I'm trying to use 3dLME for a two group, two timepoint analysis with covariates. It's basically a repeated measures ANOVA in which I want to look for longitudinal stability in the control group and decline in the patient group. My strategy is to look for stability in the controls by using 3dICC_REML, and then to look for change in the patients vs the controls (group X time interaction) using 3dLME with a mask that is the control ICC map, thresholded. My questions:
1) is 3dLME the right choice for this, or is 3dMVM just as good?
2) does the model and glt look correct? I've tried running this with or without the "time : 1*time1 -1*time2" portion of the GLT and get the exact same results with/without it.
3) does the combined 3dICC_REML + LME approach sound statistically legit?
3dLME -prefix /data/mridata/jbrown/ptnt_hc_lme.nii -jobs 16 \
-model 'time*group+age+sex+interscan' \
-qVars 'age,interscan' \
-qVarsCenters '68.79,188' \
-ranEff '~1+age+sex+interscan' \
-SS_type 3 \
-mask /data/mridata/jbrown/nc_icc_covars_timefixed_thr4.nii \
-num_glt 1 \
-gltLabel 1 'grpXtime' -gltCode 1 'group : 1*hc -1*ptnt time : 1*time1 -1*time2' \
-dataTable \
Subj time group age sex interscan InputFile \
10886 time1 ptnt 65.65 M 202 con_0001.nii \
10886 time2 ptnt 65.65 M 202 con_0001.nii \
12162 time1 ptnt 66.20 F 183 con_0001.nii \
12162 time2 ptnt 66.20 F 183 con_0001.nii \
...
2679 time1 hc 75.77 F 202 con_0001.nii \
2679 time2 hc 75.77 F 202 con_0001.nii \
6909 time1 hc 67.41 M 164 con_0001.nii \
6909 time2 hc 67.41 M 164 con_0001.nii \
...
thanks!
Jesse
Edited 2 time(s). Last edit at 08/29/2014 02:42PM by jbrown.