Hi Tian

Your analyses examining how FC varies as a function of participant's task performance are missing the "constant" term for the regression model. The proper way to specify this in CONN is to select 'AllSubjects' and 'Performance' in the list of subject-effects, and then enter a [0 1] contrast in the between-subjects contrast. I think the "Warning!" message that you are seeing in the main CONN gui when specifying the model without this constant term (i.e. without the 'AllSubjects' term) should be telling you something along these lines as well (just to double-check click on the warning message to see a more detailed description). 

Hope this helps

Alfonso

Originally posted by Tian Lin:

Hello Experts, 

I conducted a ROI-to-ROI analysis with a task performance as a continuous 2nd level covaraite to examine how the FC between ROIs varied as a function of participants' task performance. 

As on the first page of the file attached here shows, the "Performance" is the 2nd level predictor and is a continuous variable. I selected the ACC as the see ROI to display the reuslts. This analysis showed a list of regions in the results table and I highlighted the left anterior insula in the picture. From the negative beta score, we know that indivdiuals with higher performance score showed reduced correlation between ACC and left AI. To further visualize this effect, I extract the FC between ACC and left AI by clicking the "import values" button on the GUI interface. 

However, when I tried to plot this negative correlation between Performance and ACC-AI FC, I cannot see it (see the 2nd page of the file attached). In fact, I see that the negative correlation predicted by my analysis was totally off from the raw data, especially at he right end. 

Also, when I calculated the correlation between Performance and ACC-AI FC, this correlation coefficient was even a positive score (although very close to 0).

Did I do something incorrect for my 2nd level analysis? If not, how can I explain this descriptency in the reuslts?

Thank you for your attention and insights in advance.

 

Tian