09-17-2015 04:59 AM
I ran a PROC TTEST for comparison of two groups, from the Equality of Variance test the p-value is far less then the 0.05, hence, we can say these two samples are heterogeneous of variance, in this case, why the CV is not calculated by Satterthwaite method, but still by the pooled method ? How to interprest such CV? Thanks.
The data is lognormal distributed.
09-17-2015 04:14 PM - edited 09-17-2015 04:14 PM
I would ignore the reported CV in this case.
If you really need a CV-like statistic in this case, compute it using the mean and variance of each group.