Colloquy Podcast: How Your Neighbors Shape Your Politics

We hate each other more than we used to, at least where politics is concerned. Measures of effective polarization, the animosity that Democrats have for Republicans and vice versa, have increased dramatically since the 1990s according to a 2021 study by the political scientists James Druckman and Jeremy Levy. Moreover, the most polarized folks are the ones most likely to vote in primaries, resulting in more extreme general election candidates, which polarize voters further, and so on and so on.  

Boston University Professor Jacob Brown, PhD ’22, says that where we live shapes the political party we join and the candidates we vote for. The places where we grow up shape our views and social pressure influences our affiliations. Moreover, when we change neighborhoods or our neighborhoods change around us, our party ID can change too. That fact—that our affiliations are not necessarily set in stone but can shift as the people and places around us do—may offer some hope for the future of civic life in the United States . . . if we know what to do with it. 

The following transcript has been edited for clarity and correctness. 

So set the table for us, if you can, and describe the residential and political landscape in the United States. Has it become more segregated? And if so, since when and why?  

Sure. So increasingly in the US, we have really fine-grained electoral data that allows us to explore the dynamics of political segregation across a wide range of geographies. 

You can see for any election the breakdown of Democratic and Republican vote at counties or precinct levels—so higher and lower geographies—or as high up as states. And we also have individual data on where voters live. Voter registration records in this country are, for the most part, public record. You can't see where someone votes, but you can see where they live and, in some states, which party they're affiliated with. So, all these tools that political scientists are increasingly using to try and understand the extent to which we are politically segregated or becoming more politically segregated. 

And when you look at the data, so both over time and sort of in the now, you can see that across broad regions and at kind of lower levels, even looking across, say, neighborhoods within the same city or metropolitan area, there's geographic separation happening between Democratic and Republican voters. This is kind of easy to see at a broad level. You look at an election map or a county map from the New York Times, and you see signs like coastal versus rural Midwest, like, a red-blue divide. You see this prominent cleavage of the Democratic vote being correlated with population density. It's maybe one of the most pronounced features of our geographic polarization. 

And you see these more low-level separations happening even within cities. So, I think one important caveat here is that, like, I'm gonna explain historically why I think this is high or at least higher than it's been previously. But there are still, even in a geographically polarized electorate, many examples of places where partisans do mix. If you look at the distribution, say, of Democratic vote share across counties, there are lots of counties that are close to this, like, 50 percent thing. But what you see across time, especially in the last few decades or so, is that more and more voters live in more homogeneous places, so there's a slight hollowing out of this middle that is happening. 

So, there's this nice paper in the Journal of Public Economics by Ethan Caplan and his coauthors where they look at this longtime series of presidential vote share from the eighteen hundreds to the present day at the county level, state level, and then more recently at the precinct level. And what's really interesting about some of their time series is you get this really long historical comparison. And so what they show is that the level of clustering across counties, sorting across counties that we see right now, is as high as it's been at any point in American history. And in the time series, you see that, to go back to a point where it's comparable would be, like, right after the Civil War, which, you know, sounds sensationalistic, but it at least gives a historical context of how we should think about this. The more recent increase started around the nineteen seventies or so. 

So there's this period of realignment in American politics where the parties are sorting on demographics that are themselves spatially clustered, and that this over time produced these kinds of durable increases, election after election, in slightly more people living in slightly more homogeneous places. And my coauthors and I have tried to bring more specific data to this question in recent elections. So we've looked at individual data on voters over the last fifteen years or so. And what we're trying to do in that paper is, one, sort of just bring new data to confirm these findings that are increasing over time, and we find this consistent increase. But what we're really interested in is why it's increasing. 

You can imagine a few different models for how we could get a more geographically polarized electorate. I think the most intuitive idea is that Democrats and Republicans, when they make their short-term mobility decisions, are rapidly choosing to live away from each other. You can imagine that that would produce a more segregated map over time. But you can also imagine other channels. So let let's think about a map where nobody moves, but people are changing the parties that they are affiliated with. 

If that itself is more and more spatially clustered, we'll end up with a more segregated map. Similarly, the electorate changes election after election. New voters enter the electorate. People die, and leave the electorate. 

So we're not dealing with the exact same people we were dealing with for however many years ago. So the extent to which new voters emerge in the electorate who are themselves sort of spatially clustered in a way that's correlated with partisanship, that also can produce a segregated electorate. And so what we do with our data, because you can observe people over time as they emerge in the electorate, you can observe people who move, and observe people who change parties, is we can trace trends into these sources. And what we find is pretty interesting, and it corroborates some past research that raises this puzzle—that the short-term mobility story can't really explain the year-over-year increases we see. People don't move that much anymore and it's decreasing year over year. 

