Probable Causation

View Original

Episode 67: David Phillips

Jennifer Doleac

David Phillips

David Phillips is an Associate Research Professor of Economics at the University of Notre Dame.

Date: February 15, 2022

A transcript of this episode is available here.

See this content in the original post

Episode Details:

In this episode, we discuss Prof. Phillips's work on connecting people released from jail with mental health care:

“Reducing Re-arrests through Light Touch Mental Health Outreach” by Mary Kate Batistich, William N. Evans and David C. Phillips

OTHER RESEARCH WE DISCUSS IN THIS EPISODE:

TRANSCRIPT OF THIS EPISODE:

Jennifer [00:00:08] Hello and welcome to Probable Causation, 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. My guest this week is David Phillips. David is an Associate Research Professor of economics at the University of Notre Dame. David, welcome back to the show.

 

David [00:00:28] Good to be back Jen.

 

Jennifer [00:00:29] Today, we're going to talk about your research on mental health outreach for people leaving jail. But before we get into that, could you tell us about your research, expertize and how you became interested in this topic?

 

David [00:00:39] Sure. So I worked as a research faculty with the Wilson Sheehan Lab for Economis Opportunities or LEO. And so we're a lab that focuses on anti-poverty interventions in the United States. So we partner with organizations around the country, whether that be a nonprofit organization, county, government, so on who's doing anti-poverty work and that can be a lot of different spaces and some of that's just some of its crimina justicce and some in  other spaces, like I do a lot of work in housing and homelessness, for instance, and we work with them to do evaluation type of work that's measuring the impact of their interventions. And so this particular project, I think sort of an interest in this topic for me grew out of a lot of folks, like individuals, small contacts with particular partners around the country and just sort of becoming convinced that while mental health is just so intertwined with the criminal justice system, particularly at the local level, that it's something to think about more. So like for me, a lot of these individual interactions with particular people provide a particular services or services, in particular courts or jails like it really kind of hits you in the face that mental health is closely connected to the criminal justice system.

 

David [00:01:47] So for one project, I was working with folks doing public defense in Seattle, sitting at Seattle Municipal Court with them. And, you know, I'm not a lawyer, so I don't really entirely know what's going on and people are being arraigned in court and  the public defender leans over to me  is like, "Oh yeah, so here's what's going on." It's like, there's this woman who's clearly both chronically homeless and has mental illness, and she's being arraigned right now because she became convinced incorrectly that bedbugs were in her clothes. She threw them in a portable toilet and put them on fire, and it destroyed the port-a-potty and so she's been arraigned for destruction of property and arson. And it's just like, OK, you know, like here, here's a person for has incredible challenges in the world, and she's now caught up in the criminal justice system because of that.

 

David [00:02:31] And it just sort of sit there and you watch sort of misdemeanor court go and you just see this happen over and over and over again. If people it's like the reason that you are connected with this seems like it's because of your mental illness rather than something else. And so I think those are just anecdotes, there's evidence, but it's, you know, it sort of hits you when you see it in person. And then I think the opportunity to learn something about that then came up with our friends in Johnson County, Kansas, in suburban Kansas City, where Bill Evans and Mary Kate Batistich, two of my colleagues that got connected to folks in Johnson County who are doing some practical work, there was like, OK, here are some folks, we're doing some work around how you might treat mental illness in the criminal justice system a little bit differently and what that might actually look like.

 

David [00:03:14] And so I think for me, it's a combination of both sort of seeing the individual level effects of this and seeing that happen to show up in lots of different situations and running into it and then saying, okay, here's an opportunity to maybe learn something about how we can do better.

 

Jennifer [00:03:27] Your paper is titled "Reducing Re Arrests through Light Touch Mental Health Outreach" it's coauthored, as you mentioned, with Mary Kate Batistich and Bill Evans. And you framed the study as being relevant to what you call a mental health crisis in corrections. So what do we know about the share of people in the criminal justice system with mental illness?

 

David [00:03:48] It's shockingly high. So that sort of anecdote that I mentioned is actually a very common thing. So in Jackson County, where we're going to be working in Kansas, about a quarter of the people who take a screen for severe mental illness are going to screen positive for having a severe mental illness and that's not unusual. So it turns out that in the world, as we've structured it now, like some of the largest mental health providers in the country, are actually local jails. So people like to kick around the idea that Cook County Jail is actually most of the time the country's largest mental health institution. And so it's both a large fraction of people coming to contact, particularly with local jails and also in terms of and sort of mental health services in the United States actually a shockingly large percentage of them are provided by local jails, correctional facilities.

 

Jennifer [00:04:34] And you mentioned some trends that appear to be driving this crisis that you describe. Tell us a little bit about those trends.

 

David [00:04:41] Yeah, we try to pick out three things for sorting out, like how did we end up in this situation where our default option in some ways for a lot of folks was that severe mental illness is that they would go to jail.

 

David [00:04:52] And we sort of think there's three things going on. One is that there has been documented increase in just mental health issues and mental illness overall. So even pre-pandemic obviously there's been a lot going on since the pandemice, but even before the pandemic, if you look at rates of severe mental illness and a couple of decades before the pandemic, they were increasing over that time period. So you have that going on where mental health issues maybe are becoming increasingly important in society. You also have, at the same time, mass incarceration happening right where you start picking a few years ago, but still quite high now, you know, somewhere around a million and a half people in correctional facilities.

