>> All right.  Welcome back, and in our tradition in not doing long introductions, I'll introduce Al and just kind of turn it over, too. 
>> Thank you.  I am going to hopefully do this without a microphone.
>> Oh, you're good.
>> I'm good.  Okay.  There's something about your voice being amplified that just is-- I don't know.  I thought I would start with something that is relaxing.  I'm going to get into quite a bit of detail, and I'll sort of play some of it by ear, but before we start, I have all of these pictures that I take on this annual backpacking trip, and I have to find some way to put them up there, so I've got to weave them into what I'm going to present.  And this is a picture after we've come down this very difficult pass looking back up where some people with horses were trying to navigate.  They ended up not being able to come down.  But this is looking back up.
>> They're still there. 
>> And actually there was a big-- this is last year there was a big snow storm.  We got stuck in this valley where we were.  They were fortunate.  They turned around and went back, but they could not come on down.  So, it's from the perspective of somebody who's already climbed over the pass and is looking back up at people who were trying to negotiate this very difficult way down.  So I'm up here as somebody who has been involved with [inaudible].  I'll tell you.  Okay.  That is in Wyoming in the Wind River range going through the Big Sandy trail head, if anybody's been in that area.  Beautiful, beautiful area.  Though this was in August, and we got really bad snow storm, and this whole area was whited out the next day, and we got stuck, and we had to stay there.  We were stuck with a few other groups of campers which was very interesting.  But that is, I think, a wonderful way to sort of begin.  I am going to talk about the measures, selection and construction and talk a bit about their use.  And I am doing this not only as the primary speaker but for my colleagues from LONGSCAN.  So this is again a joint effort.  You'll see a few at the beginning, more than you're going to see as we get into the specifics, but it just sort of gradually introduces us to what we're doing.  It's really important, I think, as you all know if you're going to be working with data and trying to make sense out of it that you understand the data set, you understand where the data have come from so that you can better use the information that you have to address the research questions that you would like to answer.  Des talked about the LONGSCAN study and how it began.  You've got, I think, a much better idea about the samples, the different sites, how it came together, some general principles about what we were trying to do, and you've got a little bit of information about some of the data that we've got, some of the information about the maltreatment that our kids were reportedly experiencing.  What I'm going to do, as I said, talk about the measures and some of the general principles.  I was on the measures committee from the very start, gosh, 19 years ago, and we made some decisions that we still scratch our heads about and say why the hell did we decide to do that?  And for what reason?  I'm not sure that we've been add to adequately answer that for ourselves.  We're certainly not going to be able to answer that for you.  But I'll do the best that I can and some of it as you get older, unfortunately, you got more to remember and you have a hard time remembering.  So lots of times there are questions MCS why did we do this and everybody, "I don't know."  But there are some things that we do know, and I will tell you about that.  What I'm going to also try to do-- I went through and looked at your proposals, and what you're interested in and based on that what I want to try to do is to go through some of those measures a little more in detail as they relate to some work that we'd done.  So some of the studies that we've done.  So, I'm trying to match-- there are some similarities and some differences so I'm going to try to sort of hit everybody's interest.  And after I talk about all of these measures and sort of overwhelm you, then Terri's going to really overwhelm you and talk about the actual data set and the structure of the data set and how you use the data set to address some of the questions that you have.  Oops.  When you press that button, it goes off.  Okay.  This is a journey that's got to start with the first step, and you got to be careful where you are placing your foot when you start.  It can be very messy, and it can be dangerous but we're going together-- we're going to try to find our way.  This is the LONGSCAN investigators trying to figure out where the hell we're going to go.  Huddle together.  This actually is in Oregon in the pacific trail area, Mt. Jefferson Wilderness.  One thing in conducting a longitudinal study that we need to recognize is that there are number of things that change, and this is the curse of any longitudinal study.  And one thing is that our interests may change and specifically if we're talking about violence.  Our interests when we first started in the late 80's and beginning of the 90's were quite different.  Actually, there were a lot of school shootings that had taken place at that time, and there was quite a bit of interest in what was going on and what might lead to individuals engaged in those kinds of behaviors.  Since then, we've changed our interests in terrorism to our current interest which is an internet abuse which is something that we are quite aware of.  But that's one thing that happens with longitudinal studies, and you don't always prepare or start off wanting to address certain issues, and you don't always start off by having adequate assessment of those particular interests.  But that's something you have to recognize.  And that is you're stuck with what you start with.  The other thing that happens is measures change.  We had to as Des made mention of, we had to make decisions about what measures we were going to use earlier on so that we could program them for data entry and for the age 12 in particular for the ACASI presentation, audio computer assisted self-interview.  And we make decisions and sometimes those decisions are based on the measures that are available at the time, and measures that are probably much better in terms of their reliability and validity may develop afterwards, and then there's a question of what do you do?  Do you change your measure or do you use the same measure?  And actually I'm going to talk about some measures.  One example, the child sexual behavior inventory, Friedrich's instrument, is an early version.  I think it's the '92 version.  So there aren't the same questions that you find on the official published version that's copyrighted that's out there.  And that can cause some problems in terms of how do you score it and what do you do with it.  The other thing that happens with longitudinal studies is that people change, and you got a little feel for that when Des said [inaudible] Mary Schneider who had been the Chicago PI and actually there was somebody before her, Patrick Curtis.  Now there's Richard Thompson.  But with the change of people involved, investigators with staff people, sometimes that collective memory then becomes a little fainter, and we have a harder time remembering some decisions that were made and things that we might have done.  Another major change that Des mentioned that I didn't put up here is that sometimes funding agencies change and from NCAN that was our initial funder to OCAN to who knows whether or not we can in the future.  But one thing that we must admit and we must be prepared for is that once we start a longitudinal study, we're stuck.  I mean, that's it.  And even if we know we've made a mistake, what can you do but move on?  I'm going to be able to go through some of this a little quicker because we've got some consistency after working together for these 19 or so years.  ^M00:10:02 We've got some slides that are common and some stories that are common.  But the measurement approach was one where we were trying to assess multiple domains at different ecological levels.  This is a model without some of the specifics if you've gone to the LONGSCAN website, you've seen this model.  The important arrow there is the maltreatment to child outcomes, and  this is what we were concerned about as well as the antecedents and those factors that might be related to how it is that parents come to abuse or neglect their kids.  And we were considering a number of other factors, domains that might influence both whether or not parents abuse or neglect their children and also the outcomes for those children.  This is another slide.  I'm sure that you've seen this picture on the LONGSCAN website.  The purpose of this was to sort of in a picture give you some feeling for what we thought we were assessing.  Early on, much more of the family environment in terms of its importance for the developing child, more of the extrafamilial environment as we were getting into preadolescence and early adolescence.  And the ages of interviews that we now have planned, though we've got some other interviews that we're hoping to conduct as our sample moves into early adulthood.  But we've got face to face interviews four, six, eight and 12 data available for you to look at those face to face interviews, 14, 16 and 18.  You've seen this before measurement methods using multiple methods, multiple informants from surveys to questionnaires to standard measures to those that we've developed.  Terri's going to talk a bit about the measures manual and some of the information that's there.  All of the measures are in those manuals, sometimes a little difficult to negotiate and to understand exactly what we've got, but Terri will help you in guiding you.  And as Des mentioned, we move from interviews through the age eight interview to the ACASI presentation at age 12.  One little story.  In coming up with the ACASI, there is an audio part so you've got somebody who reads the question and the information in their audio headphones that the 12-year-old has, and there was some concern about who should be the person to read.  And we had to go through and hear the different people and how bad their southern accents were.  They were out of North Carolina, yeah.  So we had to, all of us in places across country that don't have accents had to approve the person that was reading this, and we came up with somebody that's a little southern but not too bad.  But there were a couple that they wanted to use that just were terrible.  I couldn't understand them.  Okay.  Criteria for selecting the measures.  One is we were looking to try to tap all of the different domains, and we were looking for measures when appropriate that we might be able to then gather information from the caregiver and/or from the child, youth as they hit 12.  And trying to fit into those domains within the developmental ecological model.  When selecting specific instruments, one thing we were doing is looking at whether or not they were appropriate for the developmental age of our kids.  Another was whether or not they were relatively easy to use in terms of administration that they didn't require extensive training or require certification as an interviewer with some measures.  