More on Rocketdyne study

[Gary Schroeder <schroede@mail.sep.bnl.gov>]

>We have not gotten very far with our review of the Rocketdyne/AI worker
study,
>but it is evident that this study does not justify any immediate change in
our
>standards or practices.  I suspect that, for many people,  the presence of
Alice
>Stewart on the advisory panel and Douglas Crawford-Brown as a consultant
will be
>enough to relegate the report to the trash can, but we must look a little
deeper.

That may be true to people who know who Alice Stewart is, but to others,
some of the statements made in this report may appear to be credible.
Statemements such as 

"[the study] found that nine of the facility's workers died from cancers
attributable to external radiation exposures of 10 mSv or more" and 

"[the study allowed] researchers to detect elevated death rates from
cancers that had not before been associated with radiation exposures."

were made in a occupational safety publication which obviously drew some
very serious, if misguided conclusions from the report.  The average reader
probably dosen't know that specific cancers cannot be epidemiologically
attributed to radiation exposure at the 10 mSv level, and they are most
certainly unaware of the statistical shortcomings of the report.  

>My problem at the moment is in trying to determine how the investigators
found 1
>or 2 deaths in a group to be statistically significant using a 95%
confidence interval.

If you torture the numbers long enough...

=======================
Gary L. Schroeder
Brookhaven National Laboratory
Environmental Protection Office
gs1@bnl.gov