 

David [00:05:26] And so that's happening that increase in the number of people that we're locking up in criminal justice facilities. It's happening at the same time that fewer people are institutionalized in mental health hospitals. So if you go back to the 1950s, something like there's something like one bed for every 300 people in the United States in a mental health institution. There were a lot of people in mental health institutions 60-70 years ago. Now, instead of one out of 300, that's like one out of 4,300. Now that last trend, that's not necessarily a bad thing overall, right? I don't know that we want to necessarily go back to the days like "One Flew Over the Cuckoo's Nest" and there were real issues with mental health institutions in the 40s and 50s. Right. And so you have this sort of combination of trends where that used to be, how we as a society works with people who had severe mental illness. And we said that wasn't acceptable and it kind of went away and we maybe have not taken up that slack with something else as well as we could.

 

David [00:06:24] And so in response, like what happens if people fall through the cracks, they end up in jail. And so we have this situation where that's what we were doing a lot of people as we were putting them in jail in prison. And so that's what's been happening is a lot of people with mental illness.

 

Jennifer [00:06:34] Yeah. So this all means, as you said, that jails and prisons are primary places where people might need and receive mental health care. So what types of programs are typically in place in jails and prisons to treat inmates with mental illness?

 

David [00:06:49] My experience of this is that there isn't a typical response that is really widely varying through my work with LEO being heavy partners across the country. I've been in jails where what it meant for somebody with severe mental illness in jail, what it meant for them is that they were frequently in a cell by themselves, literally in a metal cage with limited access to mental health professionals and medication. I've had people who work within the criminal justice system intentionally, you know, take us on the tour on that side so that they know this is how it works and this is really problematic. And I've also been in other places where, you know, especially in recent years, where people with mental illnesses who are coming through the courts are being diverted to a totally different type of court that it's not through sort of traditional adversarial misdemeanor court. Instead, it's some sort of community court or mental health court that looks very, very different actually to the person I mentioned before in Seattle. Like if she came through court now in Seattle, she would end up in a very different situation than what she is right now.

 

David [00:07:46] And so there's like a range is like incredibly wide. And in most places, I think somewhere close to where Johnson County is, which somewhere in the middle where lots of people continue to be incarcerated, who have major mental illnesses. And then the question is within jail, like what kind of services are they getting? And so that can vary, too. So even can Johnson County over the past few years, there's some variation in how much staffing is there. Is there a psychiatrist who is actually readily available to meet with everybody who needs that kind of help? Or is that like? Well, we've got one person on call who shows up once every couple of months, like they've gone through a process of that changing considerably. I think a lot of places have sort of beefed up staffing on the side of providing help with mental illnesses. And there's also variation in terms of like what medicine people can have access to. So some places they're going to be providing people medication when they find that somebody else needs, it's OK let's get you on psychotropic medication, whatever it is to try to help you while you're in jail and other places, that doesn't happen.

 

David [00:08:45] And other times, it's just really hard to because, as you know, people cycle through jails really quickly, frequently. And so it's not always possible, right? If somebody comes in, they're gone a few days later for the jail to be able to assess who needs help and let's get them actually a prescription so that they can get the medication they need is a right. That's hard to do quickly. And so the short answer is that it's like my experience is that this is widely varying from place to place in terms of how people are addressed by the criminal justice system overall. And then what sort of services they receive once they enter.

 

Jennifer [00:09:17] Yeah. And so I like groups, jails and prisons together in my question. And but you're going to be talking about jails today, and it is I think this is all highly relevant to jails. People should keep that in mind, as is probably a little bit easier in prison since people are there for a little longer, but I'm sure still lots of variation at the prison level.

 

David [00:09:34] Yeah, that's a good point. I mean, it's getting more intensive treatment programs are much going to be much more common in state prisons than their local jail.

 

Jennifer [00:09:42] And just a quick note, the Johnson County you mentioned is Johnson County, Kansas, which we will talk more about in a moment. All right. So what do we know about the effectiveness of the various programs that jails and prisons are implementing? Do we know anything?

 

David [00:09:57] I mean, given how many, how important it is and how many people are affected, I think we know a lot less than we should, but we do know some things. I think one thing that comes out of the literature, particularly, I think, is that just community provision of mental health services, right, that can make a big difference, and so you see effects on on contact with the criminal justice system when there's just more mental health care at rounds like Monica Deza and coauthors of a paper just looking at prevalence of mental health care offices across counties and showing this when a county gets more health care officers that the crime rates go down. I think at least comments job market paper from a couple of years ago, showing that when Medicaid gets expanded to single adults, people are sort of on the margins of eligibility for Medicaid, but that can make a difference for their contact with the criminal justice system.

