I was corrected off-forum for my mistake below
regarding the distribution of U: I should have said that it
is chi-squared with k degrees of freedom, not standard
normal. Thanks for the correction.

On Sun, 10 Aug 2008, Brian Schott wrote:

+       T_k = Z/sqrt(U/k) where Z is standard normal
+ (normalrand) and U is a statistically independent uniform
+ (?0) , I believe. So, isn't this the way to generate a
+ student's t?
+
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm

Reply via email to