st: How to compute with Stata probabilities by conditioning ?

["Tiago V. Pereira" <tiago.pereira@mbe.bio.br>]

Many thanks, Roger! That makes sense.

Cheers!

Tiago


If you type, in Stata,

help density_functions##normal

then you will find that the function binormal(h,k,r) computes the joint
cumulative distribution function, not the conditional diatribution
function as you seem to believe. The .pdf documentatio gives the equation
explicitly.


I hope this helps.

Best wishes

Roger
--
Dear statalisters,

Imagine two standard normal variables, X and Y, with known correlation, rho.

P(X>t1|Y>t2) can be computed with

binormal(t1,t2,rho)

What if we were interest in computing

P(X>t1 and Y>t2) ?

Since X and Y are correlated, I am a bit confused.

I am using brute force (dumb way):

set obs 1000000000
matrix M = (1,rho \ rho, 1)
drawnorm x y, means(0 0) sds(1 1) corr(M)
count if x>t1&y>t2

P(X>t1 and Y>t2)  = r(N)/_N

Is there a smarter way of computing probabilities by conditioning (with
Stata)?

Thanks in advance.

Tiago





*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/