Re: st: How to perform GLM with nested observations

[Jeph Herrin <junk@spandrel.net>]

Yes, I had typed -xtmepoisson- but then had this memory
of seeing 1m+ bats pour out of a cave in Thailand once
and thought maybe it would be ridiculous to assume that
bat counts were small integers. So I edited out the
mention.

thanks,
Jeph


Austin Nichols wrote:
> Jeph--
> With a depvar that is a count, I think you would prefer -xtmepoisson-
> but the cluster option on -glm- or -poisson- is the place to start,
> and with 500 sites there are no worries about insufficient clusters.
>
> On Wed, Jul 23, 2008 at 12:33 PM, Jeph Herrin <junk@spandrel.net> wrote:
> > See -xtmixed-
> >
> > K.Boughey@uea.ac.uk wrote:
> > > Dear List,
> > >
> > > I have 500 sites, and each site contains twelve transects.
> > >
> > > My dependent variable is the number of bats encountered along each
> > > transect.
> > >
> > > My independent variables have been measured at either transect or site
> > > level and include:
> > > habitat type (categorical) measured at transect level
> > > distance to nearest water feature measured at transect level
> > > proportion of woodland, measured at site level
> > > density of water features, measured at site level
> > >
> > > Is there a way I can include all 6000 observations and all independent
> > > variables (both categorical and continuous and at both scales) in a
> > > GLM-type analysis that, critically, allows for the lack of independence
> > > between observations within the same site, caused by the nested sample
> > > design?
> *
> *   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/