RE: st: RE: estimate 95%CIs of the mean for different sample sizes

["Nick Cox" <n.j.cox@durham.ac.uk>]

If you bootstrap you have to invent extra methodology for contracting or
expanding to different sample sizes other than that in hand. The results
you get then conflate side-effects or variability associated with that
methodology with those of bootstrapping. That doesn't sound crazy, but
it sounds like an elaborate exercise for what seems like a fairly simple
question. Also, I think you would have to explain in a bit of detail
what you're doing. All sounds a bit over the top to me, but it's your
problem. 

With -cii- you could say I have this mean and standard deviation; what's
the effect of changing the sample size? In fact, you don't need -cii- to
do that, as it's an elementary calculation; but if a calculator would
serve in place of -cii-, then so also -cii- would serve in place of a
calculator. 

Nick 
n.j.cox@durham.ac.uk 

Miranda Kim

I had in mind something like bootstrap originally, but am not sure what
method is most appropriate in this case. The purpose is to see what
precision is obtained for different sample sizes.

Nick Cox 

> -cii-

Miranda Kim

> I have data for a sample of 300 which is approximately normally
> distributed (slightly skewed). I would like to derive 95% confidence
> intervals around the estimate of the mean for sample sizes of 50, 80,
> 100, 150, and 400.
> Does anyone have any suggestions what command I can use to do this?

*
*   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/