 We're delighted that our speaker today is Kate Cagney, an Associate Professor of Health Services Research in the Department of Health Studies. Kate's research interests include neighborhood effects on health, race, and ethnic differences in access to healthcare, in long-term care, and particularly the relationship between family structures and long-term care arrangements. Kate has an MPP, a Master's in Public Policy from the University of Chicago and a PhD from Johns Hopkins. She's also the Director of the Population Research Center, an interdisciplinary research center here at the university designed to facilitate population research. I had the occasion to review the Population Research Center before coming to do this introduction, and I can only say that it reads like a who's who of the University of Chicago faculty. Anybody who is not on that faculty list is surely missing something, and I congratulated Kate on that. Some of Kate's recent papers deal with neighborhood-level cohesion and disorder, the measurement and validation in two older adult urban populations. An earlier paper, a joint effort by a number of scholars, was a vision for progress in community health partnerships. Kate tells me that our speaker next week, Rob Samson from Harvard, who of course was here when Kate arrived as a postdoc, sort of created a kind of theoretical framework from which Kate's research departed to lead to her current work. But there is a relationship between Kate's work and what Rob will be talking about next week. So it's a delight to welcome you to the Health Disparity Seminar. Thank you. Thank you, Mark. I'm very happy to be a part of the seminar series, and so I wanted to thank Mark for, including me. I also want to thank all of you for being here today. Because I've been sort of looking around the room, I keep wanting to gravitate toward people and have a conversation, so it feels like a room full of friends. Which I think is a good way for me to say that since we're all pretty comfortable with one another, and I feel I don't need to say this in this audience, but please ask questions at whatever point you choose. I feel like if I say that, then it seems, right, that that's an invitation to all of you to begin this. It's really nice for me because this work is ongoing, and although I've been doing work in neighborhood-based research for probably the last seven to eight years, this is new research. We're looking at change over time. We're looking at biological outcomes. Well, in this case, obesity for the illustration today. And I've also taken us on the road. I've spent those last years examining the city of Chicago, and now we're going to Dallas. So I'm going to show you maps of Dallas. We're going to talk a little bit about how Dallas might differ from Chicago and whether some of these neighborhood social processes are things that are generalizable across urban context. I also want to share with you one observation by my friend and colleague, Mario Small, is that much of the research that we discuss in urban sociological world, seem to such as ASA, and here at the University of Chicago, is so dependent on research that's been done in the city of Chicago. And where I feel a lot of pride in that and pride in what's really arisen from research in the Chicago School of Sociology, I think Mario's point is well taken that we might want to do a little more research in some other urban contexts and at various levels of aggregation. So with that, I will start by saying our paper is on the neighborhood context of health disparities. This is joint work with Chris Browning at Ohio State. It's been funded in part by the OSU Initiative in Population Research. And the other thing I want to say since I did foreshadow that we're going to Dallas today is that my colleague, Leanne Carina, who has spent most of her research time thinking about human genetics and had been working with the Dallas Heart Study, that's the data that we use today, came in my office one day and she was holding the questionnaire from the Dallas Heart Study and this was developed by cardiologists in Houston and she said, oh, your neighborhood questions are in this questionnaire. You know, they're interested in research on neighborhoods and I thought, okay, sure, you know, I hadn't really talked with Leanne very much about neighborhoods. I think I presented a couple of times in the department and I thought, oh, they probably have some question on whether people are satisfied with their neighborhoods. If there's ever a general question in large scale service that usually asks people, are you happy where you live, what's the quality of your neighborhood, that kind of thing. But, of course, Leanne was right and that subset of questions that exist in the Dallas Heart Study are exactly what came from the project on human development in Chicago neighborhoods which Rob Sampson felt in Urals and Steve Rodinbush developed in the early 90s. So it was really amazing for us to have those neighborhood data in a survey, in a social survey and also in a large scale research project that included biological indicators. So this is really the first cut on this. We've also looked at blood pressure. We're starting to look at CRP but sort of in the spirit of the NSHAP study which Linda directs, we're really trying to link the biological with the social and so this is a nice opportunity for us. And the third thing I'll say about these data and our pursuit here is that we were also able to get crime data on Dallas. And for any of you who have done any work with crime data, it's surprising that it's hard data to get. It's hard data to get at a particular level of specificity. And so colleagues of Chris's at Ohio State, Lori Crevo, Ruth Peterson, have put together crime data for 60 US cities, one of which is Dallas. So we were able to triangulate these data sources in such a way that we're able to look at a change in a neighborhood outcome. What we're going to talk about is crime and I'll motivate that in a moment. And a change in our example today, obesity. So for those of us in neighborhood based research as I'll talk about now, that speaks to one of our key challenges. Which is this idea that, how do we get at this notion of causality and selection? And as you can imagine, we start talking about neighborhood context. Many people have said to me, many of whom are economists will say, it's selection. That people of particular characteristics move into particular neighborhoods. And so what we're observing, what we think is an aggregate or a meso level principle that's emerging in a community