Bookmark and Share

Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down on April 23, and its replacement, statalist.org is already up and running.


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: st: Fit chi2 as in gammafit


From   Nick Cox <njcoxstata@gmail.com>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: Fit chi2 as in gammafit
Date   Wed, 31 Aug 2011 08:58:48 +0100

I agree with everyone.

Nick

On Wed, Aug 31, 2011 at 8:46 AM, Tirthankar Chakravarty
<tirthankar.chakravarty@gmail.com> wrote:
> Maarten,
>
> That is how I thought about it too, but for the consistency of the
> parameters, and hence for the fit, results of pseudo-maximum
> likelihood theory indicate that it doesn't matter which member of the
> one-parameter linear exponential family is used (so the Poisson, as in
> my example, will do just as well) for parametric fit (as long as the
> mean is correctly parameterised), and so it is probably best not to
> mess too much with the definition and folklore of the
> chi2-distribution in deference to the conservatives Nick alludes to.
>
> T
>
> On Wed, Aug 31, 2011 at 12:19 AM, Maarten Buis <maartenlbuis@gmail.com> wrote:
>> On Tue, Aug 30, 2011 at 10:19 PM, Nick Cox wrote:
>>> Ho hum. A conservative might want the free parameter to be integer-valued.
>>
>> That pretty much depends on how you think about the Chi square
>> distribution. If you think of it as the sum of nu squared independent
>> standard normal variables than it makes sense to constrain nu to be an
>> integer. If you think of it as just a skewed distribution with mean nu
>> (and variance 2*nu) than nu can be any positive number.
>>
*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/


© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   Site index