29206: CHin (reply) RE: 29202: Durban (comment): On the Lancet Study (fwd)

[Bob Corbett <corbetre@webster.edu>]

> From Elizabeth Chin (ejc@oxy.edu)

Lance Durban writes "Well OK, I'll voice my belief that the the data
were invented.
Furthermore, although someone could redo the study and demonstrate that
the results cannot be reproduced (ie. supporting my suspicion),  this
is probably not an undertaking that Lancet would finance.  The cost of
another survey plus the discomfort of scraping egg off one's face will
probably limit Lancet's admission to regretting the unfortunate
Kolbe/Duff name obfuscation, an obfuscation that I imagine was
perpetrated quite intentionally by Ms Kolbe/Duff."

I guess anybody can decide that they think the data are invented but
that's exactly what peer review is designed to detect and prevent.  To
seriously believe the data are invented, one's sense of coordinated
dishonesty would have to throw the entire journal and its peer reviewers
under a pall of doubt.  Some on this list have indeed implied that the
Lancet itself is a biased outlet, and I suppose it's possible that the
Lancet just happens to have a stable of reviewers who care enough about
a particular view of Haiti that they would overlook invented data.  But
academia doesn't work that way, if only because everyone is so cravenly
interested in their own reputations that they are watching their asses
way too vigorously.  Yes, mistakes do happen, but rarely would they be
so obvious or simplistic as the sort of statistical miscalculation being
speculated about on the list. It is clear from the article itself that
peer review was rigorous; otherwise all the caveats in the final
sections would be much less comprehensive.

That said, we've discussed pretty comprehensively some of the study's
flaws -- but after a tiny bit of comparative research, the sexual
assault numbers aren't particularly high.  In one study of Los Angeles
(where I live) it was estimated that 10.5% of the population has
experience with assault.  Translating those numbers to Port au Prince,
where the population is topping 8 million, that would mean the Lancet
study's numbers are quite low on that score since 10.5% would be
something over 85,000, not 35,000.  Of course it's not really viable to
just move a Los Angeles statistic to Port au Prince, but as they say, it
doesn't take a rocket scientist to guess that sexual assault incidence
is probably at least as high in Port au Prince than Los Angeles, if not
higher.

As for the Lancet redoing the study, that is not how peer review works
either.  Journals do not conduct studies, scientists do.  For journals
to conduct and publish their own studies is a clear conflict of interest
that taints the results from the get go. If scientists think the
Kolbe/Duff study is wrong, there might be some who choose to attempt to
replicate the study.