On Thu, 18 Nov 1999 13:25:11 +, "P.G.Hamer"
[EMAIL PROTECTED] wrote:
snip, about Chernoff faces ...
However for a quicky online introduction
http://www.statsoftinc.com/textbook/stathome.html
press `Graphical Techniques' button
press Chenoff faces' under the `nD/Icon graphs' heading
Peter,
I just now checked out that site, and I have added it to references in
my stats-FAQ -- there is quite a bit at the site you cite, and it is
put together attractively. But I do have a suggestion, or complaint.
(I will check to see if the site offers a mail address for such
things.)
A weakness, especially for students to use it, is that it is
not-at-all judgemental. For instance, it does not mention that
Chernoff faces really don't work. Or, that no one with experience
would have expected them to work, or would have any respect for them
today.
Its comments on "stepwise" regression or discriminant function are
similarly innocent and misleading, like the early documentation from
1970 when computer programs first came available. Textbooks have
been much more negative on it, in the last 10 years or more. Some
details are in the stepwise-FAQ within my stats-FAQ. (By 1985 or so,
there had been twenty years of stepwise prediction equations in the
social sciences that "failed to replicate"; that was noticed, even by
the folks who were not strong in math.)
--
Rich Ulrich, [EMAIL PROTECTED]
http://www.pitt.edu/~wpilib/index.html
