Voters and Vaccines: The Politics of Ground Campaigns

Whether it’s trying to convince you to vote for a particular candidate or get vaccinated, the identity of the person who knocks on your door may matter. So who are the people who volunteer to do this canvassing? Are they likely to succeed?

These are all questions that Harvard political scientist Ryan Enos investigates in his paper, “Party Activists As Campaign Advertisers: The Ground Campaign As A Principal-Agent Problem.” Using a rare dataset from Obama’s 2012 presidential campaign, Enos delves into the politics of door to door campaigns, and we try and tease out some lessons for our current efforts to persuade people to get vaccinated.

Listen on Apple Podcasts or wherever you enjoy podcasts.

Transcript

Will Howell:

I am Will Howell.

Wioletta Dziuda:

I'm Wioletta Dziuda.

Anthony Fowler:

I'm Anthony Fowler. And this is Not Another Politics Podcast. 

We talk a lot on this show about campaigns and elections, and we're very interested in campaigns and elections. And a lot of politics is about interpersonal conversations, is about a friend trying to persuade you to do something or vote for a candidate, or, in fact, an official campaign volunteer worker knocking on your door and saying, hey, there's an election coming up. You should change your mind on an issue. You should turn out to vote. You should vote for my preferred candidate, et cetera.

We talk a lot about what evidence we have and what theory we have about persuasion and mobilization and how effective campaigns are, and you could even make the connection today to the campaign that we're having to try to get people to get vaccinated. We're trying to, in many cases, even knock on their doors and convince them to do something they otherwise wouldn't want to do.

What do you guys think about that? Are there any interesting connections? Is there anything we can learn from the world of electoral campaigns and apply it to vaccination campaigns?

Wioletta Dziuda:

Well, I definitely think about this problem a lot. I'm trying to talk to my family members who are not vaccinated and trying to convince them. Then I pause and think, well, I'm very different than they are. I come from a very different place. I have a different background than they are. And then I worry that perhaps that sets me up for failure.

You are asking about the connection with political combines. I think there's a clear connection. If I went to canvas for the candidate that I like a lot, I'm probably not going to be similar to the people I will be talking to, and I will be up against exactly the same issues that I'm up against right now. So I think there's a connection. It would be actually good to know a little bit more about what the campaigns are up against and how they solve those issues.

Anthony Fowler:

Will, you talked to a political scientist a really interesting paper about campaigns and about who campaigns, who volunteers on campaigns, what those people are like, and what implications that might have for the success of those the campaigns.

Will Howell:

I did. I talked to Ryan Enos, who is a professor at Harvard University, and he wrote, with Eitan Hersh, a paper that came out in the American political science review called, Party Activists as Campaign Advertisers: The Ground Campaign as a Principle Agent Problem, in which he investigates the extent to which the people who are sent out to canvas on behalf of a candidate look like, or, it turns out, don't look much at all, the people who they're trying to convince to turn out to vote. It's an interesting paper, it's got a lot of really interesting data. Let's give it a listen.

I'm hoping we can come to the conceptual framework you put forward later, and instead, frankly, lead with the findings because they're terrifically interesting. We'd love to hear, if you could just lay out, what are the data that you collected and how did you go about collecting them and who were you focusing on?

Ryan Enos:

What we were interested in is this difference between the people that speak for candidates and work for candidates, and the people that they're speaking to, and maybe some candidates themselves as well. So what we actually did is we went out and tried to survey these legions of volunteers that work for campaigns. And in particular, what we did is we surveyed volunteers, actually, anybody that worked for, but most of them were volunteers, that worked for the Obama campaign in 2012. 

If you recall the Obama campaign in 2012, they really hyped, and media hyped, this idea that they were driven by these volunteers that would come and do this interpersonal contact, knock on people's doors, make phone calls, do all that type of things to make these connections with voters. And so what we wanted to do is find out who these people are and what they think, and how it looks different than maybe what your average voter thinks.

What we were able to do was to insert a survey tool into a database platform that the campaigns made use of, this thing called the NGP Van, which is a tool that they used to organize data that these volunteers would get on voters. So if somebody would call you on the phone, which has happened to a lot of us and said, hey, you should go vote next Tuesday, when that person was doing that, they were probably looking at this NGP Van tool in some front end, that would say the voter's name and say how many times they've been contacted, and all that stuff.

What we were able to do was in partnership with them, and with the Obama campaign, every few times, about one out of every 100 times, that somebody would log into that, this little survey would pop up and we'd say: we're here, we're social scientists, we're doing this in partnership with the campaign, so we really want you to do this, and that, we think, induced them to it, and fill out this little survey and tell us about things your attitudes, what you're doing for the campaign, what you think is the most important issues out there. And we collected this on about 3000 of these volunteers, and we were then able to connect it to similar questions that we asked the mass public on a big nationally representative survey.

Will Howell:

Walk us through the high-level findings. Demographically, do they look like the larger population that they're trying to reach?

Ryan Enos:

No. And in many ways, that's one of the most important findings. So both demographically and attitudinally, they look very different from the people they're trying to reach. And in many ways, this is predictable, because one of the things we know from decades of political science research is that the people that are really interested in politics are very different than the typical voter and who they were trying to reach was the typical voter. They're trying to persuade people to vote for Obama, and they're trying to persuade them to turn out and vote for Obama. The people that don't need persuaded by that are not typical voters, they're people who are really interested in politics, and even more so the people that are willing to spend their time as volunteers to give up their valuable time, are not like typical voters. 

What we found was we found that these people had what we might think of as the resources, is one way to think about it, to spend their free time doing this, and that means they're usually higher income, they're better educated, they were more white than the typical voter, both in the mass electorate and in the democratic primary, and even among the undecided voters in the primaries. Sorry, the democratic caucus, I guess I should be saying.

Then we found that attitudinally, they were different than those voters as well, where they were more liberal, much more liberal, actually, and they prioritized different issues. They prioritize these issues that liberals and activist liberals tend to prioritize, which are things like—this is going back to 2012 and things have changed a little bit, but for example, one of the issues that they prioritized most strongly was inequality, which if you hang out at a university or you hang out around democratic activists is something, people talk about.

Will Howell:

They talk about it all the time.

Ryan Enos:

Yeah. Probably for good reason. But we talk about it all the time. But it turned out, your typical voter, they don't conceive of things in terms of inequality, or at least they don't prioritize that as an issue. They might prioritize things related to it, but they were not prioritizing inequality in the way that these activists that were out knocking on the doors were.

Will Howell:

I mean, one point of comparison, and the point of comparison that you carry through much of the paper, is, between the activists and the people that they're trying to reach. They’re standing in as surrogates for, or advocates for, a candidate, for Obama. And Obama certainly cared about inequality. And so might we say that while their policy commitments diverged from the standard voter, they were faithfully representing the candidate himself?

Ryan Enos:

That's a good question, Will. And I think a lot of that comes down to the heart of the issue. If they're faithfully representing the candidate, there's no issue. And so one of the issues we have to ask is, how is a candidate presenting him or herself? And that is a difficult issue, but I think we can measure, what is a representation of a candidate? What is a candidate's ideology? We have a lot of political science tools to do that. But one thing we know, just intuitively, and there's a lot of research that points to this as well, is that candidates, they want to appeal to the middle, the median voter if you will, and this is different between a primary and the general election if you will, but you're not going to win an election by only talking to your base. You have to bring people over.

In an ideal world, that's who the candidate is going to appeal to, and so if all the people that are representing you are not near the median, both in their ideology and they actually are different from the median in terms of their identity, then that probably presents a problem for the candidate.

I think with Obama, we could go back and have a discussion about how he presented himself, and he certainly did talk about things inequality and things that, but I think in many ways, we could see Obama as somebody that really tried to appeal to the median voter.

Will Howell:

What your paper points out is that in his effort to do that, it does not appear though that what he did is unleash a legion of door knockers who look like the people who are actually answering the door. And one of the interesting findings that comes out of the paper is that while they privilege inequality, when you asked them, what do you think the voters care about? Then inequality doesn’t appear high up in this list of priorities. That is, it looks like these activists understand that their interest in inequality doesn't match the interest of the people who they're trying to reach.

Ryan Enos:

One way to put this is, there's some learning. I mean, these activists aren't so activated that they're blind to the issues that people are talking to, and so there's some meeting there that they're learning from somewhere. Maybe they just understand that they're different. Maybe they're learning from the doors they knock on. But there still is some slippage in what they understand from the voters.

You can see that one of the issues we ask about, as you mentioned, was what are the issues most important to voters? What did the campaign workers think are most important? And for example, they said healthcare. And healthcare was an issue that voters pointed to as being important to voters. But it was almost listed twice as often by these campaign workers as other voters, and so there was still this mismatch there.

One of the issues that voters would talk about a lot, for example, was inflation. And in many ways, it's interesting that that's what was on their mind in 2012, but they said inflation. That's what voters wanted to talk about. And that was nowhere even anywhere near the top issues that these campaign workers listed of what was important to voters. So they were still missing. They got closer, but they were not perfectly representing what the voters wanted in a way that we'd like in an ideal world, if you're running a campaign.

Will Howell:

You also traced where the campaigners live and are located and when they're actually doing their work, and you traced what those movement patterns look like. You want to say a little bit about that?

Ryan Enos:

In many ways, I think this is one of the most interesting parts. I'm glad you asked about it. You can see, when we map the flows of this, just how apparent it is, is that one in four were not working in that state they lived in, where they were people coming in from out of state. People in California that really care about who wins the presidential election, but they have no essential impact of that, so what they want to do is go campaign in Nevada or Ohio, or somewhere like that.

