Hello Lukas:
The two GLT's that you listed are performing two different statistical tests.
The first GLT is testing whether the AUC's for the stimuli are equal.
The second GLT is testing whether the individual IRF coefficients are equal.
Since the GLT's are testing different hypotheses, I don't know of a theoretical
reason for preferring one to the other. In most cases they should yield
similar results (when thresholded at the same p-level).
The first test creates a single GLT coefficient, representing the difference
in response, whereas the second test creates 4 GLT coefficients. Having a
single number to represent the result may make it easier to combine results
across multiple subjects, which would argue in favor the first test.
Doug Ward
