The Kids Who Got Bused—And Became Democrats

Subscribe here: Apple Podcasts | Spotify | YouTube | Overcast | Pocket Casts

The school-busing policies of the 20th century weren’t just unpopular; the ensuing fallout to school integration was so great that many Americans have written it off as a total failure. White flight, political backlash, and continuing segregation: This is the legacy of court-ordered desegregation.

And on quite a few metrics, the pessimism is warranted. Schools across the nation remain segregated; one academic measure of the nation’s 100 largest districts finds that segregation between white and Black students has increased 64 percent since 1988.

But a new study complicates this narrative of total failure. In today’s episode of Good on Paper, I talk with the economist Ethan Kaplan about his research on the rare two-way, court-ordered busing program implemented in Jefferson County, Kentucky, in 1975. Kaplan and his co-authors traced the cohorts of students affected by this policy through the decades to see how their experience shaped them. Because the students were assigned to be bused in a semi-random way (based on the first initial of their last name), it provided an ideal experiment for researchers.

The findings surprised me. White students who were assigned to be bused into majority-Black inner-city schools were significantly more likely to identify as Democrats later on in life, more supportive of redistributive policies and unions, and less likely to believe that success is earned. But what’s even more interesting is Kaplan’s explanation for why this is happening. The political shift wasn’t driven by softened racial attitudes, but rather, by white children witnessing the deprivation of inner-city schools.

This conversation revisits an era that many Americans would rather forget—and asks what it means if one of the most controversial policies of the 20th century might give us a clue about how our adult political identities are forged in adolescence.

The following is a transcript of the episode:

Jerusalem Demsas: School integration is largely thought of as a depressing and failed saga in American history. According to a GAO report, during the 2020–21 school year, more than a third of students—that’s 18.5 million kids—attended schools where 75 percent or more of the population was of a single race or ethnicity.

And it’s not even that things are just slowly getting better. Research by the University of Wisconsin shows that in 1988, roughly 37 percent of Black students were enrolled in majority-white schools. Fast-forward 30 years, in 2018, that figure was just 19 percent.

[Music]

Demsas: My name’s Jerusalem Demsas. I’m a staff writer at The Atlantic, and this is Good on Paper, a policy show that questions what we really know about popular narratives.

When I first saw the paper we’re going to discuss, I felt a stab of hope—was it possible that the political impacts of busing policies were better than I’d believed? Ethan Kaplan, an economist at the University of Maryland, and his co-authors had zeroed in on Jefferson County, Kentucky. In 1975, a federal court had ordered the public schools there to desegregate. Now, 50 years later, Kaplan and his co-authors found that this experience had affected the politics of children who had been assigned to be bused.

According to this research—busing significantly increases support for the Democratic Party and causes important and measurable shifts in the ideological outlook of white children who experienced the deprivation of inner-city schools.

I’m excited to dive in. Ethan, welcome to the show.

Ethan Kaplan: Thank you for having me.

Demsas: So let’s go back to the summer of 1975. The Louisville Jefferson County School District was forced to implement a school desegregation plan by the courts. What did they do?

Kaplan: So there had been a couple of lawsuits that have been long-standing, and they finally got resolved in the spring of 1975. And then the higher courts told the lower courts that they had to actually implement it for the fall of ’75 school year. And so they had to very quickly merge two school districts, one that was largely the city of Louisville—and Louisville, by the way, was not a part of Jefferson County at that time; it later became a part of it—and the other was largely Jefferson County but had a few mainly white schools that were in the city of Louisville, which actually was important for the legal case. And so they, first of all, got the heads of the school districts—the city school district and the county school district—together with a judge, and they created a plan to integrate the school systems.

And in doing that, there were, you know, really three types of schools. So there were some schools that were exempt because they already were integrated enough. There were some schools—and as a sub-part of those schools, there were actually some schools that, since they were in the city, but they were in largely white areas and areas that have been carved out to be very white and have white schools, they were actually, through redistricting, able to meet the qualification for not having to have busing. And so they were also exempt.

But then there were the predominantly white schools in the county that needed to bus white kids into the inner-city Black schools and Black kids in the inner city that needed to be bused to the white schools.

And, you know, there are many different types of busing programs that were implemented across the country. This was a two-way busing program, so they really tried to integrate—they weren’t just bringing Black kids into white schools. They were also bringing white kids into Black schools. And so there was this two-way busing.

So they tried to really roughly equalize the flows, you know, so that the fraction of Black and white were basically the same in all of the schools. And so they had to come up with a kind of way to do this in a very short period of time.

Demsas: And what was kind of the tenor of the desegregation fights in Louisville, Jefferson County? Was this something where, I mean—I think we have the traditional story in our heads of, you know, parents extremely upset, pulling their kids out of the school district, rioting, protests, a lot of animosity. Black parents are also similarly concerned about their kids. Can you set the stage for us here?

Kaplan: Yeah, that’s a great question. I would say it was less conflictual than some of the most conflictual integration experiences.

Demsas: Like Boston.

Kaplan: Yeah, so that’s the funny thing, is a lot of people think of the Deep South, but Boston was incredibly violent. But nonetheless, you had large KKK rallies. They polled white parents at the time, and 98 percent of white parents surveyed were opposed to the busing plan, and it’s really hard to imagine 98 percent of people today agreeing on anything.

So there was just near-unanimous opposition to busing. Some of it maybe was just due to not wanting your kid to be on a bus for a long period of time. But, you know, some of it might also—I mean, certainly a bunch of it was motivated by racial animus, as evidenced by the prevalence of the KKK at some of these rallies.

Demsas: There’s an interesting debate—I’m not sure how big it is in the literature, but—trying to characterize why people were opposed to busing. Of course, there’s the racial animus you mentioned. How seriously should we take arguments that there were other, larger concerns that parents may have had, whether it was distance their kid was going to get to the new school or the fact that you’re interrupting where they were going to school to begin with, or the idea that, you know, a lot of parents when polled care a lot about being in their neighborhood school for whatever reason?

Are those really big reasons for people not to want to integrate their schools? Or I guess we could test this with: Are parents more happy if their kid gets to stay in their neighborhood school while integration is happening versus having to get bused themselves?

Kaplan: And we do see that. There’s around—we don’t have perfect estimates of this, and we have estimates from different sources. But, you know, roughly 15 percent of students who were scheduled to be bused differentially left the school system. So definitely part of it is that they didn’t want their kids to be bused to another school. Now, that could still be related to racial animus. Like, it might be that they didn’t want their kid to go to, you know, an inner-city school in a Black neighborhood, right? So, you know, I wouldn’t say that’s proof that there was a really large component of this that was really just about opposition to time spent on a bus.

But certainly, kids who were bused out were much more likely to leave the school system and go to other public schools or predominantly Catholic schools. Actually, in the Deep South, there were segregation academies. They were actually literally called “segregation academies.” That was quite common.

Demsas: Yeah. And do we know what percentage of white students who had their schools integrated by Black students left?

Kaplan: Yeah, we don’t have perfect estimates of that. So one way that we do this is that we look at yearbooks. So if you look at the fraction of people who, in the fall of ’74, before they knew if there was going to be forced integration and before they knew anything about the particular plan through which integration was implemented—if we say, What fraction of those people can we match to a yearbook after integration has started?, that’s 70 percent.

Demsas: Okay, so 30 percent leave.

Kaplan: I think it’s not that high, is my guess. You know, like, for instance, if you’re not there on the day when they’re doing the yearbooks.

Demsas: Yeah. Or you move for other reasons, or whatever.

Kaplan: Right. But those people actually would’ve left the schools. And you’re right that there is a background rate of leaving the school system, period. But anyway the point is, you know, most people stay. I think if you look at the aggregate data from the school system, there is a large decline right around this time. So it’s really hard to disentangle, but my guess is somewhere around 20 percent, is a guess of what percent left.

Demsas: So I want to come back to the interviews you did and the yearbook analysis. But I first want to ask you about why you chose to study this desegregation program.

Kaplan: Great. So they had a very unique way of implementation that was really perfect for causal analysis. So, you know, in economics, we are concerned a lot about causality. So we see a correlation—how do we know it reflects a causal relationship versus some type of correlation? So for example, education is very highly correlated with wages. The higher your education, the higher your wage. But does that reflect the causal effect of education, or are there some things that lead you to get education that also lead you to get higher wages? So economists spent a lot of time trying to figure out that kind of causal relationship.

And so one thing that can do that is randomization. You know, so if I randomly assign you to get bused versus not, then we think that we’re getting the effect of busing. Right? Whereas if you’re choosing whether or not to get bused, then we think, you know, maybe your political opinions might be correlated with whether or not you’re signing up to get bused.

So, you know, we really look a lot for these randomizations, and in our case, we actually found something that we think is close to an experiment. In the summer of ’75, they sit down together, and they try to figure out a way that they can send some kids every year—not the same kids in every year. So some kids get bused every year so that they equalize the fraction Black and fraction white across schools, but they don’t have the same kids being bused every year. That’s different kids. Everyone ends up with roughly the same amount of busing over their 13 years of school experience. They come up with a plan based upon the alphabet.

And they say that if your last name begins with, for instance, A, B, F, or Q, then you get bused in 11th or 12th grade. If it’s D, G, or H, then you get bused in third and eighth grade. So the first letter of your last name determines when you get bused. And the thing that’s really great about that for us, in terms of our research design, is that in those initial cohorts, if I’m in, like, eighth grade when this busing plan is announced, if I’m in A, B, F, or Q, that alphabet group, I’m going to get two years of busing. But if I’m in D, G, and H—that’s third and eighth grade—I don’t get any busing at all.

So for the initial cohorts, particularly in later years, the assignment by first letter of the last name of when you get bused leads to a difference in whether you get bused or not. And so we use that, and we say, Okay, let’s look at the cohorts who differentially end up getting bused. So if A, B, F, and Q just graduated when the busing plan was implemented, they don’t get busing, but the next cohorts would get busing. Whereas for D, G, and H, those same cohorts are both not going to get bused.

So then we can compare the change in politics 45 years later for the A, B, F, and Q group, where there was a change in whether they got bused, versus the D, G, and H groups—in the same cohorts, in the same years—but where both groups didn’t get bused. And so that kind of odd, random variation is what we use to basically figure out what the long-run impact, basically permanent impact, of being bused on politics is.

Demsas: So clearly, someone in Jefferson County wanted to design this natural experiment for you. But the question I had when I was reading your paper is that, at first glance, I was like, Okay, the last name likely doesn’t correlate with a bunch of stuff. Probably doesn’t correlate with, you know, your political opinions or whatever other things, like racial animus. But I suppose it is possible that last names are not randomly distributed, right? Like, there’s a lot of Lis that are Asian American, and that’s less likely to be, you know, white. And so did you look at the distribution of last names with race or with other things to see if they actually were uncorrelated?

Kaplan: So we didn’t do exactly that—and that’s a great question—but we did other things. And also, our design actually tries to take that into account. So suppose Z is a much more Latino name. Now, it turns out there weren’t really Latinos living in Louisville and Jefferson County at that time. But, you know, let’s say that there were, and so you were concerned that there might be more Latinos.

We’re comparing the change in A, B, F, and Q to the change in politics for D, G_, and H. So in order for there to be some type of difference, they have to be on differential trends right around those years. And we don’t find evidence for that. So one way that we show that there’s no evidence for that: We actually do also look before they get bused—we look at the difference between the groups that get bused versus not in the year that they get bused. And then we look at the years prior to that, differences across the groups, and what we see is large differences in the year that they get bused and no difference in the years before.

So we kind of take care of that because we’re looking at changes in politics across cohorts for the same alphabet groups, but we also show that, actually, it turns out that there really isn’t any difference in politics between the groups that get bused and not before busing starts.

Demsas: Okay. So this paper is about long-term political attitudes, which you forecast. And, you know, you find that white students who were bused into poor, inner-city schools develop more liberal economic views decades later. Before we get into how you found that: Was that your expectation going in? When you were kind of doing your lit review, did you think that—I mean, I think most people, when they think about the busing experiment, they think of it as, like, a failed experiment. They think of it as: The backlash was really intense. We have really segregated schools right now. People really hated being bused. And so, you know, blanket background expectation is that, you know, it would make people more conservative or more opposed to sorts of these programs that they experienced themselves. Was that your expectation going in?

Kaplan: So, you know, when you choose a paper to write, you have to think about these things. Is it likely to work out? And I think, first of all, we thought that we could find the results that we found. We also thought that it would be possible we’d get the exact opposite results, that, you know, even though the Democratic Party was the party of slavery, there’s a lot of evidence that shows, starting in 1964—when the Democratic Party implemented the Civil Rights Act, and in ’65, they implemented the Voting Rights Act—that things changed in terms of the perception, and a lot of, particularly, southern whites turned against the Democratic Party starting in ’64.

Moreover, a lot of the legal decisions that led to busing—so there’s a lot of legal innovation. So you might think that busing should have happened in 1954, after Brown v. Board of Education, or in ’55, with Brown II, which said that it had to be done quickly, and then nothing happened for a decade, until ’64, ’til the passage of the Civil Rights Act. After the Civil Rights Act, the Department of Justice starts to get involved, and there’s a bunch of lawsuits that really parameterize that this has to happen and how it has to happen.

And so, you know, we thought that there would be some chance that this process would have led to—and that legal innovation was largely done by Democratic appointees—and so we thought that there was very possibly going to be a backlash, and so we’d see a movement to the Republican Party, because of the incredible unpopularity of busing when it was implemented. That would’ve been interesting as well.

There could have been a movement towards the Democratic Party. So there’s this theory of the contact hypothesis by Allport. There’s this theory that says that when you have exposure to racial groups and there’s racial animus, that exposure will, under certain conditions, lessen the animus. And so it’s possible we could have seen that. And then it’s also possible that we could have seen nothing. And from our perspective, actually, any of those three would be interesting. And so, you know, that was a great thing about the paper.

