Promoting water purification: The roles of learning, habit formation, and social norm

Last registered on September 12, 2023

Pre-Trial

Trial Information

General Information

Title
Promoting water purification: The roles of learning, habit formation, and social norm
RCT ID
AEARCTR-0003673
Initial registration date
December 14, 2018

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
December 20, 2018, 9:39 PM EST

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

Last updated
September 12, 2023, 6:39 AM EDT

Last updated is the most recent time when changes to the trial's registration were published.

Locations

Region

Primary Investigator

Affiliation
Lahore University of Management Sciences

Other Primary Investigator(s)

PI Affiliation
Harvard Business School
PI Affiliation
Uppsala University
PI Affiliation
Harvard Kennedy School

Additional Trial Information

Status
On going
Start date
2022-05-05
End date
2023-12-31
Secondary IDs
Prior work
This trial does not extend or rely on any prior RCTs.
Abstract
Despite the potential to reduce diarrheal disease burden (Quick, et al, 2002), use of affordable point-of-use decontamination technologies such as chlorine tablets is low in many developing countries including Pakistan (0.3%, with 8% adopting any purification technology; Pakistan DHS, 2012-13). While price - even when low - can hinder adoption, free access induces only partial usage. For example, free delivery of chlorine solution in Kenya yielded a usage rate of only 34% (Dupas, et al, 2016). However, there is little evidence on ways to increase usage beyond this basic level induced by access. Akram & Mendelsohn (2017; hereafter, AM) explore an alternative hypothesis: If the households do not understand - or trust - expert opinion on benefits of usage and benefits are hard to measure, they may not be convinced of the returns to such a technology. In a pilot RCT, AM use household recordkeeping of children’s diarrhea incidence to help them learn the benefits of chlorinated drinking water (along with sharing information on the diarrhea rate from a comparable population not exposed to chlorine). Compared to a base policy of free access to chlorine tablets and expert advice on why to use them, this simple intervention increased chlorine use after one year by a remarkable 56 percentage points. It was highly cost-effective as well, with the marginal cost per DALY-averted at $495. We propose to build on this pilot by a) testing at a larger scale and b) investigating mechanisms, particularly those related to learning and social norms, to inform design choices for the next stage(s) of experimentation and eventually scale-up. The experimental design is further structured to test learning against habit formation: both processes require intense initial engagement and can yield sustained behavioral change, but economists know little about which is most effective for behavioral change (and technology adoption) broadly. Particularly in a space where returns are not obvious and existing information campaigns have largely failed, investigating which mental process is most operative regarding long term adoption of preventive health behaviors is crucial to policy design: do we, as policymakers, invest in subsidizing an activity repeatedly so people develop a habit, or do we invest in improving our information campaigns so individuals can better understand the returns to a behavior?
External Link(s)

Registration Citation

Citation
Akram, Agha et al. 2023. "Promoting water purification: The roles of learning, habit formation, and social norm." AEA RCT Registry. September 12. https://doi.org/10.1257/rct.3673-3.3
Former Citation
Akram, Agha et al. 2023. "Promoting water purification: The roles of learning, habit formation, and social norm." AEA RCT Registry. September 12. https://www.socialscienceregistry.org/trials/3673/history/192408
Experimental Details

Interventions

Intervention(s)
We propose a randomized controlled trial (RCT) across 1,800 households. We first describe the various components of our interventions, and then outline the treatment arms explicitly.

Monitoring: Biweekly health worker (HW) visits to: a) collect data on diarrhea incidence over the last 2 weeks; b) test chlorine in the water once per month; and c) offer free chlorine tablets (2-week’s supply) and expert advice on use. This mimics standard public health campaigns on chlorine distribution.

Info-tool: HW helps the caregiver record diarrheal incidence (unique episodes across children) and total diarrhea days (the sum of all diarrhea days across episodes and children), and creates a bar graph to visually represent the total diarrhea days experienced by the children in the last 2 weeks. At the end of each month (during the second biweekly visit in a month), the diarrhea days from that month are added up and colored into a month-level bar graph. This is a simple, visual paper-and-pencil tool, and AM (2017) and our recent pilots have demonstrated that low-literacy-numeracy caregivers are comfortable using it.

Benchmarks: HW shares information on the diarrhea days expected in households (over the last two weeks) that do not use chlorine (estimated using data from the experiment; more on this below); and

Habit formation with financial incentive: Caregivers are offered small daily rewards (tokens redeemable for child/household goods) if the caregiver can show empty chlorine tablet wrappers as proof of usage. Each daily reward for proper chlorine use is equal to approximately 5 US cents (with ‘proper use’ calibrated to household’s pre-intervention water consumption).

Our four experimental arms are as follows:

Comparison (C): Monitoring minus the offer of free chlorine tablets and chlorine-testing
Treatment 1 (T1): Monitoring
Treatment 2 (T2): Monitoring + Info-tool + Benchmark
Treatment 3 (T3): Monitoring + Habit formation with financial incentive

To determine the benchmark in T2, we will use the diarrhea results of C from the previous two weeks. HW will share this information during the bi-weekly visits starting from the second visit (using data from the first bi-weekly visits to C from the previous two weeks).

Comparisons of arms and corresponding research questions are:

T1 vs T2: What are the short and long-run effects of facilitating learning about health returns on tablet take-up and usage?
T2 vs T3: Which mechanism is more effective in generating both contemporaneous and sustained change in behavior - active learning (via Info-tool and benchmark) or habit formation (with temporary incentives)?
C vs T1: How do the effects above compare to the standard public health effort of free distribution alone?

The interventions will span months 1-6 of the experiment (Phase 1), during which HWs visit all households once every two weeks for data collection and intervention execution. Note that, for the three treatment arms, tablet distribution and testing of water (and hence incentives for T3) starts in the third month as we reserve the first two months to help the caregivers develop some pre-tablet-access record of diarrhea incidence in their households. For months 7-16 (Phase 2), HWs will visit all households once each month to observe changes in long-term behavior. In month 16, HWs will offer a final 3-month’s supply to all households (except C). An endline survey will be administered 1 month after this visit.
Intervention Start Date
2022-06-10
Intervention End Date
2022-11-30

Primary Outcomes

Primary Outcomes (end points)
Presence of residual chlorine in household drinking water; Acceptance of offered chlorine tablets; Diarrhea prevalence
Primary Outcomes (explanation)

Secondary Outcomes

Secondary Outcomes (end points)
Height-for-age Z-score; Weight-for-height Z-score
Secondary Outcomes (explanation)
Z-scores will be constructed using the WHO child growth standards (WHO 2007).

Experimental Design

Experimental Design
We propose a randomized controlled trial (RCT) across 1,800 households. We first describe the various components of our interventions, and then outline the treatment arms explicitly.

Monitoring: Biweekly health worker (HW) visits to: a) collect data on diarrhea incidence over the last 2 weeks; b) test chlorine in the water once per month; and c) offer free chlorine tablets (2-week’s supply) and expert advice on use. This mimics standard public health campaigns on chlorine distribution.

Info-tool: HW helps the caregiver record diarrheal incidence (unique episodes across children) and total diarrhea days (the sum of all diarrhea days across episodes and children), and creates a bar graph to visually represent the total diarrhea days experienced by the children in the last 2 weeks. At the end of each month (during the second biweekly visit in a month), the diarrhea days from that month are added up and colored into a month-level bar graph. This is a simple, visual paper-and-pencil tool, and AM (2017) and our recent pilots have demonstrated that low-literacy-numeracy caregivers are comfortable using it.

Benchmarks: HW shares information on the diarrhea days expected in households (over the last two weeks) that do not use chlorine (estimated using data from the experiment; more on this below); and

Habit formation with financial incentive: Caregivers are offered small daily rewards (tokens redeemable for child/household goods) if the caregiver can show empty chlorine tablet wrappers as proof of usage. Each daily reward for proper chlorine use is equal to approximately 5 US cents (with ‘proper use’ calibrated to household’s pre-intervention water consumption).

Our four experimental arms are as follows:

Comparison (C): Monitoring minus the offer of free chlorine tablets and chlorine-testing
Treatment 1 (T1): Monitoring
Treatment 2 (T2): Monitoring + Info-tool + Benchmark
Treatment 3 (T3): Monitoring + Habit formation with financial incentive

To determine the benchmark in T2, we will use the diarrhea results of C from the previous two weeks. HW will share this information during the bi-weekly visits starting from the second visit (using data from the first bi-weekly visits to C from the previous two weeks).

Comparisons of arms and corresponding research questions are:

T1 vs T2: What are the short and long-run effects of facilitating learning about health returns on tablet take-up and usage?
T2 vs T3: Which mechanism is more effective in generating both contemporaneous and sustained change in behavior - active learning (via Info-tool and benchmark) or habit formation (with temporary incentives)?
C vs T1: How do the effects above compare to the standard public health effort of free distribution alone?

The interventions will span months 1-6 of the experiment (Phase 1), during which HWs visit all households once every two weeks for data collection and intervention execution. Note that, for the three treatment arms, tablet distribution and testing of water (and hence incentives for T3) starts in the third month as we reserve the first two months to help the caregivers develop some pre-tablet-access record of diarrhea incidence in their households. For months 7-16 (Phase 2), HWs will visit all households once each month to observe changes in long-term behavior. In month 16, HWs will offer a final 3-month’s supply to all households (except C). An endline survey will be administered 1 month after this visit.
Experimental Design Details
Randomization Method
Randomization done in office
Randomization Unit
Household (with at least one child between the ages of 6 months and 5 years old at baseline)
Was the treatment clustered?
No

Experiment Characteristics

Sample size: planned number of clusters
1,800 Households
Sample size: planned number of observations
1,800 Households
Sample size (or number of clusters) by treatment arms
450 Households
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
Our census has identified 1,800 eligible households (those with at least 1 under-5 children), who will be individually randomized into the 4 arms, with each arm comprising 450 households. Given 10% attrition, 5% significance level and 80% power, for any pair-wise comparison of arms, this implies a minimum detectable effect size of 9.3 pp (in residual chlorine presence in drinking water; baseline mean, 29%, from our piloting data) and 0.2 SD (in diarrhea days; 76% of baseline mean in Hussam, et al, (2021)). For the anthropometric outcomes such as height-for-age or weight-for-height z-scores, note that many households (45% in our pilot sample) have two or more under-five children. Taking this into account, we estimate a minimum detectable effect size of 0.158 SD with the following underlying parameters: 720 under-five children per arm (using the pilot study estimate of the mean number of children per household), intra-household correlation in height-for-age z-score of 0.05 (from a recent study by Akram, Khan et al. from another informal settlement in the same city), as well as 10% attrition, 5% significance level, and 80% power.
IRB

Institutional Review Boards (IRBs)

IRB Name
Interactive Research & Development (IRD) IRB
IRB Approval Date
2020-03-04
IRB Approval Number
IRD_IRB_2019_12_009
Analysis Plan

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

Request Information

Post-Trial

Post Trial Information

Study Withdrawal

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

Request Information

Intervention

Is the intervention completed?
No
Data Collection Complete
Data Publication

Data Publication

Is public data available?
No

Program Files

Program Files
Reports, Papers & Other Materials

Relevant Paper(s)

Reports & Other Materials