Hi Alfonso,

Sorry for the later reply, I wanted to double check a few of the things you mentioned. Your response was very helpful! I had not realized that there were these slight differences between computers. I know I spent a lot of time double checking I used the same steps on each computer, but I did not process them in tandem, which may have left room for human error.

I looked into the differences in the .mat files you pointed out and thank you for pointing them out! I believe that on computer 1 I had processed session one first, then created a new conn.mat file for both sessions, but only pre-processed session 2 in that file because I had already processed session 1. I could see how that may affect the ART files, but I did not think it would cause such gross differences in results. After running pre-processing again on both computers, I got similar results from both machines at the denoising step, first level, and second level. I made sure that at each step the same regressors were present on each computer. While I don't remember selecting the extra session effects on computer 2 for my denoising, I must have accidentally selected it. Does that make much of difference the analysis? We had 3 other computers run that had different results. 

I am curious, does the ROI matrix matter? When I looked at my second level results, I made sure that I restricted my analysis to the 164x164 matrix for both computers and excluded the extra 4 ROIs from computer 2. Does the presence of extra ROIs in the matrix change results, even though I am not selecting them in the second level analysis? Let me know if that question does not make sense.

I did as you suggested I have put together the earliest detection of differences between the two computers. I noticed there were some slight differences in the QA Timeseries values in the first level covariates. The next place I noticed differences were in the QA values for Gray, White, CSF volumes/eroded volumes and GCOR, starting around the 1000th place. These values were the first places I have noticed differences. Otherwise, everything seemed comparable.

I have attached the conn.mat files for computers 1 and 2 again. While the results are much more similar, there is still some variability in the ROIs and the values. 

Thank you for your time!

Audreyana