Covariate centering options in 3dMEMA

AFNI Message Board
1

I’m doing a one-sample 3dMEMA analysis with two covariates (contained in a text file with the recommended format). Could you please clarify the following from the help file, under the -covariates_center option:

If you want all covariates centered about their own mean, just use -covariates_center mean. Be alert: Default is mean centering!

So if the -covariates_center option is not called, will each covariate be centered about its own mean by default? If so, why would one need to “use covariates_center mean” as stated above? And does one simply type the word “mean”, or the actual numerical means of each covariate? All the examples provided seem to be for two-sample scenarios.

I’m also not sure whether I need the -covariates_model option for a one-sample test.

2

I agree those statements are indeed confusing.

So if the -covariates_center option is not called, will each covariate be centered about its own mean by default?

Yes, that is correct.

If so, why would one need to “use covariates_center mean” as stated above? And does one simply type the word “mean”,
or the actual numerical means of each covariate?

It looks like that the default is the same as “-covariates_center mean” (literally with word “mean”). If a center value different from mean is desired, use something like

-covariates_center age = 25

I’m also not sure whether I need the -covariates_model option for a one-sample test.

The option -covariates_model is only for comparing two groups.