TFCE statistics for CAT12

Hello everyone. I am working on a project. Subject group includes healthy controls (n=50), presymptomatic (n=70) and symptomatic (n=18). I used CAT12 to obtain cortical thickness in these patients. I used TFCE values for intergroup comparison. According to the results I obtained, while there was no difference between the symptomatic group and the control group and between the presymptomatic and control groups in some areas, there was a difference between the presymptomatic group and the symptomatic group. How can we explain this? Does patient numbers affect TFCE values?


I’ll be cautious here because I do not know the nature of your data, but it could be that one group was below controls, albeit nonsignificanlty, and the other was above controls, albeit nonsignificiantly. You would need to explore the data more to see. Although, you should have begun with an ANOVA to see if any difference exists between the groups, before beginning the pairwise comparisons.

Yes, usually. In a basic sense, TFCE judges significance by weighting the magnitude test coefficient by it’s spatial extent. In other words, a small cluster with really high test coefficients will be significant, as well as a large cluster with only moderate test coefficients. With more participants, there is usually a lower variance in the observation distribution, so this can lead to higher test coefficients for between-group comparisons.



Thank you for your response Steven.