 Rob Samson and I met, I don't know, 50 and 18 years ago here at the University of Chicago when he was just launching this extraordinary neighborhood project. He continues to be the scientific director of the project on human development in Chicago neighborhoods. And I just chatted briefly with Professor Samson to learn that the MacArthur Foundation has funded a new phase of that study to do further longitudinal studies here and in Los Angeles, which will allow, this is staggering, a 16-year follow-up for the birth cohort who were originally enrolled in the project. I was saying to Rob that when I came to the University of Chicago, Shep Kellam and his colleagues had just launched the Woodlawn Mental Health Project, another extraordinary longitudinal study that also is in the field for 20 or 25 years on the mental health of children born and raised in the Woodlawn community. Rob, when I first met him, was here at the University of Chicago in the Department of Sociology, had previously been at the University of Illinois in Urbana-Champaign and has chaired the Sociology Department at Harvard for the past five years. He's a senior research fellow at the American Bar Foundation and I believe this year is visiting at the Russell Sage Foundation in New York working on his latest book. I'm just delighted to welcome him to Chicago to invite him here. You want to know the title of the book? What's the title of the book? I'll tell you at the end. Those are the end. Rob's talk to us today is going to be on the social reproduction of health disparities, lessons from the Chicago Neighborhood Project. It's a delight to welcome you, Rob. Thank you, Mark. So I guess I have to shout and talk loud. Let me know if you can't hear me. So the delight to be here today, and as Mark noted, I taught here at the University of Chicago for about 12 years. So in a way, this is like coming home intellectually. So that's very nice. I see a lot of new faces, which is also nice. But it's also returning home, so to speak, in terms of a city that I've studied quite a bit, along with a number of colleagues. And what I thought I'd do today is to walk you through some of the results from the Project on Human Development in Chicago Neighborhoods, which started in about 1995, give you an overview of some key results, and then focus in on a set of more recent findings that take advantage of some of the longitudinal data to address some of the controversies and issues that arise when one studies neighborhood effects. And I've long been interested in that, and I can't imagine you got a nice grounding last week with Cagney's presentation on neighborhood effects, so I probably don't need to give you sort of a theoretical overview. But let me start with just some sort of meta-theoretical thoughts here in terms of how I see the literature. It seems to me that there's two general perspectives that dominate the social sciences, actually, and beyond. And one starts and ends, actually, with the individual as the unit of analysis, the unit of inference, and even the unit of treatment. Dare I say, the medical model is focused heavily on the individual. And in that paradigm, individuals are thought to decide and select various outcomes autonomously, such that neighborhood context and pretty much every other social context is seen as an outgrowth of those individual selection decisions. So you can take a rational choice, although that's not the only thing. I think it's a more general approach that focuses in on individual choice. Now, to the extent that individual choice is important, that has led to a critique of a large number of studies that look at social context with the idea that there is selection bias, namely that individuals are differentially selecting into environments. I want to probe that today and show you some information on actually what selection looks like. Basically, what I argue is that this dominant perspective has it wrong. Selection bias is not a nuisance or some kind of a statistical problem, but rather it's a fundamental social process that has social property, social causation. In fact, in the end, I argue that selection is a neighborhood effect. Secondly, at the other end, you have a critique that says something like large-scale structural forces overwhelm individuals and neighborhoods. Think globalization. Think large-scale economic changes. You've heard the narrative. It doesn't matter who we pick. It's pretty much the same thing. If it's Thomas Friedman, the world is flat. If it's theorists of social phenomenon from this perspective, it's the idea of placelessness, right, that we're rendered sort of, you can be anywhere and, of course, everyone is on their iPhones walking down the street and therefore you can be anywhere than your particulars of your somewhere. It don't really matter. That critique sort of comes at it from top-down. That is structural factors overwhelm both. Now, what I'm going to argue is that the intermediate level of social organization in the contemporary city is fundamentally important. It links these different levels. One is not subservient to another. It's not above another. I'll talk about selection bias from the ground up and I'll talk about how some of our research is trying to deal with some of these structural or beyond neighborhood factors. With that little preamble, let me try to motivate a little bit why one would be concerned with some of these issues, in particular and relevant to the title of the talk. Why should one be concerned with the social reproduction of inequality or the social reproduction at the neighborhood level? Well, there's some good reasons and that is the fact that there's a deep structure, a deep continuity to neighborhood social organization, as I will show you. It's not to say the same neighborhood never changes, but what I am going to argue is that there's a fundamental process that links neighborhoods and has a continuity over time along a number of different dimensions. And by the way, theoretically, this should not be surprising at some level if we think broadly and even go back to, you know, outside our individual disciplines. I thought Lewis Mumford, some of you may have read, the great student of the city, culture and city in 1954 and other works, one said, quote, neighborhoods in some primitive, in-co-it fashion exist wherever human beings congregate, which gets at the idea that there's, in a sense, a social selection or a sense of affiliation and homophily, really, by neighborhood. And last year, actually, this year, 2010, half a century later, the archeologist who does wonderful work, Michael Smith, based on research around the world, claims that, and I quote, the spatial division of cities in districts or neighborhoods is one of the few universals of urban life from the earliest cities to the present. This bases on archeological work going back to the ancient cities, showing spatial division is a fundamental social form. All right, social form, Chicago, that's where we're at. That's where I study. I can talk about other cities. You know, you can say, well, LA's different and Bogota's different and so forth. I'll try to answer those questions, although I'm not sure we have time. So let's start with a little bit of historical perspective. Now, first thing. I don't know if people in the back can see this. Is it high enough? These are maps of Chicago, of course. And this is from a famous work, Drake and Caton's The Black Metropolis. I don't know if anyone has heard of it, or read it. It's a very thick, classic book, sort of rooted in the Chicago School of Sociology. And in the pre-war era, this was published in 1945. They were collected a bit before that. They showed that a number of different phenomenon were concentrated ecologically. So you can see, for example, insanity, tuberculosis, deaths from tuberculosis, infant deaths, clustered. In areas close to what was then known as sort of the inner business district, near