3dLME: GLT codes for interactions of continuous variables

HI AFNI Experts,

We were trying to create GLT codes for interactions of continuous variables, and we keep getting the generic 3dLME error (I’m assuming we’re specifying our GLT for continuous variable interactions incorrectly).

For example, the following code without the interactions works (script 1):
3dLME
-prefix $out_file
-jobs 4
-model “con_linearage_w1lag_years”
-qVars “con_linear,age_w1,lag_years”
-ranEff “~1+con_linear”
-SS_type 3
-mask $mask_file
-num_glt 3
-gltLabel 1 ‘linear’ -gltCode 1 ‘con_linear :’
-gltLabel 2 ‘age’ -gltCode 2 ‘age_w1 :’
-gltLabel 3 ‘lag’ -gltCode 3 ‘lag_years :’
-dataTable @${data_table}

However, the following code with interactions does not work (script 2):
3dLME
-prefix $out_file
-jobs 4
-model “con_linearage_w1lag_years”
-qVars “con_linear,age_w1,lag_years”
-ranEff “~1+con_linear”
-SS_type 3
-mask $mask_file
-num_glt 7
-gltLabel 1 ‘linear’ -gltCode 1 ‘con_linear :’
-gltLabel 2 ‘age’ -gltCode 2 ‘age_w1 :’
-gltLabel 3 ‘lag’ -gltCode 3 ‘lag_years :’
-gltLabel 4 ‘linear_x_age’ -gltCode 4 ‘con_linear : age_w1 :’
-gltLabel 5 ‘lag_x_age’ -gltCode 5 ‘lag_years : age_w1 :’
-gltLabel 6 ‘linear_x_lag’ -gltCode 6 ‘con_linear : lag_years :’
-gltLabel 7 ‘linear_x_age_x_lag’ -gltCode 7 ‘con_linear : age_w1 : lag_years :’
-dataTable @${data_table}

  1. What is the correct way to specify GLTs for the interactions of continuous variables?
  2. Is there a way to specify GLTs after 3dLME has already been run?
  3. The results from script 1 return what I think are estimates/regression coefficients and a z-test (rather than a t-test), is this correct?

Thank you!,
Ekarin

Ekarin,

Since all the three predictors are between-subject quantitative variables, I suggest that you do the following:

  1. center each of the three variables properly (e.g., around their mean)

  2. create four new variables - one for each interaction - and label them

  3. perform a multiple regression analysis on the seven variables with, for example, 3dttest++ or 3dMEMA

Oh, sorry, Gang, I should’ve explained the analysis a bit more.

This is a longitudinal analysis where the variables are:

  1. con_linear (continuous-within) - a contrast of our within-subject functional task. Easy, medium, and hard conditions for each subject were coded as -1, 0, and 1.
  2. age_w1 (continuous-between) - their age at wave 1
  3. lag_years (continuous-between) - the difference/lag between their wave 1 and wave 2 in years

Thank you,
Ekarin

Ekarin, try the same variable coding strategy and then feed them into 3dLME.

  1. Is there a way to specify GLTs after 3dLME has already been run?

Unfortunately no.

  1. The results from script 1 return what I think are estimates/regression coefficients and a z-test (rather than a t-test), is this correct?

Correct.

Thank you, Gang, for your suggestion (that solution worked) and for the clarification of my questions.

Best regards,
Ekarin