#114 – Maha Rehman on working with governments to rapidly deliver masks to millions of people

Of all the incentives tested as part of a groundbreaking RCT in Bangladesh, the only thing that impacted mask wearing was their colour — people preferred red over purple!

It’s hard to believe, but until recently there had never been a large field trial that addressed these simple and obvious questions:

  1. When ordinary people wear face masks, does it actually reduce the spread of respiratory diseases?
  2. And if so, how do you get people to wear masks more often?

It turns out the first question is remarkably challenging to answer, but it’s well worth doing nonetheless. Among other reasons, the first good trial of this prompted Maha Rehman — Policy Director at the Mahbub Ul Haq Research Centre — as well as a range of others to immediately use the findings to help tens of millions of people across South Asia, even before the results were public.

The groundbreaking Bangladesh RCT that inspired her to take action found that:

  • A 30% increase in mask wearing reduced total infections by 10%.
  • The effect was more pronounced for surgical masks compared to cloth masks (plus ~50% effectiveness).
  • Mask wearing also led to an increase in social distancing.
  • Of all the incentives tested, the only thing that impacted mask wearing was their colour (people preferred blue over green, and red over purple!).

The research was done by social scientists at Yale, Berkeley, and Stanford, among others. It applied a program they called ‘NORM’ in half of 600 villages in which about 350,000 people lived. NORM has four components, which the researchers expected would work well for the general public:

N: no-cost distribution
O: offering information
R: reinforcing the message and the information in the field
M: modeling

Basically you make sure a community has enough masks and you tell them why it’s important to wear them. You also reinforce the message periodically in markets and mosques, and via role models and promoters in the community itself.

Tipped off that these positive findings were on the way, Maha took this program and rushed to put it into action in Lahore, Pakistan, a city with a population of about 13 million, before the Delta variant could sweep through the region.

Maha had already been doing a lot of data work on COVID policy over the past year, and that allowed her to quickly reach out to the relevant stakeholders — getting them interested and excited.

Governments aren’t exactly known for being super innovative, but in March and April Lahore was going through a very deadly third wave of COVID — so the commissioner quickly jumped on this approach, providing an endorsement as well as resources.

When working closely with governments, Maha says that you need to first find champions within the bureaucracy who have both the political capital as well as the required resources to pull this off. She also says it’s vital that you’re proactively following up to ensure that nothing gets dropped at any stage before it is actually launched.

Together with the original researchers, Maha and her team at LUMS collected baseline data that allowed them to map the mask-wearing rate in every part of Lahore, in both markets and mosques. And then based on that data, they adapted the original rural-focused model to a very different urban setting.

Lahore is a big, dynamic city, so the intervention needed to be designed to reach as many households as possible. And information is consumed and processed in a very different way in urban environments; for example, it’s unrealistic to think you can go door-to-door in a big city, and you don’t need to worry about cable TV and social media so much in a small village.

The scale of this project was daunting, and in today’s episode Maha tells Rob all about the day-to-day experiences and stresses required to actually make it happen.

They also discuss:

  • The results and experimental design of the Bangladesh RCT
  • The challenges of data collection in this context
  • Disasters and emergencies she had to respond to in the middle of the project
  • What she learned from working closely with the Lahore Commissioner’s Office
  • How to get governments to provide you with large amounts of data for your research
  • How she adapted from a more academic role to a ‘getting stuff done’ role
  • How to reduce waste in government procurement
  • And much more

Get this episode by subscribing to our podcast on the world’s most pressing problems and how to solve them: type 80,000 Hours into your podcasting app. Or read the transcript below.

Producer: Keiran Harris
Audio mastering: Ben Cordell
Transcriptions: Katy Moore

Highlights

The NORM model

Maha Rehman: A lot of different interventions were tried in the field, but then the original research team found out that there were four things that work the best in a community setting to encourage people to wear masks. They’re now using the acronym of NORM: N stands for no-cost distribution, O stands for offering information, R stands for reinforcing the message and the information in the field, and M stands for modeling. You ensure enough supply of masks in a community setting, and you offer information as to why it is important to wear masks. But then, at the same time, you also reinforce the message periodically in markets, in mosques, and the reinforcement is done through role models, but also through promoters in the community setting.

Maha Rehman: These four interventions together form a very powerful model that in Bangladesh, if you talk about the results, it increased mask wearing up to 30%. This was done across 600 villages, about 350,000 people. I think this is very strong evidence in terms of how can you initiate behavioral change in a community, how can you get people to start wearing more masks.

