Dear experts,
I have a question about the threshold value used when running film_gls.
I ran 2 FEAT 1st-level analyses. The first analysis was the full analysis that included all the steps. In the second analysis, I broke the full analysis to 2 steps. First, I ran just the pre-stats. Then I used the filtered_func_data created by pre-stats as input file to run stats + poststats. All setups were done using the FEAT gui.
I found that when running the full analysis, the threshold value for film_gls was set to 1000. But when running just stats + poststats analysis, film_gls used the minimum value of the filtered_func_data as threshold.
So despite using the same input data (filtered_func_data) to run film_gls, the autocorrelation estimates between the 2 analyses were slightly different, and so do the pe.
I am wondering why the 2 analyses that were supposed to generate the same results use different thresholds when running film_gls. This seems to be the cause of the slight difference in results between the two. Also, why is it that the threshold was set to 1000 when running the full analysis?
Any thought would be greatly appreciated.
shih-wei
|