Hi everyone and welcome to the third session of the second annual NDACAN summer training series. Last summer if you joined us we focused specifically on NYTD which is one of the data sets we'll be talking about today. And then at the end of the summer we sent out a survey to anyone who participated and one of the main feedbacks we got was that people wanted to hear more about the whole administrative cluster. So that is the focus of the summer. If you were with us last week we talked specifically about the NCANDS data set and this week were going to be talking about AFCARS and NYTD. And this is posted by the national data archive on child abuse and neglect although we have presenters from NDACAN we presenters from Children's Bureau we have presenters from the WRMA so we're really lucky to have such participation. And the archive is posted in the Bronfenbrenner center for translational research at Cornell University. Here's an overview of our summer so we are heading session number three. We are about halfway through our series. At the end of the session will be done going over the overviews of the data sets. Next week will be jumping in for strategies for managing secondary data especially large secondary data sources. So we thank you for joining us for session 3. I am the host Erin McCauley I'm a graduate associate at the the Center and so if you have been attending then you hear me every week I mainly do the introductions. And now I'm going to pass it's over to Tammy White and she's going to be doing the first half of the session and then she's going to be passing over to Michael Dineen whose an analyst in the archive and he'll be doing the second half of the session. Hello everyone I'm Tammy White I am the NYTD data analyst at Children's Bureau and I do some work with our AFCARS datasets as well my AFCARS co presenter wasn't able to make it so I'll be filling in for her. So I'll give you a little bit of an overview of AFCARS but predominantly I am the NYTD person over at CB. And so we'll start out with just giving you a quick overview of our Adoption and Foster Care Analysis and Reporting System or as we say AFCARS. And then I assume most of you on the phone know but in case you don't, AFCARS is our case level information that we collect from all 50 states plus Puerto Rico and District of Columbia. So we have 52 what we call states who report information to us. And it's information on all the youth and children in care that Title IV-E is usually responsible for. It's mandated through Social Security Act and we are now starting to get some tribes. We have I think one or two tribes that it just started to report as well as one of the Virgin Islands. That's if they are not publicly available yet but were just starting to get that and so we're excited to have them come on board. It's an old system that started collecting data in 1994. We collect them in to six month periods and the first one as you can see on the slide came in around 1995. So it's a old system it's come a long way. You'll see a little bit of our work that we've done on data quality to try to get a lot of this really usable for a lot of researchers and and states themselves. The types of information that it's reported to our AFCARS is demographic information and really a lot of information on the actual foster care placement history while the child is in care. We've got things like that date of birth and Michael will talk to you about what we the archive does with that. Information on the caretakers and the foster parents, race identification and dates of removal, dates of discharge, dates of placement, number of placements, case plan goals, and TPR dates. And as I mentioned before data are submitted to us on two six-month files. We start federal fiscal year of October 1 and that file ends March 31. And then the second six months starts April 1st through September 30th. We collect AFCARS through electronic case systems and it's usually the states SACWIS systems that reported to us and Children's Bureau Angelina Palmiero our AFCARS program analyst really works very closely with states on how to improve their data quality and really to work with them to correct any errors states are allowed to resubmit past files at any time and we work really closely with states on improving data quality and have done so over the years quite well. And part of that is over the years I think it's been in the last 20 years almost since AFCARS reviews our division of program implementation which Angelina works has conducted AFCARS reviews in all the states. Part of that is to go in and look at how the data are collected, talk to the managers and supervisors, and have an understanding of how they may use the data themselves. And part of that is you go in and look at the data systems but we also pull random samples of cases to look at the case reviews to see the information we see in the case reviews matches up with what gets reported to us. Throughout that process an improvement plan is developed and that gets monitored along the way and as states knock off things on that improvement plan, that ends up complete. There's final reports on our website on those you can see which states have gone through and which states have improvement plans that are that are pretty much completed. Another part of the data quality is AFCARS data is states use it in their CFSP's or Child and Family Services Plan as well as their annual Progress and Services Reports And the Child and Family Services Reviews CFSR. And so a lot of that data quality looking and reporting comes out through those reports as well as those reviews, the CFSR reviews. If and this one is if a state the SACWIS is being many states are going into this CCWIS which is the Comprehensive Child Welfare Information System. They have to have it data quality plan to address all the aspects of that new system that they've opted to to take funds to implement that system. Those are those are new I'm not sure when those systems will be up and running but when the when the data is in their then a lot of the data quality pieces will will come up in that