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 at the end of May, and its replacement, statalist.org is already up and running.


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

Re: st: too good to be true : lr test in mlogit?


From   John Litfiba <cariboupad@gmx.fr>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: too good to be true : lr test in mlogit?
Date   Fri, 13 May 2011 11:21:08 +0200

Thank you very much, Maarten,

So... everything is fine it seems ;-)
Well except for the fact that I assume independence in a panel ...
with multiple measurements for the same individuals... ;-)
But after various tentatives I think that it is not computationally
feasible (at least on my PC) to fit a random or fixed effect logit on
such a large dataset

Have a good day
Best Regards,

On 13 May 2011 10:43, Maarten Buis <maartenlbuis@gmail.com> wrote:
> On Fri, May 13, 2011 at 10:12 AM, John Litfiba <cariboupad@gmx.fr> wrote:
>> 1) It is normal to obtain such a negative log likelihood when we use
>> very large sample, right ? (n=2 millions here)
>
> Yes, that is perfectly normal. It is the log of the probability of
> observing the data given the parameters. This probability is the
> product of the probabilities of observing each individual given the
> parameters (assuming independent sampling...). A probability is a
> number between 0 and 1, so the product of two such probabilities
> results in a smaller number than either of the individual
> probabilities. Do that multiplication 2 million times and you end up
> with an extremely tiny number indeed. Take the log of such a tiny
> number and you end up with a very negative number.
>
>> 2) if the association (for example given by tabulation) show that
>> there is strong association between yvar and xvar1 then it is
>> plausible to obtain this fastastic LR statistic of... 140000 ??
>
> I would not call that fantastic. It just test whether the effects of
> all covariates are all equal to 0, so you would expect that with such
> a large sample size you would thoroughly reject that hypothesis, which
> you do.
>
> Hope this helps,
> Maarten
>
> --------------------------
> Maarten L. Buis
> Institut fuer Soziologie
> Universitaet Tuebingen
> Wilhelmstrasse 36
> 72074 Tuebingen
> Germany
>
>
> http://www.maartenbuis.nl
> --------------------------
> *
> *   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/
>

*
*   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