 

David [00:10:44] And one way that that matters is through giving people access to mental health care. One thing we know is that mental health care can be effective in reducing contact with the criminal justice system if that's possible. And then the question is sort of how to implement that and actually do that in practice from the sort of more criminology literature. There's a number of studies related to, for instance, mental health court that tries to provide an alternative to traditional criminal court for people with severe mental illness. And there's a number of studies of those types of programs. I'd say that what we get out of that are typically evaluations of very intensive interventions that let's change the entire criminal justice system, build a totally different court for these folks.

 

David [00:11:28] That's a really big change that may or may not be a good thing, right? But it is really hard to do. It's a little bit different from what we're going to be looking at, which is much more light touch. I think in the literature where we see a lot more evaluations of sort of very light touch outreach type of interventions. There is a number of papers around affecting failure to appear rates including some that have been featured on your show so right, like leaders work on contacting folks by text message to try to remind them of their court dates, right? There's a lot of evidence in that space. Now I think about how to use light touch interactions to get people in contact or to affect their contact with the justice system and make it more positive. I think there's there's less that we know about sort of the intersection of these two things.

 

David [00:12:10] So looking at mental health, looking at sort of lighter touch type of interventions and the interaction between those two, that's kind of the space that we're going to live in. That's going to be where I think there's not much at this point.

 

Jennifer [00:12:22] So why don't we know more than we do? Why is this such a difficult issue to study?

 

David [00:12:27] Yeah, that's a great question. So I think there's a couple of things. I mean, one is the combination of data that we need to take even a first pass at the question is pretty challenging. I mean, the first thing you really need to know is you need to know here is a person who's in contact with the criminal justice system who have a reliable indication that they have a mental illness. And that's just hard to gather, right? A lot of administrative datasets don't have that due to the variation in how the criminal justice system treats folks with mental illness or doesn't see folks with mental illness. There's also because of that, a lot of variation in how much we even just know about does this person have a major depression or not? Does this person have schizophrenia or not? And so I think the first thing you need is you need some jurisdiction that's screening people for mental illness in a systematic way that's trying to take some of the subjectivity out of it, that's trying to deal with all of the very hard shades of gray between, you know, the categories of mentally ill or mentally ill, right? That it's like, there's that there's a lot in between there. And so you need someone who's doing that in a systematic way for the recording. So identifying folks with mental illness is hard.

 

David [00:13:33] And then right then that data has to be linked to some measure of, okay, here's another intervention that the person is receiving and be linked to some measure of recidivism. And those are maybe more typically measured, but then you have to actually have those connected to information on mental illness. So I think the data requirements, I think, are pretty high and that's probably the biggest barrier.

 

David [00:13:52] The second barrier I'd point out is that I do think that sorting out identification, trying to measure the causal effect of these programs can be quite difficult. They're typically trying to target people specifically because they have lots of challenges, right? So they're targeting people who have been screened for severe mental illness. And so we're going to send them a mental health court or something like that. And you might imagine that somebody who has all of this stuff going on, it's going to be different in terms of the recidivism outcomes, both in terms of just sort of overall risk of recidivism and also in terms of what direction that might be going to on later on their life. Because we're looking at a person at the very moment in time, it's like, okay, you're in a crisis moment right now and it might be the case that, for instance, your risk reverts back to the main or something like this.

 

Jennifer [00:14:42] Yeah. So just comparing people who are going through the mental health court with people who aren't isn't going to tell you the effect of the mental health court because they're different people, right? As you said, there might be this brief crisis moment where they would have recovered anyway without the intervention, such as looking at like an event study, like a time series or something of an individual won't tell you and even like comparing a jail with certain programs to a jail without might not. All that informative because the jails could be so different on other dimensions. And so it's really tough to find a good comparison group here. But you guys figure it out. So let's move in that direction. So in your paper, you consider the effects of an outreach program in Johnson County, Kansas. So tell us a bit about Johnson County.

 

David [00:15:27] Sure. Johnson County, Kansas is a county in Kansas, it's in the Kansas City metro area. And so I think it's useful to think it's I think most people are maybe less familiar with Kansas City than they are with some of the larger cities in Chicago, New York, Los Angeles, Kansas City metro area in a lot of ways and a lot of dimensions is kind of a microcosm of the United States. So it has a median income very similar to the US median income, a similar poverty rate, similar sort of age distribution is a little bit less racially diverse than a typical ones. So that's one thing to keep in mind. But on a lot of dimensions, it looks kind of like a microcosm of the United States, and it's kind of in this in-between range between, you know, being a good size metropolitan area, but not so big to be fundamentally different from a lot of places that are smaller. Johnson County itself, is the more affluent part of that metropolitan area and so it's going to have higher incomes. It has a higher fraction of folks there, who are white and so on. If people want analogy, I I reference lots of things to Chicago myself. Like if you think sort of north side of Chicago, that's a better reference for some folks. That's sort of the role that Johnson county kind of plays in the metro area like a lot of metro areas.

 

David [00:16:32] There's a lot of sorting by into the race and so on, and just becomes pretty affluent portion of a pretty standard metro area.

 

Jennifer [00:16:39] OK. And in 2016, Johnson County began using a brief jail mental health screen. That was the official name I gather to assess inmates mental illness. So what types of questions does it ask and when was it administered?