I would like to push back on the perception of the failure of busing. Our paper doesn’t speak to whether busing was a failure or a success. You know, certainly, it was not incredibly popular. And I think that continues to be true ’til today. And, you know, in terms of the politics, you know, there’s no way to assess. It depends on your politics as to whether you think this is a good thing or a bad thing.

But my co-author Cody Tuttle actually has a separate paper that he’s working on where he uses this to look at the impact on economic outcomes. And what he finds is that, again, 40 years later, if you were white and you were bused, there’s no impact on your income. But if you were Black and you were bused, I think it was around 3 percent per year. And I would call that a success of busing.

Demsas: That’s a good point. I think it’s, like, a very pessimistic view we have. Because, you know, I was in South Carolina, actually, working for Kamala Harris’s 2019 presidential campaign when she made the That little girl was me moment on the debate stage and pointed out that she had been a part of Berkeley’s voluntary-busing program to Biden.

And I remember this: I grew up in Maryland, and so when I was in South Carolina and I was traveling around the state, it was the first time I’d really, really seen real rural deprivation before and that level of segregation. I mean, it’s not to say that D.C. isn’t segregated, but because it’s denser, you kind of have a lot of, like, normal integration happening by the fact of people going to their shopping or, you know, Metro or whatever. But you know, you could go your entire day and not see someone of a different race when you were in some of these counties in South Carolina. And it really struck me during that time just how it felt like, Wow, this has been a complete and utter failure. Like, you know, that was my perception of integration.

Kaplan: Can I make two comments about that in terms of the law?

Demsas: Yeah.

Kaplan: So I think you’re right. So actually, one of the legal decisions that comes out in the early-to-mid-’70s that almost leads Jefferson County and Louisville not to be integrated is that they decided that, really, you should only integrate within counties. And so if you have very segregated counties, then you’re not going to have integration, just because the courts had decided that integration should only happen within counties.

Now, there was an exception made for Louisville because they had already drawn districts to incorporate these very-white schools in the city. They argued that they could actually merge the school districts and have a cross-county school system. And there actually was a separate request made of the state, and the state actually agreed to this, to merge the school districts between the city and the county, independent of the busing. That was a separate thing that was going on at the same time.

The other thing is that there was a lawsuit that made its way to the Supreme Court where they ended up deciding that busing is not forever, that busing was until places achieved what’s called “unitary status.” So once you had unified the district so that there wasn’t segregation anymore, then the Department of Justice could not be involved anymore. And there wouldn’t be any forced integration anymore. And so there has been, over the last, you know, 30 years, a reduction in integration.

So there’s a really nice graphic on Vox, the news outlet, that shows the peak of integration. They look at the percentage of Black students going to majority-white schools in the South. And, you know, it’s a misnomer to think this is only about the South. It’s not. And actually, Maryland and Kentucky are both kind of odd cases because it’s not clear: Are they the South or the North? Because they’re both slave states that sided with the North in the Civil War.

Demsas: In Maryland’s case, forcibly.

Kaplan: And so what you see is that there’s a peak right around the time of that Supreme Court decision, in the late ’80s, early ’90s, in the fraction of African Americans going to majority-white schools in the South. And there’s been a large decline since then. And I think some of that is done by cross-county segregation. That busing does lead to white flight. And so some whites do leave the county. And we actually have evidence of this in our paper. We document that there are some kids who leave to go to other, more all-white public-school districts.

A far greater percentage actually stay where they are and go to Catholic schools. But as I said, in the South, they actually created these segregation academies as well. That was another source. But there definitely are people who leave the county because of this. And so some of the white flight out of cities, you know, is due to busing.

Demsas: So I want to dig in on the finding, though, because I think it can be hard to imagine isolating the effect of an event that happens in someone’s adolescence or when they’re a kid, and saying decades later, that affect their economic views, their likelihood of voting for a Democrat versus a Republican. So just walk me through what you did and how you figured that out.

Kaplan: Right. So what we did is we basically compared the changes across cohorts. We take the universe of voter-registration data from the United States, and we look at the changes across cohorts who were differentially bused—so an older cohort that wasn’t bused compared to an earlier cohort that was bused, and compare them to other kids at the same age, and see if there’s the same kind of differences across cohorts in politics. And we see that there’s basically very little difference for the kids who were not differentially bused, but decades later—this is 45 years later, so this is essentially a permanent political effect—we find a three-and-a-half percentage point movement to the Democratic Party.

Demsas: And that’s big.

