It is well established that most of the burden of diarrheal disease is preventable with better sanitation, water quality and hygiene. This study focuses on households living within compounds in slums of Dhaka, Bangladesh. Compounds are clusters of households, typically located around a small courtyard, sharing a common toilet, water source and cooking facilities. In this setting, the use of treated drinking water is quite low and hand washing with soap is also uncommon.
The study aims to test whether behaviour change messages designed to elicit disgust and shame can promote treating drinking water and hand washing with soap in low-income urban housing compounds more effectively than classic public health messages based on germs.
This project uses the concept of 'naming and shaming' and convey to people the presence of human feces in water and on hands that are not washed with soap. The hypothesis is that this information will increase safe water and hygiene behaviours as people would want to avoid being 'shamed' in front of their neighbours. The project also tests the effects of group incentives and solutions like installing a chlorine dispenser near the courtyard's tap and soap near the courtyard's latrine.
a) Test whether behaviour change messages designed to elicit disgust and shame can promote treating drinking water and hand washing with soap in low-income urban housing compounds more effectively than classic public health messages based on germs.
b) Measure willingness to pay for compound-level chlorine dispensers using new group-versions of the Becker-DeGroot-Marschak (Becker, Degroot, and Marschak 1964) procedure.
c) Measure the effect of providing low-cost compound-level soapy bottles on handwashing behaviour.
Outcomes: For water treatment the primary outcomes are H2S tests for bacterial contamination of drinking water stored at home; treatment with chlorine, as measured by chlorine residual tests of drinking water; and willingness to pay to subscribe to the chlorine dispenser. For hand washing the primary outcomes are direct observation of hand washing during a structured observation (during the free trial) and a cleanliness score of hands (as observed by enumerators).
All intervention compounds received a free trial of a chlorine dispenser followed by a sales meeting where they could choose to subscribe for the following year. Half the intervention compounds received a traditional behaviour change message focusing on germs and health, while half received messages designed to elicit disgust that untreated drinking water had feces in it, and fear of shame if they did not treat drinking water. Orthogonal to message assignment, two-thirds of compounds also received a soapy bottle that could be refilled with water and inexpensive laundry soap and left near the latrine or water source. Compounds receiving the soapy bottle received additional behaviour change messages that emphasised either germs and health (for the compounds with similar water messages) or that hands that only rinse (and do not use soap) after leaving the latrine still have feces on them and that this is shameful.
Theory of change
The theory of change for this study requires people to care about what their neighbours think of them, because sanctioning takes time, can be unpleasant, and risks further unpleasant confrontations. Social sanctions may include correction and rebuke from the observer and criticism, mockery or ostracism from both the observer and from other neighbours who hear about the behaviour from the observer. Sanctioning to enforce norms is more likely if there is a 'meta-norm' that people in the community sanction norm-breakers. With that meta-norm in place, someone who sanctions a norm breaker receives social approval, while failing to sanction can lead to risks of lower status and incurring sanctions oneself.'
'For this intervention the key assumptions are:'
1. People in this community perceive eating feces is disgusting'
2. People do not want their neighbours to see them doing disgusting things, as they will be ashamed and will fear social sanction and loss of status.'
3. People do not currently think that there is fecal contamination on hands after rinsing with water alone, nor in drinking water from the community tap.
This study uses an experimental design with random allocation of courtyards to three treatment arms. The control group comprised those receiving household-level training and a sales visit offering low-cost water treatment and hygiene products. Treatment group 1 received courtyard-level training and a sales visit. Treatment group 2 received courtyard-level training and a sales visit with social incentives for households (enrolled households received a discount for signing up neighbours and were enrolled in a monthly lottery) and courtyard-level technologies were available to high-enrolment courtyards.
The authors randomly selected several households to receive household-level marketing in control courtyards. All households within the treatment courtyards received the treatment marketing, thereby avoiding bias due to self-selection of participants. The research team collected data from the salespeople, conducted a longitudinal household survey and took physical measures. The longitudinal household survey collected baseline and 3-month follow-up data from a random sample of one or two households at each courtyard. In addition, qualitative researchers performed in-depth interviews with several households per month.
Six hundred and fifty compounds, 215 to control, 220 to the standard intervention arm and 215 to the disgust and shame intervention arm were randomised.
Usage rates of the chlorine dispenser were low. During the free trial there was detectable chlorine in household drinking water within 24 hours of reported treatment at 8 per cent of the homes. At the end of the free trial, only about a fourth of households agreed to participate in the auction, where compounds stated their willingness to pay to subscribe to the chlorine dispenser. Among this subset, the mean willingness to pay was about $0.10 / month per household where the auction was at the household level, and $0.83 per month per compound when the auction reported compound-level collective willingness to pay.
Adding the hand washing messages to either study arm and providing the soapy water bottle increased the share of hand washing stations that also had soap from 18 per cent at baseline to 43 per cent at endline (seven months post-intervention). During structured observation during the free trial, about 9 per cent washed both hands with soap after toileting in study arms without the hand washing intervention, and 13 per cent (P = 0.03) in arms with the hand washing intervention.
There is no consistent evidence that disgust and shame improved demand for or usage of the chlorine dispenser or soapy bottle relative to a traditional behaviour change message focused on germs and health.