Actually, one of the more difficult measures we used was the my family and friends at age eight, which is a very difficult interview, but tell at age four, but tell developmental screening which is also fairly difficult.  I hope nobody's planning on using that.  That's one of the worst measures.  I'll give you some-- oops, I shouldn't say it.  Some insight on some of the measures we've tried to look at and tried to make sense out of and very little variability actually in the [inaudible], and almost all the kids come out abnormal.  It's really-- and there's some other problems with that too.  But we tried to find instruments that were relatively easy to use and didn't require extensive training of the interviewers.  Of course, that changed when we got to ACASI.  Then the criterion was a little different.  It was-- measures then that could easily be used with the ACASI system, those that they could do on a computer.  We were interested in measures that were culturally sensitive given our diverse range of participating children and families.  We wanted to identify measures that were repeatable that we could use over multiple time points.  And this is a real challenge in a longitudinal study when you're not doing a short term longitudinal study.  If you're doing a two year study, you're usually within a similar developmental range, and it's pretty easy.  When you're talking about a 20 year study from four to 18, you're passing through a number of different developmental periods.  There are different kinds of issues that you would be concerned about and different competence in terms of the kids and  what they're able to respond to.  But we were trying whenever possible to identify measures that we could use more than once.  So we would have repeated assessments.  And we, of course, were interested in measures that had some reliability and validity.  And when not, we made up our own and then addressed those issues.  Okay.  As I said, what I am going to do is to try to touch on some of your interests.  In particular, I know you're all interested as we all are in abuse and neglect and maltreatment so I'm going to focus the first part of my presentation on maltreatment and the specific measures that we've used and some examples of how we've used them in the work that we've done.  Some things that have been published as well as some things that we're currently doing.  Some of the things we're currently doing actually include data through age 14, 16 that you don't have access to, but it'll give you some idea about our thinking and how we're approaching use of some of these data.  Also going to talk about outcomes.  Everybody's interested in outcomes.  In particular, how the kids are doing.  A number of you are interested in trajectories.  This is-- gosh when we started in the late 80's, early 90's, not quite as much available in terms of doing this kind of modeling and it's really developed over the time that we've been involved and now we knew that it was important to have multiple assessments of the same measure over time and now we know why.  That's really important in trying to model some of these trajectories over time.  Some of you are interested in behavioral problems and specific problems and again, as I said, I'm going to try to identify the measures we've used that relate to some of your interests, some interest in physical health, resilience.  Talk a little bit about resilience.  And then some additional risk factors that I'll talk about-- we've got a lot to talk about, don't we?  From mental health to history of victimization.  I know a number of you are interested in that, victimization of the caregiver.  And maybe we'll get to issues of caregiver stability or placement changes, which I'm not sure we have those data specifically, but we can talk about changes in caregiver.  So, some of the things that you're interested in we'll talk about how we've addressed them, which hopefully will help you.  And well I hope make more sense then this.  
>> It makes sense to me.
>> Okay.  Please, explain.  Okay.  So, we're going to start with maltreatment.  We have been reviewing Child Protective Service records from the start.  ^M00:20:00 We're going to talk about the methods and give you some examples of publications, and we at age 12 got reports from the kids about their abuse and neglect experiences and going to talk a bit about that.  And I'm going to talk a bit about some of the work that we have been doing of late where we've applied some of the late and variable modeling approaches to try to make sense out of those data and the experiences that the kids have had.  Okay.  CPS reports.  It is based on the maltreatment classification system of Barnett, Manly and Ciccheti.  It is a system that is used to review specifically records, official records and to code those.  The methods involved first identifying a report summary, a narrative and this is where in terms of consistency across sites and consistency across any study where there may be some issues, and that is where you might get the report narrative and how thorough that might be.  So, the narratives then once they were identified, there was the allegation narrative as well as the summary narrative that included the description of the conclusions, the investigation and the conclusions from the investigation.  And the MCS then first codes the type of maltreatment, and they're five types that we were interested in.  We do have some additional types, moral, legal, education, but the frequencies were so low that really nothing you could do with them.  So, we've been focusing on the five major types.  Neglect is broken down into lack of supervision and failure to provide.  So, that's one thing identifying type.  Within type, then each mention in each report allegations were coded for their severity, and the MCS has a system that's been developed for severity codings that go from one to six, I think, though physical abuse six is death.  So, we didn't have a lot of sixes.  But typically a one to five is what you're going to see in terms of variability.  And that's for each of the types.  There is an indication of whether or not each allegation for a type was substantiated or not.  We have the specific date for the report.  For a report, more than one type could be coded.  I think it was up to six.  Thank you.  I got confirmation.  Up to six types for each report.  And we had some criteria about whether or not it was a new report.  It had to be more than two weeks, ten days.  I can't get away with anything.  Okay.  So we had some criteria for what was a new report and then what was coded.  Okay.  That's the MCS.  Now, we modified this in a couple of ways and mostly what we did was to expand it.  For physical abuse, there are general questions for whether or not there was physical abuse and then the severity.  What we did is we focused coding on the part of the body that was involved from the head and neck to the extremities to the abdominal area, but we specified the area of where the abuse had been directed.  And we had a code for severity also for that when possible.  For the emotional maltreatment, we had very specific categories that were coded, and actually that was one way that we got witnessed domestic violence was coded within that, was given a number.  So what we did is we spent a lot of time collecting more data then we'd probably been able to-- the specificity is well beyond what we've been able to use.  But we've got all that stuff if you're really interested.  And for the emotional maltreatment, how many different codes?  We had just-- 
>> There's about five and within the five, there's [inaudible]. 
>> Yeah.  It gets into all kind of detail if you want to go there.  Okay?  We have not gone there.  We've sort of lumped it together.  A couple of important things.  We do have the specific date of the report.  And we have the date of the alleged maltreatment, the incident.  So we have that also.  That's important.  Okay, that's the modified maltreatment coding system.  So really more detail, more specificity.  We did a little [inaudible] reliability study.  We trained all of the coders to criterion first and then we did a little reliability study where we took a number of narratives and at each site the coders coded them and then we looked at the range of agreement and the [inaudible] were 0.7 and above.  
^M00:25:51
[ Inaudible Audience Comment ]
^M00:25:54
So you can rely on the data.  And we have a little statement in a number of our publications where we talk about the reliability.  You probably want to look at that.  If you're going to be using this, you probably want to use that as a-- in your paper and what you're working on.  Okay.  Most of the work that had been done previously looking at maltreatment really simply looked at those who were maltreated versus those who were not.  And it was recognized from the beginning that what we needed to do is to expand our perspective, expand our definitions of maltreatment.  And this led to one of our first special series of papers that came out in child abuse and neglect in a special issue in 2005.  And what we attempted to do is to look a these CPS reports and how we might identify better ways of specifying or characterizing the reported maltreatment that these kids supposedly had experienced.  Remembering again these are reports.  They do not necessarily indicate specifically that this is what the kid experienced nor if there was not a report does it indicate that the individual had not been abused or neglected.  These are based on official CPS reports and all of the problems that go with those reports.  Remember that.  Some of the first issues that we dealt with in this series of papers were the basic taxonomy from the maltreatment classification system, how that related to simple CPS designations and related also to the National Incidence Study definitions and that's that paper that Des led, and I'll talk about that.  We also were concerned about issues of allegations and substantiations, something that Des mentioned, and I'll talk a bit more about that.  And after we dealt with these first issues, whether or not the maltreatment classification system as we modified it seemed to be a reliable way to code maltreatment.  We then had to deal with the issue should we look at allegations or substantiations?  Then we started looking at dimensions, some of the specific dimensions.  And we looked at these dimensions as they related to some behavioral outcomes.  And the outcomes that we included were the child behavior checklist, a measure that has been applied used in all assessments.  Actually in the age 10 assessment also.  The age 10 for all sites except for ours was a phone interview, and the CBCL was administered to the caregiver.  We actually did a face to face interview in San Diego and enough said.  The CBCL as you all know has three broadband scales and nine narrow band scales.  Something that we've debated over time is whether or not we want to use T scores for the broadband scales or raw scores.  As you know, it is a normed instrument, but it's normed in a crazy kind of way.  There-- it is for the-- definitely for the narrow band scales you do not use the T scores.  It is truncated.  There are no T scores below 50.  They stop at 50 and so you got to use the raw scores.  We've-- I've lost the argument, I think, in most of these discussions and we've used T scores.  But they're almost the same in the broadband scales as the T scores and the raw scores. 