piece. Oh and I'm sorr I know we we're doing questions at the end but this one is actually really appropriate now: the dates of data collection are again we six month data collection they start October 1 in any federal fiscal year and it runs through March 31st. States have 45 days after that March 31 deadline to submit their files. The next six months data collection starts April 1 and ends through September 30. And again states have 45 days after that September 30 date to submit their files so we usually get that in like mid-November. And that's just a real quick view of our of our AFCARS information Michael will go into a little bit more detail at the end of this presentation on what the archive does with the data elements that are specific to that data set. I'm going to focus a little quickly on the National Youth in Transition Database which is NYTD. Similar to AFCARS we collect case level information on young people. The nice thing about this is it's really a the first time in a national way to collect data on young people after age 18. It's a federally mandated data collection system by that John Chafee Foster Care Program for Successful Transition to Adulthood which we call the Chafee program. Data collection began back in October 2010 and similar to AFCARS all 50 states, DC and Puerto Rico reports to NYTD. We currently don't have tribes or virgin islands yet but we expect those down the road. Just to give you an idea for those of you who may not know what types of information we collect in NYTD: we collect demographic information which includes date of birth, sex, race, ethnicity, foster care status, tribal membership, any history of youth being adjudicated delinquent, educational levels, and then independent living services which is a large part of our data collection. And with we we collect them in certain there's 14 categories of independent living services and those include examples of a academic educational independent living services, career, financial independent living services and some health and housing information. Part of the NYTD is also to collect information on youth who complete a survey and I'll talk about that in a little bit. We ask youth to complete a survey that look at a few outcome areas including financial self-sufficiency, educational attainment, and there is a question on homelessness and some high risk outcomes as well as maintaing a connection to an adult and access to health insurance. Similar to AFCARS we are on the same reporting schedule as AFCARS we get NYTD information in two six-month files. We have compliance standards that we use for timeliness and quality that's built into our CFSR regulations which the regulation numbers up there. States as similar to AFCARS states have the opportunity to look at their files and submit a corrected file at any time. We unlike AFCARS actually we we leverage penalties against states that don't meet certain data quality standards and we work very closely with the states on how to how to get them to a point where they are not they don't have penalties assessed. So a lot of what we do is working very closely with the states on improving their data quality. And it's similar to the AFCARS we actually modeled our reviews on the AFCARS reviews. We conduct NYTD reviews with states. We started out with some pilot states and if there's any of them on the phone thank you for that. You really helped inform the current federal process that we do. And it's a weeklong on-site review that we do and we go to states. We started out with states on a voluntary basis so we've have a lot of wonderful states come forward and say yes and want to go through this. And the intent is really to look at their policies and practices on how they collect and report timely and accurate data. We work very closely we go element by element and looked in their systems to see how those pieces are reported to us and then validate and verify through looking at their system but also in the case review process how that information translates from what we see in the case review process to the files that we get. We look at how the state administers the survey to the young people who are eligible to take that survey. And then a large piece of what we do is really working with states and talking with the states on how they use the data to improve their quality of services as part of their continuous quality improvement framework. And if they are not what we can do to help them think about how to do that. We want to really make sure the field is is working to that this is not just reporting data to us and kind of walking away but really how can they take a look at the data as a whole and use it to improve their delivery of services to young people. To give you an idea of where we are on the data collection schedule we art in FY 19 which is that last column. We are for all states that submit information on independent living services they report that to us every single federal fiscal year in both of those six-month files. So we have services information from FY 11 and we're almost through FY 19 which will and September 30. For the survey youth we've completed to cohorts. The first one started in 2011 and the second one was in 2014 and that just ended in 2018. Cohort three we're we're almost finished with the 19-year-olds we're in the second part of the 19-year-olds follow-up population for them. And so I I've been talking our survey population and we survey youth who have turned 17 while in who within 45 days of turning 17 while they were in foster care and we survey the cohort of youth every three years as you as you saw on that last slide. The first cohort started in 2011 and then every three years we start a new one and then we follow all of them until age 21 so it's two years after that and the youth that are eligible for follow-up are those who took the survey at age 17 and those who did not take the survey at age 17 are not part of those follow-up populations. And just a little bit more detail again at age 17 to be eligible for being in the cohort you must have participated in the survey within 45 days of