Help: Which test to use for comparing variability

#1
Hi:

I am wondering if I can ask a bit of help.

I have patients undergoing two different procedures (binary factor). One of the outcomes is is a continuous variable (range from 0 to 2, but heavily skewed to the right). I am only interested comparing the variability between the two groups, not the means.

What is the best test to use for comparing the variability between the two groups? Is Levene's test for homogeneity appropriate? Or is there something else more "standard" in the literature?

I am using SPSS.

Many thanks in advance for any help.

Joe
 

CB

Super Moderator
#2
Hiya,

What is the motivation for comparing the variances? Is this something you want to do because it's of intrinsic substantive interest, or as an assumption check for another analysis?

Levene's test assumes normal distributions of the variables, which may not be reasonable here. You could use a permutation test for equality of variances (anyone know of an R function for this?)
 
#3
Hiya,

What is the motivation for comparing the variances? Is this something you want to do because it's of intrinsic substantive interest, or as an assumption check for another analysis?

Levene's test assumes normal distributions of the variables, which may not be reasonable here. You could use a permutation test for equality of variances (anyone know of an R function for this?)
Hi: thanks for the reply.

I am comparing how two different surgical techniques affect an organ. This effect is used to plan surgery, and the less variable the effect the better, because we compensate for the effect by modifying our surgical planning. That's why the variability is much more important than the mean. Having a consistent effect helps to make the surgery precise and predictable.

If you can advise a test in SPSS it would be fantastic. It's been almost two decades since I last used Splus, R or SAS!

Many Thanks

Joe