Probable Causation

View Original

Episode 13: Ariel White

Jennifer Doleac

Ariel White

Ariel White is an Assistant Professor of Political Science at the Massachusetts Institute of Technology.

Date: October 1, 2019

Bonus segment on Professor White’s career path and life as a researcher.

A transcript of this episode is available here.

See this content in the original post

Episode Details:

In this episode, we discuss Professor White's work on the effects of short jail spells on subsequent voting behavior:

"Misdemeanor Disenfranchisement? The Demobilizing Effects of Brief Jail Spells on Potential Voters" by Ariel White.

OTHER RESEARCH WE DISCUSS IN THIS EPISODE:

Transcript of this episode:

 

Jennifer [00:00:06] Hello and welcome to Probable Causation and a show about law, economics, and crime. I'm your host, Jennifer Doleac of Texas A&M University, where I'm an Economics Professor and the Director of the Justice Tech Lab.

 

Jennifer [00:00:17] My guest this week is Ariel White. Ariel is an Assistant Professor of Political Science at the Massachusetts Institute of Technology. Ariel, welcome to the show.

 

Ariel [00:00:27] Thank you for having me.

 

Jennifer [00:00:29] We are going to talk today about your research on the effects of a misdemeanor jail sentence on subsequent voting behavior. And with the next election looming, I'm extremely excited to have you here to talk about how our criminal justice system interacts with the democratic process. To start out, could you tell us about your research expertise and how you became interested in this topic?

 

Ariel [00:00:51] Mhm. Well, I am a political scientist and I often say that I study the bad or the punitive experiences that people have with government and the ways that those can shape people's willingness or their ability to participate in politics. And I guess I usually trace this research agenda back to the job that I had before I started grad school, which is that I worked at a legal services office, meaning that I I answered phones, I did intake. I did some paralegal work on cases around things like housing, evictions, public benefits. And basically I was working there within the civil, not the criminal legal system like we'll be talking about today. But I got to talk with a lot of people who were having really infuriating or scary and life altering experiences with government in ways that were- seemed clearly to be shaping the ways they thought about what they could expect from government, but also that were not always very visible to people who were lucky enough to not have to interact with these systems.

 

Ariel [00:01:56] And so this was on my mind when I got to grad school and when I found that, of course, there was this whole literature on how people's lived experiences matter for how they think about politics. But also at this time, people were, especially in political science, just starting to do work on incarceration. And this was I mean, you come in as a grad student and you're like, wow, this is I mean, this is one of the most serious things that the state can ever do to you. And it seems just wild that political science went so long as a discipline, not thinking that it was something that was worthy of systematic study. And so, you know, clearly I felt like it was worth focusing on, at least in part of my research.

 

Ariel [00:02:37] And, you know, it's also the case that I had seen family members and people I grew up with serve time, though I don't actually know that I would have come to this research agenda had I not also had these other experiences that I just described before and during grad school. You know, I think my family, like a lot of other families, tend to think of incarceration as a sort of individual experience that one thinks about as as kind of a private thing and not something that would be part of a broader system of mass incarceration or something that would be part of a pattern of experiences that would that would relate to one's political or civic life. So, you know, I came to this through, I guess, all of those things.

 

Jennifer [00:03:18] That's really interesting. Yeah, so your paper that we're going to talk about today is titled "Misdemeanor Disenfranchisement? The Demobilizing Effects of Brief Jail Spells on Potential Voters." And as that title suggests, you're focused here on the effects of short jail cells for low-level offenses. But let's start with a bigger picture of you. So before your paper, what had we known about the effects of criminal justice involvement more broadly on political engagement and voting behavior in particular?

 

Ariel [00:03:46] Well, I think we had a good sense descriptively of the headline finding that people who have had contact with the criminal legal system are generally less likely to participate in politics, at the very least descriptively. So I think, for example, of Traci Burch's work merging criminal records data to voter files and finding that people with felony convictions — even in contexts where they were eligible to vote — tended to vote at very low rates. I think of the work by Amy Lerman and Vesla Weaver, both in their book and in in an APSR article associated with that, that combines survey data finding that people who report having been arrested or convicted or sentenced to some or a lot of prison time report — as you move up that ladder of exposure — report less and less involvement in politics. Report thinking that participation in politics is less important. Report less kind of civic trust. Also report being less likely to register or to actually vote. But they also combine this with really rich kind of qualitative interview data where people describe the the civic lessons that they take from incarceration really vividly. So this book, I think, gives a really rich sense of the ways in which people think about incarceration and other kinds of contact with the legal system as a primer for what you can expect from government if you interact with it and the ways in which people learn not to interact with it, including through things like participating. So that, I think, gave us a really good sense overall of the ways in which interaction with the legal system could at least be associated with people being less likely to participate.

 

Ariel [00:05:36] But then the question of causality was was kind of an open one. There was work by by Weaver and Lerman certainly trying to figure out the causal effect of incarceration on political participation. There's a nice paper by Alan Gerber and coauthors that focuses on people serving prison time for felony convictions that I think lays out this debate and the set of concerns about establishing causal effect really clearly. Lays out concerns about measurement error in survey data, but also in just- that lays out just the kind of broader set of concerns about how people who are incarcerated may be different, both observably and unobservably from the people who don't end up incarcerated. And how trying to tease out the causal effect of incarceration from the causal effect of all of the other experiences that often lead to incarceration can be really hard. So I would say when I started on this project, this trying to figure out the causal impacts of these kinds of experiences was definitely top of mind.

 

Ariel [00:06:37] And also, you'll have gotten a sense from from the pieces I just described that there was a fair amount of survey based work that had struggled, I think, to really distinguish between different types of contact simply because surveys often don't ask very detailed questions about what exactly people were charged with and what they ended up convicted of and what kinds of sentences they served. And the work that we had seen that was not survey-based, that was based on administrative data, had often focused on people serving relatively long prison sentences for felony cases. And so I was particularly curious about people who were serving shorter sentences in local jails for misdemeanor cases, which would not legally disenfranchise people but that I thought might still have some effect on their willingness or their ability to vote.

 

Jennifer [00:07:23] Yeah, and so along those lines, the public conversation around this issue really does focus on direct legal disenfranchisement most of the time, so the fact that many states have policies that bar people with felony convictions or people who are currently incarcerated from voting. And it's straightforward to see how those types of policies can reduce voting behavior, people are actually banned from registering and casting a vote. But it's perhaps less straightforward to see how a short jail spell over a misdemeanor sometime in your past could affect your voting behavior in the future. But in this paper, you lay out a couple of mechanisms through which that might happen. So talk us through that. Why might we expect misdemeanor jail sentences to affect future voting behavior?

 

Ariel [00:08:03] Absolutely. I think we can think about at least two big classes of story here. And the first one is what I would call the resource story. And this is maybe the most familiar one to your listeners in that we might imagine that people who spend some time in jail would then have other changes in their life that would result from that, right. There's work in sociology and in economics that highlights the ways in which spending even a short period of time in jail can translate into people potentially losing their job, maybe losing their housing, having other kinds of life disruptions that might mean that on Election Day, people might be less able to vote. They might be handling other life crises. They might not have the ability to get to the polls. Maybe they don't have transportation. Maybe they don't have a place to sleep. Maybe there are just a series of other things that are top of mind rather than voting when Election Day rolls around. So that's one story about the ways in which incarceration, even relatively short jail stints, can reshape people's lives in ways that would make it hard to vote.

 

Ariel [00:09:04] But there's also a second story that I would call the socialization story. And this draws on some of the work I was just describing — particularly by Vesla Weaver and Amy Lerman — on the ways in which people learn things about government from time spent behind bars. And that in particular, the kinds of things they learn are that government doesn't really care what you have to say and that it's better to kind of keep your head down and avoid contact with it if you possibly can, including in cases like voting. So that would mean that people would perhaps even if they were able to vote, might not be willing to vote because of their experiences with jail.

 

Jennifer [00:09:43] All right, so let's talk about the empirical challenges to studying this issue. You alluded to some of this earlier, but I definitely want to dig in more. So in some ways, it seems like this is sort of an ideal setting because criminal records tend to be public record and voter registration and voting behavior — at least whether you voted, not how you voted — are public record. But that doesn't make it easy to tell what the causal effect of criminal justice involvement is on voting behavior. So what are the remaining data challenges that you faced here and what are the identification challenges that you needed to overcome to study this topic?

