If you specified a t contrast, then you get the positive correlation with a 1 and negative correlation with a -1 over the regressor of interest. An F contrast of 1 would show you the areas that correlative positively or negatively with the regressor.
note that the correlation maps dont have a 0 referent. Y is free to vary above or below zero. its the relationship that is being tested.
cheers
scj
>>> <[log in to unmask]> 10/17/06 2:06 PM >>>
Hello,
We have a concern with our simple correlation resulting map.
From our fixed effects model we have a contrast image resulting from a
condition a>b. The resulting contrast images for each subject are fed into
the simple regression analysis with a string of correlates. Does the
resulting positive correlation map show us:
1. correlations where a>b only ? (Y >0)
2. where a>b and b>a ? (Y> and < 0)
Simply does the original contrast image, a>b include voxels that have a
positive and negative value or only those having a positive value.
Thanks in advance for you time and thoughts.
Maria
|