topic Bootstrap precedure in SAS Procedures

Bootstrap precedure

deleted_user — Thu, 13 May 2010 03:51:44 GMT

Dear SAS users:

I'd like to get some help about the bootstrap of a data set.
The data set is like following:

ID, OBS, TIME, HEIGHT, WEIGHT
5001, 10, 0.1, 180, 71
5001, 3, 1, 180, 71
5001, 1, 2, 180, 71
5003, 15, 0.1, 170, 60
5003, 3, 1, 170, 60
5003, 2, 2, 170, 60
5004
5004
5004
.
.
.

The total number of subjects is 36 with ID numbered from 5001 to 5050 (some ID number was missing). I'd like to bootstrap the data set by subject ID to form a new data set containing 100 subjects with ID numbered from 1 to 100.

Thanks a lot for help and have a nice day!

Re: Bootstrap precedure

Doc_Duke — Thu, 13 May 2010 20:33:07 GMT

Go to support.sas.com and search for bootstrap to get lots of guidance on it's use.

Usually, bootstrapping does sampling with replacement to get 'sample' that is the same size as the original. I'm concerned that changing the sample size will artificially reduce the variability and bias the inference. Efron & Tibsurani (An introduction to the bootstrap) is a classic reference.

Re: Bootstrap precedure

deleted_user — Tue, 18 May 2010 00:24:05 GMT

Hi Doc@Duke:

Thanks for the good point.