turning age 17 although we do encourage states to continue collecting information and reporting late surveys to us because it's important that that the youth know that their voices are being heard at least by the state even if if we can't use the data officially we still encourage states to collect information from from youth. They have to have been in care within 45 days of taking the survey of turning 17 within taking the survey and then have to have had at least one valid answer which means they can't answer all declined or not applicable to those that are not applicable. To be eligible in the follow-up population at 21, you had to have been eligible at 19 so even if you don't take the survey at 19 and you were eligible you were still eligible and required to take the survey at age 21 if the state can get you to do that and you are obviously if you're not reported to beat deceased at age 19 you are are in the follow-up population. We have given states the option to sample of follow-up population of youth because some states have very very large survey baseline populations and it's in then in some ways burden some for them to follow up all those youth so we gave states the option to sample and it's for all the youth who are eligible to be in that cohort at age 17 and I pull a random sample. It's at 90% confidence level with 30% attrition rate that's a little if you need the technical information it's appendix C in technical bulletin five. But we have 12 states that chose to sample in cohort one and you can see them on the slide. We have 15 states who sampled for cohorts two and three which was those original 12+ Colorado, Maryland and Montana. And the nice thing about NYTD is it's one of our first national approach to understanding some of the experiences that young people have when they leave care or when they are in care as they get older. We have a lot of states who have extended foster care to age 21 and so the nice thing about NYTD is it captures the information on those youth as well as long as their in those follow-up populations. For almost you know 20 years we've collected information on maltreatment which is that NCANDS information and hopefully you were able to listen to that webinar last week. And then the AFCARS information is about the experiences of youth in foster care so this kind of allows us to complete the picture a longitudinal picture of follow some of those same youth from the report up through age 21 if if they participated in the survey up to that long. So it really gives us a nice longitudinal look at kind of what's going on at the and of the time we collect that information. Michael will get into this later but we have a unique identifier that used in all three data sets so they can be linked which is a really great feature for those researchers that really want to do a lot of good information on your state or just in a national research as well. You can you know you can just look at demographic and characteristic point in time things the percentage of young adults who reported in NYTD that having given birth to or fathered a child, how many placements they have compared to those who did not report having a child. You can also do some longitudinal analyses like regressions and try to look to see if length of stay are related to any particular outcomes when controlling for other variables. There's a lot you can do with this with these three data sets in particular and the archive is really great at providing technical assistance with that. And I believe this is where I hand it off to my co-presenter Michael Dineen from NDACAN. Hello everybody this is Michael Dineen so far we Tammy's been talking about the Children's Bureau and of and the states end of the process. The states create the data and send it to CB Children's Bureau. Children's Bureau does a lot of work on improving the quality of the data and the completeness of the data and then when it's in good shape they send it to us usually on an annual basis except for the Foster Care File which gets sent every six months. And we do further processing on the data in the interest of security and confidentiality for the children. So we do have to take some information out of the data sets in order to make sure that the confidentiality is protected. So one thing we do is in AFCARS is to remove FIPS codes which is a code for a county. Any county that has fewer than 1000 cases in the Foster Care File is removed is set to a value that's not the value of the FIPS code. It's a value that means it's been removed. So then we take the child's date of birth and set it to the 15th of the month so that's the dates of birth are to the nearest month in the publicly available data. Then we once we've sets the date of birth, we adjust all the other dates so that the time Between any two dates is true even if the dates themselves are not exactly the date that like for the removal dates and so forth they are adjusted to preserve the time span. And those of the confidentiality protections that we do in AFCARS. The archive that's new variables to the Foster Care File. These are all variables well not all but in most cases they are variables that you could compute your self from the data but I wanted to put them in their so that well it would save you some trouble as a researcher for one thing, of doing the computation and it will also make sure that everyone who does the research will have the same value for the same variable and won't mis-compute it. One of the variables we add is called "St" state and that's a mnemonic "St" for state because it's two characters and it signals that the variable is is the two character familiar postal service code for a state like NY for New York and so forth, TX for Texas, FL for Florida you you're all I'm sure familiar with those. Because foster care doesn't come with a state ID that you can readily understand unless you know the codes for the states they are numeric like 12 is Florida and that sort of thing so you don't know the numbers so I put in the names of the states or the two letter codes for the states so you'll easily recognize the