opinion | hks policycast

Measuring the power of protests, propaganda, and religion

David Yanagizawa-Drott.
Martha Stewart
David Yanagizawa-Drott.

The Boston Globe presents the Harvard Kennedy School PolicyCast, a weekly podcast on public policy, politics, and global issues. HKS PolicyCast is hosted by Matt Cadwallader.

Did the weather on tax day in 2010 influence the Tea Party’s success in midterm elections later that year? How much power could a single Rwandan radio station have in influencing violence between neighbors? And if economic growth is understood to reliably increase happiness all over the world, what happens if Ramadan hampers that growth?

If these seemingly disparate questions share anything in common, it’s that they’ve all been asked and answered in studies by Harvard Kennedy School Professor David Yanagizawa-Drott.

This week on the Harvard Kennedy School PolicyCast, Yanagizawa-Drott explains how he and his colleagues formulated each inquiry in order to find answers to more fundamental questions: Do protests influence elections? Can hate speech lead directly to more violence? Do religious activities increase happiness?


The following excerpts have been edited for length and clarity. You can listen to the full interview in the SoundCloud player above, or on iTunes.

Get Truth and Consequences in your inbox:
Michael A. Cohen takes on the absurdities and hypocrisies of the current political moment.
Thank you for signing up! Sign up for more newsletters here

MC: Is there a common thread that connects all of your areas of interest?

DY: Essentially, all of my research is using data to address research questions for which we don’t have really good answers to yet. There’s also some conceptual connections across the different sorts of research projects that I have – political economics and development economics.

MC: One of your papers from last year analyzed the Tea Party movement: you were trying to understand whether protests inform public policy or just reflect public sentiment. I was particularly struck by a conclusion that you had about the weather at the first Tea Party protests.

DY: The weather is a methodological trick to get at broader questions. There’s a long history of political change associated with political protests. But it’s really unclear whether political protests cause or contribute to political change. Political protests can potentially just reflect political opinions of the population, but not necessarily affect anybody else’s view. Maybe policy makers aren’t really affected by the fact that people are out protesting. So it’s challenging to disentangle the causal effect of people protesting, and that’s where the weather comes in.


The first big, nationwide Tea Party protest happened on Tax Day, April 15, 2009. People really dislike protesting when it’s raining, so we collected data on weather and rain across the US on this particular day. We found that, first of all, if it rained, many fewer people showed up to protest. But then we traced local politics over time and we saw, eventually, the movement started to grow in those places that had a big turnout during the protest. So you see more local organizers, for example, signing up through social media, and more political donations.

The Tea Party movement was very much a conservative movement, so it’s natural to think that it could influence a right-wing shift. And that’s exactly what we saw. In those places that had large protests, you saw a pretty big boost for the Republican Party in the midterm elections.

MC: Were there other variables that happened to even the field? Like, obviously in Nevada there was probably going to be pretty good weather, but there’s also a fairly sizable conservative movement there. Whereas, in Seattle, it’s more likely to rain, but there’s also a fairly liberal population there.

DY: This is the classic “omitted variable bias” problem: maybe the places that saw good weather are different from the places that saw bad weather. And maybe we saw higher Republican voter turnout for reasons other than the protests. So we do a few things to get at that. First of all, we control for the likelihood of rain in April. Next, we can look at voting patterns before the protests and see if they were any different across those nice weather/bad weather places.

MC: In another paper, you looked at the effectiveness of propaganda, and hate propaganda in particular, to promote violence by looking at Rwanda in 1994. How did you go about this?


DY: I was really interested in understanding this genocide, which was horrific. During a hundred days almost a million people were killed, and what’s so very disturbing about the genocide is that often this violence would have a local nature: Hutu neighbors going out and massacring their Tutsi neighbors. I was very interested in understanding how that could come about.

One of the factors that you often see mentioned is the role of propaganda in this genocide, and the role of the radio station RTLM in spreading hate propaganda, encouraging people to participate in the killings. A lot of fear, a lot of hate was spread through radio. This question is broader than Rwanda: can this kind of hate speech actually induce violence?

Because of the topography in Rwanda — it’s actually called “the land of a thousand hills” — it more or less gives us random variation in who could listen [to the hate speech] and who could not. Basically, if there’s a transmitter in one place and there’s a village 15 kilometers away, a lot of the variation in radio signal strength occurred if the signal was blocked by hills.

MC: In the end, was it established that being able to listen to these radio broadcasts actually resulted in more violence than otherwise?

DY: The magnitude of effects was quite large. If you compare a village with no radio coverage to a village that had full radio coverage, you saw about 60-70 percent higher participation in the massacres in those villages.

MC: Wow.

DY: I also ran a calculation, trying to assess overall in the country how much of the violence can be directly attributed to the broadcasts because the radio didn’t broadcast in all of the country. About 10 percent of the participation was directly caused by the radio station.

MC: In yet another recently published paper, you focused on the role of religion on economic growth as well as happiness and used Ramadan as your case study. Can you tell us a little bit about that?

DY: Most religions have a religious practice similar to Ramadan in some shape or another. Either you’re required to go on pilgrimage, like the hajj, or you’re supposed to fast, or you’re supposed to go to church every Sunday, and so forth. And these practices are typically community-wide. So the theory is that this would affect economic outcomes by having a direct trade-off with materially productive activity: if I go to church, I can’t work. If I’m praying five times a day, that’s going to take time from other materially productive activities. At the same time, it’s interesting to look at happiness because it may not be that money’s everything. So even though you may expect a negative effect on incomes and economic growth related to these religious practices, maybe people are not less happy, as a result, but could potentially be more happy.

MC: And how do you measure happiness?

DY: We looked at the World Values Survey, which is done in 100-plus countries, polling people for a couple of decades about their happiness levels. Because there are multiple surveys from each country, you can actually compare within country over time. So, maybe the notion of happiness could differ from country-to-country, but within countries over time we could look at the impact of Ramadan.

Ramadan is longer in some years and shorter in some years, and by that I mean the length of day from sunrise to sunset depends on what time of year Ramadan occurs. So, it gives us a natural experiment in the strictness of this practice.

We found Ramadan had a pretty strong negative effect on economic growth, while at the same a pretty strong positive effect on happiness and life satisfaction. So I don’t want to go against this view that money buys happiness, but at least in this context, it seems that people become poorer but happier, nevertheless.

MC: So, how are you able to isolate Ramadan, in particular, and were you looking just at Muslims all around the world?

DY: For the economic growth analysis, we used data from all Muslim majority countries. For the happiness data, we looked at all the surveyed Muslims in the world in this dataset, more than 100,000 individuals.

We had a few potential mechanisms in mind for why there could be negative growth effects. The productivity hypothesis would be that [during Ramadan] you’re getting dehydrated and nutrition goes down, maybe you’re less productive as a worker and income suffer as a result. The other hypothesis is that people may choose to work less, not only during Ramadan but throughout the year, and we find evidence in favor of the second hypothesis. We look at survey data in terms of values, and we find that people report that in years when Ramadan is very long, people prioritize different things in life. So, they prioritize work less, they think family is more important. So, we see this shift in values regarding the work-life balance at the same time as we see people working less. They’re happier as a result, but they’re also poorer.

MC: If higher incomes mean greater happiness, there must be some kind of balance or at least some kind of point at which there are diminishing returns on the value of religion to promote happiness.

DY: We can’t really answer that question. But I hope that at least we inspire people to look into those questions in the future.