But I think the least unsatisfactory options
are
1. To omit zeros and to indicate them by a rug of ticks
on the other axis.
2. To plot downward-pointing arrows at say
log(0.5). Whatever constant is used should
be less than the smallest positive value observed.
Nick
[email protected]
> -----Original Message-----
> From: [email protected]
> [mailto:[email protected]]On Behalf Of Nick Cox
> Sent: 08 October 2007 21:09
> To: [email protected]
> Subject: RE: st: RE: GLLAMM error: log-likelihood cannot be computed
>
>
> So, a log of 0 sometimes means that the data are 1
> and sometimes that they are 0?
>
> There's no neat solution to this one.
>
> Nick
> [email protected]
>
> Leny Mathew
>
> > Thanks Nick. For the purposes of the graph, I created a new variable
> > with the zeros changed to 1 and then took the log;
> effectively setting
> > them as zero in the log graph. I guess I could scale the
> variable by a
> > very small value and then take the log also.
> >
> > On 10/8/07, Nick Cox <[email protected]> wrote:
> > > -gllamm- I leave to experts on it.
> > >
> > > -glm- produces predictions on the scale of the response,
> > > whatever the link. It can also be quite sensible to use a
> > > log scale for subsequent graphing. Indeed I've found
> > > log link and log scale for graphs invaluable in some cases.
> > > The results are not equivalent to transforming the response
> > > because the log of the mean is not in general the mean
> > > of the logs (and similarly for any nonlinear transformation).
> > >
> > > However, you can't show zeros on a log scale. If you
> > > try this, Stata just gives you a dopey graph. That's
> > > its way of saying "Isn't that rather a silly thing
> > > to ask for?"
*
* 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/
