[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]
Re: st: vce(boot) - clustering?
Scott Cunningham <firstname.lastname@example.org> asks about documentation for the
> I'm estimating a Poisson with fixed effects model using -xtpoisson-.
> To correct for overdispersion, I am using -vce(boot)- to bootstrap
> the standard errors. Where can I find a description of the algebra
> used for -vce(boot)- when attached to the -xtpoisson- command? I
> tried -help vce(boot)- and -help xtpoisson- but didn't see anything.
> Also, am I correct that -xi: xtpoisson depvar indepvar, fe i(id) vce
> (boot)- will create robust and clustered standard errors?
The -vce()- option is documented in -[R] vce_option-; type
. help vce option
in Stata to see the corresponding online help.
This will point you to -[R] bootstrap- for the background documentation on
When -xtpoisson- is called with the -vce(bootstrap)- option, it automatically
clusters on panel. Your original command
xi: xtpoisson depvar indepvar, fe i(id) vce(boot)
will fit an -xtpoisson- model using the bootstrap resampling method for
variance estimation, where the bootstrap samples are composed from resampling
the panels with replacement (as opposed to resampling observations). I'm
assuming you are using the 'i.' operator somewhere in 'indepvar' above;
otherwise, -xi:- in not necessary.
For reproducibility, don't forget to use -set seed- before calling -xtpoisson-
with the -vce(bootstrap)- option.
If you are concerned about overdispersion, there is also -xtnbreg-.
* For searches and help try: