Riley Acton is an assistant professor of economics at Miami University where she specializes in the economics of education. Much of her current research focuses on how students choose where to apply to and enroll in college and how colleges can affect their local communities. She’s also the lead author on Political Views and College Choices in a Polarized America, a working paper out with the Annenberg Institute at Brown University
Check Out Her Full Report Here
Episode Description
Fall is admissions season at universities across the United States. As colleges work to recruit a new class, a new working paper suggests that politics might impact where students decide to go to school. That’s the focus of this episode of Stats and Stories, with guest Riley Acton.
+Full Transcript
Rosemary Pennington
Fall is admission season at universities across the United States, as colleges work to recruit a new class, a new working paper suggests that politics might impact where students decide to go to school. That's the focus of this episode of stats and stories, where we explore the statistics behind the stories and the stories behind the statistics. I'm Rosemary Pennington, stats and stories is a production of the American Statistical Association in partnership with Miami University's departments of statistics and media, journalism and film. Joining me, as always, is regular panelist, John Bailer, emeritus professor of statistics at Miami University. Our guest today is Riley Acton, Assistant Professor of Economics at Miami University. Acton is an applied micro economist who specializes in the economics of education. Much of Her current research focuses on how students choose where to apply to and enroll in college, and how colleges can affect their local communities. She's also an author, three, two. She's also an author on political views and college choices in a polarized America, a working paper out with the Annenberg Institute at Brown University. Riley, thank you so much for joining us again on stats and stories. Yes, thank you for having me on again. I'm just wondering what made you and you have two other co authors decide to tackle this issue?
Riley Acton
Yeah, great question. It actually started a little bit from the paper I was on this show last time to talk about, which was about colleges implementing covid 19 vaccines. One of my co authors on that paper was Emily Cook, who's also a co author on this paper, along with Paolo galde And Emily and I, as these covid vaccines that colleges were rolling out. And she was in Louisiana, Tulane at the time, I was in Ohio. We're just talking about it and speculating, oh, the covid vaccine is becoming very politicized. Is this affecting whether students want to go to a particular university or not along political lines? And then I think we got sick of writing covid papers, so we didn't write that particular paper, but we said there's actually like a bigger question here about just how do politics affect where students go to college? To what extent has that maybe been changing over time with political trends? And then we saw in other literature that increasingly political views, political identity, seems to be affecting things people do, choices people make, outside of the political arena. So we thought there was scope for that in the college choice space as well. You
John Bailer
know, in your paper, you started with this, an assertion that students derive utility from attending institutions aligned with their political their political leanings. So that was sort of the start, and you're just curious about that impact. So, so that was sort of your, your claim the test that you're going to test Exactly So, so what were some of the ways that you started to approach trying to test that that
Riley Acton
claim? Yeah, so this is actually a hard question to answer, because in a lot of data sets we have, we either have information on students, college choices, their their admissions applications, their enrollment often, we don't even have that often. We just see where do they end up. And then in other data sets, we have information about, you know, the politics, political views of young people. But we often don't have these data sources really linked together in any sort of systematic way. So thinking through, how could we actually ask this question? Answer this question was was hard, so we use two different approaches in the paper. I'll start with the first, and then we can move into the second. So the first is, we're using pre collected survey data. This is from a survey called the freshman survey, very creative name, which has been administered since the mid to late 1960s by the Higher Education Research Institute, which is based out of UCLA. But this is a national survey. So they have administered this survey to millions of college freshmen over time. Miami's actually participated in the survey quite a lot. I think it's 50 plus times over this time period. We're a frequent user of the of the survey. Colleges opt in to administering the survey to their freshmen, typically at like a college orientation session, and this survey asked students a lot about them. It asked them about their background, their high school experiences, their expectations and aspirations for college. These are all valuable pieces of information for colleges to have. But it also asks them about their political views. Varies a little bit. All the questions they ask every year, but they always ask one question that says, How would you characterize your political views on a five point Likert scale, ranging from far left to far right, with middle of the road or moderate in the middle? So we use that data to think about what does the political composition of colleges student bodies look like across time and across different types of institutions over time.