Kaplan: That’s big. And two and a half away from the Republican Party. What’s a little confusing and interesting is that we find much larger effects in our survey on whether you vote for Obama in 2012 and Biden in 2020. We actually didn’t look at 2016, just because we thought that there might be some confounding things about having a woman running. And so we just looked at those two elections, and we find something around a 20 percentage-point flip, which is astronomical.

And we were pretty skeptical. But then we also asked, Are you a Democrat or a Republican? And then we see very similar numbers to the numbers we have in our paper. So when we ask them, Are you a Democrat or Republican? we see something in around 3.5 percent, which, as you said, is a large effect for something that’s 45 years later. When we look at who you vote for, it’s actually a lot larger. And so one thing that might be happening is, you know, it could be that registration is stale. And so, I’ve flipped to being a Democrat or a Republican, and I just haven’t changed my registration so that voting is more reflective of my true partisan beliefs.

But again, in the survey, we see them saying that they’re Democrat or Republican at very similar rates to what we see in the voter-registration data. So we think that part of what is happening is that a lot of independents become more Democratic-leaning independents if they were bused.

Demsas: I want to distinguish here, because when you’re comparing the cohorts and you’re finding that there’s a three-and-a-half percentage point difference, that is coming from the comparison of looking at voter registration and voting data in modern day compared to the names of people who were bused in the 1970s. But the survey, I feel like, is maybe, like—you tell me: I have a little bit less confidence in the survey because I think that there’s a lot of people who might be embarrassed about their behavior or views from the 1970s, and there’d be a huge selection effect of who’s going to respond to that survey.

If I was bused as a white student or, you know, I chose to skip out on that school because I knew there was going to be integration, there’s a lot of wanting to give you the answer that feels socially responsible and the social-desirability bias that is inherent in all surveys. But I think it might be bigger in this one because it’s just, right now, seen as so bad if you were someone who was opposed to busing, and it’s so embarrassing for you to do that. So are you concerned that you’re finding these big effects in the survey because of that?

Kaplan: This is great. Are you a referee on the paper?

Demsas: No. (Laughs.)

Kaplan: So that’s a fantastic question, and it’s definitely something that we’re concerned about, and I think you’re absolutely right that it’s plausible. And so maybe those estimates are too large. So we thought maybe they’re too large because of the social-desirability bias. We also thought that the registration results could be too small because registration might be stale. And then we also thought that maybe the independents are just more Democratic leaning.

And to be perfectly honest, we can’t perfectly address that issue. It’s a very fair point. What I will say, though, is that the evidence for shy Trump voters is pretty limited. Like, you know, the polls are not that off. They’re a little bit off. So I think the shy Trump voter is real, and the polls are a little bit off, but they’re not that off.

Now, let me give a counterargument to my argument, which I think would bolster your concerns. If you look at the response rates, there’s been an incredible drop in response rates, and particularly from the groups who are likely to be supporting Trump—you know, whites with lower levels of education—in recent years. But response rates are really, really low on the phone.

Demsas: Yeah.

Kaplan: So we did a multimedia approach, a multimode approach. So we did letters. We did online possibilities. We did phone calls. But we also incentivized it: We paid. So we gave people money up-front, and then we gave them extra money if they finished the survey. And so we got a much higher response rate. I think our response rate was, like, 30 percent, 25 percent. And so maybe among those people who are less likely to respond but will for incentive purposes, maybe they are more likely to be shy Trump voters. That’s possible. I can’t rule that out.

Demsas: Well, I guess, like, for you, though, you said when you first saw the results that you felt like 25 percent is just, like, really massive. Are you concerned that this is too big? Or do you find it plausible?

Kaplan: Yeah, I am. I mean, I’m not saying it’s totally implausible, but it’s really large. You know, I work a lot on, you know—so for instance, I wrote a paper on the impact of Fox News on voting many years ago. And you know, we found that between 1996 and 2000, the impact on the aggregate vote share was about 0.5 percentage points.

Demsas: Which matters for elections.

Kaplan: Which matters for elections. Absolutely. But you know, this is a lot larger than that. Even the registration numbers are much larger. And maybe that makes sense because (1) it’s a more formative part of your life. Political opinions are much more malleable when you’re younger. There’s a lot of evidence for that. And the second thing is: It’s a very intense experience.

Demsas: You mentioned this, and this is something that I’m really interested in, which is this notion of times in your life when you’re particularly susceptible to forming your identity, or particularly susceptible to ideas or experiences that help shape your future identity.

And you said there’s a lot of research on adolescence and childhood as being a formative time. Can you tell us a little bit about what that research tells us or what it says or how they find that out?

