Compute Change Score

[bnd]

I am trying to compute a change score for a variable that has two different time points (Month 0 and Month 4). I am having issues with how this should be done in SAS. I tried using an LS Means statement, but only class variables can be used. The data looks like this in my dataset. I am trying to figure out what would be the best way to see a change in these scores (Month 4 - Month 0). But, I am not sure the best way that this could be done. Any thoughts? Suggestions?

Psi timepoints	Psi
Baseline_arm_1	125
Month_4_arm_1	88

[PaigeMiller]

You mention LSMEANS, could you please show us your code? Could you also show us a larger portion of your data, it's really hard to determine what to do if you only have two records.

[bnd]

Sorry about that. My code and more of the data is provided below. 

proc mixed data=be.time04 method=reml covtest;
class acrostic;
model weight = spmrkt_grcy psi time county spmrkt_grcy*time spmrkt_grcy*time*psi /solution ddfm=sat;
random intercept / subject = acrostic type=un;
lsmeans spsis / e diff cl adjust=bon;
run;

Psi timepoints	Psi
Baseline_arm_1	125
Month_4_arm_1	88
Baseline_arm_1	110
Month_4_arm_1	110
Baseline_arm_1	59
Month_4_arm_1	76
Baseline_arm_1	103
Month_4_arm_1	113
Baseline_arm_1	84
Month_4_arm_1	92
Baseline_arm_1	117
Month_4_arm_1	92
Baseline_arm_1	84
Month_4_arm_1	92
Baseline_arm_1	123
Month_4_arm_1	118
Baseline_arm_1	123
Month_4_arm_1	131
Baseline_arm_1	90
Month_4_arm_1	87
Month_4_arm_1	125
Baseline_arm_1	123
Month_4_arm_1	131
Baseline_arm_1	90
Month_4_arm_1	87
Baseline_arm_1	125

[PaigeMiller]

Got it, thanks.

Make PSItimepoint to be a CLASS variable, then an LSMEANS statement with PSItimepoint ought to work.

[bnd]

I changed the variable from numeric to a character variable and I keep receiving the ERROR: Only CLASS variables allowed in this effect. I am not sure why this is happening.