Rosemary Pennington
So my question for you, you said, I think 15 million students have participated in this survey at some point. I'm assuming you did not look at data for 15 million students. So how did you and your co authors approach, kind of breaking down the data in a way that was usable for
Riley Acton
you? Yeah, so we do have access to the full 15 million so we have like that. That sample, we had to request data from the survey provider. They actually make a lot of it publicly available for older versions of the survey, but for the newer versions, you have to do a data request and pay them to get the get the data. So our primary sample ends up being from 1982 to 2019 that's because they don't provide information on where students come from geographically until 1982 and we think the where you come from, it's probably an important predictor of your political leanings as a college freshman. So I think that that brings us down to seven or 8 million students when we kind of get the sample size down, but then you're right. I mean, like you have all this data, what are you going to actually do with it? So we actually do with it? So we do a couple different things in the paper with this survey data. We first just look at how have political views of students just changed over time. So just aggregating across across years, and we find that since the 1980s students have gone in somewhat more liberal. You have a slightly higher share of students identifying as liberal. Most of that's coming from a smaller share of students identifying as middle of the road or moderate. You don't see really much change in the share of students that identify as conservative on college campuses. It's a pretty flat 20% throughout the time period we're looking at. And we think those are kind of interesting facts on their own, but really what we're interested in is how is the distribution of these views changed across colleges. So then we do a lot of things to look at different types of colleges, how have the political views of their students changed over time? And we see things, for example, like selective colleges have gotten a lot more liberal in the share of their students that identify as liberal than the average college has over time. And we also see that the colleges that at baseline in the 1980s had a higher share of students that identified as liberal. That's where you see an even bigger rise. So you see this kind of fanning out effect going on, where the most liberal colleges have gotten more liberal, the most conservative colleges, to some extent, have gotten more conservative. Or one way we think about this in the paper is that the paper is that the distribution of political views across colleges has widened or polarized over time.
John Bailer
Yeah, sort of there. There are two things that I was I was looking through your some of your figures in your paper. I was also intrigued at the data before 1982 because, particularly if you look like 1971 was a much larger gap between the students that were identified as as liberal and far left and versus conservative or far right. So that was kind of interesting to me, to see that that pattern there and then, when you at some of the other displays, when you're looking at how the distribution of this percentage of identification is changing, it almost looked like the distributions were becoming multimodal.
Riley Acton
Yeah, that's possible for sure. Where you're kind of having this bulk on the right side of the distribution of percent liberal, where you're having some colleges. In the 1980s there were very, very few colleges that had a majority of their students identifying as liberal, whereas today, it's not most colleges, but you do have a non trivial number of colleges where a majority of their students are identifying as liberal, and to some extent, we see that with with conservatives as well.
John Bailer
So there are lots of factors that go into play and for a student to choose a school now. Now these are students that have already selected a school, right? I mean, so this, it's always an interesting question is, once you've made a decision, you know, do you not try to validate it based on kind of how you might respond to a survey? So let's, let's put, let's stick a pin in that for a second and and let's talk about some of these other factors. And you mentioned that as sort of students, everything from weather to to college sports success to other things. What are some of the other things that help drive this? Yeah, so
Riley Acton
this is where we kind of partly see this paper as fitting in the literature, and we're a little bit surprised that people haven't thought about politics, because the empirical literature on college choice is really large at this point, and economists and other social scientists have thought about so many different things that affect where students go to college. You have the probably most traditional, which would be cost. And then academics right? This is if you you as a faculty member, write down a little model of how do you think people choose where to go to college? You think they're maybe trading off cost with like, academic opportunities. But we know from the literature, lots of other things affect where people go to college, including like distance from home. I have a paper on this. Lots of people looked at this. People tend to go to college closer to home. We know that things like how successful. Sports teams are in a given year, affects who applies to the college. There's a paper that I love called College as Country Club that shows that as colleges invest more in their dorms and their gyms and their dining halls, these kind of amenity features, they get more applications, more students enroll when the campuses are bright and shiny. And then we also know that students tend to seek out environments that are familiar to them, or have students that are similar to them. So for example, for underrepresented minority students, an important predictor of where they go is the share of students on that campus that are the same race or ethnicity as them. There's a series of papers in different contexts different countries, showing that students are really prone to follow their their older siblings college decisions. So if your older sibling just barely goes to one college versus another, you're more likely to go there. So we we know that students seek out these familiar environments, these places with people like them. So we think seeking out a college where students might think similarly to you, have similar values or political views to you could, could make some sense as well.
