Re: Comparing two variation coefficients

Dear All,

What test (possibly non -parametric) can be used to
test
the differences between two variation coefficient
(std dev / mean) ?

Thanks

Fredo


If the two distributions are normal, then the sample CVs (coefficients of variation) are nearly inversely proportional to noncentral t's. This is provided that the Prob(xbar < 0) are negligible.

The problem then reduces to the ratio of the noncentrality parameters being proportional to the ratio of the square-root of the sample sizes.

Most approximations to the noncentral t approximate s by a normal distribution, so that xbar + Ks is approximately normal. This requires that the sample sizes be of at least moderate size.

From that point on, you can come up with an approximate test for the equality of two CVs.

Jack
.

Loading