help
help > RE: How to set the parameters
Mar 31, 2016 08:03 PM | Alfonso Nieto-Castanon - Boston University
RE: How to set the parameters
Dear Lin,
Regarding question (1), you are exactly right, if you want to look at an F-test of main group effect (i.e. find those regions where the connectivity differs between groups) you may choose the contrast labeled 'any difference between gA&gB&gC' (or equivalently, enter a between-subjects contrast [-1 1 0; 0 -1 1]). For any significant cluster that you find in these analyses you are right that typically you would want to do post-hoc analyses to characterize the group differences in those clusters. There are many ways to do this; if you are, for example, doing seed-to-voxel analyses you could simply click on 'display values' in the results explorer window, and that will display two bars for each cluster characterizing the groupB - groupA and groupC-groupB differences, respectively.
Regarding question (2), assuming that both gene1 and gene2 are between-subject effects, then for an analysis of gene1 by gene2 interactions you need to define 6 groups (not five), namely
gene1level1_gene2level1
gene1level1_gene2level2
gene1level2_gene2level1
gene1level2_gene2level2
gene1level3_gene2level1
gene1level3_gene2level2
Then, you would select all those 6 effects and enter a contrast [-1 1 1 -1 0 0; 0 0 -1 1 1 -1]. That will show you those regions where the difference in connectivity between gene2-level subgroups is different between the different gene1-level groups.
Hope this helps
Alfonso
Originally posted by liu lin:
Regarding question (1), you are exactly right, if you want to look at an F-test of main group effect (i.e. find those regions where the connectivity differs between groups) you may choose the contrast labeled 'any difference between gA&gB&gC' (or equivalently, enter a between-subjects contrast [-1 1 0; 0 -1 1]). For any significant cluster that you find in these analyses you are right that typically you would want to do post-hoc analyses to characterize the group differences in those clusters. There are many ways to do this; if you are, for example, doing seed-to-voxel analyses you could simply click on 'display values' in the results explorer window, and that will display two bars for each cluster characterizing the groupB - groupA and groupC-groupB differences, respectively.
Regarding question (2), assuming that both gene1 and gene2 are between-subject effects, then for an analysis of gene1 by gene2 interactions you need to define 6 groups (not five), namely
gene1level1_gene2level1
gene1level1_gene2level2
gene1level2_gene2level1
gene1level2_gene2level2
gene1level3_gene2level1
gene1level3_gene2level2
Then, you would select all those 6 effects and enter a contrast [-1 1 1 -1 0 0; 0 0 -1 1 1 -1]. That will show you those regions where the difference in connectivity between gene2-level subgroups is different between the different gene1-level groups.
Hope this helps
Alfonso
Originally posted by liu lin:
Dear all
I have two questions for you. Would you like to give me some suffestion. Thanks.
1. I have three groups (group A,group B and group C). I'm not sure how to do F test. Should I choose the item " Any difference between (gA gB gC) in the 'Between-subject contrast'"? If it shows significant difference after F test, then should I do 3 times of T_test between group A and B, between group A and C and between B and C?
2. I am really confused about interaction analysis. Now I want to do the interaction effects of two gene polymorphisms on the network connectivity. In gene 1 , there are 3 levles. And in gene 2, there are 2 levels. So in the 'Subject effects' box, there are 5 groups:
gene 1_level 1;
gene 1_level 2;
gene 1_level 3;
gene 2_level 1;
gene 2_level2;
So what is the contrast matrix in the 3 x 2 interaction analysis？ Moreover，if there is significant interaction effect, then how to do the post hoc analysis?
I'm sorry for disturbing you with so many questions.
I hope it won't take you too much time. Thank you very much.
Best wishes.
yours
lin
I have two questions for you. Would you like to give me some suffestion. Thanks.
1. I have three groups (group A,group B and group C). I'm not sure how to do F test. Should I choose the item " Any difference between (gA gB gC) in the 'Between-subject contrast'"? If it shows significant difference after F test, then should I do 3 times of T_test between group A and B, between group A and C and between B and C?
2. I am really confused about interaction analysis. Now I want to do the interaction effects of two gene polymorphisms on the network connectivity. In gene 1 , there are 3 levles. And in gene 2, there are 2 levels. So in the 'Subject effects' box, there are 5 groups:
gene 1_level 1;
gene 1_level 2;
gene 1_level 3;
gene 2_level 1;
gene 2_level2;
So what is the contrast matrix in the 3 x 2 interaction analysis？ Moreover，if there is significant interaction effect, then how to do the post hoc analysis?
I'm sorry for disturbing you with so many questions.
I hope it won't take you too much time. Thank you very much.
Best wishes.
yours
lin
Threaded View
| Title | Author | Date |
|---|---|---|
| liu lin | Mar 18, 2016 | |
| Alfonso Nieto-Castanon | Mar 31, 2016 | |