Please fill out this short user survey of only 3 questions in order to help us improve the site. We appreciate your feedback!
Back to History Current Version
Communal Sanitation Solutions for Urban Slums
Last registered on December 22, 2017


Trial Information
General Information
Communal Sanitation Solutions for Urban Slums
Initial registration date
December 19, 2017
Last updated
December 22, 2017 11:19 AM EST
Primary Investigator
CESS Nuffield - FLAME University
Other Primary Investigator(s)
PI Affiliation
Yale University Department of Economics
PI Affiliation
Yale School of Management
Additional Trial Information
On going
Start date
End date
Secondary IDs
This project aims to create a scalable model of shared toilet infrastructure and management that can be replicated across cities to ultimately reduce open defecation and improve health among the urban poor. In doing so, this project aims to address the design, management and operational challenges of shared toilet facilities commonly found in and around urban slums of South Asia: shared toilets. We define these as toilet blocks serving a fixed population, or in other words: shared toilet facilities for residential areas. Our study location and government partners are the two largest cities in the eastern Indian state of Orissa: Bhubaneswar and Cuttack.
External Link(s)
Registration Citation
Barnhardt, Sharon, Judy Chevalier and Ahmed Mushfiq. 2017. "Communal Sanitation Solutions for Urban Slums." AEA RCT Registry. December 22. https://doi.org/10.1257/rct.2101-1.0.
Former Citation
Barnhardt, Sharon et al. 2017. "Communal Sanitation Solutions for Urban Slums." AEA RCT Registry. December 22. http://www.socialscienceregistry.org/trials/2101/history/24380.
Experimental Details
The study has three types of interventions: demand generation, initial free pricing, and habit formation.
Intervention Start Date
Intervention End Date
Primary Outcomes
Primary Outcomes (end points)
Usage of community toilets
Primary Outcomes (explanation)
Secondary Outcomes
Secondary Outcomes (end points)
Secondary Outcomes (explanation)
Experimental Design
Experimental Design
The overall objective of this project is to develop a replicable model for urban, shared toilets and provide rigorous evidence on its ability to significantly reduce open defecation.
Experimental Design Details
Not available
Randomization Method
Note: A slum cluster is a CATCHMENT AREA and may contain two facilities in such close proximity that they must be randomized to the same treatment.
For household-level randomizations (vouchers and reminders) we first use STATA to determine the number of households in each slum cluster that should get each combination of interventions (which are cross randomized), then we use a public lottery in each slum to assign those chits to households.
Randomizations at the slum-cluster and slum-cluster-gender level are done in our office using STATA to assign our demand generation and reward interventions.
Randomization Unit
We have three randomization units:
1. Slum-Clusters for Demand Generation (A slum cluster may contain two facilities in such close proximity that they must be randomized to the same treatment.)
2. Slum-Cluster-Gender for the Reward (hand faucet)
3. Household Level for Vouchers and Reminders
Was the treatment clustered?
Experiment Characteristics
Sample size: planned number of clusters
Planned Slum Clusters for Demand Generation Randomization: 52
Planned Slum Clusters where we build toilet facilities: 52 (a total of 55 facilities being built)
Planned Households under study if 55 sites are built
2200 - if there are on average 40 households per site (55 * 40 = 2200)
2750 - if there are on average 50 households per site (55 * 50 = 2700)
3300 - if there are on average 60 households per site (55 * 60 = 3300)

Sample size: planned number of observations
Planned individuals under study if 55 sites are built: 6600 (if there are 2200 households) 8250 (if there are 2750 households) 9900 (if there are 3300 households)
Sample size (or number of clusters) by treatment arms
For the 55 catchment areas in the demand generation trial, we have 27 getting a demand generation program and 25 not getting it, as that is the maximum number available across the two cities.

If we are able to build all 104 gender sides (52 male sides and 52 female sides) of toilets --
For the reward trial, we plan to have:
a. 26 that get a reward on the female side of the facility
b. 26 that get a reward on the male side of the facility
The result is that some facilities will receive rewards on both sides, some on the male or female side and some will not receive rewards at all.

Our interventions are stratified by slum, and start in batches. For e.g., as soon as a batch of toilet facilities is constructed, we launch our interventions in those slums. We don't know the exact sample size by treatment arms until we actually conduct our in-field randomization at each slum. If we get all 3000 households we would have approximately 500 in each treatment arm:
a. No voucher + Calendar
b. Free Voucher good every day for 45 days + Calendar
c. Free Voucher good MWF for 45 days + Calendar
d. No voucher + No Calendar
e. Free Voucher good every day for 45 days + No Calendar
f. Free Voucher good MWF for 45 days + No Calendar
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
IRB Name
Yale Human Subjects Committee
IRB Approval Date
IRB Approval Number
Analysis Plan
Analysis Plan Documents