 

Ariel [00:10:17] Certainly, I mean, I think you're right that administrative data is one of the few ways we can think about addressing a question like this, because, as I mentioned, surveys don't tend to ask precise enough questions or to talk to enough people that we would ever have an angle on this. But even with administrative data, there are certainly difficulties just around, you know, linking up various data sources — which I'll maybe talk about in a minute — as well as just getting your hands on the datasets that you would need. And the availability of these kinds of records varies immensely across geographic areas, which is part of the reason that we'll be talking today about one big county court system that does have these records available.

 

Ariel [00:10:57] But then of course, I think the big challenge, as you've mentioned, is about pinning down the causal effect of something like spending time in jail on voting. Where even if we have all the data in the world, establishing that one of these things causes, the other is going to be really difficult, because as I think we understand, descriptively, people who spend time in jail often have various characteristics that are different from people who don't spend time in jail. Right. They tend to be younger. They tend to have lower incomes, fewer years of education. They're less likely to be white. And so, you know, there are all sorts of things that can be correlated with voter turnout, some of which are observable to us, some of which I think we can imagine, but that are unobservable to us such that if we if we see descriptively that people who have been to jail vote less often than people who have not been to jail, it's very hard to say, like, oh, yeah, that's definitely jail and not all of these other background characteristics that might be different between these groups.

 

Jennifer [00:11:58] Right. And yeah, let's talk about the hoops you had to jump through to to merge these datasets, because I think that's probably not obvious to folks who've not done it themselves.

 

Ariel [00:12:06] Oh, sure. So I would say this is maybe one of my more straightforward administrative projects, which is maybe not saying much. I'm using two sources of records here and I should maybe tell you a little about each of them. One is a set of misdemeanor case records from, as I said, Harris County, Texas. So this is the county that contains Houston. It's four years of records, so 2008 to 2012. This brings us from one presidential election to the next one. And I observe here everyone that's been charged with a misdemeanor for the first time. So this is- when I talk about misdemeanor cases in Texas, this is anything that carries up to a one year sentence. So things like DUIs, theft, certain simple assaults, a range of charges that could lead to up to a year in jail. And so for each of these observations, I see some details about the person's case. I see how it was handled. I see how it was disposed. And I see whether they're ultimately sentenced to any jail time. So this gives me a lot of information about what's going on on the criminal legal side of things. I see this for about 113,000 people.

 

Ariel [00:13:15] But then for all of these people, I want to know whether they voted in the next election after this experience, so this is in the 2012 presidential election. And so to get at this, I end up bringing in another dataset, which is the the state voter file of Texas, which keeps track of every registered voter in Texas and whether they voted in a given election. And so I end up merging that to the court records data that I just described, using people's names and birth dates. And the birth dates really do a lot of work here, right, because that really helps narrow down potential matches between people who have otherwise similar names. In practice, I end up merging mainly based on last name, birth date, and then doing some fuzzy string matching on the first name to allow for some differences in how people might be spelling their name or recording their name across datasets. And that maybe starts to give a flavor of the many ways these things can can get tricky when dealing with administrative records across different sources of data. But ultimately, after that, I can see, you know, the set of people who were charged with misdemeanors over this period. I can see whether or not they get sent to jail. And then ultimately for everyone I see whether or not they vote in the next election, because I either see, you know, whether they voted based on the voter file or if they don't show up in the voter file, I assume that they did not vote.

 

Jennifer [00:14:36] And so what does this population look like on average? So I think you said about 100,000 people. So what share of those are sentenced to jail? How long are those sentences? And then what does their voting behavior look like at baseline, that sort of thing?

 

Ariel [00:14:52] Sure. So about half of the people in the sample end up sentenced to some jail time with the vast, vast majority of these sentences being a couple of days to a couple of weeks. I think the median among people sentenced to some time in jail is about 10 days. So that's substantially longer than I would want to spend in jail, but substantially shorter than what we often see when we're thinking about felony cases. Thinking about voting behavior at baseline, if we think about before people were arrested, so in 2008, the data gets a little bit trickier. Because I collected this set of voting records after the 2012 election, there's always some concern about going back too far in time because people might drop off of the voter file for various reasons. But I would say approximately turnout here was on the order of 17 percent or so. This may be conservative, but it's not likely to be too far off. So relatively low voter turnout for people even before they were arrested. Though, there's some variation across race in that that we can talk about later.

 

Jennifer [00:15:55] Yeah, and how does that turn out compare to just the general population that might be observationally similar to this group, do you know?

 

Ariel [00:16:05] Well, I mean, it's certainly lower than the general population not conditioning on any personal characteristics or-.

