Re: sampling bias

["Karen Wade" <Karen.Wade@xxxxxxx>]

Evan,

Well said.  I completely agree with your concern and outrage about the political
nature of this attack and its potential to "silence" research, and publication,
on controversial topics with findings that may produce discomfort and cause us
to rethink our assumptions.  That is a function of science, I believe, to
question assumptions so that more generalizable, more accurate understandings
can emerge.

Hopefully, the members of this list can distnguish between the moral, clinical,
and scientific issues which are at the heart of this disturbing attack on Rind
et al.  Are there others on this list who are concerned about defending
scientific inquiry in the face of political opposition?  And, is there anything
we can collectively do about it?    I would think some type of statement from
child maltreatment researchers that supports the notion that the merits of the
findings of a scientific inquiry are most legitimately undertaken by the peer
review of other scientists could go a long way toward putting some perspective
on the issues in conflict here.  Not that science exists in a social and
political vacuum and is "above the fray".  But, I believe that part of our job
is to educate the public/politicians on the context in which these findings
should be taken (sampling bias, other studies, etc.).

My two cents,

Karen Beck Wade
Claremont Graduate University
Claremont, CA


Evan Harrington wrote:

> On Wed, 16 Jun 1999 09:28:32 -0500 Mark J. Chaffin said:
> >Regarding the Rind, et al. study and the issue that this represents a less
> >biased sample because it exclusively used college students, it is important
> >to note that recent data presented by Duncan has found that 1) abused
> >students have markedly higher freshman year college drop-out rates than
> >non-abused students, 2) freshmen with more abuse drop out at higher rates
> >than those with less abuse, and 3) among abused freshmen, severity of PTSD
> >symptoms is significantly associated with dropout.  I think it is difficult
>
> [snip]
>
> In my opinion there is no doubt that using clinical and legal samples
> provides biased estimates of the effects of CSA if your aim is to
> generalize to all people who experience CSA, which many CSA researchers
> do.  In using college samples there will naturally be a selection bias
> in terms of higher functioning individuals being represented.  One might
> ask how much college subjects are "higher functioning" than noncollege
> subjects.  Some who have had the opportunity to teach college classes
> might respond "not much."  Admittedly, there is the possibility that
> more seriously maladjusted individuals are selected out when dealing
> with college subjects.  However, if we wish to restrict our generalization
> to those who have experienced CSA and who have gone on to college, we
> would be dealing with a MUCH higher proportion of the United States
> population than we would if we dealt only with those who experienced
> CSA and are in therapy.  One question to be asked and answered when
> considering how much college samples are biased is: Are the results
> of college samples different from NATIONAL PROBABILITY SAMPLES? It has
> been pointed out several times that in another paper Rind examined
> several national probability samples and achieved results similar to
> those found in college samples.  Additionally, Allie Kilpatrick presented
> a review which included a number of convenience samples which found
> results similar to Rind's.  A number of other authors (e.g., Ney, Fung,
> & Wickett, 1994) who controlled for background variables found results
> similar to Rind's, in that there are a number of factors which
> correlate with CSA, making it difficult to assume that CSA causes
> the maladjustment which is observed.  This latter point could be made
> with regard to the above mentioned study by Duncan: How do we know that
> it was the CSA and not some other correlated variable that made CSA
> subjects drop out of college?  My point here is that there is a varied
> literature available which is consistent with the results of Rind et al.
> Rind's results are not an anomaly, but many have been acting as if they
> were.
>
> There is one other point:
>
> The present attack on Rind et al. appears to be highly politically
> motivated.  Conservative Congressmen, backed by family values groups,
> threatened to publicly denounce this research, making the APA backpeddle
> in their support of the study.  I am shocked and appalled at the political
> nature of this attack on a scientific paper.  I am similarly shocked and
> appalled that so few scientists appear to be concerned about it.  I
> appreciate that the comments within this forum have been restricted
> to debate of the methodology of the study.  Such debate is as it should be.
> But I am shocked that so little has been said of the political attack
> on an untenured professor for publishing results that conservatives
> are uncomfortable with.  Even when I have disagreed with the interpretation
> of a study I defend it against those who unreasonably attack it.  I have
> never in my life seen such an outragous politically motivated attack
> on science, and the silence of so many is truly disheartening.
>
> Evan R. Harrington, Ph.D.
> American Health Foundation
> New York, NY