Kaplan: Yeah. So, you know, there are other papers that look at your risk-taking behavior differentially—if you grew up in the Great Depression, for instance, a paper by Ulrike Malmendier. She’s at Berkeley

I’ve also done work looking at 18-year-olds in Chile. Pinochet was kind of—in 1980, there was a lot of isolation of Chile from international markets because of the repression. He then—

Demsas: He’s a dictator.

Kaplan: Yeah, he’s a dictator, and he commits to having a process for a vote in eight years on whether to return to democracy. And so that eight years later comes, and he tries to renege, but even the military thinks that they can’t not do this and still participate in the international economy. And so he goes through with it, and he loses. But there’s this massive mobilization.

First of all, they’ve destroyed all the voter-registration records, and so they have to reregister everyone. And so we look at kids who turn 18 just in time to be able to be eligible to vote, versus not. We literally look at the day before versus the day after. And there is a massive difference. There’s something like, you know—I’m probably going to misremember the exact number, but there’s something like—a two-or-three-percentage-point difference in turnout.

I could be overstating that. But turnout’s really low in Chile, like 30 to 40 percent. So it’s, like, a 10-percent impact of being born just one day apart on voting 30 years later. And there’s actually a large literature in political science and in economics—both—documenting the effect of important events. And by the way, we then look at subsequent elections, and we don’t see similar things. So this is an incredibly, you know, formative event.

[Music]

Demsas: After the break: Busing may have made kids more likely to vote for Democrats, but it didn’t make them less racist.

[Break]

Demsas: So one thing I wanted to ask you about is: You’re only studying compliers, right? So there are a lot of people, as you’ve mentioned, that choose to leave the school system. Or we don’t know why exactly, but there’s a large reduction in them.

So when you’re seeing these effects over 45 years of people who are bused or who experience busting in their own schools, you’re not seeing those effects on the people who have left. So is it fair to say that what we’re actually observing in this paper is: People who were not so full of racial animus that they would remove themselves from the institution they were going to, those people see durable shifts towards Democrats? But it’s possible that the people who had to leave maybe even offset that because they are forced to leave their institution, and they’re pushed towards Republicans as a result of this experience?

Which, you know, we don’t have data on that, but I just mean, there’s a whole section of people who are most likely to have the most racial animus, most likely to be upset about what was happening to them, who—because we know about heritability of political beliefs—likely have parents who are full of racial animus themselves. We don’t really see those people in your research.

Kaplan: So we actually do. And so that’s why it’s important that we collect the yearbooks in the fall of 1974. So this is before the final decision on whether or not there would be, you know, forced busing in Jefferson County. And in fact, I don’t remember the exact timing, but it’s possible that by fall ’74, I think this other decision had already come out—it was from Michigan—that said you could only integrate within counties. And in fact, the judge that was overseeing the case initially decided against busing, based upon that decision. It then gets reversed at a higher level because they had already crossed these county lines to go into the city and grabbed these three mostly white schools in the outskirts of the city.

So we are looking at everyone who is assigned to be bused. We’re not looking at people who necessarily were bused. And so some of them did leave the school districts. We’re still looking at them. Some of them did go to private schools; some of them did go to other counties. We’re still following them.

So you can think of two different things. You can think of the effect of being assigned to be bused. That’s what we’re estimating. If you want to then convert that to the effect of being bused and for the people who were likely to stay in those school systems, then you would have to divide by the fraction of people who actually were bused.

And so if you think that fraction who stayed was, like, say, 70 percent, then you have to divide our estimates by 0.7, which means that they are about, you know, 30 percent higher.

So it could be that different types of people choose to stay, as you rightly said, and so if you get a correlation between those who stayed versus those who didn’t in terms of their political outcomes many years later, maybe that’s because of preexisting political differences. Or maybe it’s because of the impact of busing on one group and not in the other. We don’t have a good research design to disentangle that.

But we can say that being assigned to be bused increased the Democratic Party support by three and a half percentage points 45 years later. And as long as you don’t think that going to a Catholic school, a segregation academy, or a mostly white public school makes you more likely to support the Democrats many years later, then the effect of being bused is probably larger.

Demsas: So one thing I find really interesting, though, is that you don’t find strong evidence that being bused changed explicit racial attitudes. So you find that it makes people more likely to vote for Democrats, but not that it changes wildly people’s racial animus, etcetera. And I wonder what the mechanism here, then, is.

Because we would expect, like you mentioned earlier, that contact theory, that people are coming into contact with people who are different from them—they’re observing their interactions with Black students and seeing, like, Oh, I had these preconceptions about who Black people were, and now I am revising these in light of: I have a friend who plays with me, or we worked on a project together and they made a really smart point. So clearly not all Black people are stupid. Like, you know, these sorts of interactions make people more likely to reduce their biases, is how the contact-theory paradigm goes. But if you are more likely to be a Democrat but not more likely to revise those racial attitudes, then what’s going on here?

