Re: st: Obtain p-values for GEE models after bootstrap

[Maarten Buis <maartenlbuis@gmail.com>]

On Wed, Jan 22, 2014 at 9:39 AM, Lan Cliff wrote:
> I'm trying to obtain bias-corrected p-values for GEE regression with
> bootstrap estimation. I used to obtain p-values after bootstrap by
> following steps outlined by Buis
> (http://www.stata-journal.com/article.html?article=st0137), using the
> commands:
> local t = _b[var1]/_se[var1]
> di 2*ttail(e(df_r),abs(`t´))
> But this approach failed after GEE because stata does not save the
> residual degree of freedom as e(df_r) for GEE regression. So is there
> a way to obtain p-values after using bootstrap for a GEE regression in
> Stata?

In some situations you can use the bootstrap to compute p-values, but
not via the method discussed in my article. See for example chapter 4
of (Davison and Hinkley 1997). It is probably much easier to just
report the confidence intervals instead.

Hope this helps,
Maarten

Davison, A.C. and Hinkley, D.V. 1997. Bootstrap methods and their
application. Cambridge: Cambridge University Press.

---------------------------------
Maarten L. Buis
WZB
Reichpietschufer 50
10785 Berlin
Germany

http://www.maartenbuis.nl
---------------------------------

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/