Clean water, more schooling? Experimental evidence on solar water purifier distribution and child education in Tanzania

Last registered on August 20, 2026
View Trial History

Pre-Trial

Trial Information

General Information

Title
Clean water, more schooling? Experimental evidence on solar water purifier distribution and child education in Tanzania
RCT ID
AEARCTR-0019427
Initial registration date
August 17, 2026

Initial registration date is when the trial was registered.

It corresponds to when the registration was submitted to the Registry to be reviewed for publication.

First published
August 20, 2026, 9:20 AM EDT

First published corresponds to when the trial was first made public on the Registry after being reviewed.

Locations

There is information in this trial unavailable to the public. Use the button below to request access.

Request Information

Primary Investigator

Name
Christa Brunnschweiler
Affiliation
Norwegian University of Science and Technology (NTNU)
Contact Primary Investigator

Other Primary Investigator(s)

PI Name
Paolo Falco
PI Affiliation
University of Copenahgen
Contact Investigator
PI Name
Gustav Agneman
PI Affiliation
Norwegian University of Science and Technology (NTNU)
Contact Investigator
PI Name
Samwel Nassary
PI Affiliation
Norwegian University of Science and Technology (NTNU)
Contact Investigator
PI Name
Nathan Remcho
PI Affiliation
University of Gent
Contact Investigator

Additional Trial Information

Status
In development
Start date
2026-08-18
End date
2026-12-10
Keywords
Education, Gender
Additional Keywords
Safe water; health; time allocation; field experiment; Tanzania
JEL code(s)
I25, I15, J16, O15, O12, Q56, C93
Secondary IDs
Prior work
This trial does not extend or rely on any prior RCTs.
Abstract
Access to safe water and inclusive and equitable education remain two fundamental development challenges. In its 2026 update on the Sustainable Development goals, the United Nations reports that 2.1 billion people still lack safely managed drinking water, and 273 million children and young people remain out of school. The latter problem is particularly stark in sub-Saharan Africa, which has the lowest school completion rates in the world.

Our main question for this study is whether access to safe drinking water at home improves children’s schooling. Our study area lies in Uyui district in the region of Tabora in Tanzania, which is characterized by very poor access to clean water and the country’s lowest school enrollment and literacy rates. Within a household-level randomized trial in six villages of Uyui District, 900 households are assigned to one of three arms: free provision of a Solvatten solar water-treatment canister by World Vision Tanzania (360 households); an unconditional cash transfer calibrated to the local cost of obtaining safe drinking water or the local value of the Solvatten canister, including the value of collection time at prevailing local wages (180 households); or control (360 households). The cash arm holds the resource value of the technology roughly constant, so the device-versus-cash contrast isolates the time-savings channel from the pure income channel. A baseline survey will precede the intervention, followed by an endline study roughly 11 weeks after intervention. Only households with children aged 7-17 will be included in this study. The design means that we can capture within-household and age-specific short-run schooling effects of clean water provision at home.

Our primary outcomes capture survey-based school attendance and absenteeism at primary and lower secondary levels, as well as equal schooling opportunity for children of genders in relevant households. These are supplemented by secondary outcomes measuring schooling attitudes and social norms elicited from both parents. We also examine the mechanisms through which safe water supply can affect schooling outcomes, namely health effects, time use of children, and education expenditure of households.


External Link(s)

Registration Citation

Citation
Agneman, Gustav et al. 2026. "Clean water, more schooling? Experimental evidence on solar water purifier distribution and child education in Tanzania ." AEA RCT Registry. August 20. https://doi.org/10.1257/rct.19427-1.0
Sponsors & Partners
Experimental Details

Interventions

Intervention(s)
Arm 1 — Solvatten device (T), 360 households. A free Solvatten household solar water-treatment canister, distributed by World Vision Tanzania, with a short, standardized demonstration of use, cleaning and storage at handover. No cash, no employment information and no gender-related messaging accompanies delivery.
Arm 2 — Cash (M), 180 households. an unconditional cash transfer calibrated to the local cost of obtaining safe drinking water or the local value of the Solvatten canister, including the value of collection time at prevailing local wages. Delivered with the same number of enumerator contacts as the device arm. This arm holds the resource relief constant, so device-minus-cash identifies the time-savings channel net of income.
Arm 3 — Control (C), 360 households. Survey and economic games only. No device and no transfer.
Intervention Start Date
2026-09-05
Intervention End Date
2026-12-10

Primary Outcomes

Primary Outcomes (end points)
P1. Primary school attendance of relevant school-age children (measured by attendance dummy and school days missed in past four weeks).
P2. Lower secondary school attendance of relevant school-age children (measured by attendance dummy and school days missed in past four weeks).
P3. Equal schooling opportunity. Dummy variable for number of school-age daughters attending primary and/or lower secondary school equal to or greater than number school-age sons.
Primary Outcomes (explanation)
Primary school attendance (P1) is the main outcome of interest, as primary school is the only compulsory level of schooling in Tanzania. Lower secondary attendance (P2) is the second outcome of interest. P1 and P2 will also be measured by gender and pooled for total school days missed. P1-P2 will be elicited through survey questions of the mother at baseline and endline. P3 is calculated from responses to P1-P2.

