help > Multiple conditions for a contrast
Showing 1-4 of 4 posts
Display:
Results per page:
Mar 24, 2021  06:03 PM | Y. Chang
Multiple conditions for a contrast
Dear all,

I am new to the CONN toolbox and have a question about setting a between-condition contrast at the second level analysis. I have four conditions (A, B, C, and baseline). A, B and C are the reading task. After the ROI-ROI connectivity analysis, I would like to check if there is an overall task effect in reading versus baseline (i.e., (A, B, C) > baseline). But I am not sure how to set the contrast. Shall I give a contrast, [ 1 1 1 -3] or [1 1 1 -1] or [1 0 0 0;0 1 0 0; 0 0 1 0; 0 0 0 -1] or [1 0 0 -1; 0 1 0 -1; 0 0 1 -1]? I have tried all of these options, and the results are different so I would like to make sure which one is correct (or none of them!)? I think that I have not had clear understanding about how the conditions can be combined so I would appreciate your help and suggestions!

best,
YaNing
Apr 5, 2021  12:04 AM | Alfonso Nieto-Castanon - Boston University
RE: Multiple conditions for a contrast
Dear YaNing,

The last one is the correct one ([1 0 0 -1; 0 1 0 -1; 0 0 1 -1]), that will evaluate whether (A>baseline) OR (B>baseline) OR (C>baseline). An alternative option would be a contrast like [1/3 1/3 1/3 -1], which evaluates instead whether (A+B+C)/3>baseline. In general a contrast like [a b c d] evaluates an hypothesis of the form a*A + b*B + c*C +d*D >0 (so [1/3 1/3 1/3 -1] for example is evaluating A/3 + B/3 +C/3 -D > 0, which is just the same as (A+B+C)/3 > D), and a contrast with multiple rows simply combines the hypotheses of each row with OR conjunctions (so [1 0 0 -1; 0 1 0 -1; 0 0 1 -1] is just evaluating A-D>0 OR B-D>0 OR C-D>0, which is just the same as A>D OR B>D OR C>D)

Hope this helps
Alfonso
Originally posted by Y. Chang:
Dear all,

I am new to the CONN toolbox and have a question about setting a between-condition contrast at the second level analysis. I have four conditions (A, B, C, and baseline). A, B and C are the reading task. After the ROI-ROI connectivity analysis, I would like to check if there is an overall task effect in reading versus baseline (i.e., (A, B, C) > baseline). But I am not sure how to set the contrast. Shall I give a contrast, [ 1 1 1 -3] or [1 1 1 -1] or [1 0 0 0;0 1 0 0; 0 0 1 0; 0 0 0 -1] or [1 0 0 -1; 0 1 0 -1; 0 0 1 -1]? I have tried all of these options, and the results are different so I would like to make sure which one is correct (or none of them!)? I think that I have not had clear understanding about how the conditions can be combined so I would appreciate your help and suggestions!

best,
YaNing
Apr 6, 2021  02:04 PM | gucy121
RE: Multiple conditions for a contrast
Dear Alfonso,

as you said, a contrast with multiple rows simply combines the hypotheses of each row with OR conjunctions (so [1 0 0 -1; 0 1 0 -1; 0 0 1 -1] is just evaluating A-D>0 OR B-D>0 OR C-D>0, which is just the same as A>D OR B>D OR C>D). Is there any possibility to consider the AND conjunction in CONN when conducting the FC at the network level? 

Your reply will be much appreciated! Thank you!

Best,
Gucy
Apr 6, 2021  04:04 PM | Y. Chang
RE: Multiple conditions for a contrast
Hi Alfonso,

Thank you so much. This is very helpful for me to understand how the contrast works!

best regards,
YaNing

Originally posted by Alfonso Nieto-Castanon:
Dear YaNing,

The last one is the correct one ([1 0 0 -1; 0 1 0 -1; 0 0 1 -1]), that will evaluate whether (A>baseline) OR (B>baseline) OR (C>baseline). An alternative option would be a contrast like [1/3 1/3 1/3 -1], which evaluates instead whether (A+B+C)/3>baseline. In general a contrast like [a b c d] evaluates an hypothesis of the form a*A + b*B + c*C +d*D >0 (so [1/3 1/3 1/3 -1] for example is evaluating A/3 + B/3 +C/3 -D > 0, which is just the same as (A+B+C)/3 > D), and a contrast with multiple rows simply combines the hypotheses of each row with OR conjunctions (so [1 0 0 -1; 0 1 0 -1; 0 0 1 -1] is just evaluating A-D>0 OR B-D>0 OR C-D>0, which is just the same as A>D OR B>D OR C>D)

Hope this helps
Alfonso
Originally posted by Y. Chang:
Dear all,

I am new to the CONN toolbox and have a question about setting a between-condition contrast at the second level analysis. I have four conditions (A, B, C, and baseline). A, B and C are the reading task. After the ROI-ROI connectivity analysis, I would like to check if there is an overall task effect in reading versus baseline (i.e., (A, B, C) > baseline). But I am not sure how to set the contrast. Shall I give a contrast, [ 1 1 1 -3] or [1 1 1 -1] or [1 0 0 0;0 1 0 0; 0 0 1 0; 0 0 0 -1] or [1 0 0 -1; 0 1 0 -1; 0 0 1 -1]? I have tried all of these options, and the results are different so I would like to make sure which one is correct (or none of them!)? I think that I have not had clear understanding about how the conditions can be combined so I would appreciate your help and suggestions!

best,
YaNing