>> Doing longitudinal, you really want to use the raw scores.
>> Yeah, yeah, yeah.  So, thank you.  ^M00:30:01 At age eight, one of the other outcomes we looked at was emotional functioning of the kids, and we used the trauma symptom checklist, Briere's measure, five scales from that.  At age eight, we did not have the sexual concerns scale.  Sorry.  But we have-- at age eight, we have the CSBI.  So, they're-- and why did we not-- there was some concern that IRB's might not like us asking those questions about sexual concerns of eight year olds and kids who had not experienced trauma.  So, okay.  That's the excuse.  But it's not there.  Adaptive functioning-- we had the Vineland screener and specifically adaptive behavior and socialization scale.  So these were the outcomes that we were looking at.  First, in looking at the different taxonomies, what we found and Des reported was that the MMCS and the National Incidence Study definitions were comparable, and they both differed from typical or traditional CPS designations.  CPS designations typically just with a report one type and they would usually come up with-- I guess the easiest type that they could, they felt justify investigating and/or engaging in some kind of action, removing kids or substantiating.  But the-- and then I asked and MCS was much more specific, gave us a lot more information than typical CPS reports.  So, we were feeling good about using MCS in the way we modified it.  And in terms of its relationship to outcomes, the MCS and NIS did a better job specifically with physical abuse and sexual abuse in terms of its prediction of the outcomes that we were looking at.  In terms of allegations and substantiations, what John Hussey and others who were involved in this paper first of all found was that there were differences as far as outcome between those that had a report versus those that did not have a report.  But there were no differences between those that had a report that was substantiated versus those that had a report that was not substantiated.  And that was consistent with some prior work that others have published indicating that in terms of outcome allegations and substantiations don't really seem to differentiate individuals.  We also actually in a followup study Master student was looking at the services that the kids received from Child Protective Agency and whether or not they were different based on whether or not there was a substantiated versus-- report versus just an allegation and did not find any difference.  So, not only is there no difference in terms of outcome for substantiated versus allegations, there's no difference in terms of services they receive.  There is also some work that suggests as far as recidivism or re-reporting there's no difference, and an individual that has an allegation is just as likely to have a subsequent report that is substantiated as somebody who had a report that was substantiated initially.  So, given all of that, we have made the decision, though we look at substantiations in some cases.  But for the most part, we've made the decision to look at allegations.  Okay, looking at the multidimensionality of these reports so after deciding allegations, there were three studies specifically that were looking at the dimensions of severity, type and timing.  What we-- timing or chronicity-- what we attempted to do was to utilize approaches that others had recommended and looking at these dimensions and trying to identify what seemed to make the most sense conceptually and in terms of explaining most of the variance as far as outcomes, the outcomes that I mentioned.  And I'm going to talk about the severity paper because that's the one that I read and I know more about, though, the other ones too in terms of type.  We were looking at different categorization of types based on sort of hierarchy of sexual abuse was ever mentioned, and that's the way it was coded.  If it was physical abuse, then that's the way-- well, the typical way that others have categorized on a single dimension or single type.  And we've moved beyond this, but that's something that we looked at as far as type.  Timing looked at developmental periods and when the reports occurred and whether or not there were reports across developmental period or a single report or one developmental period and then not until a developmental period.  So, intermittent reports versus continuous reports and the constructs that have been used to identify those are extent and continuity.  But that's what came from that.  In terms of severity, I'm going to give you some specifics.  We included all the kids in these three studies who had a report prior to age eight.  So-- and this is something that's important in terms of the research questions that we ask, we may take sub samples and look at them.  We don't use the entire sample, and there are other studies where we will control for certain factors by taking certain sub samples.  This was one of the examples where we only took kids who had a report.  So we were interested in whether or not we could differentiate amongst this group who had a report based on these different dimensions.  Did they give us some additional explanatory value?  Did they tell us something about the outcome of the kids?  So, we had-- as I said, the dates of the reports looking at type.  We had the five different types and the severity that was coded.  In terms of the severity example, used a longitudinal design.  This was something-- we were sort of-- I think it was in San Diego actually at one of our meetings at the Pacific Terrace downstairs where we started talking about dimensions and what we wanted to do, and we came up with this, I thought, quite ambitious sort of plan.  And I went back-- since I was in San Diego, I think that night and started playing with some of the data, and said to myself, "I'm going to put this in a hierarchical stepwise regression controlling for a long of different things, and if these differences really come up, then I would be impressed."  I mean, I wanted to, I guess, Elliot talking about reading grants for the first time and you're impressed.  So I went and found these differences, and I was really impressed.  I thought , "We have something here."  I think I still believe we have something here but not quite as much as-- not as impressed as I was initially.  But we had-- what I did was looked at the severities between birth and age four interview and the severities were from one to five in each of the three areas, physical abuse, sexual abuse, emotional abuse, lack of supervision, failure to provide.  We then had the CBCL at age four interview, and we had the [inaudible].  And we were really interested in controlling for these earlier experiences and functioning and interested in between age four interview and the age eight interview whether or not severity of the different types of maltreatment allowed us to explain more of the outcome in terms of CBCL trauma symptom checklist and Vineland, those outcome measures that we were looking at.  And okay, so birth to four for age adjustment to maltreatment between four and eight to the outcomes, and I was, as I said, impressed with the finding that later maltreatment was as a block predicting outcomes and in particular physical abuse and sexual abuse, and this was severity of physical abuse and sexual abuse predicting those specific outcomes.  There was some indication that severity of maltreatment early on between birth and age four also in the final model was predicting the outcomes.  And you can see the specific finding.  You've got the handouts.  You don't have to take notes, and you can go back and look at it.  So this is more, yeah-- make sure that you understand what is there.  ^M00:40:08 What we were looking at with severity was what's the best way to operationalize severity?  And there's the issue and we looked at a number of different ways.  One was whether or not you looked at the sum of severities for a type, whether or not you looked at the mean severity, whether or not you looked at the maximum severity, whether or not you could then look at severity across the different types, and what we came up that seemed to make the most sense and gave us the most action was the maximum severity within that particular time frame.  So looking at physical abuse.  There may be multiple reports, but looking at-- of those reports, which one in terms of severity was the maximum?  And so for that period of time, birth to four, the maximum severity for each type.  Same thing from four to eight.  What we're interested in is again we know maltreated versus nonmaltreated.  Maltreatment is not good.  Bad outcomes.  