Hi, apologies if this has been asked before however I need some clarification on my resting-state contrasts. I have two groups [autism spectrum disorder (ASD) vs. typical development (TD)]. I am wanting to look at how behavioral covariates are associated with functional connectivity while controlling for average motion across the [u]entire[/u] sample (I will look at the groups individually after), however depending on my approach I get radically different results.
[b]Approach A[/b]
[i]3 second level covariates:[/i][list][*]sample_ids (1's for both ASD and TD)[*]behavioral_measure[*]average_motion (average of the 6 motion parameters generated in pre-processing for both ASD and TD)[/list]
Contrast: [0 1 0]
[b]Approach B[/b]
5 second level covariates:[list][*]ASD_ids[*]TD_ids[*]behavioral_measure[*]ASD_average_motion [*]TD_average_motion[/list]Contrast: [0 0 1 0 0]
Approach A gives me a huge amount of results (suspiciously large amount) whereas approach B gives me few results but potentially more reliable. Does anyone have any advice on which is the correct way to do this? My intuition is that approach B makes more logical sense as it controls for the potential differences between the two groups.
Any help would be appreciated. Thanks,
[b]Approach A[/b]
[i]3 second level covariates:[/i][list][*]sample_ids (1's for both ASD and TD)[*]behavioral_measure[*]average_motion (average of the 6 motion parameters generated in pre-processing for both ASD and TD)[/list]
Contrast: [0 1 0]
[b]Approach B[/b]
5 second level covariates:[list][*]ASD_ids[*]TD_ids[*]behavioral_measure[*]ASD_average_motion [*]TD_average_motion[/list]Contrast: [0 0 1 0 0]
Approach A gives me a huge amount of results (suspiciously large amount) whereas approach B gives me few results but potentially more reliable. Does anyone have any advice on which is the correct way to do this? My intuition is that approach B makes more logical sense as it controls for the potential differences between the two groups.
Any help would be appreciated. Thanks,