Hi Mayron,

Typically when referring to "effect-sizes" we do not mean the statistics (e.g. T- values) but rather the sizes of the effect-of-interest in our analysis. For example, if you were looking for group-differences corrected by age and site, the effect-sizes in these analysis would refer to the average differences in connectivity between the two groups when compared at the same level of age&site. In CONN those can be directly obtained form the results explorer GUI by clicking on the "Plot effects" button (the button with a bar-plot image). That will compute and display the average connectivity within each individual cluster in each of your two groups (the "adjusted means", i.e. computed at the zero-level of your covariates), which, if I am interpreting correctly, is what the reviewer is likely requesting.

The only thing to keep in mind in these "effect-size" plots is that, even if the statistics are not, these displays are sensitive to your choice of zero-levels of your control covariates, so I would suggest to (if they are not already) mean-center your age and site covariates before creating these displays so that these displays show you the average connectivity within each group at the average age/site level across your sample

That said, if you really prefer or need to compute the average T- statistics across all of the voxels within each cluster you can do so for example using rex and selecting within your second-level folder the results.ROIs.nii file as ROIs and the spmT_0001.nii file as Sources, then switch the 'ROI-level' option to 'cluster-level' and clicking on 'extract' to compute those values. 

Hope this helps
Alfonso

Originally posted by Mayron Pereira Picolo Ribeiro: