Tomek:
I don't quite understand your problem. You are testing (or at least would
like to test) for the multinomial distribution and you are talking about
normal errors. What is the relationship to the errors? and why would one
need any assumption on normal errors in order to perform a Chi-Squared
test in the spirit of Pearson or Fisher (when the parameters of the
distribution are estimated from the sample)?
Maybe by answering these questions you will solve your problem.
Janusz.
** Janusz Kawczak **
** UNC at Charlotte, Department of Mathematics, Room 350F **
** 9201 University City Blvd. **
** Charlotte, NC, 28223-0001, U.S.A. **
** Tel.: (704) 687-2566 (W) (704) 921-0273 (H) Fax.: (704) 687-6415 **
On Tue, 23 Apr 2002, Tomek wrote:
tlr25>Hi,
tlr25>
tlr25>I want to perform goodness of fit test for multinominal distribution.
The
tlr25>easiest way would be to use Chi2, but the measurment errors are not
tlr25>normally distributed. I thought about using some bootstraping method to
tlr25>perform the analysis. How can I do it in R?
tlr25>
tlr25>Tomek
tlr25>
tlr25>-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
tlr25>r-help mailing list -- Read
http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
tlr25>Send "info", "help", or "[un]subscribe"
tlr25>(in the "body", not the subject !) To: r-help-request at
stat.math.ethz.ch
tlr25>_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._
tlr25>
-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !) To: r-help-request at
stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._