And while there is partisan bias in mobility when Democrats move, they tend to move to a slightly more Democratic area, and the same for Republicans. It's not strong enough or happening in high enough numbers to explain these changes. So what we find then is that in Democratic-trending areas, so mostly cities, higher population areas, that what is predominantly driving those places more blue is the new voters entering the electorate, either young voters coming of age and entering the electorate or adults registering for the first time, some who have just become citizens, some who just haven't registered yet. And their rates of democratic partisanship are at a rate even higher than the already democratic places they're emerging into, and that's sort of slowly driving these places to be a little bit more blue election over election.  

Republican areas, though, have a different story. These are the lower-population areas. In some places, they're literally losing population, more rural areas. And there you see that party switching is the predominant story, at least in our data, that generally older, generally white voters are switching, either defecting from the Democratic party or switching to the Republican party. At a clip, while party-switching is relatively rare, at a clip at least high enough to sort of push these places a little more red. In both Democratic and Republican trending areas, we find second- or third-order impacts of mobility, so there is some contribution from the kinds of short-term moves people are making. 

But I think, similar to past research, it's not the major driver of these more recent trends. So I think the general takeaway from that, one, is that there's a phenomenon that is increasing. It's increasing at a bunch of levels. You can see this across neighborhoods, and across high regions. And also the reasons it's increasing are more about kind of the macro-level processes that govern who the parties represent more so than Democrats and Republicans rapidly fleeing from one another every time they choose to move. That's the way, at least, I would think about it.  

So you're saying we have these two different things going on. In one part of the country, particularly the blue areas, it's expansion: young people, adults registering for the first time, and immigrants who become citizens. In the red states or red counties, it's people actually changing political affiliation. But does that make the polarization itself, how people feel about their politics or relate to the party they identify with, different in any way in the two different places?  

I think so. So I think these three I think if we include the residential mobility story, which again is sort of a second-order thing, they have these different implications for how you should think about polarization. I think the most dire one would be if, like, where we chose to live was another symptom of, like, political discord. 

I think that was the original interest in some of this residential mobility story, and this one of the original research on it is because we thought it was maybe another symptom, another indicator of, like, more partisan affect. So if that were the story, I think we'd have these in either one place or both places—would have these more dire implications of how we think about that. But for new voters and the electorate of people switching parties, I mean, you can view people switching parties as a slower process of people either changing their attitudes or just the party shifting and people kind of, like, catching up and that now these are the parties that represent me. So it's less of some sinister thing happening and more just sort of the sorting of the electorate into the parties that best represent them. And you can maybe view the generational turnover story in kind of a similar way. 

For whatever reason, the Democratic Party has captured young voters in these specific areas at a high rate and in every place, higher than the people that, like, they're replacing. So there's kind of like two ends to this generational turnover story. It's not just young people entering. It's also that they're replacing people who are dying off slowly over time as well. So there's sort of these generational differences as well. 

So, yeah, I think it's a little bit more about sort of, like, a representation story, which for me carries less of a, like, negative symptom of polarization story in that case. Although, we do see that perhaps regardless of the reasons why a geographically segregated electorate emerges, there are some documented consequences. For example, it's harder to draw districts in a balanced way when we have an imbalanced electorate. It's harder to coordinate across regions to implement public policy like public transit, when cities are polarized compared to their suburbs. So sometimes, regardless of the actual reasons, these things can still have sort of exacerbated consequences. 

Although you could imagine a world where those things were even more difficult if it was, for different reasons. 

What does it say about the nature of polarization, what's driving voters in the different parties, and whether or not they can bridge the gaps? 

Yeah. So one thing I'll say about that is that you can view people switching parties, say, or even new ones that enter the electorate, as part of this long process, people sorting across the parties on different demographics. And so there's some work showing that the more and more the party labels become aligned with other demographic or identity indicators that people care about, that this might exacerbate partisan conflict because now it's not just that you're on different political sides. 

It's that all these other things that are important to you, like your religion, sexual orientation, et cetera, are correlated with this. And so I think one way to think about these, as I said, these, like, macro processes that govern who is and who is not a Democrat or Republican, and these things are spatially clustered, is the sorting along these things. So that maybe that gets to your question of what it speaks to sort of, like, the current state of polarization is it's another way of seeing that we're increasingly separated by politics mapping onto a bunch of different, demographic markers. One thing I will say is that the 2024 election saw sort of a receding in some of that sorting. You know, you saw gains made in the Republicans among voters of color, some gains made in young voters. 

