Search All of the Math Forum:

Views expressed in these public forums are not endorsed by Drexel University or The Math Forum.

Topic: simple question
Replies: 11   Last Post: Jul 6, 2012 3:54 PM

 Messages: [ Previous | Next ]
 Luis A. Afonso Posts: 4,274 From: LIsbon (Portugal) Registered: 2/16/05
Re: simple question
Posted: Jul 4, 2012 6:20 PM

HOWEVER
Though we ignore the Populations std deviations values X~N(muX, sigmaX): nX____ Y~N(muY, sigmaY)
(BUT ASSUME THEY ARE EQUAL) we use:
t0 = [(meanX - meanY) - d0] / sqrt( sp*(1/n+1/m))
sp^2 = [(n-1) * sX^2 + (m-1) * sY^2]/(m+n-2)
and sX , sY the std deviation estimators from data, meanX,
meanY the sample mean values.
The condition H0: muX - muY > d0 with probability greater than 1 - alpha leads to the no rejection condition
______________t0 > t(1-alpha),
Student t distribution, n+m-2 degrees of freedom, right tail probability alpha.
When sigmaX and sigmaY are not equal we get the Behrens-
Fisher problem to which there is no finite solution (as long I notice)

Luis A. Afonso

Date Subject Author
7/4/12 juicymixx@gmail.com
7/4/12 Stephen Wolstenholme
7/4/12 Art Kendall
7/4/12 Bruce Weaver
7/4/12 Herman Rubin
7/4/12 Richard Ulrich
7/5/12 Herman Rubin
7/5/12 Richard Ulrich
7/6/12 Herman Rubin
7/6/12 Gaj Vidmar
7/6/12 Herman Rubin
7/4/12 Luis A. Afonso