3dLME or 3dLMEr or 3dLME2? and multiple comparisons

AFNI Message Board
1

AFNI version info (afni -ver):

Precompiled binary linux_ubuntu_16_64: Sep 6 2024 (Version AFNI_24.2.05 'Macrinus')

Hello,

It has been a few years since running 3dLME and noticed that there exists a 3dLMEr package and now a 3dLME2. Which one should we use and is there a difference between 3dLMEr and 3dLME2.

Related to the 3dLMEx question, is the how to run a multipple comparisons test with it. Noticed in the 3dClustSim recommendations that the 3dttest++ is the most accurate in controlling for false positive rate. However, how could we use 3dttest++ when we have a few values missing with our 3dLMEx setup?

If we were to use the 'Newer' acceptable method of 3dFHWMx afc values, would using the -resid option 3dLMEx suffice in generating the input needed for 3dFWHMx? Our 1stlevel analysis files were run in SPM.

Thanks.

Mike

2

Hi Mike,

3dLMEr is generally preferred for cases involving within-individual variables due to its greater flexibility compared to 3dLME, as detailed in this blog post. Meanwhile, 3dLME2 is available for more specialized cases where detailed lower-level structures are required, though such situations are uncommon.

Regarding multiple testing, this issue remains debated. You can find my perspective here.

Gang Chen

3

Thanks,Gang. Interesting perspective and helpful. Struggling with this phenomenon recently.

Mike