| Organization | Evidence Action |
| Location | Uttar Pradesh (UP) and Bihar, India |
| Amount | $369,038 |
| Approval date | October 2025 |
| Funded by | Unrestricted funds designated for grantmaking |
Summary
This grant will fund IDinsight (through Evidence Action) to implement and analyze a baseline survey measuring how many children aged 6 to 59 months in Uttar Pradesh (UP) and Bihar, India, receive iron and folic acid (IFA) supplements. The survey will serve as the basis for the monitoring and evaluation of a multi-year grant we recommended to help Evidence Action expand its technical support for the government's IFA program in these two states.
Published: September 2026
The challenge this grant addresses
Iron deficiency is the most common cause of anemia, a condition in which the amount of hemoglobin, a protein in red blood cells that helps transport oxygen, is diminished.1 People with anemia may feel tired, weak, and short of breath, or may experience other negative physical and developmental effects.2 Iron deficiency is common among children in low- and middle-income countries; for example, the Global Burden of Disease study estimates that 58% of children aged 6 to 59 months in India are anemic.3
In August 2025, we approved a $7.8 million, five-year grant to support the expansion of Evidence Action's technical assistance for the Indian government's IFA supplementation program for children in UP and Bihar. The program aims to reach children aged 6-59 months old, and school-attending children 5-19 years old.4 We estimate that children aged 6 to 59 months account for roughly two thirds of that program's expected impact,5 but we have limited evidence about whether the program increases the IFA coverage of that group.
For example, Evidence Action’s previous surveys of its IFA supplementation program in India sampled children aged 6 to 59 months from the registration lists of anganwadi centers (AWCs),6 which are rural health and child care centers in India,7 but IDinsight estimates that less than half of young children in UP and Bihar are registered at the centers.8 In addition, past surveys were not fully independent of Evidence Action;9 we believe that despite the best efforts of our grantee partners, there are many ways monitoring and evaluation processes conducted by implementing organizations can be biased.10 As a result, we are uncertain about the baseline coverage of IFA for a group that we estimate makes up the majority of the program's impact.
How this grant could help
This grant funds IDinsight to implement and analyze a baseline coverage survey for children aged 6 to 59 months in UP and Bihar prior to the start of Evidence Action’s technical assistance to the government in those states.11 The grant will cover staff, data collection, and overhead costs.12
At the time we made this grant, IDinsight proposed that the survey measure IFA supplementation coverage among young children using a dual-frame sampling methodology that combined AWC registration lists with population-representative household sampling (based on voter rolls or GPS-based household selection).13 Statistical weights would then be applied so the final estimate reflected all children 6 to 59 months, not only those registered at an AWC.14 As of August 2026, IDinsight and GiveWell agreed to instead use only GPS-based household selection, dropping the selection from AWC registration lists, to simplify survey logistics while maintaining comparable technical rigor.15 The survey will also collect process indicators from frontline health workers at AWCs.16 Overall, IDinsight will survey 650 AWCs, 650 frontline health workers, and 2,600 households across both states.17
This survey would establish the baseline for a future endline survey—which we expect to fund through a separate grant and to conduct around 2028—to assess the impact of Evidence Action’s work on IFA coverage. When we originally made this grant, we expected data collection to take place in late 2025. As of August 2026, surveys in Bihar had taken place, but delays in launching the program in UP mean surveys there have not yet been implemented. We are uncertain whether the UP surveys will happen, but are open to picking them back up in the future.18
Our reasoning
GiveWell has made more than $27 million in grants for IFA programs19 . The majority of the program benefits accrue to children aged 6 to 59 months, but we have limited data indicating what proportion of young children receive supplements. For example, we previously made a grant to Evidence Action to conduct baseline surveys of IFA supplementation in UP and Bihar, but we later learned that the coverage information for children aged 6 to 59 months may not have been representative since only children from AWC registration lists were sampled.20
As a result, we think an independent, representative baseline measurement is a useful next step for improving our coverage estimates for children aged 6 to 59 months in UP and Bihar, and ultimately, for our cost-effectiveness estimates of Evidence Action’s technical assistance to the Indian government in those states. Several factors increase our confidence in this grant:
- By adding a population-representative component, the survey is designed to capture children who are not registered at AWCs and were thus systematically missed by earlier surveys.
- Having IDinsight implement and analyze the survey increases independence and allows us to compare findings against Evidence Action's own monitoring.
- We have a positive track record of working with IDinsight,21 and it has analyzed coverage data for this program before.
- We are funding the baseline survey now and expect to fund the endline separately. This preserves flexibility on the cost and structure of the later survey.
Creating cost-effectiveness estimates for value-of-information grants is more challenging than for direct delivery programs; we did not produce one for this grant and are recommending it primarily because we expect it to provide useful information about the much larger technical assistance program and potentially other IFA programs in the future.
What we could be wrong about
Our main uncertainties are:
- The 2024 coverage survey we funded found very low coverage among children aged 6 to 59 months.22 We considered skipping this additional baseline and comparing the eventual endline against a rough assumption that coverage among children aged 6 to 59 months was very low. If this new baseline confirms that assumption, then in hindsight that rough assumption may have been enough to learn about the effectiveness of Evidence Action’s technical assistance among children in that age range, and we might not have had to fund this new baseline.
- We are uncertain whether we should have also funded an additional baseline survey for school-aged children in UP. The 2024 baseline for school-aged children covered 50 districts in UP, but Evidence Action subsequently proposed to implement its technical assistance program in 27 districts to avoid overlapping with other implementers.23 We expect the eventual endline survey to cover those 27 districts, which means there will be a relatively wide margin of error in the analysis of Evidence Action’s technical assistance for school-aged children in UP.24 We ultimately chose not to fund an additional baseline survey for school-aged children in UP due to timeline constraints, but we may have overestimated the risk adding this baseline to the grant posed to the timeline.
This page summarizes our research at the time of approval and was reviewed by the grant recipient.
Sources
- 1"Chronic iron deficiency frequently turns into iron-deficiency anaemia. While iron deficiency is the most common cause of anaemia, other causes such as acute and chronic infections that cause inflammation; deficiencies of folate and of vitamins B2 , B12 , A, and C; and genetically inherited traits such as thalassaemia and drepanocytosis (sickle-cell anaemia) may be independent or superimposed causal factors (WHO 2001; WHO 2015a)." Peña-Rosas et al. 2015, p. 7.
"Anaemia is a condition in which the number of red blood cells or the haemoglobin concentration within them is lower than normal. It mainly affects women and children. Anaemia occurs when there isn’t enough haemoglobin in the body to carry oxygen to the organs and tissues." World Health Organization, Anaemia fact sheet, 2025. - 2"Anaemia causes symptoms such as fatigue, reduced physical work capacity, and shortness of breath. . . . The consequences of anaemia can vary. It can affect school performance (through developmental delays and behavioural disturbances such as decreased motor activity, social interaction and attention to tasks), productivity in adult life and overall quality of life in general." World Health Organization, "Anaemia fact sheet", 2025.
- 3
- "The WHO Regions of Africa and South-East Asia are most affected with an estimated 106 million women and 103 million children affected by anaemia in Africa and 244 million women and 83 million children affected in South-East Asia." World Health Organization, "Anaemia fact sheet", 2025.
- Our World in Data reports anemia rates among children as 59%, 53%, and 23% for low-, lower-middle, and upper-middle income countries in 2019. Our World in Data, "Share of the population with anemia in higher risk groups, 2019".
- See also the Global Burden of Disease search results here (requires login). Institute for Health Metrics and Evaluation, GBD Results, Anemia prevalence in India in 2023 in all children under 5: Value 58.56 (accessed August 3, 2026).
- 4
- See the page for our 2022 grant to Evidence Action to support IFA supplementation in 5 states in India: ". . . [T]hese states plan to target . . . children, ages 6 months to 19 years old, for IFA supplementation. This includes: children 6-59 months old (reached through ASHAs and AWCs), children 5-19 years old who attend schools (reached through schools), and out-of-school girls 10-19 years old (reached through AWCs). Evidence Action provides technical assistance for IFA supplementation at the state and district levels." GiveWell, "Evidence Action — Iron and Folic Acid (IFA) Supplementation in India (August 2022)", 2024, "Evidence Action-IFA Activities" section.
- Note that for the 2025 grant for IFA supplementation in UP and Bihar, GiveWell agreed with Evidence Action to drop out-of-school girls from the target population as they make up a relatively small proportion of the overall population. Evidence Action, Comments on a draft of this page, September 4, 2026 (unpublished)
- 5At the time we made the grant to Evidence Action to expand its IFA program in India to Bihar and UP, we estimated that the program would be roughly 18.6 times more cost-effective than our benchmark. As a sense check, we tested out a scenario where Evidence Action’s program would have impact among school-age children but no impact at all on IFA coverage in children age 6 to 59 months. This dropped the program’s expected cost-effectiveness to 5.7x our benchmark. 5.7 / 18.6 = 30.6%, or 1 - 0.306 = 0.694 (69.4%) of the impact of the grant comes from its effects on children aged 6 to 59 months. GiveWell, "Evidence Action IFA TA renewal and expansion grant investigation," June 2025 (unpublished).
- 6"Yes, our endline coverage survey selected children based on AWC registration lists for endline survey results. This means any children not registered with an AWC would be excluded from our coverage survey but that we include registered children who do not attend/do not attend regularly. This could lead to some upward bias in our estimates, assuming that unregistered children are less likely to receive IFA through AWCs. We followed this approach for both our baseline and endline surveys." Eliza Walwyn-Jones, Evidence Action, email sent to GiveWell, June 6, 2025 (unpublished)
- 7"The Integrated Child Development Services (ICDS) scheme is the largest program for promotion of maternal and child health (MCH) and nutrition in India. The scheme was launched in 1975 in pursuance of the National Policy for Children. The beneficiaries are children less than 6 years, adolescent girls, pregnant and lactating women, and women in the age group of 15–44 years.[1] The package of six services under the scheme includes supplementary nutrition, pre-school non-formal education, immunization, nutrition and health education, health checkup, and referral services. These services are made available at Anganwadi centers (AWC) and are delivered by Anganwadi workers (AWWs) at grassroots level." Sengar, Kunwar, and Mishra 2022
- 8"This PIB Report from 2021 states that in Uttar Pradesh, 11.58 million children and in Bihar, 7.95 million children between 0-6 years are registered with Anganwadi Centers. If we use the number of rural child population from the 2011 census as denominator.
- In Uttar Pradesh, about 46.3% of rural children aged 0–6 years are registered at Anganwadi centres (25.04 million rural child population – Census 2011 as denominator).
- In Bihar, about 45.7% of rural children aged 0–6 years are registered at Anganwadi centres (17.38 million rural child population – Census 2011 as denominator)."
IDinsight, email to GiveWell, August 29, 2025 (unpublished), citing "Anganwadi centres, beneficiaries and funds released under the scheme," Press Information Bureau Delhi, December 1, 2021. We have not vetted this assessment.
- 9Previous coverage surveys were carried out by Evidence Action with data analysis by IDinsight. For example, the budget for our 2022 grant to Evidence Action for IFA in India included "Four coverage validation surveys, which will estimate the IFA coverage in Evidence Action-supported states in 2022 to 2025: $899,280; Advisory services from IDinsight, to advise on and analyze results of coverage surveys and monitor data tracking Evidence Action’s key levers of change: $206,832." GiveWell, "Evidence Action — Iron and Folic Acid (IFA) Supplementation in India (August 2022)," 2024.
- 10For example, with Evidence Action’s Dispensers for Safe Water program, we found that independent monitoring came to substantially different conclusions than monitoring conducted by Evidence Action. This provides a general case for funding more independent monitoring for the programs we support. See GiveWell, "Podcast Episode 25: Following the Data on Dispensers for Safe Water," 2026.
- 11"IDinsight will collect two rounds of data - both pre-program and post-program. The survey instrument will be designed in consultation with EA to track key outcomes of interest - coverage for government schoolchildren and pre-school aged children, and process indicators on program administration... The sample will be stratified across the 27 districts in UP and 38 districts in Bihar that are targeted to receive EA's TA." IDinsight, "Evaluation and Process Monitoring for Evidence Action’s Technical Assistance in UP and Bihar," August 2025, pp. 2-3. Note that we ultimately chose to fund neither a) a baseline survey of school-aged children nor b) an endline survey with this grant because of timeline constraints. We expect to fund a separate endline in the future.