People, in many ways, especially in certain parts of the world, you can take New Hampshire. I live near New Hampshire now so I go up there pretty frequently. New Hampshire is very sensitive to who lives in New Hampshire who doesn't. They can almost smell them out by some sixth sense or something.

Will Howell:

Here comes a Harvard professor from Boston coming to tell me how to vote.

Ryan Enos:

Exactly. I've done campaign research exactly in New Hampshire, and they know when you're coming in from Massachusetts, and they're used to seeing people from Massachusetts, so imagine somebody from California flies into New Hampshire and is knocking on their door. They know that person's not from New Hampshire. Not 100% of the time, but enough of them do that it suboptimal from the campaign's perspective. But it's a predictable consequence of both our electoral structure, where only some states matter, and the fact that some people care more about politics than others.

Will Howell:

The notion that what the door knocking looks like, is neighbors reaching out to people who they regularly run into on the street. You're suggesting: not so much. There's slippage in terms of the demographic, their slippage in terms of their ideology, and then also just where they're from.

Ryan Enos:

That's right. And to underscore why we think this is important is, there's been a fair amount of research of what makes this door knocking and this phone calling and all these other things affect it, and I shouldn't say a fair amount, there's been a lot, so this is something in terms of social science evidence. This is pretty rigorous. There's been a lot of randomized controlled trials on this and we have a pretty refined sense of what works. And a lot of it is pretty intuitive, and the campaigns tried to do this, is based on this evidence, is they said, well, we want people to talk to people they know. Because you're more likely to be receptive to messages from people that you know. And if it's not people you know, you're more likely to be receptive to people that are like you.

There's a lot of reasons, psychologically, we're built that way. We're more responsive to people from our ethnic groups, from our income group, from our hometown, things like that. And essentially because these were relying on these masses of volunteers, a lot of that went out the window. They had to rely on these people that just did not meet those ideal criteria in the way that the social science evidence told them that they should.

Will Howell:

Great. So let's talk a little bit about then the conceptual framework that you bring to bear in trying to make sense of the meaning and the stakes involved in these findings. You suggest that the way to think about the relationship between somebody who's knocking on doors and the candidate, him or herself, is one of a principle-agent relationship. And, to the extent that there are differences between the principle and the agent in their motivations, in their character, or type, that relationship may be a problem. Those differences may point towards a problem. So why don't we think about this in those kinds of terms? What does that mean and why don't we think about it in these kinds of terms?

Ryan Enos:

Sure. So a principle agent problem is a very general type of problem where you just have a principle, somebody that is the central figure in trying to accomplish something in building a business, running a campaign, whatever, and then they have to employ agents to do that for them. So they have to hire people. They have to get volunteers. And it could be that the principle and the agent have different incentives, and that leads to the suboptimal situation for the principle.

If you're a company, for example, if you're Uber, what is good for the company might not be good for that driver. So you could have a driver that would go and they would shirk, if you like, where they might say, “Let's turn off this app and you can just give me 20 bucks when we get to the location, and then that way 10% of it, or 50% of it, I don't even know what it is, won't go to Uber, I can keep it,” and that driver has an incentive. If all the drivers did that, Uber wouldn't make any money.

But in a similar respect, these campaigns have perhaps a principle-agent problem, where Obama can't get elected on his own, and so he has to employ agents. And one of the points we make is that in previous campaign worlds, let's say for the history of modern campaigns up until about 2008 or 2004 or so, that was much less of a problem because the agents were people very close to the principle, very close to the candidate. There were people that he hired on his own, or his campaign manager hired.

But then Obama comes up with this idea. It preceded him a little bit, but he really made it big, was this idea that they were going to have candidates do things through this mass of volunteers, and in many ways that really changed the incentives, or it changed how they did these campaigns. It became a huge principle-agent problem, and these volunteers maybe had different incentives because of their different ideologies, the fact that they didn't know exactly what the campaign was trying to do, they didn't know what their strategy was, and, I should say, they maybe didn't have a good way to control them because there weren't things like ways to monitor what they were doing, they didn't have a possibility. There were a lot of different things.

Will Howell:

There is a core source of alignment, which is the people working on behalf of Obama, and Obama himself, wanted to see Obama elected. But, as you point out in the paper, at a micro level, conditioning upon that, there are nonetheless some schisms: that they may also have an interest in raising issues that don't align with the interests of the campaign and under the guise of an Obama advocate, they see this as an opportunity to spread the good word about, well, inequality, as we discussed, or some such.

Ryan Enos:

That's right. And not only that, but they may not know that this is what the campaign wants. So one of the things, and we can see this in public opinion data, that, for example, activists often think that campaigns, the candidates, are more liberal than they actually are, if you're a democratic activist. This is probably the opposite for Republican activists, they think that the candidates are more conservative they are. And so, since I am an imperfect agent, I don't spend my time figuring out exactly what he wants and I just talk to the voter about what I assume he wants, and that's different than the message that the campaign would be putting out if they had perfect control over their messaging.