 

David [00:16:54] Yeah, so this is a screen that was developed by sociologist Henry Steadman and a big team led by him. And the goal of it is to really quickly screen for severe mental illness. Right. So the place where it's administered, it's at central booking. So when someone comes in to Jackson County Jail, they've just been arrested by the police or dropped off. They're being booked. They haven't been arraigned for anything yet. They're just being dropped off in jail and they're doing their basic intake of basic information for folks. What they did is they inserted this big question stream into that very rapid, high volume process. And so it has to be really quick to be practical. And so it's really quick questions. There's an initial set of questions of the first and six of the eight that are trying to get at symptoms of major mental illness. I think severe depression, schizophrenia and bipolar, and it's trying to get at those really quickly. And then there's a couple of questions at the end that just ask about previous contact with mental health care. So whether it's somebody who's currently on medication for a mental health issue and whether they've been hospitalized before. And so it's sort of really rapid. They asked these eight questions and you get a really quick measured. It's not going be perfect, but it can give you a quick sense that this a person have a major mental illness or not.

 

David [00:18:12] They started doing this after not having done it for a long time, and a lot of jails don't have a good sense of like, OK, they have some subjective sense or some sense from staff or trying to identify people who have mental illness issues, either because they would treat them differently or because they're concerned about negative events like suicide attempts in November 2016, Johnson county, starrted doing this more systematically like, OK, we're going to ask these eight questions to every single person who comes through booking,.

 

Jennifer [00:18:36] OK. And then in 2017, so about a year later, Johnson County started a new outreach program for people with severe mental illness who are released from jail. So how did this program work?

 

David [00:18:47] Yeah. So they had this foundation of having already done the screening and so that that was fundamentally because if you don't have this screening, then you don't know who is at risk. Right? But yeah, like you said a few months later, having already done this thing for a little while, they said, OK, we could use this information for something, and they they started with sort of the simplest thing that that could be done, which is just to try to contact people, say, OK, people are cycling through local jail quickly. This is a moment in time where we come into contact with them. We know that they have a need for mental health care and we have contact information for them so that as they're exiting, try to connect them with services. And so that's for the folks at Johnson to do. This is a partnership between Johnson County Jail and T44 Mental Healthcare working for the county. And so what they do is basically just try to connect people with mental health care services when they exit.

 

David [00:19:36] That's usually one of two things. The county has a publicly funded mental health care center, so physical place that people can go for mental health care. And so one big goal of it is just to connect people with an appointment there or get them to walk in because it has a walking option as well to shop the place and get connected with the mental health care provider. The other thing that they do is sometimes they didn't talk to the service and they find out, Oh, you have a history of getting care for your mental illness. Let's connect you with your previous provider so that you can continue that relationship. Let's to do what we can to facilitate that being picked up again. And so, yeah, it's usually one of the two, either something for the private sector connected to you or getting connected to new care that's funded by the company.

 

Jennifer [00:20:19] And how did those outreach workers actually contact people? Was the contact information pretty good?

 

David [00:20:24] Yeah, they worked really hard to meet these folks, and they basically worked really hard to get in touch with folks, but working off relatively limited contact information. So a lot of it is really just manning the phones. It's just calling folks more than once. So they kind of they call everybody three times and so call them, they're going to call them again or call another time after that to make sure, okay, can I get in touch with this person and then do the connecting to services? We talked about that's most of it for most people. They also use the screen to identify some folks who have particularly severe mental illness or people that they have a past history with that they think might have particularly severe mental illness.

 

David [00:21:01] And for those folks, they also might attempt to an in person follow up. So most of the time it's three phone calls, but then occasionally it's ok we're going to drive, we're going to door and see if we can get someone to open up their.

 

Jennifer [00:21:13] And finally, in order to refer people to mental health care, there, of course, need to be providers with availability for new patients. So what was mental health care availability like in Johnson County during this period?

 

David [00:21:24] Yeah. So in Johnson County, there's better provision of mental health care than an average county in the United States. If you go to the county, business patterns data in the Deza et al paper I mentioned before and you just look at number of mental health care offices Johnson County has about 50 percent more mental health care offices per person than the U.S. average. And so they do have somewhat more extensive mental health care, and that's important. That's the type of intervention we're talking about. Picking up the phone and calling somebody presumably is going to be totally ineffective in a place that doesn't have mental health care resources available. So this this is complementary to some existing resources being out there, if those resources weren't out there, than these interventions is probably a bad fit. That said, the Kansas City metro area again is pretty similar to average, but Johnson count does have better mental health care in terms of just raw numbers than most places.

 

Jennifer [00:22:14] All right. So in order to measure the causal effect here, you need something akin to randomization to measure the effect of this outreach program. And it turns out that the way the program was implemented gives you a useful natural experiment that divides people into similar treatment and comparison groups, which is exactly what you need. So who is treated by this outreach program and who is not?

 

David [00:22:36] Yeah. So it turns out that the way that they initially started rolling out this outreach in Johnson County was based off of where people live. So it's really where people lived and presence of mental illness. So they right, they did the screening that was mentioned before and among the group of people that screened positive for having severe mental illness, they didn't reach out to everybody they had staff limitations. So on limited resources, what they decided to do is they reached out to people who were residents of Johnson County. They didn't do outreach with people who are not residents of Johnson County. That turns out to be important because a lot of people end up in Johnson County Jail, who we're not Johnson County residents. This is true in a lot of metro areas, so you get sort of the affluent county in the metro area. It's not unusual to have a large number of people who are in their jail, who are residents of the urban core or other parts of the metro area.