We're interested in do we learn more from then identifying and characterizing the maltreatment more specifically.  And so within the-- I would argue within the maltreatment group if we can do that then these are useful and tells us more.  If you're really interested, then just go read the paper.  But one of the things that we did is we followed this up because we-- a lot of the differences were likely due to no maltreatment of that type versus some, and we were interested in whether or not the severity actually led to differences and for the physical abuse it seemed to be operating that way.  That was really important.  But yeah, okay.  Okay.  So that's enough on the first series of papers.  We have-- and this was done in the early part of this century that came out in 2005.  Since then in the latter part-- no.  There have been some studies that have continued to look at maltreatment reports and to sort of clarify some of the issues of timing, and timing and type.  One is a study by Jonathan Koch and the rest of the LONGSCAN group, and what he was interested in doing, and this is a study that has come out in pediatrics just recently, it was looking at neglect and early neglect specifically, and that's early neglect versus sort of proximal neglect and abuse early and late.  So he ended up categorizing the types into just two types, abuse and neglect.  And looked at early was between birth and age two, and this was over ages four, six and eight looking at the aggression subscale, the narrow band scale from the CBCL.  We used raw scores.  And using a number of controls and a hierarchical general linear mixed modeling approach, GEE, basically where you are looking at over time the same subject.  We're looking at the four, the six and the eight and predicting aggression at four with the early abuse and neglect and then the proximal abuse and neglect between two and four.  And then at age six, the early neglect and abuse was zero to two and the proximal was four to six.  Okay, so that's basically conceptually what was going on.  And as I said, using this approach and with these various controls, and I'm going to talk about site as a control.  That was an issue I know in between that we were talking about, and I'll tell you some of the ways that we've thought about this and how we'd handle it in some ways.  In other ways, we haven't.  This was just as a main effects control, and almost all the studies will use it that way.  Okay, you've got again the handout.  You can look at this.  The finding just for interest was that the early neglect was the only predictor of aggression, and it was not differentially predictive across the ages, and again early abuse, later neglect and later abuse were not predicting and there was-- with early neglect, increased likelihood that the individuals with early neglect were evidencing more aggression at four, six and eight.  Another study, Melissa Merrick, one of my former graduate students looked at specifically sexual behavior problems at age eight, and what she did is break down maltreatment reports into those that occurred early between birth and age four and those that occurred between age four and age eight.  And took out all of those that had an allegation of sexual abuse.  So again, trying to see whether or not those who have not been reported for sexual abuse whether or not there are other maltreatment experiences that were predictive of sexual behavior problems, and this is from Friedrich's measure.  As I said, we used a 1992 version.  There are some items that are different then the published version.  There are ways to score it.  I don't think we have that in the data set, do we?  No.  We scored that and the different subscales, looking at the items and which ones went into the different scales.  She looked specifically at five scales, boundary problems, exhibitionism.  Those that seemed to be indicative of sexual problems.  Sexual interest and intrusiveness as well as sexual knowledge and found that both early and later physical abuse were associated with more problems sexualized behaviors.  And what was interesting is a pattern of relationships differed by gender and physical abuse was related in the girls to more boundary problems and then the boys to exhibitionism and sexual intrusive behavior.  So, again, some things that we start looking at boys and girls.  You may want to-- in a lot of the studies we may include gender as a control.  In other cases, we'll actually look at it specifically.  In some cases, we'll look at it to see whether or not there are any differences and if not, then we'll combine them.  Some cases, especially looking at sexualized behavior, you're going to be interested in differences between boys and girls and really what we're talking about is some moderation here.  Okay.  I told you we're getting there.  We're going to take a break in just a little bit.  But this is keeping our eye on sort of the target and the prize.  Okay, and recognizing that there may be different perspectives.  This is really convoluted the way I'm [inaudible].  And there is that perspective and that perspective.  We were taking pictures of one another.  And in terms of perspective, moving beyond CPS reports, and this is to report specifically from the youth.  We do have some data from some other measures that you might use such as the CTS, which I'll mention maybe later.  We have lots of stuff on the CTS, but whether or not that indicates abuse, physical abuse is a question.  We do have in the CSBI there are questions specifically about whether or not the caregiver believes the child has been sexually abused, whether or not they've seen a doctor about suspected abuse, whether or not they've been investigated, whether or not it's been substantiated.  So there's some questions on the CSBI, and if you looked at CSBI, there's questions also about whether or not the child has observed adults in the family walking around without their clothes on, if they've seen pornography on TV and so there's some other questions in there also in terms of exposure.  Okay.  But we're now moving to self-report, age 12 self-report.  And at age 12, we developed a number of measures that are administered through the audio-CASI system, and they include screener items, very specific items as an example for physical abuse.  ^M00:50:13 Start with a question.  Has-- and it starts with if you've gone through the measure or something about kids growing up will sometimes-- those that are responsible for their care might do things that-- I have some wording, sort of preparing them for this is not something that is unusual, but it's not something everybody experiences and just tell us whether or not any of this has happened to you.  And we start off with has it ever happened?  So these are the screener items.  And for physical abuse, there are 18 items.  And they will start with things like has anyone ever hit you?  Like with a hand, I guess, corporal punishment.  And then there's has anybody ever hit you with something more dangerous like a bat or shovel?  Has anybody ever kicked or shoved you or punched you?  Has anybody bitten you?  Tried to-- okay.  So, these are all very specific items that are asked.  There are 12 screener items for sexual abuse that go from anybody ever show you pictures?  To anyone ever touch you in a sexual way?  To penetration.  Twelve specific items.  With those items, if the youth endorses that item, it goes to a number of followup questions, and those followup questions will ask about when did this occur?  And then we will ask did it occur before you started school?  Did it occur since you started school?  So, 12-year-olds we were expecting, prior to age five was before school.  School was once they started kindergarten to current.  And then there was a question in the last year.  So we got ever, before school, since school, in the last year.  Okay.  For the psychological abuse, there are 26 items, and again it is-- I'm sorry.  Following endorsement besides when we had who the perpetrator was, all the perpetrators, but that's ever.  It's not specifically.  The number of times that it occurred at each of those intervals, and there's a general question about the impact overall given these experiences of how much you think it's effected you or impacted the way you feel.  And then there's a question of about attribution.  Overall, how much-- what degree was it your fault or did you ask for it?  So there's some question about that, but that's for all incidents of the report here.  We really haven't done much with that.  We really just looked at the reports.  For the psychological abuse, there are some follow up to the items, but not as much in-- I'm trying to think the number.  I don't think-- do we have the number?