west side, then south side, then further south side, sort of see a pattern running down northwest side, southwest side, relatively free of these sorts of things. This is 1945. Not just tuberculosis or insanity. Let's look at Families on Relief, Indicator of Poverty. Same thing. Illegitimate births. But also, Rob, the extraordinary demonstration of the highest and the lowest incidence of being an adjacent community. Exactly. But yet this clustering, which is an excellent point, what it shows is the differentiation by neighborhood. So that's the key thing I want you to get out of this, but also a bit where these things are located, because I'm going to show you how it's changed and how it's remained the same. Twenty years later, in the famous or infamous, depending on your perspective, Moynihan Report, he talked about the quote, I'll set aside the word pathology, because I think that's incorrect, because it gives it a medical diagnosis. I'll do respect. I think it's a social problem. But what he said is, in our view, the problem is so interrelated, one thing with another that any list of program proposals would necessarily be incomplete and would distract attention from the main point of interrelatedness. Things are going together. I just think of it as things go together ecologically. He had a very prescient point, I think, which is where we should break into the cycle and how is one of the fundamental questions facing the U.S. I think this is still the case. The Moynihan Report, despite its flaws, identified a problem, interrelatedness, and his thesis, as I argue elsewhere, really suggested that to break the cycle required a macro intervention, which basically could only come from some combination of the state and city-level policies, but I'll set that aside for a minute. So that's 1965. Okay, 50 years later, after the Drake and Caten Report, these are data that we put together. Each dot here is an incident of homicide. Yes, there's a lot of them. This is roughly in the 1993 to 1995 period. These are incidents of low birth weight babies. What you see is concentration, differentiation, but also a spread now toward the west side and a spread further south. So you see both stability and change from that earlier map, but clearly, even in this era, it's nothing like you would expect from any kind of random distribution or placelessness. Now that was the 90s, the bad old 90s. I remember when we first started this study, the homicide rate was the highest it had been in decades in Chicago. The violence was a severe problem. Let's go later and now let's look at what happened over time. These are data that I've mapped on the change in poverty over 40 years, and basically it's not really a surprise to you, but I want you just to remember this as we go through. The Black Metropolis was written basically about the neighborhoods right here north and west of Hyde Park in terms of Grand Boulevard, Oakland, what you see is the dark areas are shaded by racial composition in 1960, and then the pluses or minuses are the changes in poverty that occurred independent of a baseline. In other words, changes, unexpected changes, and you can see basically a shift moving south and moving west. So for example, areas like Austin, which were predominantly white in 1960 are now predominantly black and become poorer. Notice the great stability up here, not only maintaining a certain racial composition, but poverty levels not just not going up, but actually doing better than expected. That's poverty, so we can see kind of a stability. Racial composition I think is very interesting. I call it the asymmetry of change. There's change, but it's in a fundamental direction. This is 40 years now. What we're doing is this very simple map. It's saying if I know the racial composition of a community 1960, what's its racial composition in 2,040 years later? What we see is three fundamental patterns. One, segregation then, segregation now. No change. Segregation then, segregation now. The differences in this diagonal are all white or non-black and predominantly black. So the stability of those corners. Then we see a lot of communities running up this axis and ending up here. And this is communities that transition dramatically from white to black. So I just mentioned Austin, for example, went from basically completely transformed its population, Woodlawn and others. But what you noticed is it's asymmetric. It doesn't run in the other direction. In 40 years, and by the way, I did this for the United States as a whole. You'll see in a minute for other things. Why wouldn't it be? We think of gentrification and so on and so forth. Not one community transition from black to white. That's an important point to bear in mind. Now you're going to say, well, there's gentrification. Everything changed with regard to gentrification. Well, let's look at it. Here we are, community areas in Chicago. There's a concentrated disadvantage index which picks up poverty, unemployment, constellation of factors we're talking about. And here it is in 2000. So the decades of the 90s is often thought to be the period of intense gentrification. Well, here's what happened. If I know the level of disadvantage in 1990, I know it in 2000 with a very high degree of accuracy. A couple outliers, but not very much, that did a little bit better. I'll tell you, you'll see where some of these areas are. And things improved. You can have a secular change, but the positioning of neighborhoods, the status of neighborhoods, how people think about them perhaps is staying the same. People always say, but yeah, Chicago is unique. It's different. Yeah, it is in some respects. Not with regard to fundamental social processes. I know it looks like a little bit of a blur here, but this is every single, and I'm using here census tracts as a proxy for neighborhood right now, is every single census track in the United States. 64, 65,000. So this includes rural areas, suburbs, LA, Savannah. Pick your favorite place. It's in here. And this is a simple relationship. Same thing, concentrated poverty 90, concentrated poverty 2000. Correlation 0.89, correlation 0.88. Notice out of 65,000 tracts, not that many have upgraded. There's a great degree of stability. Okay, that's the dominant picture. It's just undeniable. All right, let's fast forward. Post 2000, bringing you up to date, new data. So what we did here is to gather all the incidents of low birth weight babies and infant mortality from the period 2000 to 2005. And what I've done is adjusted by poverty, right? And then, because I believe in trying to show things in a very straightforward but not simplistic way, low, medium, high, just caught at thirds. So low child health would be areas that have high rates of infant mortality and high rates of low birth weight babies. Defined here is less than 2,500 grams at birth. And then homicides, only now you might have thought, well, you know, those homicides, what about population size and so on and so forth, aggregated over several years, 3,200 homicides. The stars are proportional to the population. So it's essentially a risk analysis. So the larger the star, the more homicides relative to the population. And what you can see is this is, you know, Drake and Kate and all over again, right? Look at this. Areas here, here. Key community areas sometimes, as Mark noted, adjacent to areas that are both doing better in child health and have low risk of homicide. Things are going together here. Finally, I just want to do a few more to, because people say, well, okay, the economic crisis changes things. Okay. These are data now as of 2009. I don't have a census yet, but what I did is to get the index of Chicago Housing Authority relocate tees plus voucher unit holders. You have to be below a certain income to be a voucher holder. So it's an index of poverty. And this is poverty in 2000. And again, a high correlation. And you see a few communities. I'll come