It seems that there are quite a few group analysis programs already in the AFNI suite: 3dttest++, 3dMEMA, 3dANOVAx, GroupAna, and 3dLME. Theoretically speaking, 3dLME could handle almost all of the complex situations of FMRI group analysis. However, there are a couple of thorny issues with the LME approach in practice even for some relatively simple analyses: flexible but difficult-to-standardize interface with the underlying R function, and controversial assignment of degrees of freedom to the testing statistics. For most experiments typically encountered in FMRI, it feels like taking a spear to kill a fly when using the LME method to handle the conventional AN(C)OVA-type data structure. This grey area between 3dLME and the other AFNI group analysis programs was the motivation for a new approach: multi-variate modeling.

Written in R, 3dMVM performs the typical ANOVA as well as ANCOVA. In other words, it can take both categorical and quantitative variables. More importantly, there is no bound on the number of explanatory variables that could be incorporated into the model. Similar to most group analysis programs (and unlike 3dMEMA), 3dMVM only takes the effect estimates (no t-statistics) as input.

3dMVM has the standard AFNI-style interface, and most of the usage information can be found at the terminal:

3dMVM -help | less

Three examples are also provided in the help. Feedback is always welcome. More information about 3dMVM can be found at the following website:

[afni.nimh.nih.gov]

Gang