>> For psychological abuse?  I don't think--
>> I think there's not as much follow up, but it's the same.  There are some follow up, but not as extensive as with the sexual abuse and the physical abuse.  But there are 26 items, questions about caregiver, blaming them for something that they did, humiliating them in front of others, teasing them, kept them at home for school for no good reason, kept them at home to take care of them or something like that.  But there are 26 items that we identified as capturing-- potentially capturing psychological maltreatment.  And a lot of these-- the screener items came initially from the-- impetus for most of those came from the maltreatment classification system and how they classify them.  So that's sort of where this developed.  Okay.  Mark Everson and others led actually Mark and Liz Knight [assumed spelling] were the ones that developed these measures, did some piloting with these measures in an inpatient facility on the campus at UNC with kids who had reported histories.  We spent as Des mentioned quite a bit of time not only talking about should we do this, then we developed it and then there were questions about whether or not we should actually ask the kids across all the sites and whether or not there were issues that we needed to deal with human subject issues as well as individual health issues and well being issues.  So we did with kids who had a history of sexual abuse did do a pilot.  We ended up administering this, and you guys have the data available, the self-reports of these different kinds of abuse.  And in the first study talking about this, Mark and others identified kids from two sites, and these were the two sites, Baltimore and North Carolina.  No.  And these were the two sites that had kids who were not selected.  North Carolina had part of their samples selected because there was a maltreatment report.  Baltimore did not.  So, a lot of the kids who had not been reported from those two sites.  And looked at again the outcomes, CBCL that we had, the YSR, the youth self report companion to the caregiver report and the trauma symptom checklist, and this was at age 12.  So you can see we have the trauma symptom checklist at eight and at 12.  And don't get Terri started.  On the trauma symptom checklist, there's some strange things going on with that measure at age 12 and still trying to figure out because the kids in terms of reports it's like almost nothing from where they were at age eight.  And trying to figure it out.  We don't know yet, do we?  What's important now is we asked all of these questions, the screeners, the 12, 18 and 26 for this study whether or not a child reported that they had been sexually abused was determined by whether or not they endorsed any of the 11 items that we had identified.  There was a group.  We came to concensus and we identified those items that we thought would be considered by CPS as evidence of that type of abuse.  So that includes sexual, physical and 18 psychological, so you see 18 of the 26 psychological items.  15 of the 18 physical items.  One of the important points that I'm trying to make here is that when we say the kids are reporting that they were sexually abused, physically abused or psychologically abused it's a yes, no, though a little later I'm going to talk about an approach that we have used where we've actually looked at the number of endorsements.  A little different approach.  But when we start comparing CPS to self-report, remember we're talking about CPS report not-- they don't know everything and it's a report based on what the-- as opposed to what the kids actually experience.  What the kids are telling us we've identified what we believe is indicative of the different kind of abuse and then give it a yes, no.  So, one possible explanation for the lack of concordance.  Of course, some of the CPS reports occurred when the kids were younger.  Questions about whether or not they remember and whether or not they really could conceptualize what had happened to them in terms of these specific questions.  But from there, go ahead.  Do what you want, but just keep that in mind.  Okay.  And as Des showed you, the overall agreement was pretty good, but most of it was because of the agreement between the CPS and self-reports that nothing had occurred, and when there was more that occurred then there was less agreement.  Okay, probably enough on that.  Okay, some of the current stuff that I promised some of you that I would talk about and the way that we are looking at these dimensions.  One thing that I might mention in using CPS reports it is possible to look at two year intervals of whether or not there was a report for a particular type and if there was a report what the-- or multiple reports-- what the maximum severity for each type might be.  So, two year intervals starting at birth through age 12 for you guys.  It is either based on chronological age so actually every two years in the kid's life or it's based on the age of the interview.  Depending on what you are interested in and the questions that you're asking and some of the studies we were interested in maltreatment that occurred especially when we were controlling for early performance we were interested in maltreatment that occurred after that measure of early performance up until the next measure.  And so, and this is called the SD database now?  
>> The [inaudible] SD.
>> Yes.
>> We'll just say MSD for short.
>> MSD.  That's because I'll take credit for that.  That's the SD because we said how about-- if there is a missing data point in terms of missing interview then it's based on the set point is the chronological age.  ^M01:00:54 But there-- okay, two different databases then.  One based just on chronological age, one based on the age of the interview.  Again, since we have the date, we can do anything.  You could go back and do it yourself, but we've already got it in the data.  So, make it a little easier for you.  We have moved from an approach where we've been looking at dimensions and how we might locate people along a dimension to looking a person centered approach and looking at mixture modeling, in particular, finite mixture modeling approaches specifically latent class, latent profile approaches, growth modeling.  And what I'm going to do is to give you some examples of how we've used this using both self-reports and using CPS reports and then when we've tried to put them together.  Aha.  Okay.  Some things that you might want to take a little bit further.  So this is some current stuff.  The general latent modeling approach is looking at individuals.  This first paper is based on self-reports and paper that's in press coming out in child abuse and neglect and actually is the focus of a special section that we're starting in child abuse and neglect that is to identify promising methodologies that can advance the field, something that I agreed to be senior associate editor only if I could do something that made a difference, and this was one thing I talked to David Wolf about, and so we got this special section that's-- and all of you are invited.  If you've got some things that might methodological advances, issues related to assessment, things that really have not been applied routinely to research in this area.  But this is one of the featured papers that is-- have come out in the initial special section.  Kate Nooner is the lead author on this.  And what she did is took kids who had a self-report at age 12, kids that had the sexual abuse and physical abuse items, she-- and this is in contrast to approaches where we're doing a yes-no of whether or not the kids reported physical or sexual abuse.  This was when where we looked at all the items, the 18 physical abuse items and the 12 sexual abuse items.  We didn't eliminate any of them, included all of them.  And based on-- okay analysis and identified four classes of kids based on their responses to these 30 screening items, 12 sexual, 18 physical.  One was a no abuse group, a group that was high physical.  Some sexual, primarily looking at dirty pictures.  So low sexual.  Moderate physical and sexual and then a high physical and sexual.  These were the four groups.  And these groups were moderately related to CPS reports of abuse looked at some other factors, some other demographic factors also.  This is in your handout.  I don't know whether or not you're going to be able to make it out specifically, but this is the probabilities of responding, the proportion of kids for each response and the different groups.  And by looking at the pattern, you can get some indication of how that group is responding to the specific items, and that's the way in which you then can label or identify the pattern that the group is reporting.  And it's hard to read this, but refer you to your handout, and you can see with the less dangerous object, all of them were about 0.3 to almost 0.95 were indicating that they had been hit with a less dangerous object, but there are different proportions along that.  When we look at pushed, you see there's two groups that are reporting quite a bit of having been pushed and two groups that aren't.  Then we've got shot, but I don't have many of those.  Physically hurt, bruises and then broken bones and then some of the-- okay, here are the sexual ones.  Touch, touch you.  You made-- made you touched them.  Mouth on your private parks.  Okay.  But you can look at the pattern, and that was the way in which we could define or identify the different groups that had similar patterns. 
>> Has this been published yet?
>> No.  It's in press.
>> Okay.
