>> I'm sorrry that this posting earlier got truncated. What I would like to
> find is a PC based program for carrying out Logit transformations. I
> assume this might be part of a more extensive package of statistical
> routines, but an inexpensive stand alone program would be fine too. Also
> any information about the use of this plot and/or pitfalls would be
> appreciated.
>> Phil Carl
Several statistical packages do logisic analysis if this is what you
want, and not just a simple transformation to a logit. A logit is
defined as m=log(p/1-p) where 0<=0<1. Logistitc analysis tries to fit
a linear model, say a+bx, to m, which means that the fitted value,
p=e^m/(1+e^m) ranges between 0 and 1. The fitting is iterative, since
there is no closed algorithm for the problem. Sorry about the verbosity,
but I wanted to maker sure we were on the same wavelength.
SAS, of course does logistic analysis, as does SPSS, and ECHIP. The
original program, and a very satistfactory one it still is, is GLIM
from NAG. (Note ECHIP is an experimental design and analysis program,
and not a general purpose statistical program as are the others.)
Bob Wheeler, ECHIP, Inc.