 

Jennifer [00:23:25] And that's because they committed a crime in that county.

 

David [00:23:28] That's correct, yes. So you have people who live in, say, Jackson County, Missouri, which is the Missouri side of Kansas City. You end up in jail and Jackson County, Kansas, because they were arrested on a charge that they're located in Johnstown. And so, yeah, then what you have is you have a group of people who were all arrested by the same police departments booked into the same jail and screened with the same tool for identifying mental illness all screen positive on that. But some of them are going to be getting outreach and connection to mental health services because they're residents of Johnson county and other people who are going to be residents of a couple of nearby counties that are other parts of the metro area that have all of those other things to be true, but it's not going to be true that they're going to get connected to services and have this outreach.

 

David [00:24:12] And so that's a comparison that we're going to make. We're going to consider Johnson County residents, our treatment group here and residents of two other neighboring counties as the control groups that we can compare to.

 

Jennifer [00:24:22] OK, so tell us a little bit more about that. How exactly do you use this natural experiment to measure the causal effects of the mental health outreach program?

 

David [00:24:31] We're going to do this as a difference in differences design, so we're going to look at changes over time. So you can imagine following these residents of Johnson County, Kansas, and see what their recidivism rates looked like before outreach started. So you go back in 2016, say, OK, how likely were these folks who were being screened for mental illness screened as positive for mental illness but weren't yet receiving services? Okay, how many of them, or how many of them were turned back to jail? And then we're going to follow those folks over time and see how did that change? Did it fall when we started, we just county started doing them, and then we're going to compare. We know that other things could be going on over time recidivism rates change with the seasons of the year and everything else. So what we're going to do is we're going to benchmark those changes in time for Johnson County residents against people who reside in other counties. But we're booked into the county jail, so we're going to follow those folks as well. We're going to see how they changed over time, and that gives us a sense of just how was the world changing over time? How is policy in Johnson County changing over time and so on? And use that as a benchmark and say, OK, the difference between those two differences, the difference in trends between Johnson County residents and nonresidents. That's how we're going to identify the effects of this outreach.

 

Jennifer [00:25:41] And you also do something with neighborhoods that are just on the border of these counties, right?

 

David [00:25:47] That is right. So we do a couple of other things, right. So the comparison across two counties is not perfect, right. In my ideal world, right, we would it would be wonderful to run. This is a randomized controlled trial a hundred percent confidence is strrongly unlikely with really high degree of confidence, right ok, this is due to the program. It's not anything else going on to compare across counties, right. You're going to have some other things going on, right. The residents of Johnson County, for instance, are much more likely to be white than the residents of the neighboring counties. That's important when we're talking about the criminal justice system. So we do some other things to try to narrow down a bit to make sure that we're not picking up other ways these two different groups of people are changing over time. One of those is what is what you said, which is we look at zip codes that are on the border of the two counties. And so you get, you know, for instance, right on the eastern border of Johnson County, where it borders Missouri and you get folks who live on either side of that line they look much more similar to each other than sort of residents of Johnson County in the Kansas City metro area due each other in general. And so you can get a little bit cleaner comparison there. It gets a little bit noisy because fewer people in that comparison. There's a little bit less precision in the estimates. We're going to get out, but we get similar resuls that's encouraging to us that if you sort of narrow down a little bit to folks running a little bit more similar neighborhoods that you get similar results.

 

Jennifer [00:27:04] OK, so what data do you use for all of this?

 

David [00:27:08] So we're going to heavily rely on data from the jails themselves. So Johnson County itself is going to do most of the work with its bookings data. So we've got that centralized booking process that's going to give us a few things. Everybody's going to use the mental health screen during that process. And so we'll see there whether somebody screens positive or not for mental health issue. That's going to be linked already to information on whether that person returns to jail for another booking on another charge in the future, which is going to give us an outcome measure. We're going to do a little bit more work because again, we're making this comparison across counties and a lot of these folks reside in another county and particularly our comparison group resides in another county. So we want to make sure that we have recidivism measures from those counties as well. So we're going to pull in data from the two neighboring counties as well in our recidivism measures that we're measuring not just do you end up getting rebooked in Johnson County, but do you get rebooked  into sort of three county area.

 

Jennifer [00:28:03] Just as a side question. How difficult was it to get the data from all three counties? When I first read this, it was like, that's feels very lucky that everyone cooperated, gave you their data.

 

David [00:28:14] This paper would have existed a couple of years earlier to 2016 and 2017 are very recent, right. Oh yeah. This paper would'vee existed a couple of years ago if we had been content with one county, but we really thought it was important to get three counties. Now I will say then how we ended up getting it is less about us. We did some, you know, of the hard work in there.

