help
help > RE: Post-hoc tests for ANOVA
Apr 16, 2015 02:04 AM | Andrew Zalesky
RE: Post-hoc tests for ANOVA
Hi Corinna,
Apologies for the delayed response.
I suggest performing three t-tests independently, assessing for differences between each of the three pairs of groups. This is a very simple approach and creates a multiple comparisons problem, however, it should clearly reveal which groups are driving the effect.
Alternatively if the network comprises only a few edges, you may want to consider Tukey, etc, at each edge, but I suggest trying the above first.
You might also want to investigate specific connections comprising the larger network. See the manual for instructions about extracting connectivity values for particular connections.
Originally posted by Corinna Bauer:
Apologies for the delayed response.
I suggest performing three t-tests independently, assessing for differences between each of the three pairs of groups. This is a very simple approach and creates a multiple comparisons problem, however, it should clearly reveal which groups are driving the effect.
Alternatively if the network comprises only a few edges, you may want to consider Tukey, etc, at each edge, but I suggest trying the above first.
You might also want to investigate specific connections comprising the larger network. See the manual for instructions about extracting connectivity values for particular connections.
Originally posted by Corinna Bauer:
Dear All,
I was wondering if anyone has experience utilizing post-hoc tests for significant group differences when using ANOVA in NBS? I am comparing three different groups whereby each subject is set up in the design matrix as one of [1 0 0; 0 1 0; 0 0 1] and the contrast I used was [1 1 1]. This provides a large network, but I was wondering how best to go about doing post-hoc analyses?
Would it be best to do a series of t-tests on new matrices that only include two of the groups and repeat this three times for each of the possible subject group combinations?
Any advice would be appreciated!
Thanks
Corinna
I was wondering if anyone has experience utilizing post-hoc tests for significant group differences when using ANOVA in NBS? I am comparing three different groups whereby each subject is set up in the design matrix as one of [1 0 0; 0 1 0; 0 0 1] and the contrast I used was [1 1 1]. This provides a large network, but I was wondering how best to go about doing post-hoc analyses?
Would it be best to do a series of t-tests on new matrices that only include two of the groups and repeat this three times for each of the possible subject group combinations?
Any advice would be appreciated!
Thanks
Corinna
Threaded View
| Title | Author | Date |
|---|---|---|
| Corinna Bauer | Feb 10, 2015 | |
| Andrew Zalesky | Apr 16, 2015 | |