>> Okay.  Another approach that is being used looking at CPS reports-- that was the self-reports.  But using a similar approach and using actually growth modeling is a study that is under review.  Debra Jones who's at UNC and has a career development award from CDC led this paper, and she was looking at trajectories of different types of abuse including neglect and witnessed violence.  And she was particularly interested in HIV risk behaviors.  She's interested in developing interventions to prevent HIV and looking at two of the outcomes in this case that are risk factors for HIV, substance use and sexual activity and are 14 year olds.  14 year olds?  Yeah.  She in terms of measures was looking at maltreatment in two year intervals yes or no for a given type.  She was looking at witnessed violence that comes from the life event scale which is adapted from Coddington's scale, life event scale.  A lot of you I'm sure have looked at that.  There are all kinds of questions there that you're probably interested in.  Caregivers respond to what has happened in the last year in the child's life.  This is another measure that we have at each time point and actually have at each of the phone contact interviews also.  So, every year we have data about events in the child's life, and they go from in this case witnessing violence to changes in the family, divorces, separations, marriages, a new child.  And actually I'm going to talk a little bit more about the life event scale and some of the things we've done with that. But here she was looking specifically at witnessed violence.  And the risky behaviors came from the-- you don't have this, the [inaudible] age 14, the diagnostic interview scale and whether or not they were engaging in sexual activity and alcohol us.  Okay.  Looking at longitudinal patterns, she came up with trajectories for each type.  This was again part of her case.  She was working with Daniel Nagin [assumed spelling] from Pittsburgh using [inaudible] and basically came up with two groups for sexual abuse, two groups for physical abuse and so it's interesting sort of suggests that even looking at the patterns over time just a yes, no.  But sort of the pattern was of interest and then the physical abuse remitting over time and then with emotional abuse, neglect and witnessed violence, three groups, you've got the handouts.  You can look at those in more detail.  Just to again give you some idea of what can result from approaching it this way, and this is the outcome of the relationship between the different classes, the different trajectory groups and the outcomes of interest.  Either the S pluses mean that the particular patterns differentiated whether or not kids had ever engaged in sexual activity or in use of substances and then whether or not the combination of the two, the patterns differentiated kids having engaged in both sex and in substance use.  ^M01:10:42 And as I said, you can look at that in more detail at your leisure.  Basically sexual abuse predicted sexual activity and drug use both individually and combined.  She was interested in whether or not there was any reason to look at other types of or other patterns of maltreatment after we had already accounted for the impact of sexual abuse.  The expectation being that sexual abuse was clearly related to these outcomes and the question was one of whether or not that accounted for all the variants and in fact it did not.  There were some evidence that physical abuse and emotional abuse also did predict those outcomes after accounting for sexual abuse.  Okay.  That was with all of the sample.  LONGSCAN sample looking at the maltreatment over time.  Laura Proctor who is early career development recipient from NIDA who's going to be here later.  Also if people want to talk to her, you'll have an opportunity.  She looked specifically-- and this is a paper that she-- poster that she presented SRCD and is now actually we're going to talk about this study in a little more detail.  There's some other things we want to do, but just give you this part of it which was an example of looking at trajectories and a little different question here.  The question here was one of what happens to kids who have already been reported?  And are there different trajectories for the kids?  Another way of looking at this is if we're looking at individual events, we could do survival analysis in terms of-- and this is typically what's done.  But here what she did is took the San Diego and Seattle sites.  All of the kids had been selected because of maltreatment, the maltreatment report.  So looking at then subsequent reports for maltreatment zero-- excuse me, not zero, from four to six, six to eight and through 14, and she modeled trajectories of the different patterns and unfortunately this doesn't come out.  But there is a group that doesn't have any reports.  It's a group that has a high level though they go down 10 to 12.  There's one that seems to increase over time, one that starts high and levels off.  And those are the labels for the groups, and the interest then is in what factors might be predictive of those different trajectories, those different classes.  And look at a number of possible predictors including the type of placement or caregiver.  Those that are interested in foster care.  Whenever you ask one of us about doing something with foster care, we always hesitate, don't we?  And that's because it's not that clear when we talk about foster care.  And if you think we're not clear, I'll tell you that CPS at San Diego, all of our kids were foster care sample, foster sample cohort.  They all had been removed and in substitute care for at least five months.  There are some of the kids that never saw their biological parents again.  There are some that have been adopted, actually some interesting things with the adopted kids that were sort of serendipitously finding.  There are-- some kids have gone back to their bio parents and then back into the system.  There are some that have been with relatives, some that have been with nonrelatives.  We don't-- we initially assume that those that were with kin or nonkin were in foster care.  But we since learned that the system didn't know for sure if they were in foster care or not or guardianship or what the situation was.  And so we have stayed away from talking about foster care.  And instead, what we do is we talk about the living situation, placement situation, who the caregiver is.  So, what we will talk about is the kids living with the bio parent, living with an adoptive parent, primary caregiver with kin or with nonkin.  And you can get that information from who the respondent is.  You can get that information from the household composition measure.  Now, we actually in San Diego have additional data sets where we identify whether or not it's the same-- the kids are with the same caregiver over time.  And you can get that from the data that are available at the cross site data.  But that's something that we've come up with.  You got to be careful.  You got to be careful.  One of the things that we've done in San Diego is we've got access to a lot of additional data, and we-- looking at placement changes which people are always interested in-- we have official records, and we've got some data on the number of changes that the kids have had and what the official placement was.  But even that, as I said, when we went back-- as I was mentioning to somebody, we were told when the kids were about eight that we had to go and get informed consent from all of the biological parents if their parental rights have not been terminated.  And these are kids that we have been following for four years already.  And we said, "Aw, you got to be kidding me."  And said, "Well, make good faith effort at it."  We started looking at their records and where the kids were supposed to be officially.  And found out that their records varied.  Where the kids were supposed to be, where they actually were was not always the same.  And we stopped-- because initially we started making distinctions between foster care guardianship and then we said, "That doesn't make any sense at all."  Given they don't know and so we just went to labeling the caregiving environment, the caregiving situation.  You're going to be a lot happier if you-- and even that's not the easiest thing.  But you will be a lot more satisfied with what you do if that's the way you approach it.  Okay.  In addition to looking at the placement, we looked at caregiver characteristics such as alcohol abuse and the cage we have at age four.  The caregiver responds to that, and you get an indication of whether or not they are having problems with alcohol use.  The depression, the CESD that we have at four, six and 12.  Why, and we got the brief symptom inventory at eight.  Go figure.  I think what--
^M01:18:09
[ Inaudible Audience Comment ]
^M01:18:12
At the time.  Exactly.  Exactly.  And now-- and I think some of-- what we were thinking and crazy us is that the CSD only captured depression, and the BSI captured a broader range of symptoms and so we said-- and depression was included, but it's caused some problems.  But there you go.  That's okay.  And then-- okay, oops.  Okay, yeah, ethnicity.  And we also looked at type of maltreatment, the birth to four in this sample.  What type of maltreatment had they been reported for and whether or not that predicted the trajectory group that they were going to be in.  And okay.  Summary of the findings.  I'll let you read that.  I hope that is clear. If they were living with the biological parent, they were more likely to be in the higher report group.  If they were living with the depressed caregiver, caregiver had an alcohol problem and if it was a caregiver who identified as African American or multiethnic or other-- we-- I don't know if you looked at the ethnicity, race ethnicity variable which is another story.  Okay, if anybody wants to talk about that, we can talk about it.  I'll save that.  If you've got questions, maybe the panel, we can-- or let everybody else talk about that.  We have this crazy group of multiethnic that tells us absolutely nothing.  
>> Would those-- the predictor measures--
>> Yeah.
>> Would those measure longitudinally or were those measured at--?
>> They were at four, age four, at age four.  Yeah.  ^M01:20:06 Okay.  So--
>> Have these been published or is this--?