back to these that are suffering from increases other than what we would expect. Now, my assessment here is that Washington Park and also Englewood and West Englewood are seeing a lot of influx of people that have moved out of places like the Robert Taylor Homes. So we'll come back to the shift. But the point is there is a real gradation here that's not changing. You know that the crime rate has declined a lot. I purposely showed you 1995 data, 1994, because that was the height of the violence epidemic in Chicago. It's declined a lot, as you know. But this is the violence rate, 1995 to 2000. This is the violence rate this year. I gathered every incident of violence from January through July of 2010. So it's the latest data we can have. It's a straight line, 0.90 correlation. The rate is going down and I can't draw it. I don't want to draw on your screen. That would be horrible. But if I did, and if I drew the secular relationship, it would go like this. So it's a little hard to get your head around. But the point is that secular change is not incompatible with fundamental stability in the relative sorting or the social order of community structures. And that's what you see here. And finally, a few more. It's not just poverty. It's not just crime. I've done a paper. I don't have time to talk about it today. What I'm thinking of is other regarding or altruistic behavior and giving the medical audience. I thought you might like to see this map from the book, which actually takes a study that was done here at Chicago back in the late 80s, 90s, Nicholas Christakis and colleagues on CPR rate given to strangers on the street. That is conditional on having a heart attack. And what I did is took into account the race, the age, the sex of the victim, the racial composition of the community, other factors of the community. And essentially, it's the rate of CPR, sort of condition-adjusted CPR. Think of it as helping behavior. And then the other thing we did as part of the PhDCN was a field experiment of sorts. What we did is to essentially go around to each neighborhood and we had self-addressed stamped envelopes with fictitious return addresses, JNL financial services addressed to so-and-so with a stamp on it, varied the addressy things about it. Like one was a business, one was personal. Inside the letter it said something like, I paid that bill, please check to make sure your records are correct and so on and so forth. Drop something like 3,200 over 3,000 letters on the streets. Okay. Then took into account where they were dropped, where they were near, whether it was raining, whether it was windy, what time of day, and so on and so forth. Because in fact, it's raining like today, I came in, it was pouring, and if I saw a letter I might say I'm late for my lecture, I can't pick it up, or whatever. But the point is that there's random distributions and the idea is that, is there variability? Well, there sure is. And not only that, there's variability across types of phenomenon across, in this case, 14 years. The dark areas are areas that have high CPR giving rates, if you can think of it like that. They're more likely. And Hyde Park's doing pretty well. You should be happy. Kind of there's this corridor here. And the triangles are the letter return rates where they're the highest. And there's some anomalies here, right? You can think, well, very heterogeneous areas, for example. Or the north side, the loop. I mean, and I repeated this this summer. And loop, people are everywhere. This is Anome, right? Who gives a damn about picking up a letter when there's all these people? It turns out that that had one of the highest return rates. So something's going on. There's differentiation. There's other regarding and altruistic behavior. And this is the relationship between the 2002, and this is a subset of communities that we did, and 2010 where the correlation is 0.74. And this is the array with the loop and near north side, the highest. I don't want to focus on that too much because this is all motivational to get us into some of the analysis. Although that is analysis in the sense that it's showing you, I mean, a finding that I would basically say that neighborhood patterns cannot be explained away simply by compositional differences. And there's social quality of life issues here. So what do I want to do today? Talk about three things, although mostly one in three. I'm going to argue that, again, the selection bias is, and I put it in quotes because I think it's been misinterpreted, selection, I believe in selection. I believe in choice. But every choice we make is embedded in a context, and those contexts affect our choices and our individual choices affect the context. So we have to understand that as a social structure. I want to argue for the importance of what I'm here calling social climate or the social processes and community well-being. And I think perhaps you can relate to that to the idea that there's something fundamental about thinking of the rate of well-being in a community or a society that goes beyond the gross national product and even goes beyond the crime rate, right, in terms of trust, in terms of altruism, in terms of collective action and so forth. And thirdly, the higher order structures that link these various processes. And it's going to get a little abstract at that point, but you can tell me later how you don't like that. All right, so a little bit about the project that Mark noted. I'm just going to fly through this and I apologize, but there's really no other choice given our time constraints here in that this is a multifaceted project that involved a lot of people. We're working here in Chicago, we had 200 people working for eight years collecting the data. And it was designed in a simple way to capture the idea which is in the title, that is to study development in context, namely development in Chicago neighborhoods. Neighborhoods are the only context, just an important context to study on its own. The argument there in a paper that Steve Raudenbush and I published in 1999 was the notion of eco-metrics, that is the development of measures for the study of ecology that are independently valid that go beyond just the individual level characteristics. So here's in one slide, sort of the major components. Think of it as two studies in one. So the argument at the beginning was we're going to study individuals, we're going to take them seriously, we're going to follow them through time, but we're also going to study their context independently, not just what do the people say about them or think about them, but independently assess the context. And what we did was community surveys, independent samples of community members, social observations, we rode down the streets or the cheap jerry-keys or something like that with tinted windows with video cameras and videotape the sides of the street to be able to measure various aspects of social order, disorder, graffiti, garbage and so forth. This will come up in a minute as being important. We did field experiments like the Lost Letter Drop. We did key informant networks. I'll talk about them a little bit. We sampled organizations such as churches, businesses, Alderman, the religious sector and so forth and looked at the relationship among community organizations and their leaders. And then, of course, census data and crime data and all the other stuff I've been looking at. And then longitudinal cohort studies starting with birth 369, 12, 15 and 18. This is technically a sequential cohort or accelerated longitudinal design. As Mark noted, we're following this up, actually not just the birth cohort. We're taking all the birth cohort and then a random sample of the 369, 12 and 15 and we'll be tracking them and following them up starting next year. So I'm excited about that. This is just a distribution. You can see it's quite diverse in terms of immigration status, racial status of the 6,200 children. I should say that the study of children and their caretakers. That's important. So we had a unit being the child or subject and the caretaker adult. The sampling design was based on an initial neighborhood stratification. I don't know if you can see this, but basically these were stratified random sample of neighborhood clusters and I won't get into that, but we did cluster analysis and used maps and so forth to try to define aggregations of census tracts that were socially meaningful. And we arrayed it by racial composition of the major groups in Chicago, namely African American, European American and Latino American or Hispanic and then a category of mixed. It turned out there were seven categories that is homogeneous on these and then four categories of mixed, white, black, white, Hispanic and so forth and then low, medium and high socioeconomic status and this is the distribution. So we have a representative neighborhood sample and then within each we have a representative longitudinal cohort samples. Sorry all the details, but people care about that. People move. This is what made it so hard actually for the interviewers. 