 

Jennifer [00:16:12] Like income or yeah.

 

Ariel [00:16:12] Yeah. But if we think about it being a relatively young group of people, if we think about it being a lower income group of people, it's I would say still much lower than you would expect if you were to just look at a set of people with the same kind of set of background characteristics but who are not interacting with the legal system. Yeah, it's it's not as surprisingly low when you start thinking about those characteristics, but it's still lower than you would expect.

 

Jennifer [00:16:38] OK, and so like other studies we've talked about on this podcast, you're going to exploit the randomization of cases across judges as a natural experiment here. So walk us through that and tell us how it allows you to measure the causal effect of a jail sentence for misdemeanor defendants.

 

Ariel [00:16:56] OK, I should tell you a little here about the setup for misdemeanor cases in Harris County, though I suspect it will sound familiar to listeners who have been exposed to other random judge assignment papers. But when someone is charged with a misdemeanor case in this county, the Clerk's Office enters the case information into their computer system. And then that system also randomly assigns that case to one of the courtrooms that could hear that case. At the time, that's 15 different misdemeanor courtrooms. And so without consideration for the characteristics of the case or of the person that's facing charges, these cases are being sent to courtroom 1 or courtroom 2 or courtroom 15. And each of these 15 courtrooms is set up to hear misdemeanor cases. So they have a judge and other courtroom staff and assistant DAs. And each of these courtrooms hears thousands of misdemeanor cases a year. They're hearing very similar cases. They're handling them through a similar process.

 

Ariel [00:17:53] But these courtrooms do differ from each other in just how often they send people to jail. For example, if you were- if you were charged with a misdemeanor in 2011 and you happen to get sent to courtroom 7, then you would be walking into a courtroom that sent people to jail in about half the cases that came before it, 50 percent. But if you instead ended up in a courtroom 2, well, then you would be walking into a courtroom that sent people to jail in 62 percent of their cases that year. And you'll remember that because these cases are being randomly assigned to courtrooms, these courtrooms are handling basically the same caseload, the same volume of cases, the same types of cases. But some of these courtrooms are just more likely to send people to jail than others.

 

Ariel [00:18:41] And this type of random courtroom assignment set up — which, as you mentioned, is going to sound familiar to some listeners — it's going to give us an instrumental variable setup. So we're going to instrument for whether a person gets sent to jail or not using information about the courtroom that they get sent to. So I'm going to predict whether or not someone receives a jail sentence using the harsher, excuse me, the harshness of that courtroom, the proportion of cases where they rendered a jail sentence in the year of that case. And then I'm going to use those predictions to estimate the effect of jail on voting in the next election. So this is going to give us a Two-Stage Least Squares setup. But the basic intuition here is that, you know, if you run the naive regression, just predicting voting with somebody's jail exposure, you run the risk of bias because as I mentioned, there are going to be all these kind of observed or unobserved differences between people who do and don't get sent to jail. So it's hard to interpret that as a causal effect. But here we're going to use only the variation in jail time that comes about not because of people's personal characteristics, but by chance, right by the luck of the draw in the courtroom lottery to try and get at the effect of jail on voting. And so for the set of people whose sentencing outcomes happen because of the courtroom they get sent to, this setup is going to let us look at the causal effect of jail on voting, setting aside all of those other concerns.

 

Jennifer [00:20:09] And as you clearly describe in the paper, this empirical strategy allows us to measure the effect of a jail sentence on the people for whom it matters which judge they're assigned to. So there are some defendants who will never be sentenced to jail and there are some who will always be sentenced to jail, regardless of which judge is deciding their fate. And so the results of your study tell us about the effects of a jail sentence on those in between, or the compliers in research terms. So what do the compliers look like here? What types of offenses should we have in mind? What are their demographics and so on?

 

Ariel [00:20:42] Yes, this is a local average treatment effect, and so it's definitely worth thinking about who's driving these estimates and how they compare to everyone else in the system. And I look into this somewhat in the online appendix for this paper if people are curious. And now, of course I can't say, "This individual person was a complier, and this one was not," right. But I can take the Angrist and Pischke recommendation of taking the ratio of the first stages for people with a given characteristic compared to the first stage for everybody. And when I do this, it suggests that compliers look close to but not exactly the same as the whole sample. They are slightly more likely to be women. They're a little younger than the overall sample and they're a little more likely to be facing Class B misdemeanors rather than Class A. so this means somewhat less serious charges. Class A misdemeanors carry up to a year in jail, while Class B carry up to six months.