>> That is just being prepared.  Yeah.  And there's some other-- one of the-- we're going to talk about it at our meeting while we're here.  And one of the things we're going to do is look at another analysis, typical survival analysis because the issue is one of re-reporting.  We were looking at really trajectories of reporting over time, which probably gives you more information, but we're going to try it the other way and see what happens and that's our thinking right now.  So you guys are getting some of the current cutting edge stuff that we're just now talking about, a couple of papers where we've combined the CPS and the self-reports.  And one is Maureen Black paper that's coming on pediatrics where she's looked at sexual activity at 14 and 16, but what she's done is used CPS reports and self-reports for each type.  Again, defining from the self-reports those items, the 11 sexual abuse items, the 15 physical abuse, did they endorse those, yes or no, and whether or not there was a report from CPS and identified a kid as having experience maltreatment of that particular type if they had either CPS report or a self-report.  And looking at sexual activity and she'll be here too I guess if people want to hear more about it.  She can tell you about that.  Another paper in preparation is, okay-- oh yeah, this one.  This is the one I'm responsible for, and I've been a little slow on this one.  Looking at latent profile analysis, and this is when combining the self-reports and the CPS reports, taking every two years report for each type and actually taking maximum severity for each type over two year intervals, and taking the self-report at age 12, the number of endorsed items for sexual abuse as an example.  So of the 12 items, how many did the kid endorse?  Ever experiencing?  Then, sexual abuse maximum severity over each two year interval, doing the same thing for physical abuse, maximum severity two year intervals, total number endorsed of the 18 items.  And came up with classes based on these data and I'm not going to give you the specifics of what we found, but three, four class solutions that came out and found that they were predictive of outcomes at age 12 after accounting for a number of other life stressors that were accounted for actually from the life event scale as time varying covariance, a fairly complicated analysis but again another way to look at the maltreatment data.  I think most people working in the area say that official reports don't tell you everything.  Kids don't tell you everything.  Parents won't tell you everything.  Maybe what you need to do is to get information from everybody, but how do you use that information?  The simple way was anybody says yes, the other is looking at trajectories.  And actually some of the trajectories are very interesting where you see over time what's happening as far as maximum severity and how that relates to the self-report, the kids and how they're endorsing and some real interesting sort of fine kind of discriminations between groups.  Okay.  Never buying into this guy, but everybody always asks the question.  Okay.  That was the maltreatment and got a lot of other things to go through.  Actually, this is the-- got to move beyond maltreatment.  We found that our kids have not only experienced abuse and neglect, but all kinds of other life adversities, other kinds of stressors and this is just a reminder of context.  These were words of wisdom that actually I had-- at my daughter's wedding, before the first dance, I sort of went through some of this.  And thought it was really sort of touching until I saw another interpretation of this when you don't keep the order straight or the context straight because it could just as easily be this.  So-- my daughter's the one that found this one.  So, okay.  That's to introduce our talking about other risk factors and at various levels, at the various domains.  We've attempted to assess what the kids might have been exposed to.  And some of what we have done has responded in part to the ACES study, Vince Felitti, and his colleagues who have looked at early adversities, and they've identified a number of early adversities and their relationship to subsequent health and mental health.  But these are the initial seven adversities that they identified and as you can see, physical abuse, emotional abuse, sexual abuse are included.  And then you've got substance use of the caregiver, involvement with the criminal justice system, incarceration, mental health problems and then witnessed or domestic violence, and there's some others that they've added including neglect and some others.  But basically what they find is as you increase the number of stressors, you find more problems and just this an example of attempted suicide.  They had a population in HMO actually in San Diego.  I think they identified it.  They didn't call it the southwestern site, San Diego, Kaiser Permanente.  And they had people that responded to interviews.  They got information about-- from adults about their early experiences, the adverse experiences.  They had great outcome data on medical records.  They had stuff on problems-- medical problems people were having, cardiovascular problems, visits, how many visits.  Gosh, all the medical data they had.  The problem was retrospective reports.  And we had an opportunity with the LONGSCAN data to look at this prospectively though with younger kids so also to look at the general health of the kids when they were younger.  And Emalee Flaherty from the Chicago site pediatrician has led a couple of papers that have come out.  One was just looking at the ACES with our data of the kids at age six and the outcome was a general health.  There's a question I think at each interview that the parents or caregivers were asked about compared to most kids how would you say your child is doing in terms of their physical health, excellent, good, fair, poor?  And questions about whether or not they've had a serious illness that required medical care, or those kinds of questions.  And those were the outcomes.  Wasn't quite as nice as the ACES study, but some indication that the-- without any stressors, the kids were doing better and some other fine points about that.  But that's-- you can read that one if you want to look at that in more detail.  And a more recent study looking at kids' health at age 12 and looked at early adversities and later adversities.  So what was occurring between birth and six?  And what occurred between six and 12?  And how that was related to health outcomes.  And in terms of adversities, looking at sexual, physical, emotional abuse and neglect that was reported by CPS, yes, no.  In terms of substance use, we had cage for the early period at age four.  We have a report of caregiver substance use at age eight and age 12.  So we use that for the later, that adversity to identify it.  For depression, as we mentioned, CESD at four and six and then for later, the BSI at eight and the CESD of 12.  So, yes, no for each adversity within each of those periods and then get a count.  Witnessed violence is from the conflict tactic scale to adult.  And that was at six, eight and at 12.  And family member incarcerated from the life events scale.  There's question about whether or not a family member has been arrested in jail, and so that was a yes, no within those two time periods.  ^M01:30:02 And the outcome was a composite of general health, illness requiring a doctor and semantic complaints that are on the CBCL and the YSR.  So, some little more data to work with as you see-- as we're starting to work with the first study was just looking at six year old's health, look at 12-year-old, a lot more information about the adversities.  We could look at the timing of those adversities and we could look at in terms of health outcomes a little more extensive outcomes.  Okay, and those are the results.  Five or more adversities, especially those that were later seemed to increase the likelihood that there was a physical problem.  I'm going to do less talking about the results.  You guys can look at this and about the measures that you might be interested in, sort of how we handle them.  I figured talking about the results would put into context but probably too much context given the number of measures we've got.  This was a study where we controlled for who the kids were living with.  We only selected kids that were living with biological moms through age six, so at age four and age six, they were living with their biological mom.  This is something again given the research question you have, you may select certain sub samples to look at.  And this is one way to control for factors that you can't control for, you can't make sense out of.  And we actually looked at site as a control, and looked at differences across site on the outcome, and the outcome aggression.  And found that Seattle and San Diego were very similar.  Actually, we looked at aggression and anxious, depressed narrow band scales, but I'm not going to-- but for aggression, there were similarities between certain sites and what we did is we combined them as far as accounting for them.  We also looked at potential interactions with the predictors with the sites to see whether or not there was moderation with the sites and we needed to account for those.  And in some, and actually the analysis that we're doing with the combination of self-reports and CPS reports and looking at the classes, we found some site interactions and site differences, which are of interest in and of themselves.  So, site we don't always look at as a problem that we got to deal with as much as it might tell us some interesting things.  In this, we did look at again potential for moderation across sites and did not find it so just included the control for site and it was a dummy control where we combined sites based on if they were similar or not as far as the outcome.  And-- so again, we used the conflict tactic scale, and this time it was apparent child, so we were interested in the exposure of the kids to various discipline approaches, and in the early years we didn't ask the severe-- did we ask the severe?  I can't remember later.  But you got to be careful.  We didn't ask all of the items from the conflict tactic scale.  The severe items I know early on were not asked, and so what we did is we had the minor physical and the psychological or the threatening, the verbal aggression and from the life event scale, we had the parents or caregivers report-- actually these were parents-- they were all biological moms-- reporting on whether or not the kids had been exposed to physical violence in the home involving family members or to arguing and verbal violence.  We also from the kids had the things I've seen and heard at age six, John Richter's scale, and there are two items in particular.  One asking the kids if they'd ever seen grownups in the home hit one another or if they've ever had loud long arguments.  A lot of kids endorse that.  That-- those items are asking if the kids have ever experienced this, life events as in the past year from the caregiver so you got some different timeframes, something to keep in mind.  Pay attention.  Okay.  And looked again at aggression, controlling for aggression at age four and we had all biomoms, so didn't have to worry about that and controlled for site, and then again you can see the results of the kids experience of both verbal and physical violence was related to subsequent aggression after controlling for prior levels of aggression.  And only a combination of parent and child reports for witnessed violence was related to aggression.  Okay.  Terri actually is working on a study now at age 12 looking at this gives you some I guess the idea of the possibilities with additional data.  