Thanks! I do indeed have strong skewness, with no zeros or ones, so I
will try the logits.
On Thu, Sep 4, 2008 at 10:07 AM, Nick Cox <[email protected]> wrote:
> Austin Nichols has kindly pointed out that what I wrote below is not
> correct as a summary of what was in the article. See the article itself
> for the story. Thanks to Austin for the quick fix.
>
> -----Original Message-----
> From: Nick Cox
> Sent: 04 September 2008 17:50
> To: '[email protected]'
> Subject: FW: st: RE: Dependent var is a proportion, with large spike in
> .95+
>
> A formal statement can be found within
>
> http://www.stata-journal.com/sjpdf.html?articlenum=gr0010
>
> together with explicit code for exploring the question graphically.
> Whatever the parameters, the result of a logit transform of a beta
> distribution is a logistic distribution, which is bell-shaped.
>
> Of course, real datasets might not be so well behaved, as I presume all
> agree.
>
> Nick Cox
>
> Jverkuilen
>
> #A good approximation is if that you take #logits of a beta-distributed
> #variable, the distribution looks bell-
> #shaped. That's true even for
> #highly skewed betas with modes near 0 #or near 1.
>
> Yes, so long as the distribution is not J- or L-shaped, which can happen
> with the beta. It can handle those shapes and endpoint bimodality too.
>
>
> *
> * 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/