 

David [00:28:34] I should give lots of acknowledgment to the wonderful team at LEO that did a lot of the hard work of continuing to push the ball along for many different people do that, but especially our partners at Johnson County were really helpful. They wanted this evaluation happened. They wanted to help us do something that was as good as possible. And so they were very helpful in connecting us to folks. And they were wonderful people in Jackson and Wyandotte counties that responded and were helpful in thinking about problem-solving that would actually make it work. And so it was a lot of people who had decided they wanted to say yes to this thing and helping make it work, rather than trying to find ways not to do it. So there's a huge to a team of people that make it possible to make three counties sit together, which is hard.

 

Jennifer [00:29:12] Yes, the big challenge of our decentralized criminal justice system,.

 

David [00:29:16] Especially a metro area that covers two states.

 

Jennifer [00:29:18] Yes. Yes. That's right. Oh my gosh. OK, so tell us about the people in your sample. And in particular, how often do they screen positive for severe mental illness?

 

David [00:29:28] About a quarter of the people who are coming through Johnson County Jail are going to screen positive for mental illness. And so it's going to be a significant number of people as people, but it's going to be a significant number. We're screening positive for this type of severe mental illness. And then that group is going to end up being a group of people that has some characteristics that tag them as being particularly vulnerable. So they're very unlikely to be employed or married and so on.

 

David [00:29:51] Some of the ways that they're going to be different from a general population of folks coming through the county jail, they're actually going to be more likely to be female. They're going to be more likely to be white, but they're going to have similar arrest histories to other folks who are coming through.

 

Jennifer [00:30:03] And what are the outcome measure as you're focused on making?

 

David [00:30:07] The main we're going to do is going to look at jail pockets, which is going to be a combination of things, right. It's really sort of contact with the criminal justice system that may mean that the person who committed the crime had been picked up by the police. It's really sort of a combination of those things. We can't tell guilt from it, but we can tell that here's a person who ended up back in jail.

 

Jennifer [00:30:23] And you also have data on whether people actually connected with mental health care is hat right?

 

David [00:30:28] We have some information about what their immediate contact was, so we don't have information on the full extent of what mental health care someone received. That would be wonderful. And it's something that we pursued but were unable to get. But what we do have is what happened is the immediate outcome of that call. So when people, the folks across the county tried three times to get a call to somebody, they log, Ok, how did this thing? And then we have some information of, ok, here's how many they tried to contact, whether they got in contact with them, if they got in contact and then they connected somebody with a mental health resource. Here's what it was. And so we have some sense of that. Immediate services received less about sort of a full range of mental health care that someone accessed.

 

Jennifer [00:31:04]  Ok. And so the underlying assumption of your difference in difference is strategy is that the trends in Johnson County would have continued, as in the other nearby counties, if not for this new program. And this is, of course, impossible to test directly, but the evidence we usually look for is what we call parallel trends before the intervention. So what do the trends in rebooking and particular look like in Johnson County? And your comparison counties before the outreach program began?

 

David [00:31:34] They look quite similar. So we have we have sort of two different ways we can look at that. One is we can look at these few months at the end of 2016, very beginning at twenty seventeen, when they had started doing the mental health screening before they were providing services. And we can look and see how things were trending then and there we see both Johnson County residents and nonresidents trending together. The other thing we can do is that's only a few months, and sometimes it's nice in these sort of studies right to, as you know, to see a much longer period of time before and so another thing we can do is before they started doing the mental health stream. We can still compare trends for Johnson County residents versus not residents just among the entire population at the jail, not just those who screened for mental illness and then we can see similar trends for much longer period of time for this broader group that includes people who it's been positive for mental illness, but also those who were screened negative.

 

Jennifer [00:32:21] OK, great. Yes. All that gives you more confidence that these other counties are giving you a good counterfactual for what would have happened without the program. All right, so let's talk about the results. So first, you consider the effect of an outreach referral on the likelihood that people are actually contacted and connected with mental health care. So what do you find?

 

David [00:32:40] Yes, we do find that they are successful in connecting some people with mental health care resources. And that's, I should say by itself something that's useful to know it's not obvious that it would be possible by calling people up three times that you could get a non-negligible number of people connected to mental health care, but they do so they try to contact 95 percent of the people that they're supposed to try to contact. The staff does a good job doing what they're supposed to do. They actually contact about half of them. And so about half of people just never pick up the phone or they had bad contact info for whateverr and so there's that is a real issue.

 

David [00:33:11] And then of those about half that, they actually contact twenty seven percent of them overall, out of the total people they first started trying to contact twenty seven percent end up connected with the mental health resource, either their existing private practitioner or the county's resources. And that, to me, is interesting on both dimensions stuff like the 73 percent of people they couldn't getting connected and so there's a bunch of people that are going to be missed by this. But even so, even from a relatively light touch intervention, they're getting about a quarter of people connected to services.

 

Jennifer [00:33:41] Yeah, I agree. I had the same reaction that this result alone is really interesting and important, especially for a population that, you know, we'd worry is more transient, might not have cell phones, you know, that are always on and where you can reach somebody. It's kind of amazing that they were able to actually contact half and then get half of those connected. So super promising there oK. And then next, you consider effects on recidivism or rebooking in particular. So what do you find?