Will Howell:

I want to try to say, the thing that we're characterizing as a problem might actually be a positive. That is, it was in Obama's interest to have these people out in the field. So I have two thoughts and be interested in your response. One is, we can certainly imagine instances where the mere existence of difference is an asset, that is if Obama had a bunch of really conservative people who are knocking on doors saying, look, I'm really conservative. I don't look like you, who's answering the door. I don't look like Obama. But I'm voting for Obama. That's a different kind of difference, but it's something which can persuasively be really meaningful.

There's another kind of difference, which I was thinking about. I recently bought a car, and you go to buy a car and the person who's trying to sell me a car, he didn't look anything like me. He's obsessed with cars, and he's talking ad nauseum about all the details of this car, and I don't understand a thing he has to say. And he subscribes to car magazines and he's a real car aficionado. He doesn't look me. He's more extreme on this particular dimension. But I'm glad for that: in that if I had some middle of the road person saying, as best I can tell, this is a good car. Well, all right, but this guy, who doesn't look like me, singing the praises of this vehicle that I'm contemplating purchasing, has a resonance that leads me to open up my checkbook, because I'm old and I still have a checkbook. What do you think?

Ryan Enos:

That's a really interesting conjecture. I think that what we could say is, in an ideal world, what we would say is, we'd have somebody that can somehow signal that commitment, that informational capacity, that they really know about politics, and also looks like the voter they're trying to persuade. Because unless we're going to erase human psychology, we know that people rely on those cues and those your heuristics. And so the question we're left then, is what the trade-off between these two competing interests are.

Will Howell:

This plays out too in our consideration of the problems, but also the value of sending people to states to campaign where they're not actually from. We talked earlier about how that's a problem: the people of New Hampshire saying, well, you're not from New Hampshire, what are you doing here? But there's a certain value in being able to say, yeah, I'm not from New Hampshire, but I care so much about this issue that I flew all the way across the country to come talk to you. I didn't just walk across the street because I'm your neighbor and idly convey my interest in voting for Obama.

Ryan Enos:

It's a valuable point. And we could wonder how much that was conveyed. They could say, this is something that's really important, or maybe Obama is a really important person because he got people from California to fly all the way out here to do this. The thing we would have to understand is how we balance that against the people that are like, wow somebody from California either is here telling me what to do, I just don't want to do that because of some visceral response, or maybe that means that Obama's really not like me. This guy is on TV saying, vote for Obama because Obama's for Ohio, yet everybody telling me to vote for him is from California. Maybe those ads aren't true. 

I think it's an empirical question, and one that would be really interesting to try to study, is to figure out how to balance those two considerations.

Will Howell:

Terrifically interesting. Thanks a lot.

Ryan Enos:

Thanks for having me. It was a great conversation.

Anthony Fowler:

It is worth pausing and noting just for a second, that how unusual this data is. I'm somewhat surprised that an active political organization in the middle of a campaign would let them do this. Will you let us just embed a survey for our research purposes into your software while you're in the middle of a heated presidential campaign? That's a neat thing that they were able to do.

Will Howell:

It was. This isn't just your standard random digit dial, where you hope to get a few of these people, they had a solid response rate, they were able to target precisely the people that they wanted to get. We talk a lot about differences between elites and mass publics. We don't have much information about the people who are deployed by those elites in order to reach the mass publics. And what they find is they look every bit as different from average citizens as do the elites themselves.

Anthony Fowler:

There's no direct evidence on this, but there's an argument they're making, which is that this is probably not optimal from the perspective of the campaigns. An extreme liberal is not the right person to go convince a swing voter to vote for Obama. Do you guys buy that argument? They're obviously relying on other evidence here, but they're saying something's gone awry with these campaigns.

Wioletta Dziuda:

Let me put this in terms that I understand as a theorist, so let me try to think about their framework and what framework I have when I think about this problem. From what you said, and from the interview, it's pretty clear that the number one concern that they have is that those people will not represent the campaign faithfully, that they will go rouge. I don't think this is the biggest concern that I would think about when I think about this problem, because after all, the number one objective of both the campaign and the canvasser is to have the candidate elected.

However, I think the entire problem that they are seeing is in the interaction between the agent, the canvasser, and the voter. And here, we have quite a few models in economics and political science that tell us what we should expect in this interaction, and there are a lot of models that assume that the agent is more informed, so that would be the canvasser, but she does not share the same preferences as the voter. And if that case, the less common ground they have, the less they share their preference, is the less communication there will be. The canvasser can say whatever she wants, but the voter will not believe because there's no reason to believe.

I think that's the main framework that would support the conclusion that having those agents, those canvassers, different than the audience, is actually not going to lead to any persuasion.

