st: Bootstrapping with tpm

[Pooja Stata Listserve <pdesai1387@gmail.com>]

Hello All,

I am carrying out a two part regression using the new command -tpm-. The dependent variable is cost and there are several categorical and continuous independent variables. The first part is a logit model which identifies if the cost is zero or non-zero and the second part is a gamma regression with a log link for the positive outcomes. I used the -vce(bootstrap)- option. However, when the bootstrap runs, there are several red x's (almost 40%) instead of dots and when I used the noisily option in bootstrap, I got the following error:

collinearity in replicate sample is not the same as the full sample, posting missing values

My code is as follows:
tpm othdol2 ps i.case ageatindex i.male i.ethnicity2 npcomorbids psychcomorbs psychvst1 npsyvst1 poffvst1 pervst1 noffvst1 nervst1 i.speccode2 conpstonly i.preconfirm pothvst1, f(logit) s(glm, fam (gamma) link (log)) eform vce(bootstrap, reps(100) noisily)

What exactly does the error mean? Does it invalidate the results? Is there a way I can overcome this?

Thanks a lot,
Pooja Desai
Doctoral Candidate
The University of Texas at Austin


*
*   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/