state when you're dealing with the state. Then ther're all these what I call status flags. Is the child in at the end, is the child in at the start, have they entered, have the exit. These are all the coded dummy variables one or zero and it makes it real easy you can just some across the data set "InAtEnd" to see how many kids were in foster care on September 30 or you know same thing you can see how many kids exited just by summing the variable "Exited" to see how many kids left foster care that year and so forth. "IsWaiting" it means waiting for adoption. "AgedOut" is kind of intuitive. And "IsTPR" means that the termination of parental rights that's what TPR is that means that they are eligible for adoption. So either one or zero they are or they're not eligible for adoption otherwise you'd have to compute that based on the dates of both the parents' TPR's. Then I give you the age on certain dates, the age at the start of the year or on the day they entered. The age on September 30 or the day they exited. And then the age at their most recent removal date. Then there's a series of length of stay variables. The length of stay from the last removal dates to the exit date, or the last removal date to September 30 if they're still in on September 30. Same thing with "SettingLOS" that's their current placement setting. The previous length of stay if they had been in foster care they had a previous removal episode that would be that. And then life length of stay is the sum if they have more than one removal episode it's the some of the first one and the most recent one. and there aren't many that have more than two. That it's like 95% of kids have either one or two removal episodes so LifeLOS is going to give you for almost everyone their entire career of having been in foster care will be in LifeLOS. I give you the Rural/Urban continuum code which is a I think it's got 10 values for urbanicity which is based on the population size and the adjacency to a metropolitan area. I put that in there because of how we have to remove a lot of county identifiers so this gives you at least some measure of rurality if you are interested in rural counties. You can get some estimate of that from Rural/Urban continuum code. I give you two kinds of race variables. One is just pure race, the five races which are white, black, Asian, Native American and Pacific Islander. Those five races regardless of whether Hispanic or not it just tells you the five races. And then "RaceEthn" gives you the races where Hispanic origin is treated like a race it's like they are Hispanic then it doesn't matter what other race they are they are counted as Hispanic. So it's interesting to take those two races and compare them to each other because you can see how many when you do that when you do like a crosstab of race by race ethnicity you can see how many people of who are white- Hispanic, black-Hispanic and so forth of each of the five races you can see which ones have been also had one of the five races in addition to being Hispanic. Then Tammy alluded to this state foster care ID which is a variable I came up with to make it easier to do the linking, both conceptually and practically. Conceptually in the sense that the AFCARS ID in the Child File is called AFCRSID and in the Foster Care File it's called RecNumbr. In the Adoption File it's called it's called RecNumber. So the name of the what we what's really the AFCARS ID for the child is has a different name in the different data sets so I wanted them to all have the I wanted all the linking variables to have the same name. And the other factor is that but AFCARS ID could be duplicated across states like two different states could have the same AFCARS ID so if you're linking or anytime you're trying to go by individual children across states, you have to use state in addition to AFCARS ID. I just put them all in I concatenated the two-letter postal code to the front of the "RecNumbr" and that made state foster care ID for the Foster Care File and I put St for the two letter code for the state in front of the AFCARS ID in the child file and in front of RecNumbr in the NYTD file. So they all all three of those data sets have the same name of the variable that you're going to use to link those data sets together. So those of the new variables that we add here at NDACAN. We also produce new versions of the data sets and I don't know if researchers know exactly why you'll get version 6 of or version 4 of the Foster Care File. That's because the states are able to resubmit data to the Children's Bureau anytime they make a correction or improve the quality of the data or at cases or remove cases, they are free to make of resubmission to the Children's Bureau and it doesn't really matter what year it is. So they send those in I guess at random times I don't know if there's a schedule to that but the ones that are the resubmissions that are submitted during a year are included in the annual foster care package that they send to us. So when we get a year of foster care we should be getting the 2018 Foster Care File pretty soon and that will have some be submissions from various states. So when we get these re-submissions I go through and re-process that Foster Care File and make it available to you guys to the researchers as a new version of that data set. So the new versions have slightly different data. The first version of course is going to be version 1. You're more interested in versions 2 and 3 like the early versions because then many states will probably be resubmitting data. Like version 2 might have five or six states that resubmit data. Like version 6 or version 7 might only have one state. But it's still important to have the most recent version because they are slightly different in your counts and so forth but you can't expect a version you know a later version of the Foster Care File, while it is more accurate in the sense it more directly reflects the data that the state has at their it may not reflect or match published printed documents that are AFCARS data or maybe even the AFCARS report that's published annually