43% moved at least once. You can see that all of Chicago's covered. This is basically land, there's no homes down there as you know as you drive down toward where Daily wanted to put an airport in the early 90s and this is O'Hare where I just was. So basically everything is covered and you have movement. So the point is there's constant change. So when I say there's a certain continuity or change, what I'm really trying to do is reconcile a really significant amount of change that's going on in mobility and choice and selection and so forth with stability and change. And you can see that it goes across the United States too. Down in Mississippi, a couple of East Coast, Florida, Texas, California. Not surprising. People went back to Mexico and back. We follow them wherever they went. Now, I want to tell you a little bit about mobility. What does selection look like? So what we did, and this was a painstaking analysis, but what we really tried to do is to get into the literature and really try to figure out what are the fundamental factors that influence people's decisions to move and there's a good literature on this. Peter Rossi wrote a classic book on this. Why do families move? And there are some pretty straightforward things like income, education, job loss. But we decided to probe this a little bit more and we looked at things like, I mean, not just the usual suspects, such as race, ethnicity, but employment, occupational status, welfare relief, married, cohabitated, or not, high school diploma, and on and on, immigrant generation, English proficiency. But also, for those of you, and this is where I think the medical epidemiological literature is quite rich, there's a nice literature going back actually to Ferris and Dunham and some of the Doron's work in New York on the idea of drift or how mental vulnerability, if you will, is related to either getting stuck or moving. So we actually are able to look at this. We have, we assessed depression, for example, and mental health status in the interview both clinically and non-clinically. So we have criminality, depression, social support, the number of ties that people have in their neighborhood, or their family. So we really have a fair degree of precision on those individual characteristics. We even, in another analysis, analyze what I call the big two, right, from Wilson and Hearnstein's treatise on crime and human nature, but they extended it to a lot of social behavior that said that, look, what's really going on here is IQ. So it's not just, let's say, depression or not, but measured intelligence and impulsivity or hyperactivity. Now we can only measure this at the child level, but we did, based on some of the older cohorts and then averaging within families to get a proxy and a separate analysis of IQ and impulsivity along with these other things. What I want to show you, and partly this has been published, and then I'll go into new work, is that yes, those things matter, but not as much as you would think, and it really collapses down to about two or three things. What I'm showing you here, and I just want to make sure everyone understands, because pretty straightforward, this is analysis of movement through time of all our families and kids, and the outcome here is the neighborhood level. In other words, you live at some place at time one, eight years later you live in either in the same neighborhood or a different neighborhood, and we measure it by median income, so we can actually calculate the dollar amount. So think about a mover-stayer model. So you can stay in the same neighborhood, some people can move out, and the neighborhood can change, right? So you can have a change even though you stay. It turns out the biggest change is from when you move. Whites are red, others mainly Asians are green, black, Latino, blue. These are the lines for movers and stayers. Bottom line is these slopes are parallel. They don't converge, although the one exception being that Asian-Americans converge to whites, this is all adjusted, controlling for all those things I just noted, okay? So in other words, your decisions, your IQ or depression or income or race are really all adjusted here, and what we see is about a $10,000 gap between black and white. So if you're talking about disparities, I'm going to focus here on disparities, not just in health, but disparities at a more sort of general level. Black-white gap is $15,000. And I can just show you, if you just focus on this, if you want to change, you can see in the next, I'm going to repeat it, but you can see how the individual characteristics explain some of that gap, but they don't explain the overall pattern. So this is unadjusted, right? These are the people that are moved if we just look at their patterns of moving. And that's the second, I'm sorry, there's the second. And what you see is, this is like about $73,000, moves down to $65,000. Yes, so the individual characteristics are accounting for about an $8,000 increment, but the gap is still remaining. Now, why is that? So we endeavored to explore this more. We also looked at mobility tables in terms of upward and downward mobility, right? Because the whole idea behind the selection bias critique, and often the one you hear is, well, what about the family that perseveres in high poverty neighborhood and moves to Skarsdale or Winnetka? Well, maybe not Winnetka, but Oak Park or whatever. Well, those are good stories, but the point is they don't actually happen much. The action is on the diagonals. If you just take neighborhood median income and you look at the distribution at origin and destination, what you see, for example, if you look at low, even for whites, almost 70% remain in the same category. It's higher for blacks, about the same for Hispanics. If you look at it differently, there's still a racial difference, but in each case, it's much less than I think common wisdom would allow. If you look at downgrading, that's where you see real racial differences. So, for example, especially during a time, and what I'm going to predict based on our follow-up to see this increase during the economic recession, that is to say, whites, 1.4%, basically no one, were downgraded in terms of their neighborhood status, opposed to about 12% of blacks and Hispanics. So, a lot of the action then is on the diagonals, and I want to understand that more. So, now I want to move up to another level of argument. The idea here is relatively straightforward. If I move from Hyde Park to Lakeview, that's an individual decision, an individual choice, but that move, and this goes back to James Coleman and before, doesn't just affect me. I mean, that's how we think of it in our individualistic world, but it affects Hyde Park in the sense that I've left people behind, social ties, but I've also