is really more about those people who compose the community rather than anything that the community has impacted upon them. So that's one story, one challenge we've had here. We've had issues of measurement, how best you get at neighborhood level concepts. We've relied a lot on census data, which again aggregate information in communities on income status and on things like residential stability, ethnic heterogeneity. But we don't really have so much on some of these measures like trust and solidarity, some of these things we imagined would emerge. We're also not so sure how to test some of them. Steve Rauner-Busch really led the way in thinking about how we might consider psychometric properties as ecometric properties. And he's a very nice paper on that to help us think about using some of those same statistical principles to measure meso level properties. I've mentioned data limitations, but also I think really thinking, and probably most importantly, this idea of mechanism. So what is it that's linking neighborhood context to health? What is the bridge between structural characteristics, which I'll describe in a moment, and a particular health outcome? And so this is where we're going to emphasize these ideas of social process. So because our task today is really to think about how neighborhood context could inform research on health disparities, we have to think about what are the challenges that rest in the health disparities literature. And I would say that much of the research on health disparities focuses on individual level sorts of experiences. We think about the relationship between, for instance, a doctor and a patient. We think about discrimination that might happen in the dyad or people's perceptions that might inform the way a physician makes treatment choices. So that's one story. I think we also focus a lot on demographic based disparities. So we'll think about things like gender, like race, like age, and the extent to which they may shape people's opportunities. But I think one important principle to take away, I hope, from this conversation today is that disparities may exist not just at the individual level, but at the neighborhood level. And I think for those of you who have read William Julius Wilson's work, who is also one of the key people who really helped us think about how social disorganization theory could inform contemporary analyses of urban context in one illustration was helping us think about individual characteristics like social isolation and how it matters that a socially isolated individual is in a socially isolated community. So this idea of characteristics that are nested and the extent to which it may not be, for instance, about an individual descriptor such as race, but it may be about the racial composition of a community. It's really trying to extend some of these ideas about disparities that we have generally looked at at the individual level and think about how we might aggregate those up and consider disparities related research at the neighborhood level. So if you're all with me thus far, we're going to spend a few moments talking about the model. So as I said, this is the general model that Chris and colleagues and I have developed as an extension of collective efficacy theory. And this really comes originally from social disorganization theory in the 30s and 40s, Shawn, McKay, Burgess, and others here at the University of Chicago, where they theorized that structural features of communities like economic disadvantage, the absence of affluence, residential instability, ethnic heterogeneity led to a particular kind of outcome related to crime. So they were interested in how these things might be criminogenic. What we've tried to do in this context is take this same sort of model and apply it to a health outcome. So I'm just going to show you three quick research findings and how it led to the paper we're doing today. And then I'm going to roll into that analysis. So on the structural side of the model, so if we think about three components, structural process outcome, we can think how the absence of affluence led to understanding the difference in self-rated health between older blacks and whites. We found that in Chicago-based research. We then, moving around the model, we've had some interest in trying to understand, again, another structural feature, ethnic heterogeneity, and that resonance and ethnic enclave reduce the level of asthma prevalence. We then move around this model to think about collective efficacy. So I'm going to define this more fully in a moment, but well, how many of you have heard of the term collective efficacy and not just through me? So a little, yeah, it's closely related to the concept of social capital. But the way I think about collective efficacy and the way it distinguishes itself from social capital is it's really about the ability of the community to come together for the common good. So it's more sort of action oriented in some sense. It rests on these two components of social cohesion and informal social control. In today's analysis, we're going to focus on social cohesion. But an important way to think about the idea of collective efficacy is it's this notion that you feel like everybody has your back in your community. And it's not that your friends with your neighbors, but you trust them. And if your house were burning down, you'd feel like you could go wrap on the door and somebody would provide help, provide assistance. It's that feeling and normative orientation at level of trust that makes you feel some comfort in the community. That's what we focus on. So that's one result related to that social process piece that's linking the structural outcome or the structural piece to the health outcome. We then have done some work on the 1995 Chicago heat wave, so we're still in Chicago. And we found that neighborhood commercial decline was associated with excess mortality during the 1995 heat wave. So this is where we're starting to tease out what might be this mechanism. What are the stressors, if you will, that relate these structural and social process features to these outcomes? Because of this work, I think in the heat wave, we're really interested in this idea of drawing out what is disorder apart from commercial decline. So in the heat wave analysis, we're really interested in trying to characterize communities in a way that, I guess in some sense, would make them more or less inviting. So it's not just that older adults couldn't get outside. There might not have been anything appealing to them in that outdoor space, or any