on the CB website. For accuracy you should always use the most recent version. When we do make a new version of the data we announce that to our mailing list and I assume you're all on our mailing list because you got the announcement that we are having these webinars. So when you're working with multiple years of the AFCARS data, you would just stacked them like one year on top of the other. But when you do that, if you're in just one year AFCARS the AFCARS ID, RecNumbr, or state foster care ID whichever you want to either way is going to be unique in that file. So that a child only appears once in the data set in a particular year. But if you stack data sets then obviously that child could be duplicated because a child is going to have a record in the Foster Care File each year that they are in foster care. So until they exit then they will have a record in every year. So if you're trying to resolve to one record per child it's almost always best to use the most recent year because you'll know that they're either in at the end of that year or they exited the year they still have the year they exited will be the last year the last the most recent year for that child. I can't tell you right more right now about resolving a stacked file to the child level but you you're always welcome to write to our help email address and if you need some assistance on that. The NYTD files on our end when we get the files from Tammy it's an annual file. They submit them every six months to Children's Bureau but we get one every year. Tammy puts those together. The two annual files that are submitted by the Outcomes File which is the survey that Tammy's talking about and the Services File. We had a couple variable names as we tend to do. It's good to have a variable that has wave when you're dealing with the data when it's not natively in there. So I have a variable called "Wave" and it's either one two or three depending on the wave. Wave one is age 17, two is 19, and three is the 21 survey at age 21. Then again the state foster care ID is the linking variable which which I described earlier. Then the St is the same one that I add to AFCARS, the two character postal code. And then "Responded" is a variable that's just to save you some work if you're looking for because otherwise you'd have to write a code that says do they have a valid response to this question, or do they have a valid response to the next question, or do they have a valid and so forth through all those I think there are I don't know how many 17 or something like that. This is just down to one variable where it's either one or zero and if it's the one that means they had a valid response on some question. That comes in handy sometimes. Then variables we add to the Services File again state foster care ID and St to make it easier to see what state you're talking about. "FY" for the federal fiscal year of the data. Again "Race" and "RaceEthn" with and without Hispanic ethnicity, age at the midpoint "AgeMP" tells you the youth's age at the midpoint of the reporting period not the well the outcomes survey everybody's 17 at wave one, but in the Services File not everybody is 17 the services can be provided to to kids younger than 17 depending on what the states'rules are for that or older than 17 or 18, 19, 20, 21 they can be the can still receive services and I think that depends on the states definition of who is eligible. And then I put this I thought this would be handy in the Services File I put a variable that tells you whether they are in one of the outcomes cohorts. So it'll have 2014 in that as a value for a child in the Services File who was in the 2014 cohort. So it just seemed to me that that would be useful in some cases if you're looking in the Services File for people who were in an outcomes cohort. Four confidentiality protections we any county that's identified in the Foster Care File is also identified in in NYTD and then the dates of birth are set to the 15th just like in the Foster Care File. So the NYTD the data sets that we have available now are the 2011 cohort, the three years that survey and then the 29,000 people who work age 17 in 2011 53% of home responded to the survey and then the two the waves two and three. And then for the 2014 cohort where there were 23,780 kids who were 17 in 2014, had a 69% response rate which is significantly better than the first time. And then the 2017 cohorts where you get similar response rate. So those are the data sets that are available. Actually the 2017 isn't white up yet but it'll be up soon. Those are the data sets that are available from us. The weighting is done by by the Children's Bureau or by a contractor for the Children's Bureau and they weight for nonresponse and this is just telling you this theory of weighting is like if in your sample more people say just as usually the example is male/female, if more females respond then their proportion in the population then they're going to be overweighted in the survey and if they are overweighted in the survey and if they have different kinds of outcomes than males do, then it'll be skewed in the direction of the female. Same thing with race or any other variable, if they are both skewed and their skewness their non-proportionality or their the proportion in the surveys different from the proportion in the population, if that factor plus the factor that that difference is meaningful, like they actually are different in some sort of way that's going to show up in the survey outcomes then your survey's going to be skewed. So so one thing you can do to correct for that is to re-weight the responses to be the same proportions as they are in the population. And we are lucky with the NYTD data because we have a lot of information on the people who don't respond because we know who the nonresponders are and we had their not only their NYTD outcomes demographics but we also have their all the AFCARS foster care information for them by linking them to the Foster Care File. So you can get that's why the Children's Bureau was able to use a lot of variables, the 32 in the first wave one