Kaplan: That’s a great question, and that is something that we grappled with a lot. And to be perfectly honest, this was probably the biggest surprise of the survey—though in some senses, I think it also makes a lot of sense. So I think the thing is that contact among peers was not that different, right, because of the two-way busing.

So if I’m white and I left to go to a formerly mostly Black school in the city, I’m going to have, you know, 20 percent African American and majority white. If I stay in the county, it’s going to be the same thing because of the two-way busing. So it’s not really clear that there’s differential exposure to peers. Now, still, it’s possible that being exposed to different-raced peers on your different-raced peers’ home turf might be a different experience and might alter opinions.

Demsas: So if you’re a white student going to a Black inner-city school, that’s a different experience than if you’re a white student now having a Black student in your class.

Kaplan: Bingo. Right. And so we still thought that there might be some effect because of that. It turns out that there wasn’t. And actually, you know, the Allport hypothesis does have certain conditions from it, and this does speak to kind of the scope of that Allport hypothesis. Because it is possible that we could have found a differential effect of being a white student going to a Black school, having the same fraction of Black peers but being in a formerly Black school versus in a white school. And we don’t really see tons of evidence for that.

But what we do see a lot of evidence for is and—oh, let me tell you about my pet theory before we did the survey that also was wrong. I thought that it was all going to be about teachers. So I thought there were going to be higher fraction African American teachers in the city schools. We actually asked them about that. And, you know, we don’t have great evidence on this, and some of the evidence we do have goes a little bit against some evidence from the Department of Education Civil Rights Division, at the time. But I thought there was going to be a higher fraction of African American teachers, much higher in the city schools.

And I thought that (1) being exposed to—maybe you’ve been taught racist tropes, and then being exposed to a great teacher who’s African American would change your views. But also there could be curriculum differences, right? So it might be the war between the states in the county and the war to end slavery in the city. And so you might get different exposure to a narrative about history, and that could change your views. We don’t see much evidence for that.

What we do see is that if you were bused, you were substantially less likely—even with the small sample size that we have in the survey, we’re able to detect this very strongly—you’re much less likely to think that the world is fair. We also see that if you were bused, you’re much more likely—by the way, we have a free-response question, Tell us about your high-school experience, and we don’t mention busing at all—and if you were bused, you are much more likely to mention that you were bused. You were much more likely to mention busing, period. But you’re also more likely to state that—we have questions about this—you have friends who were poor, that you had friends who were poor back then. You were more likely to mention the poverty of the surrounding areas and the resources of the schools.

And interestingly, we ask a lot of attitude questions, which are common in political science. So, you know, there’s partisanship, which is, Am I a Democrat or Republican? There’s ideology, which is, Do I support higher taxes, lower taxes? Those are different. You know, historically, if you go back to the 1960s, they were not very correlated at all. Today they are more strongly correlated, but not that strongly correlated.

So we ask about ideology. And in general, we find that on most things, people who were bused are definitely more progressive. But we don’t have the statistical power to say that with any statistical certainty, because of our small sample size, which is only, like, 500-and-something people. Like, I think it’s 529.

The one thing where we find really strong effects on policy-related views is on unions. People who were bused are much more likely to support unions—much more likely. Again, the unions and the world being just, we find very statistically significant effects, even with our relatively small sample size.

Demsas: So that would indicate that a lot of what’s going on here is, it’s contact between classes, not about race contact. That kind of experience makes you much more aware of different class privileges and the general unfairness of being born into deprivation in that way. So do you think that has generalizability beyond this case? Like, do you feel like there’s a lot more to be made and studied about class-contact hypothesis, and most of the literature is about race?

Kaplan: That’s exactly what we think. So we don’t know, right? We have just this one case study. But we think that there’s strong evidence for that in this case, and that makes us wonder how widespread this is, and we think it would be worthwhile for other people to look at that in the future.

Demsas: Did you recently read, Matt Lowe conducted a meta-analysis of the contact hypothesis?

Kaplan: Yes.

Demsas: So we’ve been talking about Gordon Allport’s contact-theory hypothesis. And so, you know, I think the first time I can remember coming into contact with this body of work is when David Broockman’s and Josh Kalla’s paper on trans canvassers came out. And that one made a splash because it was in 2016, and people were really concerned about the political divide. And they essentially found that there are durable effects to canvassers having deep conversations at the doors and talking about trans issues. And people were durably moved months down the line and maintained these kinds of more progressive-on-trans-issues views.

And it’s interesting—I went back to look at it, and I realized that they don’t actually find significant differences between whether the canvasser is trans or the canvasser is not trans, which indicates to me that it’s not the contact hypothesis, that people are like genuinely being persuaded, which is a kind of a beautiful story in and of itself.

