I guess that you're going to have to program
this yourself, taking responsibility for
how you calculate df and thus P-value.
As it happens, it seems that -chi2tail()-
works as one would hope for non-integer df.
Or would a log-linear model with
importance weights be equivalent
for your underlying scientific problem?
Nick
[email protected]
Pierre-Antoine Kremp
> I'm using a dataset with "importance weights": frequency weights
> multiplied by the sampling rate n/N -- so that the sum of those
> importance weights equals the sample size n, not the target population
> size N) and I'd like to do a Pearson chi-squared test on a 2-way
> contingency table. It shouldn't be too hard, but the chi2
> option of the
> tab command doesn't accept iweights. Only integer frequency
> weights seem
> to be allowed. I could very well use frequency weights
> instead, but that
> would increase the Chi2 stat unreasonably.
>
> Any idea on how to get round this problem?
>
> Thanks a lot (and sorry if the question is too trivial!)
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