Rob Wiblin: Why did people decide to test this specific combination? Is this a very standard thing, or was it more that there was theory that these four elements in combination should work particularly well?

Maha Rehman: A lot of other different things were also tested. They also tested incentives. So, different interventions were tested and then measured. And in the end there were several interventions that did not work, but the evidence showed that these four work in a very convincing way to increase mask wearing in a community. Other things that did not work was offering people incentives, signaling, getting the police to also be a part of the reinforcement drive. But those interventions did not exhibit very strong evidence in terms of increasing mask uptake. This was really a data-driven exercise where you were trying to get to the best mix that would then initiate, or stir, this change in the community.

Maha Rehman: I think all four of them reinforce each other and all four of them are very critical when it comes to delivering this intervention. And so even when we were adapting this model for Lahore, all four of them were important. You need to ensure the mask supply. Information has to be offered, because otherwise people have different concerns as to why they’re not wearing a mask. Some are complacent, some have real concerns as to why is this important, so offering information is key. Once you have ensured access and you’ve offered information, then reinforcing that message so that it becomes a habit. So you would need to reinforce once, twice, thrice, and the fourth time you see the person, the person is more likely to be wearing a mask. Then also sharing this information through people they look up to.

Maha Rehman: I think that’s also very important. The role models are also very different in a rural setting, as opposed to an urban setting. How people get their information in urban settings is also very different. And so just in adapting this, I think all four elements are very important. Where you are not reinforcing, you don’t see as much of an increase. So, the R in NORM is also critical. So is M. How you do it in different cities, in different settings, is obviously going to vary. We are learning a lot as we are now scaling this to different cities and through different partners in South Asia.

Bangladesh RCT results

Maha Rehman: You were able to reduce the infections by 10%. This 10% I think is a huge number because it applies to this 30% increase in mask wearing. From there on, you can also extrapolate and say that if you were to ensure 100% mask wearing, or if everyone in the community was wearing masks, how much more would you see a decline in the COVID infections, in symptomatic COVID. So this was particular to symptomatic COVID in the community. The effect was pronounced for people 60 years and older. It was 35% for that group. Between the cloth mask and the surgical mask, it was more pronounced for surgical masks. I think these are very strong findings.

Maha Rehman: Let me just put that in a context. When COVID came to these developing countries in March 2020, we did not have conclusive evidence on what would work and what would not work. And to now start having evidence like this RCT in a community setting that allows us to understand what will work, I think is a big step forward. COVID really brought us down on our knees in a very significant way in last March. There was a lot of chaos at the policy level. There was a lot of firefighting going on. Everyone was trying different measures as to what would work, what would not work. So given where we were then and where we are now in terms of such conclusive evidence being put forward, I think that’s a very, very important result that needs to be disseminated far and wide.

Rob Wiblin: Yeah. Okay. Mask wearing went from 10% of the population to about 40% off the back of these. Then that reduced the number of COVID cases that they detected in the population over that time period by about 10%. That suggests, I guess, that if you manage to go from no mask wearing to full mask wearing, it would reduce the spread by basically a third, or at least it would reduce the number of cases in this kind of takeoff scenario by about a third. How does that compare maybe to what people were expecting? Because some people were arguing masks weren’t going to work at all. Other people have prioritized them incredibly hard. I suppose this is somewhere in the middle, that it’s quite effective but of course it doesn’t end the pandemic, which maybe is what we should have suspected.

Maha Rehman: In order to actually stem the transmission of COVID in a community, you need to bring R naught to a number that is less than one. So if R naught is one, that means one person is at least affecting another one person. So you would need a complementary set of interventions that’s implemented in any community to fully stem the transmission of COVID.

Maha Rehman: I think this result is just the beginning. There’s several other studies that are being planned; there are followup studies to this in Bangladesh and in several other cities. In Lahore, for example, we adapted this to an urban setting. There’s a followup study plan in Bangladesh, BRAC is scaling this up to 80 million people. So there’s several other aspects that are now also being investigated in greater detail.

Do masks give people a false sense of security?

Rob Wiblin: Last year when we were discussing whether people should wear masks or not, at least in the UK, there was this big concern that if people started wearing masks, then it would give them a false sense of security, people would say. And so it would cause them to not engage in social distancing as much, so they would stand closer to one another, and that might undo all of the gains that you might get from the masks. And this experiment tried to measure that. Do you remember the results from that one?