Secondary Outcomes

Secondary Outcomes (end points)
S.1 Primary schooling attitude: Gender prioritization for primary schooling, based on survey question ‘Who do you think should be sent to primary school in a family?’
S.2 Lower secondary schooling attitude: Gender prioritization for lower secondary schooling, based on survey question ‘Who do you think should be sent to lower secondary school in a family?’
S.3 Social norms for primary schooling: Social norm elicitation for gender prioritization for primary schooling, based on survey question ‘Thinking of other people in your local community, who do you believe they would send to primary school?’
S.4 Social norms for lower secondary schooling: Social norm elicitation for gender prioritization for lower secondary schooling, based on survey question ‘Thinking of other people in your local community, who do you believe they would send to lower secondary school?’
S5. Water and health (mechanism): water-borne disease related episodes in the last two weeks for school-age children, by gender.
S6. Time use of child (mechanism): average minutes per day of school-age children on homework, water collection, fuel gathering.
S7. Income effects (mechanism): money spent on children’s education.
Secondary Outcomes (explanation)
S1-S4 will be elicited by both spouses at endline; S5-S6 will be elicited through survey questions of the mother at baseline and endline; S7 will be elicited by one spouse at baseline and endline.

Experimental Design

Experimental Design
A three-arm, household-level randomised controlled trial in six villages of Uyui District, Tabora Region — Kalangale, Kizengi, Lutona, Misole, Mwakadala and Nyankombe — with a universal elicitation module at endline.
The six villages were selected because they contain unimproved dug wells without treatment and broken or missing pumps, so publicly provided safe water is limited and roughly comparable ex ante. The sampling frame was built from building footprints inside village boundaries; 150 main-sample households were drawn per village, giving 900. Enumerators screen each geo-located building for an eligible couple and interview two adults per household, yielding approximately 1,800 respondents per round. Kizengi is a pure-control village receiving no devices, which bounds spillovers.
The endline round will happen roughly eleven weeks after distribution. Note that our sample will constitute a subset of the sample of the parent RCT. The subset will include only those households with at least one child aged 7-17, i.e. eligible to attend primary (7-13 years) or lower secondary (13-17 years) school. According to the Sixth Census of Tanzania of 2022, the average household size in Uyui is 6.2, which translates to 3-4 children per household. With a conservative estimate of 2 school-age children, this yields expected 1800 (2*900) children in our sample.
Experimental Design Details
Not available
Randomization Method
Assignment is at the household level, stratified by village, with a fixed recorded seed, performed in R before any field activity. Kizengi receives no devices; cash and control households there are allocated in the same proportions as elsewhere, scaled to fill the 150-household quota. The do-file, seed and allocation list are archived with the registry entry.
Randomization Unit
The household, stratified by village.
Was the treatment clustered?
Yes

Experiment Characteristics

Sample size: planned number of clusters
900 households
Sample size: planned number of observations
900 households, 1800 school-age children
Sample size (or number of clusters) by treatment arms
360 households device treatment (720 children); 360 control (720 children); 180 monetary treatment (360 children).
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
The trial will randomize 900 households, with 360 households allocated to the intervention arm and 540 to the pooled control + monetary treatment arms. We assume that we will have an average of two school-age children per household, yielding a total of 1,800 respondents. The primary outcome is the number of days of school missed during the preceding four weeks, assumed to have a standard deviation of 3.5 days (overall average 1-2 days/month in Tanzania). Randomization occurs at the household level, and the intra-household correlation coefficient (ICC) is assumed to be 0.30. With two respondents per household, the variance of the household mean is 3.5^2[0.30+(1−0.30)/2]=7.9625. Under a two-sided 5% significance level and 80% power, the proposed 900-household sample provides a minimum detectable treatment effect (MDE) of approximately 0.54 days of school missed. This corresponds to a standardized effect size of approximately 0.15 standard deviations. No allowance for attrition is included. ICC sensitivity: With ICC=0.5, we get MDE=0.58 (SD=0.17); with ICC=0.1, we get MDE=0.49 (SD=0.14).
Supporting Documents and Materials
IRB

Institutional Review Boards (IRBs)

IRB Name
Research Ethics Committee at the Department of Economics at University of Copenhagen
IRB Approval Date
2026-03-09
IRB Approval Number
N/A
Analysis Plan