SAS PROC PLM Simple Slopes & Plot Don't Match

[madisonsmith]

Hi all,

I am running a multivariate PROC MIXED model with a combination of continuous, categorical, and spline variables (syntax below). I have observed a significant interaction between fobss_total_personmean (continuous) and TransitionRelat (categorical, 0/1). I have attempted to probe this and test significance of the simple slopes using PROC PLM, but the direction of the slopes in the table clearly don't match; one slope is positive and one is negative in the table whereas in the figure, both are clearly positive. What am I doing wrong?

PROC MIXED data=work.import covtest noclprint method = ML; 
class fab400_id partner(ref="Yes") TransitionSingle(ref="Continuance of Singlehood OR In a Relationship") 
TransitionRelat(ref="Continuance of Relationship OR Being Single") GID_Male(ref="Gender2") dem06_L2(ref="Gender2"); 
model pss_sum = Age_T1 GID_male dem06_L2 visit partner Length2 TransitionSingle TransitionRelat Time mspss_total_personmean fobss_total_personmean 
partner*fobss_total_personmean Length2*fobss_total_personmean TransitionSingle*fobss_total_personmean TransitionRelat*fobss_total_personmean
Time*fobss_total_personmean /solution; 
random intercept visit / sub=fab400_id type=vc;
Repeated / Subject=fab400_id;
store pred1;
run;

proc plm restore=pred1;
estimate 'FOBS slope, Transition Relat 0' fobss_total_personmean 1 TransitionRelat*fobss_total_personmean 1 0,
'FOBS slope, Transition Relat 1' fobss_total_personmean 1 TransitionRelat*fobss_total_personmean 0 1 / e;
effectplot slicefit (x=fobss_total_personmean sliceby=TransitionRelat=0 1);
run;

[jiltao]

I do not see your ESTIMATE statements in PROC MIXED. Please include that in your program.

Thanks,

Jill

[jiltao]

Actually I see the E option output that might represents the coefficients used in your ESTIMATE statements.

The estimates are not the same as what PROC PLM plots. The effects in the E option output table are cut off due to long names so it is not easy to understand the effects, but it seems to me that some average of some effects are computed, which is not the case for the PROC PLM slicefit plot.

[madisonsmith]

How would I go about making the slicefit plot and the estimate results match? I just need to prove what the overall significant interaction term looks like, and test the significance of the simple slopes.

I'm sorry for all my questions but I really don't understand this a lot, so I'm not sure how to resolve the issue of the average effects being computed either.