st: How to obtain-group-by-group statistics on stratified bootstrap created with -bsample-?

[Daniel Waxman <dan@amplecat.com>]

Dear Statalist,

This post is to ask how I can obtain group-by-group bootstrap
statistics on a file that is created by  -bsample, stratify(group)-
and therefore is in the form of 1 observation per group per
replication.  For example:

group 1, rep #1 adjusted outcome for group 1

group 1, rep #2 adjusted outcome for group 1

group 2: rep #1 adjusted outcome for group 2

group 2: rep #2 adjusted outcome for group 2


I would like to obtain 95% confidence intervals and achieved
significance level (p-value) for the difference between the each
group’s adjusted group-level outcome and the grand mean for all
patients for that outcome.    There are up to 10,000 groups, and a
highly variable number of patients per group.

The group-level adjusted outcomes are created by performing (on
stratified resamples) ordinary logistic regression according to the
formula “adjusted outcome = (group mean observed) – (group mean
predicted) + (grand patient-level mean).”  (I realize that there are
some issues with this adjustment formula but that part isn’t under my
control.)  The adjusted outcomes are then posted to a file after each
run.

I can see how to obtain percentile-based confidence intervals as
quantiles of the bootstrapped replications, but I’d also like to look
at bias-corrected CIs, and I’m wondering whether there is a way to use
Stata’s built-in routines for doing this, or if anybody can tell me
how to obtain them manually.

Thanks for considering,

Dan

(Stata 13)

*
*   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/