Re: [S] "Sums of squares", the taxonomy issue again.

Alan Zaslavsky (zaslavsk@hcp.med.harvard.edu)
Tue, 28 Apr 1998 21:59:41 -0400 (EDT)


I suspect that the government regulations that Bill is talking about are
FDA regulations for analysis of clinical trials required for licensure
of new drugs or treatments. These procedures are quite routinized, as
indeed they must be given the powerful interest that manufacturers of
devices and drugs have in tailoring their analyses to produce desired
outcomes for their products. The procedures that are recognized here are
fairly conservative but not immutable to change. (Somebody whose work is
closer to this area might be able to be more specific about how this
works.)

These standards have a powerful effect on the statistical tools that are
required by the pharmaceutical industry and therefore on the market for
statistical software, but there is certainly no compulsion to conform. If
there were, S would not have become a popular language on the strength of
having nice functions for graphics and oddball analyses like clustering
and the knapsack problem. However, as S-Plus moves into larger commercial
markets it does tend to try to fill some of the same conventional needs as
other packages, and this is the sense in which there is a kind of
standardization imposed.

Changes in the standard analyses, when they are needed, will come about
through the efforts of statisticians working on and before the review
boards that set the standards, not through principled noncompliance by
software vendors! However, Bill, your good work on the software front
certainly plays a part in paving the way (as will that of Frank, whose
frustration is showing in the resolute way he has his tongue jammed into
his cheek).

-----------------------------------------------------------------------
This message was distributed by s-news@wubios.wustl.edu. To unsubscribe
send e-mail to s-news-request@wubios.wustl.edu with the BODY of the
message: unsubscribe s-news