Hi Gang,

My question regarding -tcoef versus -fcoef is in regard to the warnings in the 3dRegAna manual. From my reading, the F statistic associated with -fcoef is the significance of the overall regression, not the significance of the individual parameter. By contrast, the t-statistic associated with -tcoef is the significance of the individual parameter.

In my case I want to find voxels associated with one regressor, so my instinct is to use the -tcoef.

In regard to the idea of testing to see which of the two regressors is a better fit for any one voxel, you mentioned the three possibilities: 1) no trend; 2) linear trend; and 3) quadratic trend.

For those three possibilities, I am imagining a situation where a voxel is a very clear cut linear-trend voxel. In that situation the ttest for the linear regressor would be very significant for the linear regressor (significantly above zero), and not significant for the quadratic regressor. In that situation, it would seem that obvious that possibility #2 is true for that voxel.

Now imagine a situation where a voxel has some type of waveform in between a linear and a quadratic. The ttest for that voxel might be significant for both regressors. Then the question is: Which regressor does it look most like? Does it look significantly more linear than quadratic or more quadratic than linear?

How do I test each voxel to find out which ones have significantly more variance accounted for by the linear regressor than the quadratic regressor and which ones have significantly more variance accounted for by the quadratic regressor than the linear regressor?

I know it is not right to do a ttest on the t-values for each parameter estimate, but in essence, that is what I am after.

Would a ttest on the parameter estimates themselves be correct? In that case, I am worried that the parameter estimates for a linear versus quadratic trend might be affected by the different scales used to code for the linear vs the quadratic waveforms.

Best,

Christine