Back to PLS HelpPosted on 03/07/24 05:05:56
Number of posts: 41
Hi all. We are running a very simple analysis with 3 groups, 1 conditions - ERP data. We are finding exceedingly large bootstrap ratios (>15000) and signifiance throughout the entire ERP. This does not match the visible single-subjet variability in the ERP within each group (plotted via PLS gui), and we've verified the baseline is not offset across groups.
What are some thoughts for troubleshooting? It feels like there's a problem with bootstrap distribution for some reason or other but not sure where to start digging.
Many thanks
Agatha
Replies:
Posted on 03/07/24 05:16:20
Number of posts: 41
I apologize for not providing necessary details.
There are about 150 individuals in the analysis, split between 3 groups. We ran 1000 bootstrap iterations. We're testing with more to see if issues are stability of solution. Agatha
Posted on 04/04/24 14:20:03
Number of posts: 394
I apologize for not providing necessary details.
There are about 150 individuals in the analysis, split between 3 groups. We ran 1000 bootstrap iterations. We're testing with more to see if issues are stability of solution. Agatha
Hi Agatha - sorry for the long delay. I think we'd need to see the bootstrap distribution. I assume you're doing a mean-centered PLS for this? Group main effects can be quite strong, but I can't say that I've seen ratios that high.
Login to reply to this topic.
