help
help > RE: contrast ANOVA
Nov 18, 2014 08:11 AM | Andreas Hahn - Medical University of Vienna, Austria
RE: contrast ANOVA
Dear Andrew,
Thanks for the help!
Another question regarding thresholds: I tried several thresholds for the F-test, but interestingly F=5 gave a significant cluster but F=4 did not. How can this be? I thought a lower initial threshold would just make the cluster larger?
Best,
Andreas
Originally posted by Andrew Zalesky:
Thanks for the help!
Another question regarding thresholds: I tried several thresholds for the F-test, but interestingly F=5 gave a significant cluster but F=4 did not. How can this be? I thought a lower initial threshold would just make the cluster larger?
Best,
Andreas
Originally posted by Andrew Zalesky:
Dear Andreas,
NBS differs from SPM in this respect. However, both design matrix formats will give exactly the same result. Note that for a one-way ANOVA with four groups, you could also use a design matrix in the NBS of the following form:
1 1 0 0
1 1 0 0
1 -1 1 0
1 -1 1 0
1 0 -1 1
1 0 -1 1
1 0 0 -1
1 0 0 -1
The contrast is then: [ 0 1 1 1].
Looking at the above design matrix, it can be seen that columns 2, 3 and 4 correspond to the rows of your SPM contrast matrix.
I hope this provides you with some intuition above the equivalence between them.
The initial threshold refers to the F-statistic (if you select an F-test). So if you use an F-test, you would typically select a larger threshold than if you were using a t-test.
Originally posted by Andreas Hahn:
NBS differs from SPM in this respect. However, both design matrix formats will give exactly the same result. Note that for a one-way ANOVA with four groups, you could also use a design matrix in the NBS of the following form:
1 1 0 0
1 1 0 0
1 -1 1 0
1 -1 1 0
1 0 -1 1
1 0 -1 1
1 0 0 -1
1 0 0 -1
The contrast is then: [ 0 1 1 1].
Looking at the above design matrix, it can be seen that columns 2, 3 and 4 correspond to the rows of your SPM contrast matrix.
I hope this provides you with some intuition above the equivalence between them.
The initial threshold refers to the F-statistic (if you select an F-test). So if you use an F-test, you would typically select a larger threshold than if you were using a t-test.
Originally posted by Andreas Hahn:
Dear Andrew,
Thank you very much for providing the NBS toolbox!
I have a question regarding the contrast setting for a one-way ANOVA with four independent groups of subjects with the following design matrix:
1 0 0 0
1 0 0 0
0 1 0 0
0 1 0 0
0 0 1 0
0 0 1 0
0 0 0 1
0 0 0 1
(with more than two subjects in each group).
Can you please explain why the F-contrast according to the example in the help is [1 1 1 1] to test if any of the means is significantly different? Being used to SPM contrasts it appears to test if the sum of all 4 groups is different from zero and the following would be more intuitive for me to test the differences across groups [1 -1 0 0; 0 1 -1 0; 0 0 1 -1].
Another quick question: does the initial "threshold" refer to the F-value (or t-value respectively) for the initial mass univariate testing?
Thank you and best regards,
Andreas
Thank you very much for providing the NBS toolbox!
I have a question regarding the contrast setting for a one-way ANOVA with four independent groups of subjects with the following design matrix:
1 0 0 0
1 0 0 0
0 1 0 0
0 1 0 0
0 0 1 0
0 0 1 0
0 0 0 1
0 0 0 1
(with more than two subjects in each group).
Can you please explain why the F-contrast according to the example in the help is [1 1 1 1] to test if any of the means is significantly different? Being used to SPM contrasts it appears to test if the sum of all 4 groups is different from zero and the following would be more intuitive for me to test the differences across groups [1 -1 0 0; 0 1 -1 0; 0 0 1 -1].
Another quick question: does the initial "threshold" refer to the F-value (or t-value respectively) for the initial mass univariate testing?
Thank you and best regards,
Andreas
Threaded View
| Title | Author | Date |
|---|---|---|
| Andreas Hahn | Nov 14, 2014 | |
| Andrew Zalesky | Nov 16, 2014 | |
| Charanya Muralidharan | Nov 9, 2017 | |
| Andrew Zalesky | Nov 10, 2017 | |
| Charanya Muralidharan | Nov 13, 2017 | |
| Yuan-Fang Zhao | Oct 14, 2016 | |
| Andrew Zalesky | Oct 15, 2016 | |
| Andreas Hahn | Nov 18, 2014 | |
| Andrew Zalesky | Nov 19, 2014 | |