John Bailer
So you, you approach, there's, sorry, here, my brain is paused here, Charles, so you back up on this one thing that I wonder about when I when I look at a survey that's spanning 50 years, I got to stop talking with my hands, or I hit the mic. So that's a so when you're looking at this, when people are responding to what does it mean to be this or that? When someone says, I identify as conservative or identify as liberal, does it mean the same thing now as it did in 1969 when this survey started?
Riley Acton
Yeah, that's a great question. And my my gut instinct would be, no, it doesn't. I think it has to reflect the times and reflect the political parties of the time, reflect what people in the world think of these things. We haven't dug into it too much yet that there is the potential for us in this paper to also look at how students respond to other questions about politics and public policy. So this survey also includes some questions about contemporary political and policy issues. The challenge with that is those, those issues change over time. So the questions change over time. So we haven't quite figured out how to really operationalize those questions yet, but it's possible we could try to look at some things. There's some somewhat consistent questions about like, do you think wealthy people should pay more in taxes than they do? Now, we could try to see how well that correlates with with political identity, and to what extent that might be changing over time.
Rosemary Pennington
You're listening to stats and stories, and we're talking about how politics may impact college choice with Miami University's Riley Acton. So the first part of your study is looking at this wealth of data that has been collected nationally, and it seems like what sort of has emerged is that, you know, there's been a growth in Liberal students over this, this period of time, and sort of the but the moderate shrunk, but the conservative population is not grown enormously. And you use are using that to build on to your second piece of of this study. Can you talk through like, what the second part of this is? Yeah,
Riley Acton
so the first part, we think the punchline is a lot of what you said, plus the fact that these trends have evolved differently across different institutions. And it seems like campuses, if you just look at their student bodies, are becoming more polarized over time. At the same time that in many different measures, our country is becoming more politically polarized over time. We try to do some things in that first part, to adjust for, control for in our regressions, other attributes of students sorting into different colleges over time, and we don't get a ton of bite from that, suggesting that perhaps students are actively choosing to go to particular colleges based on their political views and their perceived political views of an institution, at least on the margins, maybe not the sole factor for all or even most students, but might be contributing. So to try to test this further, Emily and I were thinking about this for a long time of like, okay, there's no perfect kind of quasi experimental design here. You can't randomize what politics what political views a student has coming out of high school, hard to randomize the political views of colleges, like there's not a lot you can do here, but we were then kind of thinking about what else we could do. And we were able to collaborate with our third co author, Paula, on this who is more of an experimental economist, so she runs different experiments, and a lot of the work she does is focused on doing survey based experiments, where you can manipulate scenarios and people can respond to them without making actual choices or actual randomization in the real world. So that's the approach we take in the second part of the paper. We do a survey based experiment with hypothetical colleges and real college students. So we have a sample of real college students who we show different profiles of colleges to, and we randomize basically everything that they see on on these profiles in terms of the attributes and then the values of the attributes, but we include some information on it of okay, this college, college, a has, you know, 40% of students that identify as moderate. And 40% that identify as liberal, and only 20% that identify as conservative, whereas maybe college B, 40% are conservative, and you know, 50% are moderate and 10% are liberal. And we show them a lot of these scenarios, and we include things about cost and distance and quality and institution size, and then, based on their answers of what's the probability you'd go to college a versus college B? We can back out their preferences for these different attributes and get a sense of how much politics might play into their decisions
Rosemary Pennington
before we move on. I do want to ask about polarization, because you use polarization in a couple of different ways. In this there's sort of the talk of polarization, polarization writ large, but then also what you call affective polarization. So can you talk us through like, what the differences are between those two? Yeah.
Riley Acton
So affective polarization, the way I understand it, not being a political scientist who did not come up with this term, is affective polarization captures differences in how you feel, particularly how positive or negative you feel about a group that you belong to versus the opposing group or the group that you don't belong to. So in the politics context, that's capturing measures of how positively do you feel about the political identity or political party that you belong to, versus how negatively do you feel about the political identity or political party that you don't belong to. And what we see, or what political scientists see in other settings, in other surveys, is the disparity between how positive you feel about your party and how negative you feel about the opposing party has grown over time. So Democrats have become more negative about Republicans. Republicans have become more negative about Democrats, and we end up seeing something a little bit similar in our experiment to this, where it seems like more so than students preferring to go to a college that has more students of their political identity. They really want to avoid going to a college with a lot of students of the opposite political identity. And this sort of, we don't know exactly what to call it, but this sort of pattern that mirrors affective polarization of avoiding the other group, even above and beyond preferring your own group, does seem strong in our setting, for both liberal and conservative
John Bailer
students, as part of this survey experiment that you're you're conducting, you have preferences for scenarios, but you also have a willingness to pay component to it. You know, so how much you're willing to get to that that preference. So can you talk a little bit about how that plays out?
