[S] Difficult newbie question

Marc Feldesman (feldesmanm@pdx.edu)
Thu, 19 Feb 1998 14:41:14 -0800


I've picked a tough problem to hone my skills in S-Plus. I published a
paper several years ago with a statistician in which we used approximate
randomization to put empirical confidence limits on the F-statistic from a
one-way ANOVA based on samples where the assignment of individuals to one
of three groups was not certain from the outset. We used STATA's
programming language and bootstrapping primitives to achieve this. It
turned out to be fairly easy to program because Stata's language allowed
access to all the various pieces of the anova, including the F-statistic.

I'm trying to replicate the analysis using S-Plus. I can't seem to figure
out at this point (and I'm only three days new to S-Plus, so this probably
isn't a surprise) how let the bootstrap function know that the statistic I
want to bootstrap is the F. If I supply the aov.obj to the anova()
function, the anova.object has the F Ratio as a column vector, but the
bootstrap function claims that it the statistic doesn't evaluate to a
vector or matrix. I may be in way over my head on this problem at this
stage in my S-Plus learning, but I thought it would be worth a try.

If anyone has any counsel whether this is a doable approach to the problem,
I'd be grateful. If anyone has an alternative approach to the problem I'm
all ears. I learn best by plunging in head first.

Thank you for any help you can offer. This group is one of the best I've
encountered in terms of the speed of response and quality of advice.

Best wishes and thanks,

Marc Feldesman
-----------------------------------------------------------------------
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