Thanks. It is good to know the difference is caused by inclusion of extent. The issue, however, remains. I was hoping to compare effect sizes from TBSS with effect sizes obtained from other methods. If the effect sizes derived from the tstat files do not reflect the p-values everyone reports, comparison of effect sizes in turn does not truly reflect the differences in sensitivity/detectability of different methods. I realise that a fair comparison in this way is going to be tricky...
Sjoerd
|