Maha Rehman: In Bangladesh, it was observed that mask wearing also led to an increase in social distancing. The fact that people would not be as careful anymore is a conclusion that was being put forward by a lot of international bodies, but they also quickly updated their stance on this. And this experiment also then gives us evidence that that’s not really the case. This message also became a key part of the messaging in Lahore, where anyone who is wearing a mask is also more likely to comply with other SOPs. Because then you’re getting into a state of mind where you’re more likely to be the first ones to get vaccinated. So we also have a lot of evidence coming up in which people who are more likely to wear masks are also the ones more likely to get vaccinated first. They’re also more likely to be the ones who are more mindful of the standard operating procedures, vis-a-vis COVID. I think this experiment just contributes to that. The international bodies that were advocating differently have also now updated their stance vis-a-vis social distancing and masks, and how does it operate in a community setting.

Rob Wiblin: Yeah, I think this experience with masks has caused this kind of reasoning, which at least in economics is called risk compensation. So there’s often this concern that if you make something safer in one dimension, then people will offset it through their behavior in another dimension. For example, if you design a car so that it’s safer, then perhaps people will drive more recklessly because they feel more secure knowing that they have airbags, or whatever. I think this was always a little bit questionable, but I think this has made people realize that this might be an idea that’s a little bit too clever by half. In fact, most of the time, just doing the obvious thing that makes things safer is actually almost always good on net.

Rob Wiblin: It’s interesting that it was always possible to make a roughly equally compelling argument in the other direction, which would be that if people are wearing masks, then that shows that they think that COVID is serious, and so they’re providing social proof to the people around them that they should be engaging in social distancing because there’s a pandemic happening. It seems about as sensible as the risk compensation argument that they’re going to stop engaging in social distancing because they’re wearing masks. I guess, ultimately in this experiment, it seemed like that effect or some similar effect was the dominant one.

Maha Rehman: Yeah, absolutely. That’s correct.

Mask quality and colours

Rob Wiblin: You mentioned that they found that surgical masks were a whole lot better, which I guess won’t shock anyone. But the design there was in half of the villages they distributed surgical masks, and then half of the villages they distributed cloth masks. They found that, I think, it was like plus 50% effectiveness to be using the surgical masks versus the cloth masks, which is nice data to have to try to persuade people to up their mask quality.

Maha Rehman: Yeah. In a lot of rural settings in some South Asian countries, people tend to wear more cloth masks because that’s also the mask that is being produced more locally. A lot of women are sitting in their homes stitching cloth masks, the cloth is readily available. And so a big shift in rural Bangladesh is now trying to convince people to move to a mask that’s going to be more effective.

Maha Rehman: You don’t see that as much of a trend in Pakistan. In Pakistan, you see people who were wearing more of a surgical mask wherever they are wearing a mask. In rural communities, not so much. But the way rural Bangladesh and even rural India are different, that a lot of people’s livelihoods depend on this cloth mask making it to the market and then being sold. And so yes, the experiment definitely showed a greater increase [in effectiveness] where people were wearing surgical masks. And we cannot say for sure of the impact on cloth masks, but the next trial will also investigate the type of masks further. That is now being planned for Bangladesh by the original research team.

Rob Wiblin: Yeah. Another nugget from the results table was that, like you mentioned, there were various other things that they tested including incentives for, say, officials in a village to try to get people to wear masks. They found out that didn’t work. But actually the thing that most impacted mask wearing was the colour of them. People are just very responsive to the appearance of things that they’re putting all over their face. I think people preferred purple to green or something like that. The color was a significant result.

Maha Rehman: We haven’t tested this in Lahore, but that’s also something that’s more likely to increase uptake, if the masks conformed with certain fashion trends, or as you’re talking about, in terms of a colour. So it’s also very interesting to see how people respond differently to something that’s considered more trendy or something that’s blending in well with their overall appearance. That’s definitely a very interesting result that’s coming out of Bangladesh. I think in the next trial, we would also try and look into that in Lahore.

Adapting the findings from Bangladesh to Lahore

Maha Rehman: So in a rural setting, you could hire a team of promoters that could go from house to house, hand out a mask, and you know, give a pitch as to why was this important. A similar intervention was very expensive in an urban setting like Lahore and so the first adaptation was that the masks were then sent out through the Lahore postal system.

Maha Rehman: So one of the key partners to the campaign was Lahore Post. And they offered to send and post masks out to every household free of cost. And so then there was a challenge of where do you get everyone’s addresses from? What’s a credible data set? And so we got the addresses from Lahore Electric Supply Company as billing data because they have verified addresses. They deliver that bill every month.

Maha Rehman: And then of course there’s a major cost element to it involved, that where do the masks come from? And the masks were then collected by the Lahore Administration from various corporates and philanthropists. So not a single mask was bought. They were all donated by different philanthropists and corporates in Lahore. I think that was also something that’s very distinctive about the adaptation in Lahore.

Maha Rehman: Once that was set up, then of course you also wanted to offer information in person or reinforce it in person. And so reinforcement teams were stationed in markets and in mosques, which were a part of Lahore’s MCL [Metropolitan Corporation Lahore] Division. And in a rural setting, a person with a loudspeaker goes around a community dispensing information. In Lahore, you had trucks with mobile announcements going around the city. Announcements were also made through mosques. And one key element that is also distinctive of the Lahore intervention is flag marches. So the Lahore chief of police and the commissioner also led flag marches around the city.

Rob Wiblin: What’s a flag march?

Maha Rehman: So a flag march is when the Lahore police chief and the Lahore commissioner, the key administrators of a city, take their workforce with them. And they go around different markets in Lahore, dispensing information, talking to people, and ensuring compliance to an intervention. This was first implemented earlier, and then this aspect of the intervention was refined for this model.

Rob Wiblin: Is this a common thing in Pakistan? I’ve never heard of this before, but maybe it’s a good way to get a lot of media?

Maha Rehman: So flag marches, I think that’s very interesting because this is usually done in a war setting when there is more of a security threat to the city, that you would go around the city and ensure everyone’s safe. But that was adapted for COVID because you again need to save lives, and to save lives, you need to go into the markets and in areas that are more crowded and ensure that people are wearing masks.

Maha Rehman: So this was more of a strategy that was adapted for the COVID intervention, and this campaign in Lahore that now goes by the hashtag #LahoreWearsMasks. And so that was the hashtag that the campaign was associated with. There was a lot of marketing collateral that was put up around the city, which was again sponsored. A lot of marketing collateral was also then aired online through different cable networks, as well as put on social media, because that’s the area where a lot of people take their information from.

Working with governments

Maha Rehman: It was an emergency situation. Then a model was pitched to them by a coalition that itself has a lot of credibility in Lahore, so a coalition comprising Yale, Stanford Medicine, LUMS, IPA, and a few other partners. When something like that is pitched to them, something that’s already tested, that has a lot of credibility. And the fact that we did not just go to them with an ask for… I mean, so we were not asking anything of the government. In fact, we were allowing them to save resources by helping them do this effectively.

Maha Rehman: And so I think the fact that you need somebody who’s a little more inclined to listen to a data-driven intervention. So you need to identify champions within the bureaucracy. Not everyone would’ve pulled this off. And so identify the right champions who are able to pull this off, who have both the political capital as well as the required resources to pull this off. And then I think there was a lot of credibility of the coalition that then took the message to them.

Maha Rehman: We worked very closely, we kept the incentives aligned and I think there was a lot of listening and planning and proactively following up. One thing that’s key when working with the government is that you’re proactively following up in order to ensure that it doesn’t get dropped at any stage before it is actually launched.

Maha Rehman: It does take time, but then you learn, and the next time you do it, you will do it faster and you will be able to do it more effectively. A lot of academics and a lot of policymakers and a lot of data scientists working in this field shy away from working with the government because working with the government is hard. However, if they actually want to impact change, working with the government, making it an ally, I think is very, very critical.

Maha Rehman: And just getting down and doing that work it takes to actually see an impact, or your results actually being implemented, or your results shared with the policymakers, or helping them actually implement it, I think that’s the real gain at the end of the day. And so a lot of policymakers, a lot of academics, don’t really want to do that. And I think that behavior towards “who should we partner with and who should we not partner with” really needs to change, because when there’s a lot of hard work going into it, you also want to see those results actually being implemented. So I think that’s something that really irks me about who we choose to partner with and how we actually scale the results that we see out of an experiment.”

Articles, books, and other media discussed in the show

Mask-wearing interventions

Maha’s other projects

Related episodes

About the show

The 80,000 Hours Podcast features unusually in-depth conversations about the world's most pressing problems and how you can use your career to solve them. We invite guests pursuing a wide range of career paths — from academics and activists to entrepreneurs and policymakers — to analyse the case for and against working on different issues and which approaches are best for solving them.

The 80,000 Hours Podcast is produced and edited by Keiran Harris. Get in touch with feedback or guest suggestions by emailing [email protected].

What should I listen to first?

We've carefully selected 10 episodes we think it could make sense to listen to first, on a separate podcast feed:

Check out 'Effective Altruism: An Introduction'

Subscribe here, or anywhere you get podcasts:

If you're new, see the podcast homepage for ideas on where to start, or browse our full episode archive.