place where they could seek relief. Might not have been any place to buy a fan, or to go indoors to cool off. So it's not just that people might be afraid, but that there wasn't anything drawing them outside. That said, we're really interested, too, in trying to understand why, for instance, might fear matter. And what is it that might be closely related to fear in a community context? And so that led us to want to understand, I think, in a more detailed way, the impact that crime might have in communities. And again, sort of thinking about that as a social stressor. So we're emphasizing, as I said, that piece. But one of the other things we wanted to do, too, we want to draw out that mechanism. We also want to understand health in a more detailed way. And I think, and it'd be interesting for some of you to comment upon this in the room, one of the things that is somewhat of a frustration, one of the things I like a lot about the NSHAP study, is that it can be hard to find appropriate or more detailed health measures in a social survey that also has really rich information on networks, on community life, on lots of other aspects of social engagement. So to be able to put these two together is quite beneficial to us. But it also made us realize that we had to turn to other theories, not just to contemporary elaborations of social disorganization theory, as I mentioned, Rob's work and William Julius Wilson's work. But we also turned to the work of our colleague, Tressa Seaman and Eileen Crimmins. And this is a model that they developed, a heuristic biopsychosocial model of health outcomes, where when they originally developed this piece, it was meant to look at disparities in health and health care and to try to understand how demographic factors, socioeconomic status might lead to particular kind of health behaviors, might lead to these components of biological risk. And sort of in our conceptualization today, we're thinking about obesity, consistent with Vicky Friedman and others, sort of being one signal of biological risk and how that might lead to particular kinds of health outcomes, mortality, physical functioning, cardiovascular disease and the like. And what we'd like to do in this context particularly is we're really interested in thinking about differences by sex. I'm going to share some other information about the way that women and men react differently to crime, but there are also lots of other reasons to think that gender is an important concept to pull apart when we think about the nature of neighborhood life. But the one thing we wanted to do to this model, so what I'm trying to do is synthesize a social disorganization approach with this biopsychosocial model. So visually what I was trying to do is insert the neighborhood structural component into the Crimson Seaman model so that we see both that we can imagine neighborhood social features affecting biological risk and we can also imagine that there might be some kind of feedback mechanism. But what we noted that was missing in this model was, again, through this meso-level process, that it's not just about these individual-level characteristics, but the environment in which people are situated that might have an impact on biological risk and some of the kinds of outcomes we're interested in examining. So that's essentially where we are in the theoretical motivation. And I'm now going to talk a little bit about the literature on VMI. And as I said to you, I'm really interested in thinking about obesity as a biological risk. There are lots of reasons, I think, why obesity in particular is interesting in neighborhood context. Some of those reasons have been outlined in research by Jason Borman, by our own Virginia Chang, thinking about how some kinds of contextual factors like racial composition and economic disadvantage matter for obesity. And if we think about the short piece I shared about our research on the heat wave, you can imagine that if people are fearful going outdoors or if there's nothing appealing, drawing them down the block, then they may be more likely to gain weight. There are also some stories about the built environment. The work of Salas and Zick has showed us that in communities too where our sidewalks are difficult to traverse, if there aren't really easy ways for people to get out and about, apart from this component of fear, that it may be difficult for particularly older adults to navigate in community. So, you know, on the neighborhood social process piece, which is what we'll focus on, there's again some evidence that disorder and psychosocial hazards, the work of Burdett and Hill, again, another piece by Virginia Chang and the work of Tom Glass, indicate that if communities, apart from a built environment concept, that they might not be so inviting, but if we see a lot of litter and trash around, might see people in your neighborhood, you're not so used to seeing that that idea that you live in this disordered physical and social space might have an effect on the likelihood that one gains weight. There's also a suggestion in this work, particularly articulated by Cubbin and by Glass, that a focus on individual level treatment might be ineffective. So it's this idea, again, you know, if we think about how we are accustomed to treating obesity or to talking about weight, we really think about it as dialogue, usually between a physician and his or her patient. And so it's the idea, too, that interventions for something like obesity might be better aimed at a group level form of initiative. And so I hope we return to that point at the end of the paper. So I'm just going to reiterate these mechanisms. I really can force you out of this in the diagram, but it's really our intent here is to link theoretical approaches from neighborhood research to models of physiological dysregulation. I talked about this urban social context piece and the work of Wilson that's helped us think about racial differences in neighborhood SES disadvantage, the work of Rob and collective efficacy theory, and this idea that there is a collective capacity on the part of communities to control and to create a cohesive urban space. And then we're moving to, you know, thinking about, again, sort of these two models of combining social disorganization theory with this bio-psycho-social model of Teresa and Eileen. And we're thinking about, you know, potentially this mechanism of allostatic load, impact of adaptive physiological responses that as they know chronically exceed optimal operating ranges. And then, you know, really helping that, helping, I think, what