Search All of the Math Forum:
Views expressed in these public forums are not endorsed by
NCTM or The Math Forum.


cody
Posts:
2
Registered:
11/4/06


Re: Ftest for comparing variances?
Posted:
Nov 4, 2006 9:07 PM


I heard an entirely different answer on sci.stat.math,
"If the alternative hypothesis is twosided; that is, H1: siga N.E. sigb, then you can reject Ho if F is either too small or too large. You just compare your F statistic with the .025 and the .975 quantiles of the F distribution with the degrees of freedom you mentioned.
Jack"
It seems to contradict your claims?