But I think there's a different framework, which also leads to same conclusion, which is signaling. If I see that the people who are willing to sacrifice their earnings and their time to canvass for Obama are different than me, I conclude that those people are probably the people who really benefit from this candidate the most. They are really willing to sacrifice a lot to have this candidate elected, and they are different than me, so that must mean that the candidate is very different than me.

I think that that channel is very persuasive for me. I think it's actually very present when we talk to people and try to convince people at our own expense about some facts or some candidates or some vaccines.

Anthony Fowler:

I would think of it similarly to the way Vera just described it, but maybe a little bit differently. I mean, so if I'm an undecided voter, what are the two kinds of things I'd like to be more informed about? I'd like to be more informed about the ideology of the candidates, just what are their positions? What are their values? what are their policy leanings? And I'd also like to be more informed about their ability and their competence. 

If I see that some really ideologically extreme person like this candidate a lot, I could infer that maybe they're really competent, or I might infer, maybe they're really extreme. And if I'm a middle of the road swing voter who queues undecided, then that latter signal a bad one. Seeing the enthusiasm of all of these really extreme people might make me think, “oh man, this is the kind of candidate that gets the really extreme people excited.” Or maybe it's that, “wow, their competence is so great. But then if their competence was so great, why don't I see middle of the road people who are also really excited?” 

It seems pretty clear that if I were one of those middle of the road voters and you're trying to convince me to vote for Obama, clearly the most persuasive case would be someone who is like me, someone who seems to share my values, but nevertheless has more information than I do about the competence of the candidate, or maybe the moderation of the candidates, or something like that.

Or it might even be someone—Will, you made a reference to this in the interview, you said, what if there was someone who you expect to be maybe even more conservative than me who still likes Obama? That would be persuasive too, if someone came along and said, you know what, I actually, I'm religious and I'm pro-life, and I'm in favor of low taxes, but I still just like really Obama because he's just such a competent, high ability person and I think he's going to be a better president than Romney.

You, as a centrist, might say, “maybe that's really persuasive.” But neither of those is anything like these canvassers that are actually going out and knocking on your door. It's someone who in fact is an extreme liberal, who's going to sit there and talk about how climate change and inequality are the biggest problems, they don't look you, they don't sound like you, and presumably that's just not going to be very persuasive.

Will Howell:

Can I breathe some life into the possibility that maybe they are persuasive in a couple of ways? One is that while we learn about the ideology of these canvassers, we don't have any reason to believe that they themselves are sharing their own ideology. They are being deployed by a campaign and presumably were trained by the campaign, and they're working, as Wioletta pointed out, in a context in which their incentives are very much aligned with the campaign, which is, you want to get Obama elected. 

If you see that if I come forward and present myself as a crazy liberal, that's going to be unproductive, well, so you have reason to check yourself. It also, in the context of the survey, when asked what they think the top issues, the most important issues, are of the people who they're reaching out to, they get pretty close. I mean, there's a much greater alignment there. They don't raise that they themselves care a whole lot about inequality, but they also get that it's less important to the people who are answering the doors. 

I guess there's this other piece too, which strikes me as relevant, which is that, all else equal, I want to get somebody who looks like me, that's true, but when a 22 year old, or an 18 year old, drives from Seattle all the way to Florida to knock on doors, and somebody from Florida answers the door and sees somebody from all the way across the country and they present themselves as such and they know that they're not from around here, one thing you can interpret is, “Oh my God, you look so different, I have nothing to learn from you.”

Alternatively, you might say, “Wow, this really matters. You and all of your enthusiasm and how much you care about this is underscoring the stakes of this election”. And so, it isn't just about “do I align with you ideologically?” There's a sense, I mean, the Democrats are typically reaching out to democratic voters and they want to say, this one really counts and I'm putting in my time, I've just driven all the way across the country to meet you at your door, and that there's reason to believe that that might have some effect, it may increase the likelihood that they subsequently turn out.

Anthony Fowler:

It could. Or you might say, “Oh man, they couldn't find anyone from my community to actually knock on doors, they had to go recruit the extremists from California to come knock on my doors.” And you might say, “Man, this campaign is really in trouble.”

Will, would you be making the same argument if the ideologies and the parties were flipped around? Just as a thought experiment. Would you say, suppose we had good data on the Romney campaigners, and suppose a lot of the Romney campaigners themselves are very conservative, evangelical Christians coming from West Virginia and Oklahoma and so forth, and they're driving into Ohio and knocking on suburban doors in Ohio and saying, “You should really vote for Romney because I feel so strongly about it.” What do you think that Ohioan thinks?

Will Howell:

Actually, I think in some ways it's cleaner. They're presumably going to be targeting undecideds or trying to get out the vote among their own base. They're going to be targeting Republicans. Romney, who is a moderate Republican running in 2012, and I'm sitting in my living room on a Thursday afternoon and the doorbell rings and there's an evangelical Christian from Texas, and that comes out in the course of the conversation. I've driven all this way to talk about this campaign. What that's signaling is, “I care so much about this.” And it's informative not just about Romney, it's informative also about the opposition. It's because we have to come together as a party. This isn't my first choice of a candidate. I would have preferred an evangelical Christian representing me, and Romney is different.

