help > Post-hoc analyses for ANOVA
Showing 1-2 of 2 posts
Display:
Results per page:
Jun 2, 2024  07:06 PM | Ieva Andrulyte
Post-hoc analyses for ANOVA

Dear Andrew,


Thank you for creating this software; it has been great.


I have used NBS to identify networks that differ between three laterality groups using ANOVA. I found one significant network comprising 29 edges and 29 nodes. To understand the group differences, I then looked at the significant connections at the subject level and conducted 3 separate t-tests on the 29 significant connections using MATLAB, correcting for multiple comparisons to better understand which groups drove the differences. The results make sense, but I am wondering if this kind of post hoc analysis is appropriate for NBS, or would you suggest running 3 separate t-tests on the whole brain using NBS?


Many thanks for your help!


Ieva

Jun 3, 2024  12:06 AM | Andrew Zalesky
RE: Post-hoc analyses for ANOVA

Hi Leva,


your approach seems quite reasonable to me. Post-hoc t-tests will provide insight into the direction of the effects and these could be reported in your paper. 


Running 3 separate t-tests using the NBS is also a possibility, but your current approach is probably the most sensible. 


Andrew


Originally posted by Ieva Andrulyte:



Dear Andrew,


Thank you for creating this software; it has been great.


I have used NBS to identify networks that differ between three laterality groups using ANOVA. I found one significant network comprising 29 edges and 29 nodes. To understand the group differences, I then looked at the significant connections at the subject level and conducted 3 separate t-tests on the 29 significant connections using MATLAB, correcting for multiple comparisons to better understand which groups drove the differences. The results make sense, but I am wondering if this kind of post hoc analysis is appropriate for NBS, or would you suggest running 3 separate t-tests on the whole brain using NBS?


Many thanks for your help!


Ieva