Re: Chi-square statistic
- From: hrubin@xxxxxxxxxxxxxxxxxxxx (Herman Rubin)
- Date: 3 Jun 2007 17:56:18 -0400
In article <1180802429.817791.151170@xxxxxxxxxxxxxxxxxxxxxxxxxxxx>,
dkw <dwinsemius@xxxxxxxxxxx> wrote:
On Jun 1, 6:59 pm, Dorothy Middleton <cour...@xxxxxxxxxxxxxxxxxxxxx>
wrote:
The chi-square statistic obtained from a contingency table is unlikely
to have the chi-square distribution.
Can you explain why that would be?
It has APPROXIMATELY a chi-squared distribution if
the sample size is large. The rate of approximation
is approximately the reciprocal of the smallest
expected cell size.
Suggest you use an exact test
based on the hypergeometric distribution instead.
Wasn't the OP asking a question about sampling from a beta
distribution (continuous) and his results with a X^2 GoF test?
The chi-squared test of GoF from a continuous distribution
has very poor power, and does not have a chi-squared
distribution asymptotically if parameters are estimated.
--
This address is for information only. I do not claim that these views
are those of the Statistics Department or of Purdue University.
Herman Rubin, Department of Statistics, Purdue University
hrubin@xxxxxxxxxxxxxxx Phone: (765)494-6054 FAX: (765)494-0558
.
- Follow-Ups:
- Re: Chi-square statistic
- From: dkw
- Re: Chi-square statistic
- From: Bruce Weaver
- Re: Chi-square statistic
- References:
- Re: Chi-square statistic
- From: Dorothy Middleton
- Re: Chi-square statistic
- From: dkw
- Re: Chi-square statistic
- Prev by Date: Re: logistic regression on spss
- Next by Date: Re: logistic regression on spss
- Previous by thread: Re: Chi-square statistic
- Next by thread: Re: Chi-square statistic
- Index(es):
Relevant Pages
|