Wioletta Dziuda:

For example, if you look at the data that Ryan collected, they showed that those campaign workers, or volunteers, are actually whiter than the average person that they're trying to persuade, and he also can tell a story that some people, some white people, might have had some qualms about Obama being the first black candidate. Perhaps he would be very radical. He would just pursue solutions that would disfavor whites, and he would have white people, who have all the reasons to be as skeptical as I am, coming and saying, “Well, no vote for Obama.” So I think those people in this dimension are more likely to be persuasive than if other minority members came in favor of and canvased instead.

Anthony Fowler:

I think you're now agreeing with the Enos and Hersh argument, essentially, which is that if you're a white voter, you're more likely to be persuaded on the margin by a white canvasser. Just like if you're a moderate voter, you're more likely to be persuaded on the margin by a moderate canvasser. But these are all extreme, liberal, whites who are probably not like the swing voters that they're trying to persuade, and that's the tension. 

Will Howell:

There's another piece to this, which, I'll say, gives me some pause about the use of this principle-agent framework for thinking about this particular phenomenon. Which is that conditional upon our interests being aligned, that is that in fact these people are trying to get Obama to be elected, Obama reaching out and deploying people who look very different from him, all else equal, might be a very good thing. Different from him, because it shows that look, this is not just this one segment of folk who get behind him, they're all kinds of folks who do.

This goes back to the story as it relates to issues involving race, but you could tie it into all kinds of other things, working class folk canvassing for Obama, because sure he's elitist law professor, but he really gets the needs and interests of us union workers. That can be a powerful signal. And so in that way, conditioning upon their interests being aligned, difference in ideology, democratic profile, whatnot, can actually lead to greater persuasion downstream.

Wioletta Dziuda:

They have data that actually points in the direction of those canvassers understanding what voters want, what voters care about. So they have a graph where they asked the canvassers, what do you think are the most important issues? And then what do you think the voters think are the most important issues? And then they have data on what voters actually think that are the more important issues, and you can see that there's a convergence. So the canvassers think the economy is less important than the voters do, but when they think about the voters, they think the voters actually attach quite a bit of importance to the economy.

Assuming that they do actually get the voter, they do get whom they're talking to, it's good that they themselves are people who are really excited about Obama. And perhaps they're excited about Obama for the reasons that those median voters would not be excited, but it's good that they are excited. Because if you think about canvassing, it requires a lot of effort and you have to be excited, man. You have to be all the time on your feet and you have to be very friendly all the time with a smile on your face, and you're only willing to do that if you really care about the candidates.

Conditioned on the voters not being spooked by the fact that people look differently than them, it's good that Obama got people who are maybe different than the voters, but extremely excited about who he is, because they are going to exert more effort.

Will Howell:

And who are young. Who are-

Wioletta Dziuda:

They can stand for a long time.

Will Howell:

They can stand. I mean, if I had to do this canvassing, I wouldn't make it to the end of the block before I would bail out, out of sheer fatigue.

Anthony Fowler:

Just one little point on this issue thing. I think that's interesting, the data they have. It is true that these canvassers, there's some signal here in the sense that the canvassers, when you ask them: “What issues do you care about most versus what issues do the voters care about most?” They give different answers often, and they have some sense that their interests are not perfectly aligned with the voters. And so that's interesting.

I wish we had better measures of all of this stuff. I wish the standard way, the survey, what's the most important issue to you? What do you think is the most? And it is true that most people just say the economy, which is a big nebulous thing. And of course the economy is important to everybody. I just wish we had a better way of actually, I don't know, capturing this stuff.

What we really care about is something like, if it turned out that Obama could get this much of a better outcome in some policy domain or change his position this much in a policy domain, how many people would change their votes, and so on. Those things are pretty hard to get at.

Ryan pointed out in the interview, just as an example, he said, Tthere's a decent number of voters that actually say they care about inflation, but almost none of the campaign workers said that they cared about inflation, nor did they imagine that the voters cared about inflation.” But it's not like inflation was the right answer, because you only get to pick one choice. You just get to say, “What do you think is the most important issue to voters?” And it's only 6% of voters that actually think inflation is the most important issue. 

Anyway, all that's to say, this seems very complicated and I don't know, it seems like maybe we're going to put a pencil in that, there's going to be a better paper one day on measuring the issues that voters care about most, and we're going to cover it at some point. Somebody should write that.

Will Howell:

The economy is clearly a bundle of issues, and they don't all go together. And I guess I would say further, it's not just unemployment versus inflation, it's how you talk about it. That you could talk about inflation going down or up a point or two, or you could talk about the price of gas when you're knocking on the door, the price of milk. Which then points to, it'd be nice to have data on transcripts of what's actually communicated, what's actually said, and to see the extent. I mean, what may be happening is that these people, when deployed, are basically just reading from a script, and then finding a way to connect with the person, to have a moment where they get a laugh or they share a moment. Do you know? I mean, it may well be, it's not just running through the issues.