that model can do for us is to help us think about elucidating these biological pathways and this idea that it must work through these biological precursors. So as I said, we're using obesity as one example today. We're hoping to draw out several from this data source and be interesting to see what kinds of ideas you come up with on that piece. Okay, so we've talked mostly about this larger concept of neighborhood social context and about how it could inform the bio-psycho-social model. I want to spend a little time talking about crime. And, you know, you may remember a few weeks ago, right, there were all these B&E's in Hyde Park. Do you remember this, where people were breaking into windows? Yeah, so I'm going to kind of set the stage with that. And what's interesting about a place like Hyde Park, and we'll get back to this too in a moment, is that information about crime travels pretty rapidly. And if you are, how many of you live in Hyde Park? There's a lot of Hyde Parkers here. It'd be interesting, and we could talk about this later, how you found out about that. And how you think about the transmission of information in a community that has what the data indicates and what it feels like on the street, dense social networks and how they might matter for transmission of information. Okay, so with that, there is evidence from Samson and colleagues that cohesive social environments associated with reduced crime. And there are, you know, we can believe, or we can theorize at least, you know, this influencing us in two ways. First might be health behaviors. So we've described this idea of the use of outdoor space for recreation, changes in diet, and this idea of comfort food. So there seems to be evidence in the literature that people turn to chips. Yes, someone just did. When they're feeling sad, anxious. I don't know how many of you are feeling sad or anxious, but you do at least have those. And that exposure to stressors may increase caloric and take disproportionately saturated fat carbohydrates. So there is this idea not just that people are eating more, but they're eating a different kind of food and that that different kind of food may have implications for particularly rapid weight gain, which is what we're going to care about today. Then there's also this idea of the stress response, perceived threats trigger a fight-or-flight response that can lead to higher BMI, and this is related to the release of hormone epinephrine, and then this follow-up stage results and this release of cortisol. And there's some evidence that indicates that this particular kind of process may lead to a little bit more weight gain around the middle. So, and it's an ID, too, that over time these frequent strains can lead to heightened reactivity. You know, there's been some research here at the UFC looking at the extent to which people who gain and lose weight a lot. John Colley was one of the people who actually looked at this initially, he was in the Department of Economics now at Cornell, but that there's something, it's harder on your body when you gain and lose and gain and lose. And so when I introduced the construct of crime spikes in a moment, it's really to sort of motivate this idea that there's something that can really throw off the health of the body that is not just about gaining weight, but it's the gaining and the losing. So just to kind of think about that and the construct of why we might be interested in thinking about this idea of crime spikes. And so here we are, crime spikes. Neighborhood research tends to neglect changes in the psychosocial environment, and this is one thing that Chris and I have been particularly interested in. It's not so easy to get data that tells us anything about change. The data, and Rob actually might talk a little bit about this next week, because he's been really interested in questions of mobility. But there's some evidence that neighborhoods don't change a lot, or that the kind of data we can get from census information may not indicate the level of change that's probably really in those communities. So it's important to think about how do we obtain information about change in our neighborhoods. And so it's also important to think about the health consequences of significant increases in crimes within the neighborhood. So we're really interested, again, in this idea of spikes. So we're going to look at the extent to which neighborhoods experienced an increase in crime over a one-year interval. And we want to understand how that may independently affect BMI. And as I know, the weight fluctuation is associated with higher risk of all-cause and cardiovascular mortality. And one of the things we're interested in, I mentioned this with the idea about Hyde Park. It's not completely clear what social cohesion would do in the context of a crime spike. So the way I told the story about Hyde Park is one where someone breaks in down the block and everybody on that block by 7 a.m. knows that that occurred. And then a lot of that's happening by word of mouth and by other forms of communication because people are pretty close-knit. So most of the time we think about social cohesion as something that's protective. If you live in a cohesive community, you feel, like I was saying, everybody has your back, this idea of social control in communities, social cohesion. So you would really believe that it would bring people comfort. So that's one hypothesis. But the other, so that's really sort of as we think about the buffering piece and it might reduce fear. But at the same time, you can think about information being shared and that kind of information might cause some anxiety. So there's this idea that the connectedness of communities may actually amplify a particular kind of threat like crime. So that's going to be an important piece of the way that we approach the analysis today. And this gets us back to this idea of gender and might there be a gendered response to neighborhood stressors. The fear of crime literature suggests that women exhibit heightened fear in response to local crime. There's also this idea, too, that cohesion might be anchored on women's social networks. And I shared this recently in a workshop that Melissa Gilliam put together. And there was a lot of discussion about whether are women's social networks better? Are they more prevalent? Are they more consequential? And do women share... I'm trying to think about how best to say this. Do women share more and more information in a particular kind of social interaction? So is there really a different... Are those networks