st: bootstrap p-value and predictions

[Shehzad Ali <drshehzad_ali@yahoo.com>]

Hi listers,

I have a basic question regarding bootstrap.


For instance, you are running linear regression with the -bootstrap- procedure to estimate bootstrap standard errors and p-values. In that process you also predict outcome for each bootstrap
sample using actual values of covariates but varying only the treatment
variable (setting equal to 0 and then 1) to get the difference between treatment 1 and treatment 0 for each bootstrap sample. After say 10,000 iterations, you get the bootstrap SE and p-values and also 10,000 predictions for the difference between treatment 1 and treatment 0. 


My question is: based on the SE and p-value for the treatment parameter, would you be able to estimate the proportion of the predictions from the bootstrap samples that should be above and below zero for the difference between outcomes in the treatments?


Thanks

Shehzad

*
*   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/