 

Ariel [00:21:36] But I will say this, this difference in charge types was actually less pronounced than I expected when I went into the data. If you look through, for example, the many different offenses that people were charged with and the jail rates for those different offenses, there are actually relatively few for which everybody gets a jail sentence or nobody gets a jail sentence. I expected there would be more of these relatively clear cut, kind of extreme cases. And actually, for most commonly seen offenses, like even the more serious ones, there is real variation in whether people get jail time. And so there's room for courtroom assignment to matter here.

 

Jennifer [00:22:13] It does seem to be a common theme across a lot of these judge randomization papers that actually like the judge that you see, it tends to matter a whole lot more than than these underlying characteristics, which is kind of horrifying. So so what are the main results? What do you find are the effects of being sentenced to a short jail spell on subsequent the subsequent likelihood of voting?

 

Ariel [00:22:36] Well, looking across everyone in the dataset with the specification we've just talked through, I estimate that getting sent to jail reduces voting in the next election, the 2012 presidential election, by about 4 to 4 and a half percentage points. So think here about moving from, say, 14 percent voter turnout down to 10 percent voter turnout. Now, the simplest estimates shown in the main paper are somewhat imprecise. We can't statistically distinguish them from zero. And I get into this more in the online appendix where we can incorporate other case information into the setup, if we want to. We can calculate the courtroom harshness instrument within race or within charge type in ways that buy us more precision. And in these setups, the estimates are more precisely estimated and are statistically significant. But here I think the presentation is a little more straightforward and the takeaway is quite similar.

 

Jennifer [00:23:27] And that's a big effect.

 

Ariel [00:23:29] It is, particularly if we're thinking about something that occurs over the course of a couple of weeks, yeah.

 

Jennifer [00:23:36] Right. And then you you examine these effects by race as well, as you mentioned. So what do you find are the differential effects for black versus white defendants?

 

Ariel [00:23:47] Yes. Well, if we look at the estimated effect among black defendants still within this IV framework, I estimate that a jail sentence reduces voting here by about 13 percentage points. That's 13 percentage points. So think here about going from 26 percent voter turnout down to, say, 13 percent. And even noting here that the confidence interval is broad enough that these estimates are also consistent with the effect being 8 or 10 percentage points, that's still a really substantial effect on voter turnout from a single experience of jail. So that's a really large effect among black defendants. But meanwhile, the effect of jail time among white defendants is approximately zero. The point estimate is very small, indistinguishable from zero. So this is a story about quite racially disparate effects of jail on voting. It's a story about jail time, disproportionately demobilizing black voters, even when we're focusing only on people who have at least some contact with the legal system.

 

Jennifer [00:24:50] I don't think we mentioned this yet, but if I recall correctly in the paper, you're focused here on the population for whom this is their first misdemeanor trial. Right. So you're not looking- you know, you don't have to deal with people who've been cycling through the system. Am I remembering that right?

 

Ariel [00:25:03] That's exactly right. And that's both because I think it's a clearer interpretation, but also because the random assignment process is a little less clear if you're coming back after having had a previous case.

 

Jennifer [00:25:13] Right. And yeah, so this is- it's just a massive effect given that this is like the first time someone has been charged with a misdemeanor and they happen to get a jail sentence or not. And then it's essentially a 50 percent decline in whether they show up to vote. That is huge. So, yeah, let's talk about what's going on there, the mechanisms involved. So one potential reason for the racial differences you find is that the experience of jail has a different effect on black defendants than it does on white defendants. So that is the actual treatment effect could be different even if those individuals are otherwise the same going in. The other potential explanation is that the marginal white defendant who's sentenced is different from the marginal black defendant. And so the racial differences you find could reflect those differences in how judges treat black versus white defendants. So you talk about this a bunch in the paper, so talk us through those possibilities and what you think is going on here after all the various checks you run.

 

Ariel [00:26:10] Absolutely. So I can't rule out the possibility that people would be having different experiences in jail on the basis of race, that they might be getting treated differently by race. And I think that's completely plausible. And it could be part of the story about why we see this big drop off in black turnout, but not in white turnout. But as you noted, I also think there's another part of the story that happens up front. That's about who even gets pulled into the misdemeanor system in the first place and how that process might differ by race. So if we imagine for a minute that we live in a world where it's relatively more difficult for a white person to get arrested for a misdemeanor crime. And it's relatively easier for a black person to get arrested for the same types of misdemeanor crimes. And I think we could imagine a number of ways that this could arise. Right. Maybe there's more of a police presence in black neighborhoods. Maybe the police threshold for arresting or for charging people is lower.

 

Ariel [00:27:09] There are plenty of possible stories that would get us here, but there is some evidence that we actually live in this world that I've just described. Right. There's a pretty commonly cited example in the case of drug charges where self reported white and black drug use rates are quite similar. But the rates at which people are charged with drug possession are fairly different. And if this is the world that we live in, then we could imagine that the white people that you would see in the misdemeanor courtroom might be a little different from the black people that you would see there. Right. If the threshold for white people ending up in misdemeanor court were higher, then we might expect that a fair number of the people that we would meet would be socially marginalized in some other ways. Maybe these are people who are homeless or who are dealing with mental health or substance use issues. Maybe a lot of the white people who are charged with misdemeanors would be relatively unlikely to vote even before they got arrested.

 

Ariel [00:28:07] Whereas if it were easier to get arrested as a black person for any number of reasons, we might expect that in the misdemeanor court system, we would see basically a broader swath of the black community. We would see some nonvoters, but also some people who would have been voters, people who would have been residentially stable and engaged in the community and prepared to participate if they had not been arrested and facing jail. And so I think it's possible that racial differences in who gets pulled in here, possibly because of differences in how different groups get policed, that this could be driving some of the racial differences that I see and the effect of jail downstream.

 

Ariel [00:28:46] And there's there's some data presented in the paper that's consistent with this story. So if I go back and I look at people's voter turnout from 2008, before anyone in this dataset got arrested, it is the case that black defendants were much more likely to have been voting in 2008 than were white defendants, almost twice the rate of turnout in 2008. And that pattern of substantially higher black turnout than white turnout is not something that you see in the general population of Harris County. Right. It's specific to the set of people who end up in the misdemeanor court system. And so that suggests to me that there are some differences there in who gets exposed to the misdemeanor court. Basically, if black voters get arrested, then there's room for them to be demobilized. Whereas if most white people who get arrested are nonvoters, then there's a lot less room for them to stop voting. So I think that's part of the story as well.

 

Jennifer [00:29:37] And you do some robustness checks, as one does in good empirical papers like this, primarily in this case to make sure the effects you're finding aren't simply due to incapacitation either from that initial jail spell — so perhaps some people are still locked up on Election Day — or because that jail spell might lead to subsequent recidivism and a felony conviction or incarceration. So tell us what you do there and how you're able to convince yourself that those possible effects aren't driving your results.

 

Ariel [00:30:06] Yes, you've put your finger on it. We might wonder whether people would still be in jail on Election Day or whether they might be there again, or maybe whether they could have ended up no longer allowed to vote if they've been convicted of a felony after they got out of jail the first time. And to start with this first question of whether people are still in jail. That struck me as the less likely part of this story because, as I noted, most sentences are a week or two long. But we can also we can drop all cases handled in 2012, the election year, to really rule that out for just about everybody. And the estimates still look quite similar. I think the harder part is this question of do you maybe get arrested again and then end up either sent to jail again or perhaps with a felony conviction?

 

Ariel [00:30:50] And one thing that I do here is to run a version of my main analysis, but with a different dependent variable where instead of estimating jail's effect on voting, I'm now estimating whether getting a jail sentence in that first case in, say, 2010 or 2011 makes you more likely to get jailed again by the time of the 2012 election. Or whether it makes you more likely to end up with a felony conviction by the time of the 2012 election. And I find no evidence that jail in your first case makes you more likely to end up in jail again or to pick up a felony by the 2012 election. And I should note here that there is research that has found that jail time can make people more likely to have future contact with the legal system. But I'll note here that we're looking at a pretty tight timeframe. We're talking about people going from their first misdemeanor case, no earlier than 2008 and sometimes as late as 2012, and then to another jail time or possibly a felony conviction by the 2012 election. And so I think it's maybe not surprising that we don't see a ton of return by that next election. We see relatively low rates overall and we don't see it being predicted by that first jail sentence.

 

Jennifer [00:31:58] So taking all of this together, what do you see as being the main policy implications of this paper?

