The Rind Controversy

[Evan Harrington <EVAN-H@xxxxxxxxxxxxx>]

I just received some materials from Bruce Rind explaining the merits
of the rebuttal against his Psych Bul paper, and his refutation of
those criticisms.

Bruce assisted me with my own research while I was at Temple University
as a grad student.  My research was on a totally unrelated topic.

The primary center of criticism lodged against the Rind paper originates
from a group called the Leadership Council, which lobbied Ray Fowler at
the APA and the United States Congress.  Soon after receiving material
from the Leadership Council the APA reversed their position and called
for a second review of the paper, and Congress passed a bill (107)
condemning the study.

The Leadership Council is made up almost entirely of psychiatrists,
psychologists, and lawyers who have been active for years in the
field of MPD.  Thus, it is not a group representing a broad base
of psychologists or scientists.

The two main criticisms lodged against the Rind paper are that:

1).  60% of the data came from one study.
2).  38% of the studies were unpublished.

The document sent to me by Rind describes at length why these two
criticism are clearly invalid.  The following text was written by me
(Evan Harrington) and paraphrases what Rind wrote in a MUCH lengthier
and more detailed account.  DO NOT quote this as Rind's...it is mine.

First, the study referred to in item one is one by Landis (1956).
Critics claim that this study weighted the results toward a finding
of less trauma.  In fact, this study showed more negative self-reported
reactions than did other studies, and the weighting of the studies
MAXIMIZED these negative self-reported reactions.  Had Rind sought to
intentionally bias his results he would have weighted the studies
differently.  Additionally, this study DID NOT factor into the
meta-analysis.  Since this study DID NOT factor into the meta-analysis
the claim that the entire study is flawed because the Landis study
accounted for over 60% in self-reported effects becomes a specious
and invalid argument.

Critics claimed that Rind et al. loaded the study with papers that
presented "mild" cases of CSA.  They included all studies available at
that time.  Statistical analysis in the paper (p.33) examined if there
was a difference in effect for contact versus non-contact CSA.  There was
not.

The second point was that 38% of the studies were unpublished.  Rind et al.
used 36 published studies and 23 unpublished studies.  Of the unpublished
studies, 21 were doctoral dissertations and 2 were masters theses.  As I
recall, a number of these dissertations were completed by people who are
now researchers and/or clinicians in the CSA field.  On Page 34 Rind
et al. compared published and non-published papers and found no difference
in the magnitude of effects.  Dissertations are often conducted at
a higher level of scholarship than found elsewhere, so the argument
that they ought to be discounted because they are unpublished is not
sound.  Furthermore, as anyone who has conducted a meta-analysis knows
quite well, the researcher must search for unpublished papers to
combat what is known as "file drawer threat."  This refers to the
fact that studies with null results are unlikely to be published, no
matter how well they were designed.  If you wish to have an unbiased
estimate of an effect you must include such papers.


I hope this gives members of this list some food for thought.


Evan R. Harrington, Ph.D.
American Health Foundation
New York, NY