[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
RE: New study of foster care outcomes --comment
<x-html>
<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN">
<HTML><HEAD>
<META http-equiv=Content-Type content="text/html; charset=us-ascii">
<META content="MSHTML 6.00.2800.1589" name=GENERATOR></HEAD>
<BODY id=role_body style="FONT-SIZE: 10pt; COLOR: #000000; FONT-FAMILY: Arial"
bottomMargin=7 leftMargin=7 topMargin=7 rightMargin=7>
<DIV dir=ltr align=left><SPAN class=326032414-16072007>David: On your first
point - while the assumption of randomization is necessary for the author's
method, I don't think its correct to say that he "expects us to take his word
that he has thought of these possible influences and accounted for them."
He does not expect us to take his word on anything - he explicitly lists
possible violations of true randomization, and he also specifically mentions
that he excludes cases where sexual abuse or Spanish language problems are
involved precisely because these are the cases which are not randomized.
Using an instrument is always risky, but there are clear steps taken in this
paper to ensure that random assignment is truly random. Granted, there
will probably be some unaccounted for violations of the random assignment
of investigators, but the point is that this is still a vast improvement on
the creation of ad hoc "comparison groups" that aren't really comparable at
all. </SPAN></DIV>
<DIV dir=ltr align=left><SPAN class=326032414-16072007></SPAN><SPAN
class=326032414-16072007></SPAN> </DIV>
<DIV dir=ltr align=left><SPAN class=326032414-16072007>I would also refer
readers to Tables 2 and 3, and the discussion on pages 21-24. Far from
David's assessment that "no validation of this assumption is given", the author
goes to great lengths to demonstrate that investigator assignment is
random. I'm sure the author would welcome discussion on unobserved
characteristics correlated with assignment of investigators, but all observed
data - as well as background research on the Illinois child protective system -
seems to suggest that investigators are randomly assigned in the sample
used in this analysis. </SPAN></DIV>
<DIV dir=ltr align=left><SPAN class=326032414-16072007></SPAN> </DIV>
<DIV dir=ltr align=left><SPAN class=326032414-16072007>As for the concern about
the "marginal cases" - I agree, the paper would benefit from describing
these cases in more detail. But remember, no single observation in the
sample can be identified as a "marginal case" because each child is investigated
by only one worker. The marginal cases are theoretical cases
where two investigators would have disagreed. This margin is defined
by the difference in the underlying propensity of an investigator to remove a
child (which is calculated from a series of cases), so the range of
this margin should not be dominated by unique characteristics
of certain children. It is a theoretical concept and we can't attach
the concept to specific observations in the sample, but the derivation of
this margin from the underlying propensities of investigators should free it
from any bias resulting from what you describe as "something very
specific".</SPAN></DIV>
<DIV dir=ltr align=left><SPAN
class=326032414-16072007> </SPAN></DIV>
<DIV dir=ltr align=left><SPAN class=326032414-16072007><FONT size=2><FONT
size=1><FONT face="Times New Roman"
size=2>----------------------------------------------------------------------------------------</FONT>
<BR><FONT face="Times New Roman" size=2>Daniel P. Kuehn</FONT> <BR><FONT
face="Times New Roman" size=2>Research Assistant</FONT> <BR><FONT
face="Times New Roman" size=2>Center on Labor, Human Services, and
Population</FONT> <BR><FONT face="Times New Roman" size=2>The Urban
Institute</FONT> <BR><FONT face="Times New Roman" size=2>Office:
202-261-5394</FONT> <BR><FONT face="Times New Roman" size=2>Email:
DKuehn@ui.urban.org</FONT> </DIV></FONT></FONT></SPAN><BR>
<DIV class=OutlookMessageHeader lang=en-us dir=ltr align=left>
<HR tabIndex=-1>
<FONT face=Tahoma><B>From:</B> bounce-1603077-6840348@list.cornell.edu
[mailto:bounce-1603077-6840348@list.cornell.edu] <B>On Behalf Of </B>Finkelhor,
David<BR><B>Sent:</B> Sunday, July 15, 2007 8:58 AM<BR><B>To:</B> Child
Maltreatment Researchers <BR><B>Cc:</B> Lucy Berliner;
KOLKO@vms.cis.pitt.edu<BR><B>Subject:</B> RE: New study of foster care outcomes
--comment<BR></FONT><BR></DIV>
<DIV></DIV>
<DIV dir=ltr align=left><SPAN class=500503612-15072007><FONT color=#0000ff
size=3>I have serious concerns about this study for several reasons. 1) Its
findings are completely based on an assumption that rotational assignment
of cases to caseworkers is the equivalent of a randomization procedure. No
validation of this assumption is given. We all know that there is a crucial
need for true randomization in outcome assessment because there are so
many possible selection influences most of which we are unable to detect or
predict. The author expects us to take his word that he has thought of these
possible influences and accounted for them. 2) Any unaccounted for biases
in foster care outcome research are almost certainly in the direction of worse
outcomes for those in placement, because those being placed typically have more
severe histories and problems. This study could easily be capitalizing on this.