and 42 variables in wave two and three to adjust for nonresponse. But having done that then people who have used the data have reported to us that there is really not much difference between the weighted in the unweighted results. But the weights are included in the data that you get from us. So working with multiple years of NYTD the Services File has, I think the current one is 2011 through 2018 so you have you have nine use of data and to reporting periods for each year so you have 18 chunks of data where a child can appear. So you could have a child could I don't know if they could be in their 18 times but they could be in their a lot of times because they appear every time every six month period that they receive a service their ID is going to be in the Services File. So if you want to resolve the data to one record per child or to your analysis on one record per child you'll have to work around that. Then for the Outcomes File the data you from us unless it's just the first wave of the cohort will be duplicated, some kids will be duplicated because any child who completes all three of the surveys or two of the surveys or just the first one where they could do the first one and third one or the first one and second one. So it comes along and a lot of people who I think if you use, I know with SPSS people need to convert that from long to wide format. So that's what you'll need to do if you want to work with certain I guess it depends on the application you use your statistical application and the particular test that you're doing. And then when I do the fifth of our webinars is going to be on linking and I'll show you with the application the actual code that you can use for converting long to wide format. But that's another webinar. These are resources this slide is just to be in the online version so you can have access to these resources. and I think that's the and you can ask questions if you want. I'll turn it back over to Erin. Beautiful thank you so much Tammy and Michael that was a very in-depth presentation talking about what happens kind of on both sides of the data process and how the data comes to you guys. We have two questions already, the first question I think both questions were about the NYTD sample and how that works so the first question was for the NYTD sample in the first moment of the three data collections the question says so within 45 days of their 17th birthday if so is that data provided in months. Hi this is Tammy the data are not provided in months but it's we get the young persons birthdate and the date that the completed the survey so it would be the date of the survey minus the data set they turned 17. So that's 45 days or less than that's within the 45th day. Great and our next question I think will also be towards you it says regarding the eligibility, if the youth answered the survey at age 17 but left care in 18 is the youth eligible to participate in the age 19 survey. Yes as long as long as that 17-year-old's survey was a valid survey and he or she is in the cohort then yes. You do not have to be in foster care at age 19 or 21 to complete the survey. NYTD has we have participation level requirements so if you have a youth at 19 for the 19-year-olds who are in foster care I think it's 60% of the response rates should be of those eligible in foster care your participation rate should be at least 60% and then for those youth who are not in care I think it's 80%. Don't hold me to those because I don't have the legislation in front of me, but we do have participation requirements. Great in the next question is about AFCARS it says are youth 18+ in extended foster care reported in AFCARS. So that's a really good question and and I will take that and and hopefully there is an opportunity to answer these questions after the seminar but yes if there are certain requirements and rules around it so states that have opted for extended foster care reports to AFCARS if the youth who is 18 and older is receiving Title IV-E so so we only get those youth who are 18 and older who are 4E eligible. So many states do not use 4E funds for their extended care or some of their extended care is out of state funds and so those youth are not reported to us. I believe that will be changing in with the new AFCARS if and when that ever happens I think with the new comments and MPR that came out a while ago that that addressed that. But if you need more information let the archive know and we will have our AFCARS program analyst get back to you on that. Yes on our website we have kind of information about each data set, links to information that CB puts out about each data set and then we also have an area where you can send in a question. And we'll give it a ticket and then assign you to the best person too so if you need more information on that I highly suggest it. We have another question saying I'm wondering how the removal reasons for each of the AFCARS admission was assessed. I'll let people answer but I would also recommend checking out our website again this pretty detailed information about each coded what the potential options are. Yeah I think I'm going to echo your suggestion. We have Technical Bulletin I think One on the Children's Bureau website and that gives some information about how states are to report that. It's really the state driven so there are certain things that states use to determine whether it's a reason for a reason for removal but it's not always the only reason so if it was an important consideration in the reason for removal then the state is to report bits. It's multiple-choice for that I mean each element has its own reason for removal and then the state says yes or no so it's multiple response for that one for that one child. And again check out our technical bulletins and also as Erin said check their website and you can communicate back to us if you need more information on that. Great. For AFCARS data if the child moves between states will the have different RecNumbr's in the different states. Yes. The will. It's whatever AFCARS ID they use within those states, yes. They would have a