Re: st: simulating datasets with certain characteristics

[Richard Williams <Richard.A.Williams.5@ND.edu>]

At 09:36 AM 4/25/2006, John Novak wrote:
> drawnorm will do it.  Suppose you wanted 100 observations of each, with
> a specified mean and correlation structure.  You could use:
>
> matrix means = (10, 30)
> matrix stdev = (5, 9)
> matrix cor = (1.0, 0.4 \ 0.4 1.0)
> drawnorm measure1 measure2, n(100) m(means) corr(cor) sds(stdev)
>
> This will sample 100 observations on measure1 and measure2 with means
> (approximately) 10 and 30, a correlation of approximately 0.4, and
> standard deviations of 5 and 9, respectively.
There is an important difference between corr2data and 
drawnorm.  With corr2data, the simulated data set will have EXACTLY 
the correlations, means, and standard deviations you specify.  It is 
like having the entire population in your data set.  With drawnorm, 
you are drawing a random sample from a population with the population 
parameters you specify.  Hence, because of sampling variability, the 
correlations etc. in your simulated sample will differ somewhat from 
the parameters you have specified.

Which do you want for your purposes?  If, say, you were trying to 
replicate published analyses, where the correlations etc. were 
included in the paper, you would want corr2data.  But if, say, you 
planned to draw a thousand samples from a hypothetical population, 
you would want drawnorm; otherwise if you used corr2data there would 
be zero sampling variability.  This might be handy if, say, you 
wanted to see how many times a hypothesis was correctly or 
incorrectly rejected.


-------------------------------------------
Richard Williams, Notre Dame Dept of Sociology
OFFICE: (574)631-6668, (574)631-6463
FAX:    (574)288-4373
HOME:   (574)289-5227
EMAIL:  Richard.A.Williams.5@ND.Edu
WWW (personal):    http://www.nd.edu/~rwilliam
WWW (department):    http://www.nd.edu/~soc 
*
*   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/