- 12IDinsight, "Baseline coverage surveys for children under 5 for Evidence Action's IFA TA in UP and Bihar budget", 2025 (unpublished).
- 13"The sampling design for frontline functionaries and preschool–aged children will replicate the EA CES UP-Bihar survey methodology, with increased sample sizes at both the block and Anganwadi Centre (AWC). However, the sampling frame used for parents of pre-school aged children will differ from the previous approach. Instead of only relying on AWC enrollment lists to sample children, IDinsight proposes to use a dual frame sampling strategy. The AWC enrollment list will be complemented by a population representative sampling frame to include pre-school-aged children who are not enrolled in any AWC, in a ratio of 3 to 1. The population representative sampling frame will be based on either voter rolls or random pin-drops, to randomly select households within the village or ward linked to the randomly selected AWCs. The identified households will then be screened for the presence of a pre-school-aged child." IDinsight, "Evaluation and Process Monitoring for Evidence Action’s Technical Assistance in UP and Bihar," August 2025, p. 4.
- 14
- "The AWC enrollment list will be complemented by a population representative sampling frame to include pre-school-aged children who are not enrolled in any AWC, in a ratio of 3 to 1." IDinsight, "Evaluation and Process Monitoring for Evidence Action’s Technical Assistance in UP and Bihar," August 2025, p. 4.
- For more on multiple-frame sampling and the statistical weighting method, see IDinsight: "The magic of multiple-frame sampling: an introduction," 2020.
- 15Evidence Action, Comments on a draft of this page,September 4, 2026 (unpublished)
- 16
- "IDinsight will select approximately ten process indicators based on their importance to EAʼs theory of change, GiveWellʼs priorities, and the feasibility of data collection. After each data collection round, IDinsight will analyze the data and share the results with GiveWell and EA. This iterative process will enable continuous observation of changes in these key indicators over time." IDinsight, "Evaluation and Process Monitoring for Evidence Action’s Technical Assistance in UP and Bihar," August 2025, pp. 5-6.
- Examples of potential process indicators include the percent of frontline health workers who currently have a stock of IFA syrup and frontline health workers’ knowledge about the frequency of IFA syrup administration.
- 17"Sample sizes for AWC surveys:... 650 AWCs…650 FLFs will be surveyed…2600 households." IDinsight, "Evaluation and Process Monitoring for Evidence Action’s Technical Assistance in UP and Bihar," August 2025, p. 4.
- 18Evidence Action, Comments on a draft of this page, September 4, 2026 (unpublished)
- 19See our list of all iron and folic acid (IFA) supplementation grants we have made.
- 20"Yes, our endline coverage survey selected children based on AWC registration lists for endline survey results. This means any children not registered with an AWC would be excluded from our coverage survey but that we include registered children who do not attend/do not attend regularly. This could lead to some upward bias in our estimates, assuming that unregistered children are less likely to receive IFA through AWCs. We followed this approach for both our baseline and endline surveys." Eliza Walwyn-Jones, Evidence Action, email to GiveWell, June 6, 2025 (unpublished)
- 21"IDinsight is a strong learning partner." GiveWell, "IDinsight — Updated Beneficiary Preference Research and Pilots (2025)", 2025. GiveWell has approved 19 grants to IDinsight since 2014 for research and data analysis.
- 22Evidence Action, "2025 UP-Bihar Baseline Report", 2025 (unpublished)
- 23"In Uttar Pradesh, we have reduced the number of districts we plan to target from 50 to 27 districts. This is in response to Nutrition International’s (NI) scale up in 20 additional districts this year." Evidence Action, "Uttar Pradesh/Bihar - IFA Expansion Inputs", p. 3. June 25, 2025 (redacted)
- 24"The reduction from 50 to 27 districts in UP means only about 50% of the 2024 sample remains applicable. Using the truncated sample would significantly increase margins of error." GiveWell, Summary of a conversation with IDinsight, August 26, 2025 (unpublished)