different in the extent to which you might act upon the information you received in that exchange? So the work by Roundtree suggests that there is something different about women's social networks. And we're a continuing explorer literature that might suggest some of the other kinds of concepts that I just raised. Okay, so... We argued that... We're testing, you know, residents in crimes like neighborhood is associated with short-term weight gain. But we think probably and likely this effect will dissipate over time. We want to test whether social cohesion exerts a protective effect on weight gain. And it has, to some degree, determined the way that we're characterizing crime today. Is it true that people react to information about crime, about different kinds of crime in different ways? So one of the things that I found most surprising when I started to read this literature is that in general, people are more afraid of burglary than they are of homicide. And here's the reason. We'll see what you think. That there is a belief, and, you know, to some degree the data indicate this, that people are killed by people they know. And people are killed in risky sorts of situations. So there is some belief in literature to support this that people believe they can shape their lives to protect themselves from that kind of crime. I'm not somebody who would be in a setting where I might be murdered. There's also some literature that indicates that for women, when they hear about burglary, they believe that it might lead to other sorts of crimes, like, you know, sexual violence. So that's one reason why burglary is thought to be that type of crime which evokes the most fear. Yeah. I think about a neighborhood right now is where people live. And that has been generally a convention in this literature to think about where people own a home or rent an apartment or sleep at night is their sort of main base of activity and where they are engaged in what would be described as routine activities, grocery store, laundromat, all those sorts of things. One thing that we're really interested in doing actually, and I were talking about this the other day, for a grant that we're proposing, is to look at activity space. And so that's really what, if I'm understanding your question correctly, I would get it. It's not just where people live, but it's how they live out their lives. So where are they going? What are they doing? You know, there's a general belief in the literature that older adults' worlds constrict as they age, but we actually don't have any data to tell us whether that's true or not. So that's the motivation for this new study. But right now, I'm using a crude indicator of neighborhood which is where people live, but your point's well taken and is something we'd like to extend. Okay. And I just wanted to remind you that one of these hypotheses is that the effects of intra-neighborhood change on intra-individual change and weight are what we're looking at. So it's change on change. What Laney was asking about was really this idea of social networking sites. What are they? What do they mean? How do they change that idea even if we think about activity space? One of the ways in this new study we're thinking about measuring it was some kind of chip in the heel or something. You know, it's GPS device, you know, some way that we could track individuals. But how we track them in cyberspace, yeah, we... But that would be an interesting thing to consider for this other study. I haven't seen work like that that intersects that form of research with research on neighborhood context. That's a nice idea. This is over a very short term. So it wouldn't be long enough for the neighborhood to have undergone change. So we're looking at these crime spikes so really it's this idea that we look at how an individual's weight changes and then how an individual's neighborhood has changed. So we're using individual level data in that sense. So it's really just saying, you know, how is my environment changed? How is that affecting my health? So we're not looking at anything over the long term. Yeah. Related to this study, very few people move because we're looking over, you know, on average a 10-week period. What is the title for Rob's talk next week? Because I think he's actually talking about this issue of mobility. I'm not sure from that type, but that's one of the things I know he's been concentrating on. I think, you know, one of the things we've experimented with different ways to try to get at that kind of information and also, you know, look to use life history calendars and other sorts of things, not to just know where someone lives now but how often they've moved, or what neighborhood they lived in when they were a kid, what neighborhood they lived in when they were an adolescent. All these different sorts of context could matter and could intersect from a developmental standpoint. So there isn't great literature in that there's some promise in the LAFANS data that we might be able to look at that sort of question. But the tracking is very difficult. And you particularly see a lot of movement in low-income families. So that's the biggest concern when we're trying to make inference about that. All right, so I'm going to describe the data quickly. As I said, this is the 2000-2002 Dallas Heart Studies Area Probability Sample of Dallas. We have detailed SES, biomarker imaging data. We have a Visit One social survey with baseline biological data. We're looking at ages 30 to 65. We're invited back for two follow-up visits. We're looking at information from Visit One and Visit Three. We don't have weight status at Visit Two. And that's why we're looking at it that way. In large part, although we're not sure we would have had enough movement, but it might have been nice to have that data point in between. Analytic sample, we have a little over 800 women and a little over 700 men. And that Visit Three occurred within 24 weeks of Visit One. And there's an overlap with this available crime data that I already described. And to note here an oversample of African-Americans. At the census track level we aggregate measures of social cohesion from the DHS social survey data. So again, our neighborhoods here are census tracks with your points very well taken but that's how we're conceiving of it today. We're using the national neighborhood crime data that has incident-based crime data geocoded to the census track, 2000 decennial census and as I noted that's a nice opportunity for us. Our outcome is the change in BMI between Visit One and Visit Three. 