We got reports from the kids and the parents about witnessed violence and looking at how those are related to the outcomes, and the outcomes reported by the parents and the kids and there are ways to control for that and you can ask her.  I'm sure she'll tell you more about that.  Okay.  This is another study that what I want to talk about is the life event scale, and the-- okay.  Loneliness and dissatisfaction scale, some of you might be Asher scale that we have at six and I think that's it, don't we?  We did it at San Diego at ten.  Okay.  The life event scale.  One of the things with all of those items that we've tried to do is to summarize them and make sense and this was a study where what we did is we got a group of experts together, came up with categories of stresses and then did confirmatory factor analysis and came up with these three scales.  One that we identified as family dysfunction, and they were items of such a separation, divorce, incarceration, witnessing loud, long arguments, instability scale items kids moving, new kids.  People moving in and out.  Moving to a new home.  Changing schools.  And then harm to self or others, other family members and these were accidents, illnesses property crime, witnessed threat to the family.  And if you're interested in this and the items that are involved and I would suggest that you go to Anna Lau who is a post doc now on the faculty at UCLA and go to that paper, the '03 paper, and you can look at the specific items that are included within those scales.  We do not have those but scored in the data-- in the data set.  One of the outcomes we're looking at various latent structures, was a social isolation where we had this inventory of supportive adults in the child's life we had at age six.  Kids are asked is there somebody in your life who's, I guess, who is there for you to support-- there's a general prompt.  If they don't identify a parent, then there will be a question about do you have somebody who's like a parent who is there for you?  And once they go through for a mother, a father figure and then any other adult so they can have up to three adults who they identify as supportive and then their followup questions about how supportive are they in terms of instrumental support, informational support in four areas.  But that's another measure at age six.  Then the loneliness and dissatisfaction scale.  24 items, only 16 of them are actual items.  Eight are filler items.  Questions like are you lonely at school?  And this is the-- I had to show you this.  This is my fun slide, but there's not enough time so that was the result, and there I go again.  Okay.  And this is-- an approach to resilience that is a current approach that Laura Proctor is heading up.  And that is looking at-- excuse me-- over time, with kids and this is looking at the kids within the San Diego site, all of whom were removed from their home and placed in substitute care which we would assume is a risk, both being exposed to abuse and neglect and being removed.  ^M01:40:01 And then the issue of what did they look like in terms of their behavioral resilience over time and doing some growth modeling.  And this-- actually mentioning [inaudible], some data from [inaudible] suggesting that kids in terms of over time-- this was not the longitudinal part.  This was just taking kids cross sectionally at different ages and Barbara Byrnes WPPSI and others involved with WPPSI reported that as the kids moved from preschool to adolescence, there's less evidence of resilience, behavioral resilience.  And so what we did is modeled trajectories from six through age four.  Resilience was based on the CBCL both internalized and externalizing.  Resilient kids who were identified as those that had T scores that were less than 60, the borderline cut point.  We were interested in potential protective factors, cognitive ability, actually the WPPSI, social competence and caregiver stability as well as risk.  This was the make up of a sample.  As I said, CBCL internalizing and externalizing.  The WPPSI we have the two scales black design and vocabulary.  We've got the standard scores.  I don't believe we have the short form IQ.  We haven't calculated that now.  We don't have that.  Those are the-- for short scale for the WPPSI, those are the two scales that are recommended, and you can-- if you want an IQ, you can come up with it, but we've got the standard score for each scale.  Then the Vineland Socialization Scale, which I mentioned before, which we have throughout multiple measures of that.  And then caregiver stability, which is whether or not the kids were with the same caregiver from age six to 14 each two year interval.  So, if they were with the same caregiver, they got a one, and then [inaudible] so that the degree of stability.  And then maltreatment, we looked at earlier allegations and later allegations and this one we just used the number of allegations.  So, across studies, we've looked at different ways of conceptualizing, operationalizing maltreatment.  And some of it depends on what the questions are.  Some of it depends on sort of how many variables, how many degrees of freedom we have left, but number of different ways to approach it.  Growth mixture modeling.  Okay, the classes for internalizing.  You've got the stable resilient group.  You've got a group that's mixed, that's decreasing and then a group that started off with problems and increasing resilience over time.  Interesting.  And then when you look at the externalizing, you've got a group that's pretty stable and then you got a group that's stable disorder, and then a group that's increasing resilience over time.  And then looked at the various predictors.  You can look at the results, but what we found is that cognitive abilities seemed to be related to outcomes as did the stability of the caregiver.  Okay, and little different for the externalizing and internalizing.  Again, you can look at those specifics.  Okay.  And we're almost there.  History of victimization.  For those who are interested in that, should talk about that.  There-- for that we asked caregivers at age four, well okay, how come it's not simple?  Some of the samples identified their kids prior to age four.  All of us actually had identified our samples earlier on with prior studies, and we have unique data.  So in North Carolina, it was the stress study and their data they have.  In Baltimore, Howard's got information about kids that had been coming to the pediatric clinic for some time.  We have information about our kids once they entered out of home care prior to age four.  But at age four and six is when we set the baseline for recruitment into LONGSCAN.  Chicago and Seattle, though, that came on as separate studies and came into existence with LONGSCAN has pre-age four data and I think they're part of the data set, though, though not every site administered that.
>> Right. 
>> We did not administer it to our caregivers at age four because 2/3 of them were unrelated to the child and some of them were temporary caregivers and for various reasons, that was one reason-- it was also we did not want to-- we had enough trouble with the adoptive parents and enough trouble with the foster parents getting them to participate and ask-- to respond to questions about their personal history, and we weren't ready to ask them about their history of victimization.  So, we do not have that measure for the San Diego sample.  So, there's some unique kinds of things like that that you'll need to be aware of.  So you might find things that are missing.  That's one thing.  There is a general probe that was asked.  Have you ever felt that you were mistreated?  People then asked more specifically what might have happened?  If they didn't respond to the general probe, they were then asked the specific questions, 11 questions, two dealt with abuse-- physical abuse-- during childhood and two during adulthood and then three sexual abuse during child and teen and then one sexual abuse during assault when they were adults.  The child that was younger than 18, for the sexual abuse, it was under 14 and then as a teen and okay.  But 11 questions and their follow up with some specific questions.  Two papers, one Richard Thompson from the Chicago site and Howard Dubolitz [assumed spelling] from the Baltimore site.  And I'm not going to talk about those.  This is just in terms of the number that had been assaulted.  Of those that responded, the 608, 52% had experienced some form of physical abuse as a child or teenager.  So, a lot of history here with our samples.  So a lot to work with.  What Howard did was looked at-- those studies were looking at the exposure to violence by the caregivers and the resultant outcomes of the kids.  Some interest in factors that might be related to that and what might be mediating it.  And just in terms of the measures.  Got the CTS, CESD, CBCL, WPPSI again, general health.  Okay, at age six, nothing else in terms of the measure.  You can look at the-- you can go to the studies you want to know about the outcomes.  
>> Specific sites or--?
>> No.  So, it was just Baltimore and North Carolina I believe.  Yeah.  And I'm not sure if San-- yeah, okay.  And Richard, he was looking just at the Chicago site and looking at history of victimization and the outcome was again age four CBCL so this was with some of the early data, so this is based on the victimization-- pre age four and then looking at outcomes at age four and possible mediators.  Cage, you see some of the same measures, okay.  And the only partial mediated that he found was the CTS verbal aggression.  History of victimization, relationship to aggression on the CBCL partially mediated by mom's use of verbal aggression and disciplining kids.  This was a study that where we're looking at just in San Diego those kids that were in kin care versus non kin care, and this is a discipline methods assessment which is age eight we added more to the CTS where we had a number of vignettes where we asked caregivers how would you handle kids that are noncompliant, stealing, lying and question which was then coded and if they gave one response, if that didn't work, what else would you use?  So that's the DMA, discipline methods assessment.  The outcome at age eight was the social problem solving measure adapted from [inaudible] measure.  And also called behavioral intent assessment, age eight, and there are seven social situations that the kids are described.  You go on the playground.  You see some kids playing ball, and you asked a kid if you can join, and he says no.  What do you do?  ^M01:50:04 You're standing in line, and a kid comes and pushes you for no reason at all.  What do you do?  Seven social situations, conflict situations or entering social settings.  And those are coded by coders in terms of appropriate responses, verbal assertion, compromise and then non pro social, physical and verbal aggression.  Okay, so I'm not going to do that.  I'm not going to talk about that.  You guys if you got questions-- because I'm-- the AAPI, adult adolescent parenting inventory.  So some of you might be interested in that.  We've looked at some of that, but you guys know that's available.  And actually with our CDC study looking at the children of LONGSCAN, children of our kids.  We asked the new parents, and they respond to the AAPI.  But that's-- you guys don't have that.  Sorry.  Okay.  This is not going to-- is there-- I mentioned my family and friends which is a wonderful measure.  We haven't used it all that much.  Okay.  I don't think there's anything else.  Okay.  This is after going through the muck.  You got to stop and [inaudible].  And I think that is it except for you can read that.
^M01:51:54
[ Laughing ]
^M01:52:00
>> Thank you very much. 
^M01:52:02
[ Applause ]