If that's true, then these differences that are underscored here may not be especially problematic. It's, again, they're effectively reading from a script and there's very little variance across people when they're actually talking about the issues, and they may be instructed not to talk about their own ideological convictions. I mean, I don't know. This, as background knowledge, would be helpful to know.

Wioletta Dziuda:

There's an argument to make to that, and I know Anthony is going to get very upset, but there is an argument to make that what voters care about is a little bit indigenous to what the politicians tell them to care about. Everyone remembers the Tea Party and how important the budget deficit was. This seemed like people who I would think don't have a good sense of how microeconomy works, were saying that balancing the budget is the number one thing that they care about, and then this all went away quickly after the next election. So I think it's a fascinating topic in itself. Can we actually sway people by just taking certain issues and making them more sane than they are, or that they should be?

Anthony Fowler:

There is some interesting work that maybe we should cover at some point, people doing these conjoint analyses where you can say, suppose this candidate came out with this position instead of that position, and you can get a sense of not just what positions the voters prefer, but how much do they actually care about that, how much are they willing to trade it off relative to other things that they care about? We should come back to that topic.

If you're interested in hearing more or contrasting views on today's top political issues, be sure to check out Politics & Polls, with Princeton professors, Julian Zelizer and Sam Wang. The show features some of the best minds in politics, journalism, academia, and even the arts. The result is a respectful, intellectual, and oftentimes fun debate and discussion on the history of politics and current events. The show is part of the podcast enterprise of the Princeton School of Public and International Affairs. Listen to the latest episode at politicspolls.princeton.edu.

Should we come back to vaccines? There are people knocking on doors trying to get people vaccinated. I don't know what the demographics of these people are like, but if you had to guess, you would guess that they're similar. You'd guess that the kinds of people who are especially motivated to go try to persuade people to get vaccinated are people who are themselves probably politically pretty liberal, people who themselves think it's just completely obvious that you should get vaccinated, that might not actually be able to relate to the concerns of somebody who's on the fence.

Wioletta Dziuda:

I think I would worry about that. I think people that are more likely to go out and reach out and convince people will be medical workers, for example, and they are very different than an average American. What I like about this paper is that it points people to this problem, but if you think about Biden's administration and trying to organize some outreach, they might be a little bit less constrained and they should think a little bit more about whom they are putting out there.

There's, for example, a paper by our colleague, Andrea Dubell, that looked at Ebola epidemic in Africa, and what she found is that it's really important that the local church leaders, faith leaders, are trying to persuade people, because those are people who have a lot of trust, and also they come from the community so they are similar to the communities. 

I think that could be deployed in the US. We can't do that perhaps so much for the campaigns, but you can definitely do that for the vaccine outreach and that's more likely to be successful.

Will Howell:

There are a lot of margins off which you might work. One has to do with whether or not we look alike, another one has to do with whether or not you, when trying to persuade me, are a trusted source or whether or not you have private information that I don't have, or whether or not you have an enthusiasm that is indicative of, or speaks to, the stakes involved in a particular contest. All these things are bundled up in a persuasive appeal, and what this paper does is tease out one point of difference. But we could see how other things may be in play as well, and may correlate with, in a sense, age or how far you travel. That makes it hard to make clear predictions about whether or not these appeals are successful or not.

All of that said, Enos was, as a good social scientist, very careful not to generalize from this particular sample, this particular year, this particular campaign. My own sense is, I mean, you guys didn't say this, but that these findings probably carry over into certainly other campaigns and probably into other kinds of context.

I share your concerns that these kinds of differences may be even more alarming in the context of a pandemic, that in the context of a pandemic, it may be especially important that it be somebody you know, somebody who looks like you, suggesting that you take a vaccine about which we don't know the long-term health implications, or at least we have imperfect information about it. That difference may be especially important.

Wioletta Dziuda:

In a sense, you're right that the pandemic raises the stakes. So if it raises the stakes, then maybe it's particularly important that I trust the person who is talking to me, because otherwise I might be willing to go with whatever the person says for some behavioral reasons, because my the decision is not consequential, I'm not pivotal in the election, but here it's decision about my house, I'm going to be more attuned to what I hear and more skeptical.

In this sense, I think it's harder. But on the other hand, in some sense, and I'm saying it with a lot of hesitation, but in some sense, it's a common value environment here. You know, of course some people will like Obama, some people will dislike Obama, because not everyone is identical, but if you think about vaccines, either they work and are safe or they don't, or somewhere in between. But if we all knew the truth, there would be much less disagreement than in the political environment. So in that sense, you would hope persuasion would work better here. But I share your concerns for some reason—

Anthony Fowler:

That's not obviously right. I mean, that doesn't have to be the case. There is often, in the context of these persuasion campaigns, an arrogance of the side trying to do the persuading, which is, if only everybody had all the information that I had, then they would all make the same decision that I'm making. But people have different values. They have different preferences. You can inform people about the best available evidence on the side effects of the vaccine, the policy positions of the candidates, etcetera, but it's going to be pretty hard to shift someone's underlying values and principles and things that they care about, and so it might be the case that there's no campaign that's going to reasonably persuade people without bamboozling them.