 

Ariel [00:32:04] Well, I think it's worth citing this paper, along with the set of papers recently that has suggested that there are very serious costs of jail for both the individuals who end up sentenced to it, as well as broader ones for their their families and communities, in terms of some of the kind of lifestyle disruptions that I described earlier in terms of job and housing loss. But here, this points to a loss of civic engagement in ways that I think matter both for individuals, but also for the communities of which they are a part. Because here we're thinking about, not just demobilization, but racially disparate demobilization in ways that I think could really start to add up for some communities. We know that, you know, incarceration is not just evenly spread geographically across place. It's not evenly spread across society. And so if we think about the neighborhoods that actually end up with quite a lot of residents ending up in jail, we could think then about a substantial number of people, particularly in black neighborhoods, who would then kind of drop out of the political system because of this experience. And that would turn into not just, as I said, individual people not voting, but also, I would say real political impact on the places they live, because lower voter turnout is going to translate into basically these places being less likely to have the ear of politicians when those policymakers are deciding future future laws, future policies, including those around incarceration. And so I think this could translate into the places that are most affected by our incarceration policies, also being the places that are least likely to have any voice in them as we're setting them going forward.

 

Jennifer [00:33:53] This paper was published this year, 2019, in the American Political Science Review. Publication lags, I mean it's been out for a while, and I know you've continued to work in this area. So what other more recent research should we talk about here that helps shed light on these issues?

 

Ariel [00:34:11] Sure. Well, one thing on my mind lately has been how we think about the electoral impact of felon disenfranchisement laws, so thinking about people who are legally prevented from voting, unlike in the misdemeanor cases we were just talking about. And I have a project with Avery Nguyen who's one of our awesome MIT undergraduates looking at what would happen if- well, if — as some Democratic presidential candidates have proposed — if people were allowed to vote while they were serving felony sentences in prison. So we take a particular cut at this where we start with data from the state of Vermont, which is one of the two states where people are currently allowed to vote while serving felony sentences. And we merge — this another of these kind of administrative data projects — we match up the voter file from the state of Vermont to a census of people serving felony sentences to figure out what turnout has looked like among this population of incarcerated people in recent elections. So that gives us a sense of of how much voting we can expect when people are not only allowed but encouraged to vote while incarcerated.

 

Ariel [00:35:20] And then we can extrapolate from that a little bit. We can do some back of the envelope calculations and say, "What if other states were to adapt the same policy as Vermont of allowing people to vote while they were in prison for felony cases?" What if in those places incarcerated people voted at about the same rates as people do in Vermont right now? And what we find there is that even in a state like Vermont, where there's this long standing policy of protecting people's right to vote while they're incarcerated, in fact, very few people end up voting. So fewer than 1 in 10 in the most recent midterm elections. And so if we then look at what would happen in other states, right. Even states with fairly large incarcerated populations, the number of additional votes that you would expect to see from a policy change like this would actually not be that large compared to, for example, the winning vote margins in recent statewide elections.

 

Ariel [00:36:15] And so this exercise makes us think that, well, when we talk about incarceration and voting, two things. First, I think we should probably be having debates about felon disenfranchisement laws on the basis of of what we think is right and what we think the rights of citizenship should mean. Not so much about what we think felon disenfranchisement laws are going to mean for our preferred political party or our preferred candidate, because I think those effects of felon disenfranchisement laws have probably been substantially overstated for a number of reasons. But also — and I think this actually comes out in both this paper and in the jail and voting paper we were just talking about — I think we often focus too narrowly on whether people have the legal right to vote, which is certainly important. But I think we should also maybe be going a step further and asking whether people are using the right to vote when they do have it. And if not, why not? Why that is, you know. Is it that- is it that people who have previously been incarcerated, who have been convicted, is it that they don't know whether or not they're allowed to vote because of, you know, various changes recently and high profile cases about people having voted when they were not eligible to do so and actually facing criminal charges? Is it that people don't think they have anything to vote for? Is it that no one asked them? Is it that there are other practical barriers to voting that might need to be addressed? Basically, I would say my takeaway from these couple of projects is that I don't want us to just stop at this question of whether there's a law, but to think about what it would mean for people to be meaningfully incorporated into the political system.

 

Jennifer [00:37:56] And presumably that, you know, if people did start trying to use their right to vote at higher rates, then we could actually see impacts on the electoral outcomes. Right. I mean, so the Vermont paper's taking current turnout rates as given, if I recall correctly. If people started turning out at rates closer to what the broader population does, I would expect that would matter. Is that your take also?

