Proc mixed

[noa_m]

Hi, 

I have data with 29 participants. i used the following model:

IV1  = Time (3 time points, used as a CLASS)

IV 2 = brain activity (3 time points, **centered using baseline coding (i sbstracted the first time point from the rest).

DV = level of extraversion.

my script ran as followed: 

PROC MIXED method=ml DATA= _exp0_.Amygdala_acc_hipp_2 COVTEST;
CLASS ORDER time;
MODEL E = time| Amygdala_dev /s ddfm=bw s outpm =data covb;
random intercept /subject = ORDER;
run;

for using baseline as a refrence point, i coded the baseline time point as "2". because of the centering, it means that the brain activity values in this time point equal to zero. 

this is the output i recieve: 

 2 = before 1 = after 0 = during 

estimates: 

33.0690	1.0009	28	33.04	<.0001
-1.9631	0.8678	52	-2.26	0.0279
-2.1397	0.8883	52	-2.41	0.0196
0	.	.	.	.
Amygdala_dev - 3.2427	1.8508	52	1.75	0.0857
Amygdala_dev*time -5.3499	2.4557	52	-2.18	0.0339
0	.	.	.	.
0	.	.	.

from what i understand, the estimate in the line "Amygdala_dev" represents the simple slope of my refrence point (which is the first time point). given the fact that there not suppose to be any slope on this time point (data is set to sero at this point) - why do i get the estimate   3.2427? 

when i use different coding - such that the refrence point is set to the third time point i get the same estimate. 

please help me figure out how to interpret this output.

thank you from advance, 

noa 

[PaigeMiller]

Could you show us more of the SAS Code, showing the steps before PROC MIXED where you perform the coding of the variable? Please paste this code into the box that appears when you click on the "running man" icon when you type your reply.

Can you show us the actual output from PROC MIXED (screen capture is fine)?

[noa_m]

thank you for thequick reply.

i've coded the variable using SPSS.

the steps were: 

1. use "aggregate" in order to create a time-invaraiant baseline variable (the first time point)

2. then creating the centered variable by substracting this variable from the origincal variable. 

looking at this variable seems to look OK (i've attached a screenshot showing my columns) 

and i've attached the SAS output 

Noa