 

David [00:34:11] Yes, we do find that recidivism goes down and it goes down more in the Johnson County groups than the folks who are residents of other counties. And so that we interpret that as the effect of the outreach and so if you take that assumption right, then what you're going to get is a result that recidivism within two months decreases by about eight percentage points and then that eight percentage point decrease less out to about one year. So it's not sort of just delaying it would have been booked next month, but instead of going to book six months later, it looks like it sort of persist out to a year. And that's a pretty big effect. Eight percentage points what does that mean it turns out that recidivism rates are very high among this group, so about half of the people in our sample are going to return to jail within a year. Forty six percent of them and so knocking, that's down to eight percentage points is a noticeable change in recidivism for this group of people.

 

Jennifer [00:35:02] Right. And again, because this is just connecting them with mental health care, maybe some of these people then are fully engaged and get medication and go to weekly appointments, but maybe everybody just drops off next week. And so it seems like there's really something happening there wherer just connecting them is leading to real treatment if you're going to see effects of that size.

 

Jennifer [00:35:23] I think that's right. Yeah. So it's like in some ways we observe there's this big causal chain and we observe the first link in the chain, which is due to connected to something right away in the woods or something at the end, which is, do you return to jail? There's a bunch of things going on in the middle that we wish we could see, but we can't and that's one place where I think talk about room for future research later on. I mean, I think someone is able to observe all of those things happening would be really wonderful to know exactly what's going on in between.

 

Jennifer [00:35:46] OK and then as a placebo test, you use the same empirical strategy to measure effects on people who did not screen positive for mental illness. So why do you do this and what do you find?

 

David [00:35:56] Yes, we do this. And part of the reason is this isn't randomized controlled trial, so we don't have perfect identification of causal effects in this context. There are differences between counties where you talked about sort of this let's focus on the border strategy of dealing with that, but there's other ways you can think about dealing with that, right, if you're concerned that, say, Johnson County, Kansas, residents from their residents from other parts of the metro area are going to have other changes over time that are different between the two places or if you're concerned that there are other policy changes going on or maybe you're concerned, it's like, ok, Johnson County was just doing a bunch of stuff for its own residents, but not for residents of these other places to try to reduce recidivism. And that's just what we're picking up. Then you might want to look to say, ok, well, let's make sure that this is something specifically about folks who have been screened for severe mental illness. Let's make sure that that's where the effects are showing up and so that's what we do. So this is the sort of placebo test you mentioned of doing the same a digital strategy comparing trends of Johnson County versus not for people who have not screened positive for mental illness it sort of gets at that.

 

David [00:36:59] It's like, OK, we shouldn't see anything going on there. If this is really driven by mental health outreach and that's what we find, we don't find the sort of big effects on recidivism among the group of people who do not screen positive for themselves.

 

Jennifer [00:37:11] OK, and then you consider whether the effects of this program vary across different groups, including by race and gender, as well as whether someone had received mental health treatment in the past and what types of symptoms they reported. So what do you find there?

 

David [00:37:25] The biggest thing that we find is the differences by past mental health treatment, which we think is interesting, right. So the recall that the screen identifies some people for severe mental illness because of symptoms and in the first six questions, the screen in the last couple of questions of the screener about past use of mental health care. And so we can look at sort of a group of people who have indications of symptoms, but no self-reported past mental health care and the effects are really sort of concentrated among that group has really big decreases in recidivism and we find this really interesting. I guess you might imagine this type of it's some sense like a low cost estimate, why type of intervention. Let's screen everybody at the jail and then let's try to do some kind of contact with everybody. You may pick up some folks in that type of intervention where the returns are really, really high because they've slipped through all of the cracks before. And it seems like that's that's part of what's going on. Like, we're seeing these really big effects for people who haven't had mental health care in the past, despite the fact that they have very severe symptoms.

 

Jennifer [00:38:23] And so what are the policy implications of these results? What should policymakers and practitioners take away from all this?

 

David [00:38:29] I think there's a couple of things. I think one is that. The first step toward treating mental illness differently in a local jail may have really large marginal benefits, a relatively low marginal cost, at least to the county like you said before, we don't know the sort of cost, the health care that people are getting access to at least the county's immediate outreach cost is really low. It's some staff time to a phone bank, and that's having big effects on recidivism. And so given where they were starting from in a context where there was not a lot of psychiatric treatment going on in the jail, that first step could have a really big bang for the buck. And now they've changed some of what they do in jail now. So maybe the bang for the buck will be different now or different in another context where there's other care wrapping around, but I think it gives some indication that for communities that are thinking about, OK, we really need to take a first step at that first step may have to be hydrogens.

 

David [00:39:18] I think a second thing that I would say is that it convinces me that this large group of people in the criminal justice system who have mental illness, overlapping their contact with the criminal justice system that I want to think about interventions for them that are maybe different than what I think about somebody who's a Becker style, rational criminal actor who's like weighing costs and benefits. But there's going to be a big chunk of people who like the woman I started with at the beginning in Seattle. It's like the reason for her ending up in court is not so much about an understanding of this environment and making rational choice about what's going o it's it's related to her mental illness. And so like, if we have a big chunk of people in that category, we want to think about interventions that are particularly targeted at them. That's different that I might think of as somebody who makes their living, burglarizing homes or something like that, right. Those are different groups of people. We should think about different interventions for them.