moved into Lakeview, so there's a new intervention, and there's a tie established. Now, if you think of it in the population level, not about me, but about the flows of people, think about it like a demographer, and Michael, I'm sorry, Douglas Massey and his colleagues have looked at this with regard to Mexican migration, international migration, where you have migration flows. So, what we're doing here is to think about neighborhood migration as a network. These are community areas in Chicago. These are the nature of moves between the neighborhoods, and they're not valued here. That's why if you value them, then there'll be a lot more lines. It's hard to see. So, what I want to do is to show you a little bit about what the structure looks like and then get to some basic results. For those of you interested in network analysis, you can tell me what time to open for discussion. About 10 minutes. I'm really going to breeze through this. This is a network map, and what it shows is neighborhoods, and the circle is in proportion to the number of neighborhoods to which the focal neighborhood is sending a tie. Now, they may seem a little foreign to you. Those who do network analysis in the audience will get it, but what it's doing is moving beyond now the idea of the nodal attributes, that is just the characteristics of the neighborhood, and now taking into account the connection between, let's say, this neighborhood and all other neighborhoods. So we're looking not only to who, which neighborhood is it sending ties, but how many. And the idea, just to motivate it, is that the analysis to come is going to take all that into account, the number, the flows, directions, and so forth, and basically ask the question, what is driving these flows beyond what we would expect on the basis of chance, and what we would expect on the basis of distance, because people are less likely to move to, let's say, California than another neighborhood in Chicago, and economic status. So what explains exchange? That's the indegree, which is all the neighborhoods it received, because we started with our sample, and then they can move anywhere, and see it fills up pretty quickly. One more map that shows you the extent, and really amazing extent to which, even in this day and age, the nature of mobility in Chicago is socially structured. So, for example, if you look at whites, and this is not census data now, these are our families moving over time, whites are clustered in a particular area, and on the southwest side, the African-American sample is very interesting. It has a number of different ways of concentration, but you notice it doesn't overlap, and I'll show you that more in a minute, and Hispanics have another area. So we know there's racial and ethnic flows. What else is going on? Well, in a recent paper with Karina Greiff, a graduate student at Harvard, and in further analysis beyond that for the book, looking at three things. One is distance, directly contrary to the idea of death of distance. Structural distance, the idea being that the more similar neighborhoods are in terms of racial composition and income, the more likely there's going to be a tie in migration. That's not terribly surprising, although once you sort of take into account the, let's say, family income, it's not immediately apparent what that pattern would be. And more interesting, I think, is the social climate. That is, are some of these mechanisms that we've been, or I've been talking about relevant in terms of explaining this? And the answer is yes. And one of the things I want to mention, just briefly, because I have to move on, is the idea of disorder. And I've worked on this concept a lot. It's both simple, seemingly, and very complex. Because there's a whole theory out there, the broken windows theory, you've all heard about it, that disorder leads to crime. Well, that may or may not be true, but I think what's been understudied is really the way that perceptions and social meanings of disorder are constructed and play out. Because there's really two things that are going on. One is the actual disorder in the neighborhood. And we actually measured this through videotapes and counted up the number of broken bottles and graffiti and what kind of graffiti and gang graffiti and people hanging out on the street and selling drugs and so forth. Mark and I could live in the same neighborhood and, you know, there's garbage piled up, but if I asked him, how much of a problem is this? It's a big problem, a little problem, and he asked me, we might not disagree, agree. And it turns out that people who live in the same neighborhood don't always agree, especially if there's racial differences. So the idea is that disorder, that is the amount or however you want to think of it, is not the same thing as the evaluation, the meaning that is given to disorder. And in our research, we show that what we believe disorder, what I think of is the collectively shared understandings of disorders, a powerful predictor of a number of things. Not just crime, but now going beyond migration. And this is a little complex, I think I'm going to go over it quickly, given the time frame, but for those of you interested, what this is is an analysis that looks at all those flows, as I mentioned before, and assesses the similarities and differences based on space, racial composition, median income, population density, crime rate, perceived disorder, friend and kinship ties, organizational participation, and the collective efficacy or the cohesion in the community. To make a long story short, the collectively perceived disorder has a significant relationship, and another analysis sometimes larger than income, and for example the neighborhood level percent black is no longer significant, it's mediated through some of these other factors. So what's happening is the social characteristics are in a way part of what I think of as cultural homophily, that is that there's sorting going on in a sort of macro demographic sense among neighborhoods. So the idea of selection bias takes on a different cast, it seems, that it's not about so much or only individual selection, because that decision is never independent or autonomous as the classic theory has it, but rather completely dependent on the structure or the system of flows. Let me show you this a little bit more. I've just argued and shown that the disorder relationship is independent of the compositional characteristics, and when we break it down it's at both ends. That is to say it's not just circulation among areas that are low in disorder to other low, it's also high to high, in other words these are all the ties or flows from a neighborhood that's high in collectively perceived disorder to another neighborhood that's high in collectively perceived disorder and what you see are basically clumps spatially distinct areas of population flows that are related to this social characteristic and notice this red I put this red circle around it, if you shift this over here, I mean it's quite remarkable that this is empty. There's just nothing going on here. And again, it's not just about race, it's not just about income, that there's something about the social quality of the neighborhoods. Now my hypothesis is, and based on further work I don't have time to get into here, is that a lot of this has to do with reputations and the way that reputations influence mobility flows. We all sort of partake in this, that such and such a neighborhood is this and others that or students live here or there's good rents there foreclosures going on there that's a disorderly neighborhood and they are mediated by organizations real estate organizations with one of our factors but it's also cultural I think so that people's perceptions of disorder and what they're comfortable with in terms of street activity are sorted in a very systematic way. So what I would argue then is that spatial distance matters a lot I'm