To all: I'm doing linear regression with the clinical trial as the unit of analysis and want to weight the regression by the trial size (no of patients enrolled) because, next to having individual trial variances which I don't, it is the best factor to reflect the strength of the evidence from that trial. I then want to calculate the confidence interval for the model, which is straightforward, and the prediction interval for predicting y as a function of a particular trial with a particular value of x. Stata appropriately says this is "not well-defined" as it will depend on the size of the trial for which one is interesting in predicting. The prediction interval will be narrower for a 10000 patient trial compared to that for a 100 patient trial. (The prediction interval also will vary with the number of trials in the regression, but in my work this doesn't change). I'm told SAS does this, first asking for the particular trial one is interesting in predicting. Does anyone kn
ow how to do this in Stata?
I would most appreciative of any help here.
Thanks
Kent Johnson
University of Newcastle, AU
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
