st: means when one variable has associated uncertainty

[rmhaemh@ucl.ac.uk]

> Dear All,
> This may well have a very simple answer, but I haven't come across it in 
> my field before. I have one variable which is measured on one occasion 
> with certainty for every individual in a study, and another variable which 
> is not known precisely for most individuals, but for which there are 
> probabilities of the different possibilities assigned.
> i.e.
> studynumber     var1    var2    probability of var2
> 1               9.8     a       0.5
> 1               9.8     b       0.25
> 1               9.8     c       0.25
> 2               5.7     a       0.75
> 2               5.7     c       0.25
> 3               4.9     a       1.0
> etc...
> e.g. for studynumber 2, variable 1 was measured as 5.7, and variable 2 may 
> be "a" w.p. 3/4 or  "c" w.p. 1/4
> I want to calculate an unbiased estimate of the mean and sd for variable 1 
> by  variable2 (considering information from all individuals) and also 
> calculate whether or not the null hypothesis that var1 doesn't differ by 
> var2 can be rejected.
> Any help would be greatly appreciated,
> Many thanks
> Emma Hawe

*
*   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/