 

Ariel [00:38:20] Hmm. If we could imagine getting to a place where people were voting from behind bars at the rates that you see in the general population? Yes, I think that would be a very different calculation in terms of electoral impacts. But I will note that I mean, I think I think that in the world that we live in right now, Vermont is in a lot of ways the best case scenario in the sense that I mean, the State Corrections Department actually has a clearly stated directive on its website about how it believes that it is its job to help facilitate the right to vote. It lets in volunteers every election cycle to help register people and have them sign up for absentee ballots. Vermont itself is already a relatively high turnout state. And so, I mean, to the extent that I think Vermont is probably as thorough in helping people vote as I could imagine any U.S. state being, at least in the current moment, I think you're right that if the rates were to go up substantially, that would be a very different calculation. But it's hard for me to see how we get from here to there without some other really big changes.

 

Jennifer [00:39:28] Fair enough. That is a very fanciful hypothetical I threw at you. OK, so what what's the research frontier here? What are the big open questions in this area that still need to be answered then?

 

Ariel [00:39:40] Hmm. Well, I can think of at least a few. One is that I think there are still a lot of remaining questions about how existing findings generalize. So, for example, you and I have just spent an hour talking about basically four years of records from Harris County, Texas. And I think there are all sorts of good questions about how the experiences of incarceration or the effects of incarceration might be different in different contexts. Right. And certainly when using administrative data, I like to think about how things might generalize and to make my best guesses. But ideally, we would also be bringing to bear all of the new forms of administrative data from many other places that have been coming out recently to see how well our findings report to different contexts. So that's one thing.

 

Ariel [00:40:26] I think another thing that maybe comes out of something I was just talking about would be, you know, if we were to set aside questions of causality for a little while and just note that descriptively a lot of people who have had contact with the criminal legal system don't vote. I do think there's still a lot to be done on how to how to reincorporate or even maybe just incorporate people into the political system. So whether we can try and use some of the same kind of mobilization tactics that we see campaigns using in general or whether, you know, whether we might need to do other things to try to convince people that voting is worthwhile and that it's feasible for them. So this is another kind of angle on that point I was making a minute ago about how felon disenfranchisement laws are certainly one part of the story here. But there are also just a really large population of people who are previously incarcerated or have had felony convictions in their past who are not currently disenfranchised and yet who are not voting. Right. It's a very large group of people. And I think if we could try and figure out what's going on there with that nonparticipation, I would certainly be interested in it.

 

Ariel [00:41:36] And then finally, one more thing on felon disenfranchisement laws — even as I say we talk too much about them. You know, I've proposed one really narrow cut and you've pointed out the ways in which this is narrow, looking at just how many people might use the vote from prison based on some things we observe right now. But I also think and maybe this gets to your point about what if people voted at much higher rates? I think sometimes people have theories in mind about how the very presence of laws like this, restricting access to the franchise can shape people's relationship with the state and their ideas about whether the political system is open to them, maybe even if they themselves have never been arrested or convicted. Right. Maybe there's a broader message here about how open the system is to one's input. And if that's the kind of story that people have in mind about what's going on with felon disenfranchisement laws, then I'd also love to see more work testing these kinds of broader perceptual impacts, not just the question of, you know, narrowly how many people would take up the right to vote from prison. But what does this mean for people's broader conception of themselves as potentially participatory citizens?

 

Jennifer [00:42:44] That sounds like a tough question to measure the causal impact of. You need a really good natural experiment there. Great. Well, my guest today has been Ariel White from MIT. Ariel, thanks so much for talking with me.

 

Ariel [00:42:57] My pleasure. Thanks for having me.

 

Jennifer [00:43:04] You can find links to all the research we discussed today on our website, probablecausation.com. You can also subscribe to the show there or wherever you get your podcasts to make sure you don't miss a single episode. Big thanks to Emergent Ventures for supporting the show and thanks also to our Patreon subscribers. This show is listener supported, so if you enjoy the podcast, then please consider contributing via Patreon. You can find a link on our website. Our sound engineer is Caroline Hockenbury with production assistance from Elizabeth Pancotti. Our music is by Werner, and our logo is designed by Carrie Throckmorton. Thanks for listening and I'll talk to you in two weeks.