going to set that aside structural distance, yes income and race matter a lot but social climate similarity is important to it and I should note that this holds up in our recent analysis when we do subgroup analysis so what we did for example is to say let's take household heads that are clinically depressed and those that are not and then see where they move let's take low medium and high family income and let's take family criminality which is defined here based on both self reported data of domestic violence in the home plus a criminal record official criminal record usually of a male associated with the caretaker in the house and then we looked at criminal and non-criminal flows the basic finding is that those patterns of for example income homophily and social disorder homophily are holding up which I find to be quite interesting so in a classic sense this is what we would call a contextual analysis okay I want to try to keep to your suggestion but just motivate it a little little bit more and wrap up to push the idea of collectively perceived disorder that it's not just shaping crime in fact collectively perceived disorder is related to later poverty this is in 1995 predicting the poverty rate in 2000 and you might say well okay that makes sense but we would expect there to be a correlation between the poverty rate and the actual disorder so and this is based on a paper recently published actually it's really powerful even when we take into account poverty itself that is proportion families in poverty in 1990 so the model here is right prior poverty and then poverty in 2000 then taking into account violence racial composition immigration and the rate of measured or observed disorder and what you see is that it has that has no effect but the prediction for collectively perceived disorder along with poverty not saying it's not important and along with racial composition is there so what it's saying is that poverty of a neighborhood because it's really about change it's about the trajectory or the pathway that a neighborhood is taking is being influenced by these social flows and characteristics I'm going to skip over this but I wanted to to tell you which is why I think the analysis of the flows between neighborhoods is relevant in another matter is that these are all the key informants these are not named but I won't name him or her we then coded these all by community and in fact you can think of communities as being a network so instead of migration now think of the connections that are formed when the alderman that covers Hyde Park needs to get something done in Hyde Park and goes to someone downtown or near north side or you need a development lot 37 or we're looking at zoning we're looking at budgeting and looking at how connections map on and there's a structure to this and surprisingly enough perhaps that structure based on organizational flows is significantly related to the migration flows what that tells me is that there's such a thing as and it knows a little bit of jargon perhaps but that the sorting that's why I started with the sort of individual level way of thinking about the world in part maybe a better way to think about it is structural sorting but yes you're sorting but neighborhoods are also sorting you in a way that both are going on so the idea that we are the controlling factor is a bit misleading and selection bias I think has been misunderstood and this is an attempt to understand selection to go right to it it's not a problem it's a social process and here's how it works and that's a neighborhood effect because the selection itself is based on the characteristics of the context so it makes no sense at some level if you follow the logic to control for the context to look at the individual effect because that's in fact part of the decision importance of social quality of life and recognizing stability and change this is a way to think about the three levels of that's the working title by the way because you wanted to know the book that we'll have all of this and more coming out next summer probably University of Chicago Press can be slow but we'll see the idea is that even if I grant that individual selection in a methodologically individualistic way is sort of at the center of the bottom we think about bottom up the spatial dynamics in the higher order structure some of which I've talked about have their own causal logic and have their own measurement properties that's the idea of ecometrics that's why I think methodological individualism is wrong because you can't just start here and turn this because everything else is turning and interrelated and the neighborhood effects I'm not here to say the neighborhood effects are the most important thing in the world that would be silly I don't think we really know actually partly because it is a structure is an interlocking structure of these higher order dynamics and what the book does is to go through basically one section the first section starts with the traditional neighborhood effects and looks at the neighborhood level processes and I'll tell you a little bit more about the theory for that in a second and then has a section on motion processes and mobility and choice into account and see how that works in relation to this and then finally the higher order structure and again what I mean by that is beyond the border of the neighborhood not about an indigenous characteristic of the neighborhood but rather the relationship between the neighborhood and the rest of the city and three things that I've been working with which I think are really important for health and perhaps for this audience that I don't have time to describe but that come out of this and sort of hold up based on a lot of these analysis are the collective efficacy of a community which we've written on and has to do with the nature of shared expectations for control and cohesion in a community it's altruistic character which is not necessarily the same thing but can be measured behavioral indicators like returning loss letters CPR whatever and moral and legal cynicism which I will set aside but I think is an important cultural mechanism that has to do with the corrosive nature or perceived nature of institutions and the legitimacy of for example the police and other community norms and the collective efficacy part we've written on a lot that tries to take into account the individual characteristics and selection processes the macro level processes and how both the traditional neighborhood characteristics are mediated by collective efficacy but that there's also a reciprocal relationship but if we go back to this I think and I'll just end by saying that I think it has implications as well and it's in the book I don't have time to talk about it for policy and intervention because I think it suggests that the medical model and even experiments as they're traditionally conceived are perhaps misguided because they assumed independent actors and that's incorrect they assume an equilibrium which doesn't necessarily exist and even if one intervenes at the level of the neighborhood it assumes that again will stay the same and it doesn't take into account the nature of the relationships and if we just use for example you know intervening to take down Robert Taylor Holmes or the Moving to Opportunity experiment where we move people out of poverty that's fine we move people out of any green then it looks better but those people move somewhere and the neighborhoods where they moved are part of this interlocking structure so if nothing else policy evaluation needs to evaluate in a dynamic fashion what is going on but it also again calls into question of what is the unit of analysis and what should be the unit of intervention and I end with an argument for at least an equal time if you will for more macro or public health like interventions where the intervention is not at the individual level but rather takes into account