The American Economic Association's registry for randomized controlled trials
Creating a Toilet Habit
Last registered on October 02, 2015
View Trial History
Creating a Toilet Habit
Initial registration date
October 02, 2015
October 02, 2015 1:28 PM EDT
Contact Primary Investigator
Other Primary Investigator(s)
Berhe Mekonnen Beyene
Innovations for Poverty Action
Johann Caro Burnett
Yale University, the Department of Economics
Judith A. Chevalier
Yale School of Mangement
Additional Trial Information
Public health externalities from unhygienic sanitation remain a significant development challenge, even in areas where hygienic latrines are accessible or affordable. Our partner, Sanergy has created a network of hygienic latrines in Nairobi. However, as with many other technical solutions to a range of development problems, widespread adoption of this “clean toilet” option has been a challenge. Barriers such as distance, cost, long queues, or lack of understanding of health risks may preclude individuals from choosing the sanitation option that may be better from a public health point of view. We also hypothesize that behaviors like open defecation may persist because they represent ingrained habits that are difficult to change. Inspired by findings from psychology and neuroscience, we propose field experiments that are designed to instill a revised habit of community toilet use among the slum population of Nairobi. Habit loops have been successfully created by private sector firms to increase demand for many household products and behaviors such as brushing regularly with Pepsodent toothpaste, or spraying Febreze air freshener. We propose to create such a loop for Sanergy toilets using a combination of economic incentives and a marketing campaign that is attentive to psychological cues and rewards. The experiments are designed to separate habit formation from other closely related models of risk aversion and learning.
Beyene, Berhe et al. 2015. "Creating a Toilet Habit." AEA RCT Registry. October 02.
Sponsors & Partners
Intervention Start Date
Intervention End Date
Primary Outcomes (end points)
Main (primary) outcome variables: We will have the following main outcome variables which are related to toilet usage during the post-discount period.
1. Total toilet usage: This will be the number of times the toilet is used in a particular period of time, say a month.
2. Toilet usage during the preferred two-hour window. This can be measured in different ways: (a) the absolute number of usages during the
two hour window, or (b) usages during the two hour window as a percentage of total usage in a particular period of time.
3. Use of other substitute strategies, such as open defecation or flying toilets.
Secondary outcome variables: In addition to toilet usage, we will also gather information about the following two additional outcome variables:
1. Health outcome: Self-reported diarrhea incidence. Questions related to diarrhea will be asked both at the baseline and endline surveys.
2. Happiness and related variables: We will also look at effects of interventions on subjective measures of well-being, such as happiness and
Primary Outcomes (explanation)
Secondary Outcomes (end points)
Secondary Outcomes (explanation)
We will randomize on the following three cross-cutting dimensions:
Discounts: There will be four different discount types
Marketing: There will be marketing and non-marketing groups. The marketing group will get individual marketing treatment.
Length of discount period: There will be 30 days and 60 days discounts
Experimental Design Details
As part of the experimental design, 3000 adults (one from each sample household) will be selected around 45 FLTs. The sample toilets charge either 4 KES or 5 KES (1 USD is approximately 90 KES) per use per adult. The 3000 study participants will be randomly allocated to different groups receiving varying incentives to use the FLTs. We will have four discount groups as outlined below. Group 1 – Receive 100% discount (5 KES or 4 KES discount depending on whether they are recruited from a 5 KES or 4 KES charging toilet) for one use. They can use their discount any time of the day and for any additional uses they will get a 1 KES discount up to two times. A sample of 600 individuals will be assigned to this group. Group 2 – Receive 100% discount for one use during a two hour time window. The participants will prefer the time window during the baseline survey. For any additional uses either inside the time window or outside the time window they will get a 1 KES discount up to two times. A sample of 900 individuals will be assigned to this group. Group 3 – Receive 2 KES discount for one use. They can use their discount any time of the day and for any additional uses they will get a 1 KES discount up to two times. A sample of 900 individuals will be assigned to this group. Group 4 – Receive 2 KES discount for one use during a two hour time window. The participants will prefer the time window during the baseline survey. For any additional uses either inside the time window or outside the time window they will get a 1 KES discount up to two times. A sample of 600 individuals will be assigned to this group. Each of the above groups will be divided into marketing and no marketing sub-groups (one third marketing and two thirds non-marketing). The advantages of FLTs over the alternatives will be presented to the marketing group in a one on one meeting. Reminders about the benefits of FLTs will be sent through SMS. Gifts will be offered to those who use their discounts during the first week of the discount period. The marketing strategy also has a global component which will equally apply to all participants. A sample of 600 individuals will be assigned to this group. We will also vary the length of the discount period (i.e., one third gets the discount for 30 days and two thirds gets for 60 days).
Randomization will be done in the office. The information about the treatment allocation will be sealed in an envelop and the enumerator will open the envelop in front of the study participant and explain the treatment group he/she is assigned to. The enumerator doesn't know the treatment group the participant is assigned to before hand.
Individuals. We will first select sample households and from each household one adult (18 years or above) will be included in the study.
Was the treatment clustered?
Sample size: planned number of clusters
Sample size: planned number of observations
Sample size (or number of clusters) by treatment arms
The 3000 sample will be divided into the four discount groups according to the proportion discussed the Experimental Design section. Each group will then be divided into marketing and non-marketing sub-groups by the the ratio of 2:1. We will further divide each sub-group by the length of the discount: 60 days versus 30 days following the ratio of 2:1. The three dimensions (discount group, length of discount period and marketing (versus no marketing)) will be cross-cutting and balanced as illustrated below for discount group 1 (which gives one full subsidy that can be used anytime of the day and two additional subsidies of 1 KES).
Marketing No-marketing Total
60 days 267 133 400
30 days 133 67 200
Total 400 200 600
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
Measuring the effect as the number of additional toilet usages per month and using the standard deviation from the pilot study (7.7), with a sample size of 3000 individuals and based on the sample allocation discussed in section 4.1 we are able to detect an effect of 1.06 between our main comparison discount groups (i.e., discount groups 1 & 3). Similarly we are able to detect an effect of 0.87 between the marketing and non-marketing groups and the one month and the two month groups.
Supporting Documents and Materials
INSTITUTIONAL REVIEW BOARDS (IRBs)
IRB Approval Date
IRB Approval Number
Post Trial Information
Is the intervention completed?
Is data collection complete?
Is public data available?
Reports and Papers