4% of the Visit One sample had self-reported weight. Not surprisingly, their weight changed more on that third visit who'd have thought. And I hate to say this but I will that it changed much more for women than for men. Yeah, BMI descriptives mean BMI women 29 men 27 so they're a little heavier and I just want to note here too that BMI or drops are BMI that was reported over 6. So with list-wise deletion and other missing values we probably lost about 7% of our sample. What are our individual level variables? Age, race, ethnicity, we're interested in form-born status particularly in Dallas, marital status, years resident in the neighborhood, SES education and income are our key measures here. What else do we know on the individual level? Well, we have Visit One BMI, we have physical activity, smoking, alcohol consumption, usual source of care and then we did put in individual level measures of social support, the number of friends you feel close to relative to you feel close to just to kind of get some sense of whether somebody might have a particular orientation. As I said we use a social cohesion measure based on the work of Samson, Radovush and Earl's. Our neighbors close-knit, are they trustworthy? Do you perceive them as helpful? That final measure characterized in the same manner as it was initially in the project on human development is an empirical based residual from a three level rating scale analysis. So, okay, on neighborhood crime rates, evidence of fear of crime by type, I've really talked about that already. We look at the burglary rate by year, city of Dallas through your average rates per 100,000 and then we look at this change in burglary rate. And this is our consideration of a crime spike. There is no literature on crime spikes. This is something that we came up with. So, we are open to other ways that you think it might be useful to characterize a crime spike. We did a very simple cut this time around top 20% of neighborhoods on change in burglary rate in the year prior to visit one interview. So what we did was we took everybody so you would each have your own crime rate. So, you would carry with you your track level crime rate. And what I'm looking at over that year is I'm dividing the year prior to your visit to one interview. And I'm saying, I'm cutting that in two and I'm saying, am I seeing a spike in crime in the six months prior to the interview date from what I observed in the six months prior to that. Does that make sense? And I'll show that graphically in a second. And then on average for those, so we took the top 20% who would experience some kind of crime spike. And on average, so in that top 20%, about, you know, they experienced on average a significant increase in crime. So a pretty significant increase for those who did experience it, which is about 40% of our sample. Okay, so this is just to show you visually what we want to show you here because this is a clinic sample. So it's an area probability sample, but it's executed in a clinic manner. So we have case one, we're following that person. This is visit one and visit three. We can have variation between visit one and visit two, case three. You can see that there isn't always for all of our cases overlap in access to those crime data. So we just wanted to show you that in some instances we do miss some cases because the crime data were unavailable during that time. I'm also going to show you if there is time. I put in this little red line on the 9-11 effect. So one of the things, one of the things that has come up in conversation one of which for my health studies colleagues is this idea that we do collect data over the 9-11 period. And one of the things that we see that's most striking is that we see a lot of weight gain. So I think we're seeing a lot of weight gain nationally. We're seeing a lot of weight gain in Texas as compared to other states. So one of the things we're trying to do right now is to draw on some national data sources to see if we can validate the weight gain we observe is something that we're seeing broadly. But also because data were collected over 9-11. So some people had to visit one, then 9-11 occurred, then we had to visit three. We thought it was worth just exploring whether we see a little action there. So if there's time I'm going to show you one slide that's related to that 9-11 analysis. But this makes sense. Everybody's carrying his or her own crime spike in this period of time and interval. And I just want to illustrate here our follow-up time in visit three. So we use a multi-level linear model, individual-level covariates. We incorporate time and then person-period specific crime rates. That's at level one and then level two, our neighborhood characteristics, those structural characteristics, and then the measure of social cohesion. This is our statistical model. This is our level two, which is informing level one. And what I just want to note here is that these lengths is centered at 10 weeks. So that's how we interpret the results. And then we're going to look at differences. We're going to look at two different tables and we're going to look at differences by gender. So in the results section, I'm going to show you three maps because I love maps. I don't know how much you know about Dallas, but I thought it was worth sharing a little. So we're going to look at three maps. We're going to look at two tables and we're going to look at two figures. And then at five time I'm going to show you neighborhood level social cohesion, crime spike effects over time, and then we're going to look at that interaction term that you noted. So this is socio-economic disadvantage. The measure is scaled, but what's important to note here is that as the color gets darker, disadvantage increases. So one hint from this kind of descriptive analysis, and I think it's always a nice place to begin in neighborhood-based research, is to try to understand what we see at the neighborhood level. So these are census tracts. I believe there are 193. And you can see that there is some pretty significant clustering. So that's a economic disadvantage. That is a scale score, and that includes a number of different variables, including economic status of the community, but also a percent of persons on unemployment, percent of female-headed households, other sorts of indicators. It's a, as I said, a scale score. This is the measure of social cohesion that gets darker. Cohesion increases. So you can see that there might be some. It looks, you know, not as much sort of density as we saw in socio-economic disadvantage. One of the, I think, a