ID. Some states they don't have it within counties with the missing states at least we tried to encourage states not to do that although in some county administered states that does happen. But yes between states they would have different numbers. Our next question says I plan to study the second cohort data 2014 which years of AFCARS should I use to gets data on all the foster care experiences of this cohort? I can answer that and that's also a question for Michael. It depends on your research questions. So for the 2014 the youth turned 17 and 2014 so certainly the AFCARS 2014 data will have the experiences relevant to the year that the youth took the survey. That's the most current information and then I'll let Michael talk about if you want information when youth was younger than 17 and how that works in the archive. Also I'll put in a quick plug that we are having an entire session on linking led by Michael not next week but the week after. But yeah Michael. I've worked with people who have done this and and almost all the AFCARS foster care people you want will be in the same year. There'll be a few in the following year. I guess that's because of the 45 day things like if they their birthdays toward the end of the year the still have some time to be there 45 days into halfway through February 2 still answer the survey. But one odd thing and I want to your would Tammy says about this. There are always there are people in the Outcomes File who have no record ever in AFCARS. And you'll run into that to so you're not going to get at least with the 2011 and 2014 cohorts, you're not going to get a match for every child in the outcomes survey with somebody in foster care for any year. So that's just a quirk. That's true. I think because the states report the eligible 17-year-olds to us through their own data systems and because the AFCARS is that six months most recent there are some some youth that are short stayers who come in and out that don't end up in the AFCARS reportable population. If you do the match pits it's roughly 95 to 97% match though. The match up pretty well but that is Michael said there are certain that won't that you won't find in AFCARS and those it's just because the way the AFCARS files are constructed in the way their reported to us. Some kids just fall in and out and they don't end up in AFCARS. If you just use the same year like say 2014 and for 2014 you'll get about 90% match and you can improve that a little bit but you're fine with that. Is there a NYTD equivalent data set that looks at long-term outcomes for children who exit foster care the adoption and are in the AFCARS adoption file? There is not unfortunately. I I believe the AFCARS the adoption file you can't link to the Foster Care File because as your adopted you get a you're a new person a new ID and so. Not every state. Some states. Not every state? Okay sorry Michael Godhead. That's a you yeah the that's a tricky thing because I it's and it's very strange to me why it's like they the died and became a person when they write got adopted. And so they some states give them a AFCARS ID so than if they return to foster care then you won't be able to detect that the returned and which I think is a problem but that's the way some states do it but but there are some states who use the same and you can track them. All right we have time for probably one more question. Are there restrictions on obtaining the AFCARS six months file i.e. is IRB required now or is it the same as the public files. No it's the same as AFCARS foster care annual, there's no restriction. But I would I would like to answer the last question. Yes if aging out equals one in AFCARS then will that youth have foster care equals zero in the NYTD outcome for the same year. I'm trying to fully understand which states offer extended foster care both federally and at the state level. I'm trying to understand if this can be understood from these administrative files or only by reading legislation and statutes? Well the the answer to the actual question if they age out will the have foster care zero that answer is not necessarily. It depends on when they age out and when the answer the outcomes survey. So it's like Tammy was saying the have 45 days from their 17th birthday and then if they turn 18 and the their aged out in a state that automatically ages you out at 18 anyway it depends. And I don't think you can use that criterion age out equals one is automatically going to be foster care equals zero. Right it's foster care status at the time they take the survey at ages 19 and 21. You have to be in foster care to take the survey at the time take the survey at 17 you had to have been in foster care. But in answer to the extended part I believe we have a list of the states who have extended foster care on the Children's Bureau website. I think we're up to 29 states now. I'm pretty sure that list is on our website. So you know which states they are.. Can I ask you about that Tammy? Can I ask you a question about that? Sure. I've wondered if his extended foster care is that merely like and administrative things like the child doesn't actually have to be in a foster care home or anything that just officially in foster care in order to be able to be eligible for services? Or are they you know in a in some sort of facility or in foster care home? There's usually a placement associated with that it might be supervised independent living or something but it's under care and responsibility of the agency. Oh, okay thanks. All right so we will end with a preview of next week's we have Frank Edwards who is a professor at Rutgers and also associated with the archive coming to do a data management series so this will be some really great practical code that you can apply to working with data management especially for these large administrative data sets they can require large amount of data management. And thank you everyone for joining us and big thank you again to Tammy and Michael that was a really, really fantastic presentation and a great question and answer session and we hope to see everyone next week.