Stata The Stata listserver
[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

Re: st: discrepancy in -sdtest- ?


From   rgutierrez@stata.com (Roberto G. Gutierrez, StataCorp)
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: discrepancy in -sdtest- ?
Date   Wed, 24 Nov 2004 15:26:01 -0600

PAUL VISINTAINER <PAUL_VISINTAINER@nymc.edu> asks:

> One of my faculty brought this discrepancy to my attention.  Why is the
> 2-sided test of significance (.8004) not equal to twice the one-sided test?
> (e.g., .3749 * 2 = .7498).  She tells me that SAS gives a two-sided p-value
> of .7498.

Stata's calculation for the signficance level for the two--tailed test is 
(for F_obs > 1, say)

        sig. level = Pr(F < 1/F_obs) + Pr(F > F_obs)                   (1)

Contrast this with the equal-tailed calculation

        sig. level = 2*Pr(F > F_obs)				       (2)

Although tests based on (1) and (2) both have correct Type I error, our
simulations show that (2) generally has better power.

As a result, we will change -sdtest- to calculate (2) instead of (1).

The change will be available in the next ado update.  

--Bobby
rgutierrez@stata.com
*
*   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/



© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index