But you know, a lot of people have kind of looked back at contact theory and I think found it to be a bit more wanting in its explanatory power than we might’ve hoped. It’s a very nice story that you just come into contact with people who are different from you and you update your views about them. But can you tell us about Matt Lowe’s paper and whether it kind of downgrades your assessment of whether contact theory is really—

Kaplan: Yeah, so Matt Lowe’s paper is great. And Matt Lowe is fantastic. He surveys a lot of literature, so the Annual Review of Economics is kind of an academic review of literature. And so he does this for the contact hypothesis. There’ve been a bunch of papers written. And what he shows is that a lot of the papers that are the best done don’t find really large effects. And so that definitely makes me more skeptical of the contact hypothesis.

And I will say, I actually have another paper that I’m doing right now where we look both on race and on politics at what happens after having a cross-racial neighbor move in next door to you, someone from a different political party. And we find nothing.

Demsas: Wow.

Kaplan: So I do think, though, that this is a really good setting, that this is a really important experience, where you go to school. And it’s at a time that was very formative and, you know, school is an environment where maybe these types of relationships can somewhat be handled well. And so this might be different than, say, having a neighbor move in next door. And so there, again, if you look at Allport’s original theory, he has four preconditions where he doesn’t think that contact everywhere ends always with—

Demsas: Yeah. Not just brushing arms in the supermarket.

Kaplan: Yeah, exactly. And so maybe, you know, the scope of the conditions for the Allport hypothesis were valid in this case, and oftentimes in these studies, they are not. I don’t really know, but I think this is a great open area for future research.

Demsas: So as long as we’re throwing out alternative hypotheses, one thing I was wondering is whether it’s possible that it’s less about desegregation in class and more about being forced into an extremely uncomfortable situation. Like, I think that when I look back at my life, the areas where I think I’ve exhibited the most personal growth are, like, the most stressful moments in my life, when I was just like, Wow, I’m now doing this job that’s, like, really hard. I feel really unprepared for it. But I’m having to throw myself against the wall. Or, I’m in college. I’m taking this class. It’s quite difficult, and I feel really out of my depth, that I’m forced to really work at problems for long periods of time.

And, you know, when you’re a kid, you move to a new school, having to make new friends. These are very difficult life experiences for young people sometimes and increases your need to socialize differently, engage with people across cultures differently, maybe get into arguments and fights and have to resolve those in ways that are different than maybe you would be used to. And so I wonder how much stock you put into the possibility that the effects you’re finding have something to do with this kind of personality effect, because we know from the Big Five personality tests that people who are open to experience are more likely to be liberal. And I wonder if that’s shaping people in this way.

Kaplan: So, you know, as with most papers and economics also—I keep a quarter of a foot in political science too, and, you know, the same thing is true for political science—we’re much better at estimating treatment effects than analyzing the mechanisms, because it’s easier to get plausibly causal variation that we can use. And so I think I’m much more certain about the fact that being bused had this effect then I am about the exact mechanisms. The mechanisms are definitely more speculative.

What I would say in defense of this view—but again, I don’t want to push it too far, and I think you’re absolutely correct to think that there’s other possible things that could be going on—is that we don’t find an increase in empathy related to race, and we do on class. And we think that there was differential exposure to class and not on race. And I think that, in my mind, is probably the biggest evidence that supports the view that maybe this really does have something to do with a class-based contact hypothesis. But again, I wouldn’t want to push that too far.

Demsas: Yeah. Well, I think that’s a great place to ask our last and final question. What is something that you once thought was a good idea but ended up being only good on paper?

Kaplan: So I remember in the late Clinton administration when there was an investigation into Bill Clinton, starting with Whitewater; it ended up with Monica Lewinsky. And because of that, the Department of Justice said that the Department of Justice shouldn’t investigate the president. And you know, at the time, because of the politicization of investigatory powers, that seemed like a good idea. But in retrospect, I think that having greater oversight over public officials of both parties is a good idea.

Demsas: And this is just apropos of nothing.

Kaplan: Yes. No, absolutely.

Demsas: Well, thank you so much for coming on the show. That was fantastic.

Kaplan: Yeah, this has been a lot of fun. Thank you very much for having me.

Demsas: Good on Paper is produced by Rosie Hughes. It was edited by Dave Shaw, fact-checked by Ena Alvarado, and engineered by Erica Huang. Our theme music is composed by Rob Smierciak. Claudine Ebeid is the executive producer of Atlantic audio. Andrea Valdez is our managing editor.

And hey, if you like what you’re hearing, please leave us a rating and review on Apple Podcasts.

I’m Jerusalem Demsas, and we’ll see you next week.