Wioletta Dziuda:

Yeah, no, you're absolutely right, but all I will say is that I think, when I think about the political environment, there the common interest is less obvious thing. It's more likely that even if we have all the information, if we had all the information that's available, we would still disagree because I like the distribution and you don't.

But with vaccines, there still might be some moral issues, and I might just care about other people more and or less so I might be more willing to vaccinate myself to protect others. So there will be still disagreement, but you would think this disagreement should be smaller in some sense, it's hard to compare, but smaller in some sense, and not as obvious as it is in the political sphere.

Will Howell:

But the vaccine, at least in principle, they could be more or less efficacious for different kinds of people, and to the extent that there's risk involved, you need to weigh that risk against different environmental concerns. How prevalent is the disease? How likely am I to get COVID? And to the extent then that I'm having somebody who looks exactly me, who is operating in the same environment and facing the same risk profile, who comes and tells me, I got the shot, that, at the margin, should be more informative than somebody who looks radically different, for whom the vaccine may have a different effect, and who may be looking at a very different environmental risk profile.

Anthony Fowler:

Agreed. That's the same argument that you disagreed with in the political realm, by the way. But yes, that same argument works in the political realm.

Will Howell:

I know. I'm all over the place. That’s right.

Wioletta Dziuda:

It's a safe space guys, we can contradict ourselves.

Will Howell:

That's right. But I guess that's why I say I'm not clear on why I come out this way, but I think that this concern strikes me as more acute in the context of a vaccination campaign then in the political world, in which I think I can imagine the possibility of difference actually being useful.

Anthony Fowler:

Will, what's your bottom line?

Will Howell:

Well, look, I really enjoy the paper. It's super transparent in terms of what the findings are. It took a lot of ingenuity to collect these data. This is a population that matters a bunch for politics, about whom we knew very little. Just from the standpoint of learning about this group and how they differ from the larger population, I thought it was terrifically valuable and helpful.

I am wary of some of the conclusions that they draw about the significance of the differences that they document, just for the reasons that we've been talking about. I think we can tell a number of different stories. It would be terrific to get, again, more data on the efficacy of these different appeals, conditioning upon the different kinds of differences that are laid out here, and then to also learn more about what actually happens after the door is knocked and somebody answers, whether or not the differences that they document bear upon the content of the appeal that's been made. So lots of questions, but I really enjoyed reading the paper.

Wioletta Dziuda:

I am less skeptical than Will is about a story that they tell on the side, that the fact that the canvassers are different than the voters is a bad thing for persuasion. I actually believe that that's true, and there's some other evidence that we brought up that suggests it might be true.

Given that, I really value their paper because I think what they show, this may be something that people thought about, but really didn't have any very firm evidence of that. If we are relying on volunteers to do the persuasion for us, we are going to get people who are really aligned with us and they are super excited, but that might not mean we are going to be effective at persuading because they will be very different than the people we are trying to persuade.

I think, again, it's obvious maybe once you to state it, but I think it's just phenomenal that they were able to show this to us very clearly, to bring this to our attention. And I think once you know this paper, when you are designing the campaign to promote vaccination, you are going to be a little bit more careful and clever about how you solicit who's going to be the agent on your behalf.

Anthony Fowler:

I agree with everything that's been said. I think it's a really interesting paper, that it's really cool, unique data that we don't normally get access to as political scientists. So I thought that's really exciting that they can bring that to bear. I share the open questions that you guys have raised. I think it'd be great to have even more research to figure out what the implications of all this are, but the findings and the conclusions that the authors come to, they ring true for me.

I sit in rooms, nowadays virtual rooms, with people in the policy and the campaign world, and I look around and I ask myself, have any of you people talked to a Republican? Because that would really help, I think, because they're so out of touch with reality and they don't have a sense of what the median voter thinks or cares about that that they're trying to persuade.

The same is probably true for a lot of our public policy students at the University of Chicago. They spend a lot of time studying, they convince themselves about what the right policy is, and they think, “If only I shout it louder to the politicians,” and they don't think very much about the fact that those people have different preferences and different values, and they don't think very much about how the right effective ways to try to persuade those people. So that's a problem. I think that's a problem for people who care a lot about policy, care a lot about politics, is that they might not actually be the best persuaders themselves, and I think that's a big problem that campaigns have to deal with.

I think the findings are really interesting and I'd love for us to come back to this topic another time with some even more research.

Thanks for listening to Not Another Politics podcast.