Back to PLS Help

trouble interpreting the sign of my brain score values
astajduhar
Posted on 04/18/22 14:46:41
Number of posts: 3
astajduhar posts:

Hi PLS experts, 

I'm having a bit of trouble interpreting my brain score values. I ran a non-rotated PLS with only one condition on some structural MRI data. I have two groups and have thus specified a contrast to pull apart any group differences. My contrast datamat consists of one column, two rows, -1 for group A and 1 for group B. When I run the analysis and print my brain scores, I have all positive values, despite my salience values being both positive and negative... If brain scores are derived by multiplying the raw brain data with the saliences, shouldn't at least a subset of my bs values be negative? 

Thanks in advance for the help!

Replies:

Untitled Post
rmcintosh
Posted on 04/18/22 19:02:03
Number of posts: 385
rmcintosh replies:

The brain scores are the dot-product of the saliences and data (sum of the cross-product), so it's hard to predict the sign of a score becuase it will depend on the relative number of positive and negatives saliences and the data (voxel) values themselves



Untitled Post
astajduhar
Posted on 04/18/22 19:26:10
Number of posts: 3
astajduhar replies:

quote:

The brain scores are the dot-product of the saliences and data (sum of the cross-product), so it's hard to predict the sign of a score becuase it will depend on the relative number of positive and negatives saliences and the data (voxel) values themselves

Thanks, Randy. I have more positive salience values than negative and even the means of the raw data between the two groups are very close together, which supports my non-significant LV. Could these two reasons be why my brain scores are all positive? 



Untitled Post
rmcintosh
Posted on 04/21/22 17:15:49
Number of posts: 385
rmcintosh replies:

quote:

Thanks, Randy. I have more positive salience values than negative and even the means of the raw data between the two groups are very close together, which supports my non-significant LV. Could these two reasons be why my brain scores are all positive? 

Probably that is the reason, yes. :) 



Untitled Post
astajduhar
Posted on 04/21/22 18:01:06
Number of posts: 3
astajduhar replies:

quote:

Probably that is the reason, yes. :) 

Thanks, Randy. So am I correct to interpret my salience values in this case as ROIs that more (+ salience value) or less (- salience value) contribute to the non-significant difference between the groups?



Untitled Post
rmcintosh
Posted on 04/23/22 12:01:13
Number of posts: 385
rmcintosh replies:

quote:

Thanks, Randy. So am I correct to interpret my salience values in this case as ROIs that more (+ salience value) or less (- salience value) contribute to the non-significant difference between the groups?

Think of it this way. The saliences are similar to regression coefficients and the bootstrap ratios are like a t-value for each. If your design lv has positive weights for group1 and negative for group 2, then positive weights for the brain would indicate regions that are higher in group1 and the negative would be lower in group 1 all relative to group 2. The reason for the non-significant LV is that these differences, considered across the brain, are not high enough to permit significant differentiation of the two groups.




Login to reply to this topic.

  • Keep in touch

Enter your email above to receive electronic messages from Baycrest, including invitations to programs and events, newsletters, updates and other communications.
You can unsubscribe at any time.
Please refer to our Privacy Policy or contact us for more details.

  • Follow us on social
  • Facebook
  • Instagram
  • Linkedin
  • Pinterest
  • Twitter
  • YouTube

Contact Us:

3560 Bathurst Street
Toronto, Ontario
Canada M6A 2E1
Phone: (416) 785-2500

Baycrest is an academic health sciences centre fully affiliated with the University of Toronto