Riley Acton
Yeah, exactly. So the willingness to pay measure, and it's pretty common in this sort of setting, really comes from comparing the preference for a particular attribute to your preference for cost. So we're we're seeing all we're providing students with different costs in different scenarios, all drawn from the actual distribution of college costs that students face. And so we can see, essentially, for a given student, how much, when all else is equal, do you prefer to go to a college that's $1,000 less expensive and all else equal, how much do you prefer to go to a college with more peers of your political view? And then we can scale these by each other to understand the willingness to pay, and we find that students are willing to pay anywhere from about 2200 to $2,600 depending on on their own political leaning, to go to a college that has a 10% that has a 10 percentage point lower share of students from the opposite political party, so liberal identifying Students are willing to pay about $2,600 more per year to attend a college that has 10 percentage point fewer conservative students.
John Bailer
Okay, I'm just trying to that's, that's a interesting. I was thinking it really differs depending on the college you go to. You know, that could be a a small percentage of tuition room and for some, for some institutions, and it could be your
Riley Acton
dramatic true. And so this is a median across students and across the scenarios that we show them, and we can scale that to the like median cost that they get shown in these scenarios. But you're right that obviously, depending on where you are, that could be a lot, that could be a little, but this is all kind of at the median.
John Bailer
So you did some modeling too, right? You did you develop some some predictive models, or to characterize this, can you talk a little bit about what you were trying to predict and how you're using the information to predict
Riley Acton
it? Yeah. So in the experiment, the basic setup here is we write down what economists would call a utility model. So we say we're going to model your behavior as choosing the option that gives you the most utility, which means choosing the option that makes you the happiest, choosing the option that you like the most. We think, if we think intuitively human behavior, people should choose the thing they like, right? And so that's what we are doing in this experiment, though, we are allowing students to provide some uncertainty over which option they would prefer if they were making this decision in the real world. So they're giving us a probability, you know, 70% chance I'd choose college A, versus 30% chance I'd choose college B. And so we write down a utility model that we say is linear in the attributes of these colleges, right? So there's different factors. They're all entering this utility function linearly. And then if you're willing to make some of. Assumptions about the distribution of the error term in that utility function, you can then get a closed form probability function that represents, okay, here's the probability you would choose college a over college b based on these attributes that that we're showing you. You get a little fancier as you go, but it ends up then becoming basically a linear model that you can run and run a regression on, where the dependent variable is the log of the ratio of the probabilities that the student chooses, and we use that to back out these, these preferences.
Rosemary Pennington
It's really interesting to me the findings of this, given that so many universities are working really hard to present themselves as these, like pluralistic places where everybody is welcome, and we want you to engage in, like, you know, vigorous debate and learn from one another. And instead, it seems like in your study, you have found this like desire amongst some population of students to maybe avoid that. I guess I wonder. I wonder, sort of what you make of that in relation to this, like push to be more pluralistic.
Riley Acton
Yeah. So this has been so interesting to be working on this paper over a couple of years now we've been working on this paper, and this has just become more and more relevant in our cultural discourse as we've been as we've been working on it. I think you're right. There's a lot of colleges and policy makers who are very passionate and driven towards making colleges these places with diverse viewpoints and open inquiry and, you know, robust discussion about things, I think we have yet to see how those sorts of initiatives and those sorts of policies will potentially affect where students go. Right? It might actually be the case that for some students, that is attractive to them about a particular college, and that makes them more likely to go to that particular college. Though our results would suggest that students also kind of want to seek out like minded individuals, and so if they see those policies as reducing the number of like minded individuals on a campus, that might make them less that might make the college less attractive to them to go there. So, yeah, we think this all fits in a lot to the current conversations happening that we don't have precise predictions of how different initiatives might affect these sorting patterns going forward. You know,
John Bailer
I was curious, as I was thinking about all the factors that might play into this. And I was curious about parents influence on this. You know that whether the parents education, the legacy, if they had gone to an institution, the parents political beliefs, sort of all of how all of these might influence, did you have access to any of that information as part of this?
Riley Acton
That's a fantastic question. That is something I think we would love to have access to. Is like doing this experiment on the parents as well, potentially, and seeing how much do parents differ from their kids in this in this choice? Because I think one thing we know from from other researches or anecdotes, is that it seems like parents are increasingly playing a large role in college choice processes. They are often the ones footing the bill. So they certainly want to have some input into this decision. We don't have any information on that right now. And with our experiment, we were looking at students who are already in college, who are over 18, who are easier to do experiments on. But we are very hopeful. We're working now with NORC, which is an institute at the University of Chicago that runs a nationally representative panel of Americans, including a nationally representative panel of teenagers, and does have them linked to their parents as well. So we're hoping to do some future surveys and experiments on high schoolers to understand, you know, these are students in our survey who were giving them hypothetical scenarios. They're responding to them, but they have already made their college decision. It's not perfect. Whereas high schoolers, we'd love to understand how they're thinking about these things, how their families are thinking about these things, how they're interpreting some of these policies and initiatives that colleges are doing, as well as just how do they even think about what makes a college liberal or what makes a college conservative?
Rosemary Pennington
Well, I think you already semi answered the question I was going to ask about what's next, but I do wonder, like, what questions might have popped up for you as you were doing this that you didn't anticipate that you're thinking of as fruitful ground for research?
Riley Acton
Yeah, I'd say there's two big ones right now. So one is thinking about the decision to go to college at all. So right now we're really thinking about what college do students go to, but that's assuming you've already decided you're going to college of some sort, and then you're deciding which one to go to. We know from surveys, public opinion polling, things like Gallup, that there have been large partisan divides and a large divergence in how people perceive higher education, how they perceive the value of higher education, across political lines, and so we're interesting to what extent might those sorts of views, either if they're held by parents or if they're held by students themselves, or some combination thereof, be affecting whether students want to go to college at all. Right, we see a lot of headlines about declining trust in higher education. A lot of that is a larger decline. Among conservative individuals, might that be affecting whether their kids want to go to college or not? That's something we're really interested in trying to explore. And then the other thing we're trying to explore is, you know, how do students actually form beliefs about whether a college has liberal students or conservative students? In our experiment, we're giving them information on this, that is information you can get from some sources online. So like niche.com has their own rankings of conservative and liberal colleges. They use kind of self reported student data to get percentages, and it's a little buried on their website, but you can find it, but we think more realistically, students and parents families are getting a sense of these things from their campus tours, from the marketing materials that they get from colleges, from news articles, things other people tell them. So we're interested in how students and families are forming these beliefs that they then might be making decisions based off of.
John Bailer
do you see a role for some focus group work with students? I mean, this to kind of, so here I am talking about the qualitative part. Where did this come from? John, talk about this quantitative stuff Rosemary we got to talk about some of the qualitative work. I mean, come on, yeah. So I'm just curious about, sort of, there's, there's the, there's always this kind of, the finding the truth here, you know, yes. So how do you, how do you make sure that you're kind of narrowing in on, on, kind of where you what's right?
Riley Acton
Yeah, that's a great point. I think this is definitely an area where mixed methods, sort of research could, could be useful. None of us on this team are are qualitative people. So we are definitely on, like, the quantitative side of this. I think if we wanted to go that direction, we'd we'd need to bring in some other experts on it. But I think this is a really fruitful area of research that there are other people working on, but only kind of just starting to work on right now. I think for a long time, we just thought about college choice outside of politics, at least within the economics and education research worlds that that I live in. And I think increasingly people are realizing that politics political identity above and beyond even public policies are affecting how people view college, how people view different colleges, how they make their decisions. And so hopefully, a wide range of researchers and scholars can think about this question from different angles and kind of build upon each other's work to come to some understanding or some truth.
Rosemary Pennington
Well, that's all the time we have for this episode of stats and stories. Riley, thank you so much for joining us again. Yeah, thank you so much for having me very fun. Stats and stories is a partnership between the American Statistical Association and Miami University's departments of statistics and media, journalism and film. You can follow us on Spotify, Apple podcast or the places where you find podcast if you'd like to share your thoughts on the program. Send your email to statstories@ams.org or check us out at stats and stories.net and be sure to listen for future editions of stats and stories where we discuss the statistics behind the stories and the stories behind the statistics.