• Home
• Profiles
• News
• Scientific Progress
• Publications
• Employment
• Resources

Username Password Create an account Reset password

Back to PLS Help Post-hoc correction for confounders after Mean-Centering Task PLS Jayachandra Posted on 10/22/15 13:05:47 Number of posts: 14 Jayachandra posts: Hello, I have performed an analyses using Structural PLS to identify group differences between patients and controls. Is it possible to check if the significant results survive after correcting for confounders such as age, gender.. etc? How do i perform this correction in PLS? Best wishes Jay  Login to reply to this topic. Replies: Untitled Post I'm Online nlobaugh Posted on 10/22/15 19:59:18 Number of posts: 229 nlobaugh replies: Hi Jay.. you can't do that directly in the PLS software - if you want to look at residualized data (after influence of potential confounds are removed), you should calculate those values outside of PLS and enter the residuals as if they were the structural measures. We have a function called residualize.m that is inclued with the PLS m-files (and used for something else) that you could use if that works better for you... Nancy Post-hoc correction for confounders after Mean-Centering Task PLS Jayachandra Posted on 10/23/15 09:38:15 Number of posts: 14 Jayachandra replies: Hello Nancy, Thank you for your reply. As you have suggested, i have calculated the residualized data (after influence of potential confounds are removed) and then performed the regular PLS analyses. For some reason the the calulation of confidence intervels (CI) is going wrong when i used the residual data. I see that the confidence intervel error bars are not displayed on the bar (example: the upper and lower bound of CI are both above(outside) the bar(brains score), when i view Task PLS brains scores with CI plot). Can you tell me where am i going wrong? Best wishes Jay   Untitled Post I'm Online nlobaugh Posted on 10/23/15 09:49:11 Number of posts: 229 nlobaugh replies: That's odd... are you using the most recent version of the software?  Note, however, the CIs reflect the mean of the sampling distribution - check in the results to see what the adjusted upper/lower CIs are, and if they include the orginal. (Those *adj variables are there for use in case of extreme skew in the sampling distribution) nancy   Untitled Post Jayachandra Posted on 10/23/15 10:34:46 Number of posts: 14 Jayachandra replies: Hi Nancy, I just realised that the upper/lower CI bounds was not displayed properly because i was using MATLAB version 2014b. When i installed MATLAB version 2014a, it is displayed properly (as expected).  Best Wishes Jay Untitled Post I'm Online nlobaugh Posted on 10/23/15 20:00:54 Number of posts: 229 nlobaugh replies: OK.. thanks for letting us know..   Login to reply to this topic.

© 2011 - 2012 Baycrest. All Rights Reserved.