 
was not enough food?
-0.03 -0.05 -0.06 -0.13 -0.11 -0.12 -0.10 -0.15 -0.05
Was there ever no food to eat of any kind 
in your household because of lack of 
resources to get food?
-0.05 -0.08 -0.04 -0.15 -0.08 -0.11 -0.19 -0.14 -0.08
Did you or any household member go to 
sleep at night hungry because there was 
not enough food?
-0.04 -0.01 -0.01 -0.12 -0.05 -0.09 -0.17 -0.12 -0.04
Baseline Survey Report of the RFSA Graduating to Resilience in Uganda, Cohort 2 (Vol. I)
Appendix 39
FCS components, number of days for which food type was consumed out of past 7 days
HFIAS components, binary indicators: In 
the past four weeks… 
Main 
staples
Nuts 
and 
pulses
Vegetables
Meat, 
fish, and 
animal 
products
Fruits Milk 
products Fats/ Oil Sugar Spices/ 
Condiments
Did you or any household member go to 
sleep at night hungry because there was 
not enough food?
-0.00 -0.01 0.03 -0.09 -0.03 -0.06 -0.12 -0.08 -0.10