Sunday October 24, 2010:

I am reading a paper on sample size in a clinical study relating to bioequivalence. The values of teh paramater compared are on logarithms and it is a multiplicative model.

We have data from an ANOVA,. The residual variance is (sigma)^2 and

the CV= Squrt [exp^(sigma)-1]....this is equation 1

The paper then says The above relationship (equation 1) corresponds to to the relationship *(equation 2)

(sigma)^2 =Ln(1+CV^2).......equation.

Then the paper says : When the square of the CV is small then:

Ln(1+CV^2)= cv^2 holds.

Can this work be explained to me so I can undertstand it better.

Angus