help
help > RE: How to use "age" as covariates?
Apr 26, 2014 03:04 AM | Alfonso Nieto-Castanon - Boston University
RE: How to use "age" as covariates?
Hi Yifei,
Yes, you are exactly right. If you want to look at the effect of A, controlling for gender and B, you would typically select the 'All','A','gender', and 'B' subject-effects, and enter the contrast [0 1 0 0]. The coefficient for the term A in the associated regression equation are reported as 'effect sizes' in the toolbox. For ROI-to-ROI analyses these are reported in the column labeled 'beta' in the ROI-to-ROI results table. The effect-sizes (coefficients of the regression equation) for the seed-to-voxel results, averaged over each significant cluster, can be obtained using the approach described in this thread. The coefficients associated with the other terms ('All', 'gender', and 'B') can be obtained similarly using the corresponding contrasts ([1 0 0 0], [0 0 1 0], and [0 0 0 1], respectively).
Hope this helps
Alfonso
ps. note that depending on the form of "control" that you wish to implement, some times you may want to look at the 'unique variance' associated with effect of A (variance not shared with any of the other effects). To do this you will need to orthogonalize first the values entered for the second-level covariate A in Setup->Covariates->Second-level (make those values orthogonal to the values of gender and/or B). The regression equation that you get using these transformed variables will be in the end exactly the same as the one before, when using the original variables, but the statistics associated with the effect A will be different, reflecting your different hypothesis.
Originally posted by Yifei Zhang:
Yes, you are exactly right. If you want to look at the effect of A, controlling for gender and B, you would typically select the 'All','A','gender', and 'B' subject-effects, and enter the contrast [0 1 0 0]. The coefficient for the term A in the associated regression equation are reported as 'effect sizes' in the toolbox. For ROI-to-ROI analyses these are reported in the column labeled 'beta' in the ROI-to-ROI results table. The effect-sizes (coefficients of the regression equation) for the seed-to-voxel results, averaged over each significant cluster, can be obtained using the approach described in this thread. The coefficients associated with the other terms ('All', 'gender', and 'B') can be obtained similarly using the corresponding contrasts ([1 0 0 0], [0 0 1 0], and [0 0 0 1], respectively).
Hope this helps
Alfonso
ps. note that depending on the form of "control" that you wish to implement, some times you may want to look at the 'unique variance' associated with effect of A (variance not shared with any of the other effects). To do this you will need to orthogonalize first the values entered for the second-level covariate A in Setup->Covariates->Second-level (make those values orthogonal to the values of gender and/or B). The regression equation that you get using these transformed variables will be in the end exactly the same as the one before, when using the original variables, but the statistics associated with the effect A will be different, reflecting your different hypothesis.
Originally posted by Yifei Zhang:
Hi Alfonso,
Thank you so much for the detailed explanation! It helps a lot!
May I ask another question related to this? Now I want to do the regression analysis within the patient group (the only group) like this:
Functional connectivity ~ All(intercept)+A+gender+B, where A is the predictor. And I want to control for gender and B.
According to the FAQ, am I right if I define the 2nd-level covariates: All(all 1 for each subject), A, gender(male=1,female=0) and B, then select the four covariates in "subject effects", and enter the contrast [0 1 0 0]? Could you please also recommend a way to calculate the regression equation?
Thank you again for the help!
Best regards,
Yifei
Thank you so much for the detailed explanation! It helps a lot!
May I ask another question related to this? Now I want to do the regression analysis within the patient group (the only group) like this:
Functional connectivity ~ All(intercept)+A+gender+B, where A is the predictor. And I want to control for gender and B.
According to the FAQ, am I right if I define the 2nd-level covariates: All(all 1 for each subject), A, gender(male=1,female=0) and B, then select the four covariates in "subject effects", and enter the contrast [0 1 0 0]? Could you please also recommend a way to calculate the regression equation?
Thank you again for the help!
Best regards,
Yifei