The effect of behaviour change interventions on use of public handwashing stations in Bangladesh

How to cite

Perera, Dilhan; Kettle, Stewart; Brown, Dan; Clark, Alexander; Rahman, Atiya. 2020. "The effect of behaviour change interventions on use of public handwashing stations in Bangladesh." AEA RCT Registry. AEARCTR-0006967. https://doi.org/10.1257/rct.6967-1.4000000000000004

Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.

Dilhan Perera* , Stewart Kettle , Dan Brown , Alexander Clark , Atiya Rahman

* Primary investigator

Registry ID
AEARCTR-0006967
DOI
https://doi.org/10.1257/rct.6967-1.4000000000000004
First registered
Last updated
Status
Completed
Trial dates
2021-01-03 – 2021-09-24
Countries
Bangladesh (Dhaka division), Bangladesh (Khulna division), Bangladesh (Mymensingh division)
Registry record
http://www.socialscienceregistry.org/trials/6967

Abstract

As part of the COVID-19 response, BRAC has built 1000 public handwashing stations in several hundred villages in 20 sub-districts of Bangladesh. We plan to investigate the effects of two sets of behavioural interventions on use of the handwashing stations, compared to no additional interventions. The first set of interventions, which do not require ongoing labour (hence termed 'low intensity'), involve the installation of various 'nudges' (e.g. mirrors and signposts) on the station and in the surrounding area; these interventions are designed to attract attention and guide people to the stations. The second set of interventions, which do require ongoing labour ('high intensity'), involve 'promoters' visiting the stations a few times a week for three weeks to offer free bars of soap to station users, update a community board next to the station with cumulative usage numbers, and to encourage users to place comments or marks on the board as a form of social proof. Stations will be assigned to receive either the low intensity nudges, the high intensity intervention, or no additional intervention (beyond BRAC's planned community and communications programming) using a clustered covariate adaptive randomisation process. We will measure use of the stations for one week prior to the implementation of the behavioural interventions, and then for a further six weeks. Station usage will be measured by a combination of structured observations and tally counters attached to the foot pedals that operate the station.

Intervention

- High-intensity intervention bundle: a limited number of free soap bars and facemasks provided as a incentive to handwashing station users on specific days (5 days across 3 weeks); a community board which is manually updated to show the cumulative number of uses, as measured by tally counters attached to the station foot pedals, and encouragement by promoters on specific days (for the whole day; 5 days across 3 weeks) to use the station and engage with the community board. These interventions will be delivered for 3 weeks.
- Low-intensity intervention bundle: mirrors installed on handwashing station backboard and signposts installed and/or large arrows painted on the ground in the surrounding vicinity to guide people towards the station.
- All villages (including those in the control group) will also receive in-person community demonstrations delivered by BRAC, free soap distributed through village committees (not conditional on handwashing station usage), and small stickers/posters pointing to the stations and encouraging inclusive use.

Experimental design

Villages will be assigned to one of the three conditions using a covariate adaptive randomisation process. All stations in the same village will be assigned to the same condition. The randomisation will attempt to achieve balance on the subdistrict that the village is located in, the number of stations in the village, and location categories of stations (e.g. near a mosque, near a market, near a school).

Design details

Randomization method
Covariate adaptive randomisation implemented using the Stata user-written command rct_minim.
Randomization unit
Village
Clusters
800 villages
Observations
869 stations, with 7 weekly outcome measures per station
Arms
266 villages control 266 villages 'low-intensity' 266 villages 'high-intensity'
Primary outcomes
- Average soap pedal presses per day in the first three weeks post-intervention - Estimated average handwashing station uses per day over in first three weeks post-intervention
Secondary outcomes
- Average soap pedal presses per day in the second three weeks (weeks 4, 5 and 6) post-intervention - Estimated average handwashing station uses per day over in the second three weeks (weeks 4, 5 and 6) post-intervention

Cited by

No citation data available for this registration yet.

Keywords