 

Jennifer [00:40:12] Right. So threatening the woman you saw in Seattle with more punishment is not going to do anything to her decision, quote unquote decision to engage in criminal behavior.

 

David [00:40:23] And in some ways, that's self-evident. But then it's like, o, this large number of people in that situation. And here's an example of somebody intervening different way having big effects. And that's it. It give some meat to the anecdote of saying, ok, this might be a common occurrence, rather than something that's rare.

 

Jennifer [00:40:36]  Mm-Hmm. I think it's also really interesting, and your study highlights that even in a place where mental health care was seems fairly easily available, you have clinics there that people can just walk in. It took this nudge, right? It took these outreach workers to let this extremely high risk group that needed that care, know about it and help them access it. And so it just makes me think like even in a place where it feels like everything is, you know, going right or like everything is sort of lined up in a way where this care is available, people still aren't getting the care they need. And so you can only imagine in places where it's harder to access that this population kind of doesn't have a chance. OK, so are there any other papers related to this topic that have come out since you all first started working on this study?

 

David [00:41:28] I've been excited to see you a growing number of economists working on this. I think it seems to me like it's rapidly expanding, even like I've been doing reading lots of job market papers since that time of year. You know, just the podcasts this time of year where we hire economists and see how so there's lots of young economists. I think we're doing really cool stuff on this. There's a group of grad students from Texas who have Sam Arenberg, Neller, and Stripling that they have this interesting paper looking at Medicaid eligibility on adult incarceration that they're working on, showing that giving Medicaid to kids affects their incarceration later on. And they argue whatever mechanisms through availability of mental health care for kids.

 

David [00:42:04] Panka Bencsik is a postdoc at UChicago is doing a bunch of stuff around sort of behavioral health, more broadly construed as the intersection of both like diversion programs for people who are being arrested for drug charges, but also thinking about how crime has sort of the direction of causality, the way that crime might affect people's mental health when crime happens in the neighborhood. There's an interesting paper by a group at Baylor that's trying to sort out what's going on in mental health courts and trying to sort out. One interesting thing they find is that people are screened in mental health court and who actually does the screening matters for whether the person gets assigned to mental health care and whether what type of legal advice end up getting for public defender or private attorney. And so the idea that like how you actually work out the mechanics of these interventions for mental illness or contact with the criminal justice system, that it's a complicated thing. And it's just that it's been exciting to me to see other people. I think coming to a similar conclusion of how we see a lot more work in this area and to see it grow is really neat.

 

Jennifer [00:43:04] Yeah. So speaking of which, what's the research frontier? What are the next big questions in this area that you and others will be thinking about in the years ahead?

 

David [00:43:13] I think there's a lot less to do. I mean, I've I've had more conversations with local organizations who are doing cool things than I've been able to find. Here's a situation where we can evaluate this thing like there's lots of different jurisdictions are doing these corespondent models where they send public health professionals alongside the police to respond to call 9-1-1 calls that have mental illness attached to the reasons of a call, and an instance of lots of different places where diversion of folks has mental health and mental health courts in lots of different ways. And so some of these models we have evidence on, some of the evidence needs to be a lot better. Some of them we have like no evidence at all. I mean, there just are just very little no evidence is strong words, but very little evidence. And so I just think there's there's a lot of innovation happening at the policy level in this space. And I think it's going to take some time for the research to catch up to it and so just like filling out what we know about a lot of these are actually promising and which ones aren't I think it's just hard to know at this moment, and we really need to know a lot more. So I hope that there are lots of others who will beat me to the punch of running RCTs on some of these things and stuff because we just have so much more to learn.

 

Jennifer [00:44:19] Yeah, well, I will plug. I actually have students at Texas A&M who are working on a study on that co-responder our model you mentioned they have neat data.

 

David [00:44:26] Fantastic.

 

Jennifer [00:44:27] El Paso. Yeah, where they have these CIT, where they have a mental health professional, go out with a police officer and the results are kind of complicated, it turns out, but super interesting. And yeah, I think this is a space where to me, it just seems like there's so much great work being done by graduate students in particular, like they're able to like a lot of this just takes so much time right to build relationships with practitioners, and the grad students have more time than than faculty members do. So it's really exciting to see. My guest today has been David Phillips from the University of Notre Dame. David, thank you so much for talking with me.

 

David [00:45:04] Thank you so much. It's been a pleasure.

 

Jennifer [00:45:11] You can find links to all the research we discuss today on our Web site 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 Emerson Ventures for supporting the show. And thanks also to our Patreon subscribers and other contributors. Probable Causation is produced by Doleac Initiatives, a 501C(3) non-profit so all contributions are tax deductible. If you enjoy the podcast, please consider supporting us via Patreon or with a one time donation on our website. Please also consider leaving us a rating and review on Apple Podcasts. This helps others find the show, which we very much appreciate. Our sound engineer is Jon Keur with production assistants from Nefertari Elshiekh. Our music is by Werner, and our logo was designed by Carrie Throckmorton. Thanks for listening, and I'll talk to you in two weeks.