So, you know, if those things are sustained, it's entirely possible that some of the story that we've documented in past data could get reversed or at least slowed down in different kinds of ways, although it depends on sort of the geographic specificity of where all that happens. So let's talk about the research that just got accepted by the Journal of Politics. Talk about the big questions that you were trying to answer with this work. And, if you can say a little bit about how you went about it. There's this, I think, not yet really well-tested question of how exactly from a behavioral standpoint they were influenced by political segregation. 

There's a lot of good research on how institutions are having a hard time with the functioning of democratic governance in a more segregated electorate, whether it's at the representation level or public policy level. But while I think it's a really popular hypothesis that maybe we're influenced in all these ways by the politics of the people around us or the political norms we live in, I think a very popular hypothesis is that it's making us more polarized. In my mind, that was, you know, tested but could be more tested, especially given that we increasingly have fine-grained data on where people live, fine-grained data on where their neighbors live, and fine-grained data on some political behaviors that might be a function of those things. So what I try to do is take those measurement challenges really seriously and sort of bring that, like, individual-level data to this question. So I tried to look at people as they either change neighborhoods or as their neighborhoods change around them and see if there was a response to the changing political context. 

If you end up in a more democratic place or if your neighbors are more democratic through people moving in or other things, do you respond in some kind of way, whether that's from direct interaction or more indirect, like, social norms or maybe your neighbors are a heuristic for what kind of representation is good for you? It's hard to say, although we try to bring some evidence to that. But the main goal is to try to sort of view voters over time in a really specific kind of way and try to track the changes in their own voting and their own partisanship and see whether there's a sort of conformity effect that occurs. Do you adopt the partisanship of the people around you? And then in the data, we tend to see that. 

We see that if you live in a more Democrat—if you live in an area that becomes more Democratic or becomes more Republican, you become a little bit more likely, although I say the effects are sort of modest, to adopt the partisanship of the people around you. The effects may be a bit more dramatic. In some studies, they've looked at people moving to different kinds of places. They've looked at people moving, right before or after election season to, like, a more Democratic place. [University of California Los Angeles Professor] Ricardo Perez-Truglia has some papers on this and shows, like, democratic donations go up when you move to a more democratic area. And then [University of Bologna Professor] Enrico Cantoni and [Harvard Business School Professor] Vincent Pons looked at people moving across counties and states and lower levels as well to sort of show that when you move to a more Democratic or Republican county, you become more likely to sort of adopt that registration yourself. So there's all these reasons, and in my particular paper, trying to think about really low-level kind of context, knowing exactly who your neighbors are and seeing how when that changes, is there some kind of response we can observe? And we tend to see that in the data.  

Talk about that a little bit more. What did the data show you?  

So the data show you that, you know, you sort of look at someone across elections also, like, between presidential elections. And if it's from, say, 2016 to 2020, the year for your closest neighbors, so looking at sort of the neighbors in the voter file that live close to you and around you. So really, I'm trying to really get at the, like, residential contact potential contact that you might have. When that shifts more Republican, you become more likely to switch your registration to Republican. 

And conversely, when that switches more Democratic, you become more likely to switch Democratic. You know, the challenge in that kind of research is trying to isolate exactly, like, why that might be happening. There are all these challenges of causal inference in that kind of study. You know, is it because the political dimensions are shifting, or is that just a proxy for other things that are shifting? How well can we make those comparisons? 

And this paper tries to grapple with that. And there's also this question of, like, why this might be happening. So one other thing we do in the paper, it's kind of the second set of analyses, is we surveyed people. We surveyed people where we knew exactly where they lived because we have the information in the voter data, and we asked them, do you understand the politics of the people around you? Can you sort of accurately guess that? 

And people seem to be able to. So even accounting for, like, the higher level, like, the partisanship of, say, like, your town, you can at least directionally tell me if you live around more Democrats or Republicans in a way that tracks with objective reality. And then trying to get these reasons why this adoption might happen, we ask people questions about how comfortable they are, like, expressing their partisanship or the idea of their neighbors knowing their partisanship, how comfortable they would be putting out a yard sign that display is part of public displays of their partisanship. And this is trying to get this idea of, like, social pressure that you'd be, and what we see in the data is that people are more likely to express this kind of comfort, express this willingness to engage in these public displays of their political preferences if they're in the political majority, and that they're less likely to do this if they're in a public minority. And so, you know, we can, of course, think of exceptions to this. 

So, like, your neighbor who has opposite political views than you and is very loud about their politics. But on average, at least, there's some level of sort of, like, norms that people internalize and don't really want to violate, and that this may speak a bit to some of these other behavioral effects that we see.  

