Recipe: Youth who do age out of care

In the previous Recipe, you learned how to calculate the number of children entering care each year who are at risk of aging out within five years. This Recipe will show you how to calculate the proportion of children in this group that actually do age out.

This Recipe will take you about 20 minutes to complete. In addition to the web tool, you will need a spreadsheet program such as Microsoft Excel.

Question: Of children who are at risk of aging out of foster care within 5 years of entry, how many actually do age out?

[Note: This Recipe presumes that children will age out of foster care upon turning 18. The steps can be altered for jurisdictions that extend foster care benefits to children through age 19, 20, or 21.]

1. On the All spells page, scroll down to the Spell overview section. Under Spell started age, enter 13 in the Min box. This tells the system to only return records of children who started their spells at age 13 or older. We select 13 as the minimum spell start age, because that is the youngest age at which a child entering care can be at risk for aging out within 5 years of entry (13 + 5 = 18).
2. Scroll down to the Sample selection section and enter 01-01-2003 in the From box and 12-31-2008 in the To box. This narrows the selection further by telling the system to only return records of children who entered care between 2003 and 2008. We need to go back this far because in order to examine the outcomes of children who were at risk of aging out within 5 years of entry, we need to include at least some children for whom 5 years have elapsed since entry.
3. Scroll down to the Define output section. Under the List records section, select Age at Spell start, Spell Start Date, Duration of Spell (days), and Spell exit type. Then click Download Data.
4. Clicking Download Data will open a comma-separated file (.csv) containing your results in Microsoft Excel. As in the previous post, use the “Text to Columns” and/or “Date and Time” functions in Excel to create a spreadsheet that looks like this:

5. Each row in the spreadsheet represents a child who entered foster care between 2003 and 2008 and was 13 or older when he/she entered care. Sort the data by SPELLAGE to double check that you made the correct selections above. All the ages under SPELLAGE should be 13 or greater.
6. Now create a Pivot Table to look at the proportional representation of each exit type in each year. Refer to the Variable Names and Codes page of the User Guide to learn which codes represent which exit types. Here are the results from my sample county:

According to these results, of children who entered care in 2003 who were at risk of aging out within 5 years, 45% reunified (XRF), 22% aged out (XRM), and 23% ran away (XRY). Notice that no children in this group were still in care (ZTC) as of the censor date, which in this example county is 12/31/2011. Why? Because if the age of majority is 18, no child who entered care at age 13 or older in 2003 could still be in foster care on 12/31/2011. However, note that when we repeat this analysis for later cohorts, we find a small number of children who are still in care.

I can learn a lot of things from this table. The first is that, in this county, the largest proportion of children entering care as teens wind up reunifying with their families. Second, I learn that the proportion of children in this age group that actually do age out has more or less been decreasing over the years, possibly due to the increased use of relatives (XRL) as permanent resources. Third, I learn that more children at risk of aging out within 5 years of entry actually run away from care than age out. As I consider what kind of services youth at risk of aging out need, this finding makes me realize that runaway prevention will at least be as important as finding these youth permanent homes.

What can I do with these percentages to help me estimate outcomes for this group in the years to come? First off, I need to acknowledge that a small number of children in the 2007 and 2008 entry cohorts are still in care. It would not be unreasonable to predict that these remaining youth will ultimately age out. Therefore, for the sake of estimating the ultimate proportion of emancipations for these two cohorts, I could add the proportion of children still in care to the proportion already aged out, yielding a total of 17% and 13% of youth aging out from the 2007 and 2008 cohorts, respectively.

I could then take the average of these (15%), go back to my results from the previous Recipe, which calculated the number of children entering care in 2005 through 2010 who were at risk of aging out within 5 years, and estimate what would happen if 15% of the 2009 and 2010 entry cohorts actually aged out.

Or, I might think, “Why should I apply that average percentage those entry cohorts if the general trend in aging out seems to be decreasing?” Taking that approach, I might note our estimated rate of aging out for the most recent cohort above (2008), which was 13%, and from that reference point decide to apply 12% and 11% rates to 2009 and 2010, respectively. At this point, there are no hard and fast rules about how to estimate the outcomes of these recent cohorts; base your decisions on your knowledge about your own system and your own judgment about how conservative you wish to be.

Summary

In the previous Recipe, you learned how to calculate the proportion of children entering care who are at risk of aging out within five years. In this Recipe, you followed that up by calculating—and for more recent cohorts, estimating—how many of those children actually do age out of care. Taken together, these analyses can give you a sense of the quantity and types of services your system needs to deliver to teens entering foster care in the near future.

But what about the not-so-near future? These analyses are among the key building blocks of projection models that aim to predict the number of children who will age out of care years down the road.

Back in November 2011, Data Center director Fred Wulczyn gave a presentation to the Executive Committee of the National Association of Public Child Welfare Administrators (NAPCWA) on the Data Center’s recent work on projecting trends of youth aging out of foster care. These projection models estimate the risk of youth aging out years into the future by allowing you to consider different ways in which the caseload might change over time. For example: How many children are projected to age out over the next five years if my system continues to bring the same proportion of teens into care as it has in the past? What if we take in fewer teens? What if we take in more? What if we achieve more permanent exits for our “backlog” of teens already in care? Projecting the rate of aging out based on various scenarios allows you to estimate the expected cost of services given potential policy and practice changes that may alter the proportion of teens on your caseload.

Projection models are an example of the kind of analysis that goes beyond the current capacity of the web tool. If you are interested in working with Data Center staff on projection-related tasks, contact me at lalpert@chapinhall.org.