really important contribution of Rob Samson's work was to show in the Chicago-based data that not all poor communities have low levels of social cohesion. And so I think it's an important thing to remember that there can be lots of ways to reduce this disadvantage, but they may be socially connected. As an extension of that kind of concept, we also note that, and this comes from the work of Mary Patillo, that you can have highly cohesive neighborhoods because you have a lot of gang activity. And so you can have different kinds of social cohesion, and your social cohesion might not be my social cohesion. There may be different ways that we view it in different ways that it operates, but it may all contribute to the form of connectedness. We don't know how that connectedness necessarily unfolds in different kinds of contexts. Yeah, I mean, that's one hypothesis one might have, is that there are ways in which these communities could be connected. I'm not sure about that. I haven't done on the ground kinds of research in Dallas to really know those neighborhoods in the same way I feel like I know the Chicago neighborhoods. We can see lots of ways in which social cohesion is the same thing always in all sorts of ways. So I just think it's important to say that it could play out in a different manner, in different sorts of contexts. And this is just to show you the three-year average crime rate where, again, as it gets darker, the crime rate increases. And we see, again, there's some overlap, closer overlap with levels of disadvantage than with levels of cohesion. Okay, this is where I'm starting to show you, I'm going to show you this is the analysis for women, then I'm going to show you the analysis for men. And then I'm going to show you three figures based on this analysis for women. So this is the crime spike in the results on the crime spike. This is Model 3 of our set of models for women. As you note, we're putting in, you know, in our first set of models, age, race, ethnicity, study period length. We then incorporate a measure of cohesion. And then we're looking in this third model, the crime spike. And we're not seeing any action from cohesion, but we do see this crime spike effect. So what this is really telling us is that it looks like, you know, people are gaining weight due to the crime spike. I mean, that's that first cut. And what we learn here is that that, so that's in a BMI point what that translates into, essentially, if you're sort of, you know, a 5'5", 130-pound woman, that's going to be the same number of rounds. I think so that's above and beyond what others are gaining. And it's going to be a little easier to see this graphically, so I'm going to show you that in a moment. So I want to show you that when we look at this interaction of the crime spike and social cohesion, we saw no action from cohesion initially, but it looks like we have an amplifying effect. That's one indication. So in communities with higher levels of social cohesion, people seem to gain weight more rapidly than those without. Okay, I'm just going to show you one more quick result here, which is just to show that, you know, this, the finding on the interaction between crime spike and study length indicates that there is some dissipation over time. So what I want to show you now, and then we're going to revisit these results when we look at them graphically, is just to show you on the men, what's interesting is that we see an effect of crime rate. It's not so powerful, but there are some indications. We see an effect of crime spike for women that's significant, but not crime rate. So, puzzle over that for a moment, and then we'll look at these two figures. So this is the crime change in BMI between visit one and visit three by a crime spike neighborhood. So the top line is where communities that experience a crime spike are individuals who are resting in communities with a crime spike, and this is the non-crime spike. One important thing to note here is that everybody's gaining weight. Enjoy that lunch. Yeah, just to add. Yeah. I hadn't really fought so carefully about presenting this during lunch, and the impact it might have on how, yeah, much people cared. So, yeah, what's interesting is that everybody's gaining weight, but we do see this pretty significant increase, of course, or difference, I should say, in those crime spike communities. So I said we centered on 10 weeks, so this is where we're seeing that .46. So this is about 2.7 pounds. So everybody's gaining about 2.5 pounds over on average, a 10-week period. And, but the people who live in crime spike neighborhoods are gaining about twice as much as those non-crime spike neighborhoods. They're gaining another 2.5, 2.5 pounds from those folks. So we controlled for that because we wondered that too. An indicator in the model that controlled for whether or not there was underreporting. So this is what led us to this 9-11 question, whether there were some, you know, the demographer and me, of course, wanted to think about important period effects. And so there might be some possibility, there's something unusual about this time period because of 9-11. But, and I'm just going to show you this quickly, I don't have a website for me, but lest you think you can't gain weight that quickly. So one, you know, you grab one extra Diet Coke, you're grabbing one extra bag of chips a night. I mean, thank you. Important distinction, regular Coke. And an extra bag of chips. You know, you could gain 15 pounds over that period of time. Now, of course, you know, when I've talked about this with some clinicians, you're not gaining at that smooth rate in this sense, but this is just a heuristic to help us think through how easy it is to pick up a few pounds. And if something changes dramatically in your environment, in your community, you could imagine that your life behaviors and patterns change in concert with that. Yeah. So we did look at those communities, so I don't have that analysis here, but we looked at the communities where the crime went down. It went down not a lot, so there's not a lot of movement to see whether maybe people lost weight because they felt more compelled to go outside. We don't observe that. They stop gaining, but they don't lose. Yeah. And this gets back to, yeah, some of the work by John Colley a little bit about the churning and, you know, the extent to which, I mean, we may know this totally that once you gain, it's hard to lose. So sort of something else that's kind of interesting to consider. Yeah. So I feel proud in saying that I just I think that grant is