knowledge about how individuals make choices and the nature of moves but essentially intervenes to renew existing structures rather than only blow up the old ones I can say it that harshly so I think I'll end the one thing I didn't hear you say anything about which surprises me is education in this I'm sure it tracks with everything it may be what a lot of people think might be the lever the way to intervene can you speak about education measures of quality of education whether it's high school graduation rates whatever it is but I'm sure it's part of the mix yes it is although I would probably say it's not in the mix as much as it should be there's several people on the project that are deeply interested in education and are working on it and we've done several analysis for example one looking at school dropout which is a key mechanism in terms of life chances and what that analysis shows particularly worked by David Kirk and an article we have coming out on the relationship between neighborhood effects but also experiences with the criminal justice system in terms of school dropout so it is a piece of it that's been independently analyzed two other things I would say that in all this analysis it's there and it's important depending on the nature of what we're studying so for example in terms of the mobility housing was sort of in the background I just didn't have time to emphasize but education is very important what we didn't do is analyze education as an outcome with the following exception we did spend a lot of time in attention on the younger children's experiences and measured test taking and verbal and mental abilities depending on how you want to interpret the test so for example we published a paper on verbal ability among 6, 9 and 12-year-olds based on the Westler Bellevue and other tests and what we found and this has also been found now in two other studies in Chicago so it's been replicated across three studies is that living in or growing up in severely concentrated disadvantaged neighborhoods sets back or has a lagged effect on verbal ability and we estimate that to be about the equivalent of missing a year in school so it was about four to six points on the IQ scale that is when you norm the verbal ability IQ test you can then estimate the loss if you will educationally to living in severely concentrated disadvantaged neighborhoods the important part about that I think is that the effect maintained and was lagged so it was essentially a developmental effect so even if one moved out of poverty but had let's say grown up in that there was still a lingering deficit in terms of the ability so that's a powerful finding if corroborated because it suggests a number of things that one cannot just focus in on escape from poverty as a solution if in fact early childhood has been characterized by all the things that I've been talking about and that's part of the argument is that childhood and it's partly how I in the book is that we really need to I believe link child development policies with community intervention policies more holistically you know there are ways that we're doing this but I think for example some of the voucher programs are taking kids out but after they've been exposed for number of years to learning environments that are less than satisfactory and if there's lag effects then there may be difficulties in that and secondly the equilibrium that will then reassert themselves if people move and then new inequalities are created that we don't anticipate so the unintended consequences so I think some of the policies with regard to let's say the Harlem Children's Zone and some of the vouchers are somewhat problematic but that's a long way of saying yes education is important and that's how we analyze it but it's really tip of the iceberg there's a lot more one could do I have a question to maybe relate it we measure crime in a poor neighborhood but crime is not only physical crime is if I very simplistically say when you rob people from their health or financial or freedom and so when you get 10% to 15% of medical expenses in this country is crime of a stealing and this is not in a bad neighborhood sure it's done throughout the country by doctors and lawyers and anyone who could do it yes so we only measure here crime of a physical and one of the thing is that I found out that somebody throw out the purses with money in it in a various country but that in US and England and Mexico somewhere about only 30% of the people return it to a place while in Sweden and Belgium and those area was only about 2% so the attitude of the country and pro-ness to crime I suppose it make a difference whether it wants to be a crime in a physical in a bad neighborhood or would be financial in a good neighborhood so if you had a prescription to do something what would you do well first of all I think the point is an excellent one in terms of what we're talking about here with regard to crime so this study and some of those maps that I showed earlier on homicide and so forth these are indeed interpersonal crimes of violence chosen in part because of the fact that these types of violent crimes in particular seem to be based on the research so detrimental to children's health and to the trajectories of communities as an aside there's an interesting paper that recently came out in the proceedings of the National Academy of Sciences showing that witnessing violence or having homicide occur like on the street close to one's home for children had a really negative effect on learning for those kids using kind of a quasi-experimental design so I think the issue is what kinds of crimes are we talking about and what kinds of effects do they have so that's why we looked at more violent and street crimes the kinds of crimes you're talking about are fundamentally important I think they are in a sense at least conceptually incorporated not so much from the perspective of crime but in the work and in the model at this higher level which is to say that the way that communities for example are laid out the nature of segregation, planned communities income distributions social services all these things are more macro level that are not about the individuals in the communities making decisions and you could argue that it's a quote crime that services are diverted from one community into another that could be corruption or different kind of crime it may be that taxation policies are unfairly harming certain communities so I totally agree with you that I would even maybe think about it more generally in terms of the kinds of I guess the way I conceptualize is allocation policies that affect how that's why I think neighborhoods are I mean they're endogenous in a sense that's been one of the critiques that doesn't mean they're not important it's just that they are subject to these forces and allocation policies that have to do with a number of things foreclosure crisis maybe is the best example of what you're saying I mean some communities right here in Chicago are now being hit really hard with foreclosures and a lot of those decisions were I would say criminal I mean malfeasance incompetent if you don't want to say criminal right but they certainly weren't street crimes so totally agree with you I think it's captured though in a theoretical way to think about how policies and organizational decisions affect that and finally the cross country stuff is fascinating it's like the letter drop stuff you see differences sometimes it's not quite what you'd expect one of our studies here compared Chicago and Stockholm it was interesting that you mentioned Sweden I think something I have to look at that study more sometimes I think putting money in a person can found some things depending on the poverty level of the country but I think the cross national comparisons are are great and right now there's about four or five studies ongoing that have used some of these measures in