Peer Messaging to Reduce Covid-19 Transmission in Zambia

How to cite

Goldberg, Jessica; Karlan, Dean; Chintagunta, Pradeep; Hangoma, Peter; Prina, Silvia; Graboyes, Melissa; Burlando, Alfredo; Macis, Mario. 2020. "Peer Messaging to Reduce Covid-19 Transmission in Zambia." AEA RCT Registry. AEARCTR-0005940. https://doi.org/10.1257/rct.5940-1.2000000000000002

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

Jessica Goldberg* , Dean Karlan , Pradeep Chintagunta , Peter Hangoma , Silvia Prina , Melissa Graboyes , Alfredo Burlando , Mario Macis

* Primary investigator

Registry ID
AEARCTR-0005940
DOI
https://doi.org/10.1257/rct.5940-1.2000000000000002
First registered
Last updated
Status
In development
Trial dates
2021-01-15 – 2021-12-31
Countries
Private
JEL codes
I12, I15, O15
Registry record
http://www.socialscienceregistry.org/trials/5940

Abstract

Wearing masks and regularly washing hands with soap and clean water are strategies recommended by the CDC and the WHO to reduce the spread of the novel coronavirus that causes COVID-19. These actions may be feasible in settings where social distancing measures are difficult to implement. However, it is challenging for health authorities in low-income countries to share accurate, up-to-date information with local communities, and misinformation and mistrust can undermine compliance. We will partner with the Ministries of Health in Zambia and Kenya to deploy accurate and locally-appropriate text messages about how the coronavirus spreads and actions individuals can take to mitigate the spread, and about reducing stigma associated with infection or preventative actions. We will leverage individuals’ own networks to spread messages. Recent evidence indicates that peers may have an advantage over third parties in communicating health information, but financial incentives may be necessary to overcome the time, monetary, and psychological costs of outreach. Therefore, we will test the effect of SMS message dissemination by peers compared to from a central authority, with and without financial incentives, on behavioral change by the social contacts of cellphone users.

Experimental design

The project is a multi-arm randomized trial. Participants are reached via phone (voice or SMS). Study participants are categorized as original or network participants. Original participants (OPs) are contacted first, and provide a list of up to 5 contacts (names and phone numbers). These contact persons will be network participants (NPs). A new participant is someone who (1) the original participant has exchanged SMS messages with in the past, (2) lives outside of the original participant’s dwelling. In the study, the OPs will receive five text messages with Covid-19 factual information and including advice about face covering and hand hygiene. They will be asked to forward these five messages to the NPs.

The design uses a snowball sample that begins with a sample of 20,000 cell phone users assigned to one of four treatment arms, with additional sub-variation in the authority of the original message:
T0: Original participants (OPs) are asked for contact information of five associates (NPs), and OPs receive SMS messages about COVID-19 and steps to prevent transmission.
T0a: The original message to the OP is from the Ministry of Health Risk Communication and Community Engagement working group, which is the government agency tasked with developing community messaging strategies for covid-19.
T0b: The original message to the OP is from the Zambia National Public Health Institute, an independent, public agency that has less authority than the Ministry of Health but may be perceived as less political and more technical.

T1: OPs are additionally asked to forward the messages to the NPs they identified.
T1a: The original message to the OP is from the Ministry of Health Risk Communication and Community Engagement working group, which is the government agency tasked with developing community messaging strategies for covid-19.
T1b: The original message to the OP is from the Zambia National Public Health Institute, an independent, public agency that has less authority than the Ministry of Health but may be perceived as less political and more technical.

T2: OPs are offered a financial incentive for forwarding the messages to the NPs they identified. The incentive is calibrated such that someone who forwards all 5 messages to all 5 NPs will earn one day’s median wage.
T2a: The original message to the OP is from the Ministry of Health Risk Communication and Community Engagement working group, which is the government agency tasked with developing community messaging strategies for covid-19.
T2b: The original message to the OP is from the Zambia National Public Health Institute, an independent, public agency that has less authority than the Ministry of Health but may be perceived as less political and more technical.

T3: Messages are forwarded to NPs by the research team, under the auspices of the CE subgroup (a local health authority).
T3a: messages are sent by the Ministry of Health Risk Communication and Community Engagement working group, which is the government agency tasked with developing community messaging strategies for covid-19.
T3b: messages are sent by the Zambia National Public Health Institute, an independent, public agency that has less authority than the Ministry of Health but may be perceived as less political and more technical.

Design details

Randomization method
Randomization by computer.
Randomization unit
Individual original participant (cell phone number).
Clusters
Participants are Zambians from all regions with a valid phone number. They belong to a sample collected by a market research company, Sample Solutions. The sample was obtained through a Random Digit Dialing exercise, and as such is not meant to be representative of Zambia. Our budget allows us to purchase 20,000 phone numbers from the market research company Sample Solutions. We anticipate a 25% response rate and therefore 5000 original participants (OPs). If OPs provide an average of 2 contacts each, then we will have an additional 10,000 NPs.
Observations
Participants are Zambians from all regions with a valid phone number. They belong to a sample collected by a market research company, Sample Solutions. The sample was obtained through a Random Digit Dialing exercise, and as such is not meant to be representative of Zambia. Our budget allows us to purchase 20,000 phone numbers from the market research company Sample Solutions. We anticipate a 25% response rate and therefore 5000 original participants (OPs). If OPs provide an average of 2 contacts each, then we will have an additional 10,000 NPs. We will collect 3 rounds of follow up data from each subject (and a baseline for OPs), resulting in 20,000 observations of OPs and 30,000 observations of NPs.
Arms
T0a (1000 phone numbers / 250 OPs) T0b (1000 phone numbers / 250 OPs) No financial incentive Financial incentive Message sent by peer T1a (3000 phone numbers / 750 OPs) T1b (3000 phone numbers / 750 OPs) T2a (3000 phone numbers / 750 OPs) T2b (3000 phone numbers / 750 OPs) Message sent from official source T3a (3000 phone numbers / 750 OPs) T3b (3000 phone numbers / 750 OPs)
Primary outcomes
Behavioral outcomes will be measured by SMS for OPs and NPs. These will include social distancing (exposure to individuals from outside the household); hand washing; and mask wearing. Additionally, we will measure the dissemination of messages (an outcome for OPs, where differences between T1 and T1 measure the effect of financial incentives on peers’ willingness to engage in outreach) and receipt of messages (an outcome for NPs, measuring the policy-relevant parameter and to be used in cost-benefit calculations).

Cited by

No citation data available for this registration yet.

Keywords