being uploaded today. I'm involved in a grant with Mercedes Carnathon from Northwestern and Kristen Knitzen here on neighborhood context and sleep. And that's one of the things we're going to look at is the extent to which it matters. So they're leading this effort, but I'm hoping to develop some of the neighborhood measures, but I think it's a compelling question on the extent to which just, yeah, the disruption over time and the noise and also these levels of fear that might shape the extent to which you have restful sleep could matter for lots of different outcomes, but particularly obesity. Okay. So I'm sort of showing you this trajectory and I just want to show you this last graph. Again, this is just for the women where we're looking at neighborhoods. This is the mean level of cohesion and this is high cohesion neighborhoods and you see that they seem to track but that we're seeing in those high cohesion neighborhoods people gaining more weight. Yeah. We did look at seasonality. Now one thing, and we didn't see any effect from seasonality, one thing that is so interesting though in what we learned from talking with the Dallas Heart Study participants is that our seasons aren't their seasons and so people aren't staying inside in January. They're staying inside in July and Linda with your experience in Texas maybe you can tell us if that's the case but those summer months are when people are less active from it, from sort of the kind of activity that we're considering here. But we didn't see it. Yeah, that would be really nice and that's one of our next steps is to incorporate some kind of spatial analysis so that we're borrowing information from these contiguous communities. Sean Rear sent some nice stuff on that too. So you get if you will some kind of smoothing because you could imagine if you live if you live right here you may be more influenced by what's happening in Woodlawn than what's happening in the rest of Hyde Park but your, but the information assigned to you is going to be Hyde Park specific so this is where the smoothing becomes critical. So yeah, thanks for bringing that up. Okay so we're looking at that tracking so I just want to review a couple implications and I want to show you that 9-11 graphic. So what we find, neighborhood crime spikes are associated with change in BMI but that effect seems to dissipate over time is anticipated more cohesive neighborhoods amplify the effect of the crime spike on BMI so there's some evidence that crime volatility is more important than overall crime weight for women less for men we're kind of intrigued by that finding although it's you know as I said preliminary and not so strong for the men but it's interesting and then stronger evidence of the potential causal impact of neighborhood stressors on health so we like it again you know we were so excited to find these data because there's so few possibilities in neighborhood based research in part because these things are so expensive to launch where you get change on change where you have change in individual outcome and change in neighborhood status so for us it's been kind of a dream to be able to analyze data of this kind we can maybe think about 9-11 as a national crime spike or it's you know certainly in a critical period event that might have changed the way that people perceived their communities or and may have altered certain sorts of health behaviors in kind there are few analyses that actually examine changes in fear or temperament nationally we might get a little bit out of the GSS I think I know we get some unhappiness but there aren't lots of ways that we sort of are able to track that it is interesting you know so this is just another this was a backup slide for fun but you know we made this quickly this is from the BRFSS but people are gaining a lot of weight and they're gaining it fast and they're gaining it fast in Texas and so I'm with you yeah I'm completely with you on the general point but I don't know that it's impossible and it may be that there's been a particular trend and then something like 9-11 you know had some kind of effect it may be that there's something we are you know we did test for seasonality but there may be something on the micro level there that we didn't completely draw I don't know but but we are seeing a lot of gain and I just wanted to sort of show you this to say that there were very few people who were pregnant in this analysis and we excluded them actually okay so this was really to say but we can imagine that it may be that something like 9-11 may be particularly relevant in this group and it could be we could imagine two in Texas there might be a heightened level of concern related to immigration and maybe there's some ethnographic research that indicates that pardon oh I'm sorry that there may be something associated with skin color that could have some kind of impact so we can ask whether something like 9-11 would lead us to eat more do we eat more when we're scared and do we gravitate toward comfort food again so it's the same sort of principle again this is very new but we think it's intriguing what we did was we took you know we just took the entire analysis and we looked at people who had a visit one before and a visit one after 9-11 we compared them to visit one and visit three before 9-11 because you know we have this graded kind of entry and exit into the study and then we just looked at visit one and visit three after 9-11 so more gender age as you asked about rental status health behaviors and the like we haven't extended this analysis to look specifically at race ethnicity but one thing that we have found in the initial cut is that it seems this does seem to be particularly important for the Latinos in the sample of whom are about one third of this sample so we see them gain more weight after 9-11 than their African-American or white counterparts we're intrigued by it we also know there aren't lots of data opportunities to investigate something like a 9-11 effect I don't know at all whether there's some kind of immediate connection for people I think we were conceiving it as something that was experienced at the national level in some way so not necessarily dependent on these the dyadic kinds of relationships but that it was something that we all felt and knew and experienced there were lots of discussion and changes about the way that people certainly traveled but also how we think about immigration yeah and that might have had an effect yeah that was one of our beliefs as well thank you all very much Kate thanks for having me thanks so have you