Australia, London Stockholm, Africa and Tanzania, Moshi to be precise and then several other cities in the United States so we're going to have more cross cultural data shortly thank you so much my understanding is that there are much wider disparities in wealth compared to income is that relevant for this discussion yes so the question is that there seems to be more disparities in wealth right relative to income and I think that that is correct we have not looked at disparities in wealth although we have looked at disparities with regard to the neighborhood level in terms of more upper end kinds of characteristics such as percent in professional managerial occupations percent college educated and so forth which tends to be highly correlated with wealth the problem is that we don't have real good measures a lot of people don't actually of wealth also at the neighborhood level I think one can get at that in part through the median income although as you say median income doesn't capture at all although it is correlated I think it's a tall order I think it's an important one to get it how you can conceive of and measure community level differences in wealth the individual level there's much more advances that have been made on this some sociologists have tackled this problem so I can't really say actually how it would play out other than my hypothesis would be it would exacerbate these differences would be as pronounced if not perhaps more pronounced it's a good point I was wondering if you could say a little bit more about the developmental trajectories of children in the context of the development of a neighborhood and thinking both about the intersection of the context of that neighborhood and where a kid is in the life course but also in thinking about lag effects of neighborhoods so if one is in a neighborhood until age six how long does one carry that influence of the neighborhood through life so the question here is really about developmental effects of neighborhoods one way to think about this is to make the contrast stark is that you could have imagine two kinds of neighborhood effects it's oversimplification but one is more situational so that you have a particular context let's say in a community context let's say here in Hyde Park where you walk out after this to lunch or what not and something happens or there's a letter's dropped on the street or there's a potential crime or what is the nature of the effect of that context on what's happening then or even in the seminar this seminar is a context someone asked a question sort of a contemporaneous effect and then it's gone now that's a real causal effect and my belief is that a lot of the patterns that we see especially for things like street crimes that are more contemporaneous in nature but the different kind of effect might well be that individuals are especially kids don't just live in a bubble they live in a neighborhood but they're also in many neighborhoods during the day and it could well be that if one grows up in a certain context high crime environment sort of is mediated by learning perceptions interactions with others such that for example a suspicion a cynicism about interpersonal relations such that when you're in another context a perceived remark or an insult might be seem to be necessary for you to attack back because a lot of interpersonal events like that seem to be rooted in these misunderstandings and that is based on learning pattern so that's an example of where growing up and living in certain kinds of environments can have long-term effects I think that's what we showed with respect to some of the learning in terms of the verbal ability others using the phd sand data I believe have shown some developmental effects for other things in terms of delinquency sexual behavior I think it's still it's a research question that has not been fully articulated I don't think we know all the different characteristics but I do think what is useful is to make that analytic distinction and using the example of the kid who goes out in different neighborhoods and may be affected no matter where he or she is it's not incompatible that at the same time despite his or her impulsivity or nature of interpretation of relationships if you're in, let's say after school and you're in a park and there's drug use and there's no supervision then you may be more likely to engage let's say in drug use and in a way that's a neighborhood effect because it's the context that's affecting you but it's not your home residents so partly where I think the literature has gone wrong is to assume it's only about your home residents taking the neighborhood as the unit of analysis that our work does is more about that situational context effect the developmental one is more where you live and then how it affects you no matter where you go so it clarifies the counterfactual if you will Marshall, you have a last word last question so I was wondering when you speak to a lay audience what your message is in one sense one can say well this is just the old American dream story that you try to have your family live in the best neighborhood possible for education and so a lot of similar correlations and at the same time you may be saying well maybe the American dream is a myth and that you mentioned there's relatively little transition across different strata in one of your slides so in terms of the practical message for the general public what are the insights you have that you stress when you give the talk to the general public ah I don't like those no so a couple things this is not an anti American dream although what I would say is you look at those early slopes of neighborhood attainment and yes you move up but everything's relative and there's a a deep structure to all this that's not about you right and so that's an important message we don't like to hear but the positive side to that and more and more in my career is like a population health person and that's where I think the medical model not the individual medical model but more of a population health model makes a lot of sense because what it says is it's not so much about you or me or intervening in that level it's about changing the norms and the structures that allow those lines right to get to converge and that is more abstract but I think it's important because it does have implications depending on whether you're thinking about disease or I mean you can look at I found it intriguing I was writing an article on looking experiments and observational research the CDC came out with a very interesting descriptive report on longevity from 1900 to 2000 over 100 years life expectancy went up I mean this is a great success right by what 25 30 years something like huge and then they listed the quote top 10 reasons for that except for one they were all observationally based and they were about macro population health kinds of policies about smoking about automobile improvement fluoridation and so on and so forth that may seem simplistic but if you think about it it's not because when you change norms and you change macro level consequences like that they have big effects and population health that's not trying to change an individual so I think the same analogy works and it works when you start to put these different pieces together and thinking about intervening at that level so I say that look if you can do it for longevity for health certainly we can help improve things like education and so forth if not even you know individual lives and so that's why I'm actually perhaps surprisingly fairly optimistic about policies because it does logically anyway lead to a set of doable things that are I think more cost-effective than individual level and one of the main reasons is as you probably all know we can't predict very well at the individual level and part of what you see here is that structures are very stable so we do know where to intervene please join me thank you