3) Even if the study is correct, the conclusions only pertain to what the
author calls "marginal cases" (those where the decision could go either
way), but he has no way of specifying just what characteristics these
cases have or how many there are or what portion of the caseload they
represent. In fact, the disagreement among workers decisions may be about
something very specific, not just all cases with some doubt. Urging caseworkers
to err on the side of no placement in a general way may not produce better
outcomes at all. Meanwhile, the study is likely to be interpreted as an
argument against foster placement in general, which is what the USA Today
article implied, and that is a mis-statement of the
conclusions.</FONT></SPAN></DIV>
<DIV dir=ltr align=left><SPAN class=500503612-15072007><FONT color=#0000ff
size=3></FONT></SPAN> </DIV>
<DIV dir=ltr align=left><SPAN class=500503612-15072007><FONT color=#0000ff
size=3>I hope others on the listserv (more knowledgeable than me) will chime in
with comments.</FONT></SPAN></DIV>
<DIV> </DIV>
<P><FONT size=2>David Finkelhor* Crimes against Children Research Center*
Family<BR>Research Laboratory* Department of Sociology* University of
New<BR>Hampshire* Durham, NH 03824* Tel 603 862-2761* Fax 603 862-1122*
email: david.finkelhor@unh.edu<BR><BR><A href="http://www.unh.edu/ccrc/";
target=_blank>http://www.unh.edu/ccrc/</A><BR><A href="http://www.unh.edu/frl/";
target=_blank>http://www.unh.edu/frl/</A><BR> </FONT> </P>
<DIV> </DIV><BR>
<DIV class=OutlookMessageHeader lang=en-us dir=ltr align=left>
<HR tabIndex=-1>
<FONT face=Tahoma size=2><B>From:</B> bounce-1586498-6832158@list.cornell.edu
[mailto:bounce-1586498-6832158@list.cornell.edu] <B>On Behalf Of
</B>NCCPR@aol.com<BR><B>Sent:</B> Monday, July 09, 2007 8:53 AM<BR><B>To:</B>
child-maltreatment-research-l@list.cornell.edu<BR><B>Subject:</B> New study of
foster care outcomes<BR></FONT><BR></DIV>
<DIV></DIV><FONT id=role_document face=Arial color=#000000 size=2>
<DIV>
<DIV>
<DIV><FONT style="BACKGROUND-COLOR: transparent" face=Arial color=#000000
size=2><FONT face=Arial size=2><SPAN
style="FONT-SIZE: 10pt; FONT-FAMILY: Arial; mso-bidi-font-size: 12.0pt">By now,
many of you may have seen a <I>USA Today</I> story on a study comparing
outcomes for children placed in foster care to comparably maltreated children
left in their own homes.<SPAN style="mso-spacerun: yes"> </SPAN>For those
who have not, the story is available here: <A
title=http://www.usatoday.com/news/nation/2007-07-02-foster-study_N.htm?csp=34#Close
href="http://www.usatoday.com/news/nation/2007-07-02-foster-study_N.htm?csp=34#Close";><SPAN
style="COLOR: windowtext">http://www.usatoday.com/news/nation/2007-07-02-foster-study_N.htm?csp=34#Close</SPAN></A>
and the full study can be found here: <A
title=http://www.mit.edu/~jjdoyle/doyle_fosterlt_march07_aer.pdf
href="http://www.mit.edu/~jjdoyle/doyle_fosterlt_march07_aer.pdf";><SPAN
style="COLOR: windowtext">http://www.mit.edu/~jjdoyle/doyle_fosterlt_march07_aer.pdf</SPAN></A>
</SPAN></FONT></FONT></DIV>
<DIV><FONT style="BACKGROUND-COLOR: transparent" face=Arial color=#000000
size=2><FONT face=Arial size=2><SPAN
style="FONT-SIZE: 10pt; FONT-FAMILY: Arial; mso-bidi-font-size: 12.0pt"></SPAN></FONT></FONT> </DIV>
<DIV><FONT style="BACKGROUND-COLOR: transparent" face=Arial color=#000000
size=2><FONT face=Arial size=2><SPAN
style="FONT-SIZE: 10pt; FONT-FAMILY: Arial; mso-bidi-font-size: 12.0pt"></SPAN><SPAN
style="FONT-SIZE: 10pt; FONT-FAMILY: Arial; mso-bidi-font-size: 12.0pt">A
University of Minnesota study used a different methodology and measured
different outcomes, but came to similar conclusions.<SPAN
style="mso-spacerun: yes"> </SPAN>(Byron Egeland, et. al., “The impact of
foster care on development” </SPAN><SPAN class=f161><I><SPAN
style="FONT-SIZE: 10pt; FONT-FAMILY: Arial; mso-bidi-font-size: 8.0pt">Development
and Psychopathology</SPAN></I></SPAN><SPAN class=f161><SPAN
style="FONT-SIZE: 10pt; FONT-FAMILY: Arial; mso-bidi-font-size: 8.0pt">, (Vol.
18, 2006, pp. 57–76)).<SPAN style="mso-spacerun: yes"> </SPAN>As far as I
know, that study is not available online.</SPAN></SPAN></FONT></FONT></DIV>
<DIV><FONT style="BACKGROUND-COLOR: transparent" face=Arial color=#000000
size=2><FONT face=Arial size=2><FONT size=3><FONT
face="Times New Roman"></FONT></FONT></FONT></FONT> </DIV>
<DIV><FONT style="BACKGROUND-COLOR: transparent" face=Arial color=#000000
size=2><FONT face=Arial size=2><FONT size=3><FONT face="Times New Roman">Richard
Wexler<BR>Executive Director<BR>National Coalition for Child Protection
Reform<BR>53 Skyhill Road (Suite 202)<BR>Alexandria VA 22314<BR>(703)
212-2006<BR><A title=http://www.nccpr.org/
href="http://www.nccpr.org/";>www.nccpr.org</A></FONT></FONT></DIV></FONT></FONT></DIV>
<DIV></DIV></DIV></FONT><BR><BR><BR>
<DIV><FONT style="FONT: 10pt ARIAL, SAN-SERIF; COLOR: black">
<HR style="MARGIN-TOP: 10px">
See what's free at <A title=http://www.aol.com?ncid=AOLAOF00020000000503
href="http://www.aol.com?ncid=AOLAOF00020000000503"; target=_blank>AOL.com</A>.
</FONT></DIV></BODY></HTML>
</x-html>