Hello Randy,
I am performing a structural analyses using behavioural PLS. The following are the two analyses performed
Test1) correlate FA and MD (2 conditions) to behaviourl data using behavioural PLS with 5000 permutations and 500 bootstrappings. In the results i observed 2 significant LV's. LV1 had a p-value of 0.011 and LV4 had a p-value of 0.014.
Test2) correlate FA, FA and MD (3 conditions) to the same behavioural data. I had added the same FA data twice by mistake as 2 different conditions. So, now i have 3 conditions. And performed behavioural PLS with 5000 permutations and 500 bootstrappings. In the results i observe 4 significant LV's. LV1 with a p-value of 0.00 ,LV2 with a p-value of 0.003, LV3 with a p-value of 0.006 and LV8 with a p-value of 0.006.
Is it an expected behaviour that the p-value has changed even though the same data was added twice as two different conditions? No extra information was really added in the analyses. Can you tell me why the p-value has changed?
Best Wishes
Jay
Hi Jay - there really aren't comparable scenarios. Remember that the permutation test is an analytic solution based on an empirically-generated distribution. If you add variables to any analysis, you change the distribution, even if its in error.
Baycrest is an academic health sciences centre fully affiliated with the University of Toronto
Privacy Statement - Disclaimer - © 1989-2024 BAYCREST HEALTH SCIENCE. ALL RIGHTS RESERVED