You started off talking about the different nature of polarization in blue states and red states; how, when people move to a place where the majority party is different from their own or when political affiliation changes around them, they're more likely to change their party registration. That's the story of residential segregation and political participation in the red states. You said the same kind of thing goes on in Democratic states, but is it at the same level?  

Yeah. So the behavioral effects of real estate are generally at the same level. I think the key point when, like, comparing across the sort of these two statements I said, which is that the predominant thing driving blue states bluer is because of new, first-time voters, energy, electorate. It's really a question of, like, the numbers. 

So in these more populous areas trending Democratic, lots of voters are in the electorate. Some voters are switching, but it's relatively few, and the rates are such that it’s sort of swamped by these other factors. In these Republican-trending areas, there are fewer voters entering the electorate there because it's just generally lower population, generally lower levels of turnover, and so there's more potential for this a similar level of party-switching to have this kind of effect. Although, I think you still might see, like, slightly stronger, effects in these Republican areas. But there's also sort of, like, two things to distinguish between here. 

There is the, like, general partisan switching that is happening. Some of that might be because you're being influenced by the place around you. Some of it might be because of just other things that are sort of causing you to change your party. And what the paper is trying to do is speak to how much you're being influenced by changing context around you versus just, like, the general spatial clustering of any kind of party switching, which could happen for lots of reasons that are related.  

You talked about surveying people to see how comfortable they were with expressing their political opinions, putting out a lawn sign, or something like that. That speaks to the role of social pressure on partisan affiliation. Is it implicit in the sense that the people around you are changing or you move to a neighborhood where the partisan majority is different than the one that you come from, or do you have evidence of direct social pressure being put on people?  

So I don't have good evidence I would say in either direction. My hunch is that it's more indirect. That this is more about, like, an internalization of norms and either, like, feeling internally, even though there's not, like, a real reason to do this, some kind of pressure to conform to those norms, or those norms are kind of like a learning heuristic for you, particularly for people with, let's say, like, weak partisan attachments, which are the kinds of people you're going to see switch their party. They might view the general trend of their neighborhood as, like, a reasonable informational signal of, like, which party better represents them and might respond to that in some kind of way. 

You mentioned also that the effect you noticed in the data was relatively moderate. If that's the case, why is it still important to take note of these findings?  

Sure. Yeah. So there's this challenge here with geographic research, that the overall effects of place can be in combination, relatively substantial. But once you try to isolate, which is very hard to do, any, like, the effects of any individual characteristic and isolate that effect net of other effects, you're going to get this, like, bundle of various of, like, small effects. So I think what's important here is that these are behaviors that, for the most part, are pretty sticky. Like, which political party you have a preference for or registered to doesn't change that often. So even pretty small changes in response to these low local stimuli are interesting, like, behavioral dynamics. And perhaps, although I again, because they're small, they can't be, like, the primary driver of, you know, partisanship itself or geographic polarization itself, but they are sort of one input in one way that we should think about geographic polarization having a reinforcing quality. 

And, again, what we're trying to do in our paper is, like, look at very local stimuli and try as best as possible. There are challenges in the data trying to attribute it directly to, like, the politics of the people around you. But if we go back to this idea of geographic polarization, in part being the party sorting along a bunch of different demographics, you could imagine that it also leads to these more compounded effects that are themselves larger. So in some other work, we've looked at the effects of, like, the longer-term effects of where you grew up, And there, you see more substantial effects, in part because we're looking at a longer time span and maybe more formative period in your life, but also because in that particular analyses, we're doing less work to distinguish between specific characteristics of place, and we're looking at, like, just the overall compounded effect that place can have on you.  

What should political scientists and policymakers take away from your research, particularly if we believe that polarization is a problem for our democracy and our society? 

So I have two ways of thinking about this, and I get torn between them. So the more optimistic side, I want to think that if I am influenced by my local neighborhood in a way that causes me to, like, be more amenable to their political ideas or political affiliations, that's an example of local interests superseding more nationalized partisan polarized interests. And so that from, like, a behavioral standpoint might be nice. More generally, though, if, like, if we generally are worried about just, like, living in a geographically segregated society at all, then anything that's an input to that is going to contribute to this overall imbalance. And I really do stress that previous work on sort of, like, institutional imbalances is, like, sort of pretty well documented, some of the negative consequences of geographic polarization. 

It contributes to congressional polarization, contributes to representational imbalances, and is an obstacle to public policy. So to the extent that even if it's an example of, you know, I feel more friendly towards my neighbor and I happen to adopt their politics, it still might be contributing something that is a sort of net challenge for a democratic society.