Okay, I understand. I have seen papers use a cutoff for the ICC, and I'll follow those guidelines in my manuscript (even if they're a bit arbitrary).   EDIT: For others interested in using such a measure in a publication, I suggest citing the following paper, which discusses assessing correlations within individuals in repeated-measures data:   Liang K-Y, Zeger SL. Longitudinal data analysis using generalized linear models. Biometrika 1986; 73(1): 13-22. ... View more

Excellent. That worked. In case anyone is wondering, here is the full code. Proc glimmix data=Data NOCLPRINT NOITPRINT METHOD= RSPL namelen=40; CLASS id; model breakfast(ref='0')=restdur_PMC*restdur_PMC restdur_PMC restdur_PM*restdur_PM restdur_PM school_PMC / ODDSRATIO SOLUTION DIST=BINARY ddfm=bw;store TST_quad_break; RANDOM Intercept / TYPE=AR(1) Subject=id; run; ods output fitplot = Fit_TST_Quad_Break; proc plm source=TST_Quad_Break; effectplot fit(x=restdur_PMC); run; ODS GRAPHICS / RESET HEIGHT = 4.9in WIDTH = 6.83in; proc sgplot data=Fit_TST_Quad_Break noautolegend; series x = _XCONT1 y = _PREDICTED / lineattrs=(color=black pattern=SOLID thickness=3); band X=_XCONT1 lower=_lclm upper=_uclm / fillattrs=(color=black transparency=.85);xaxis values = (-4 to 4 by 2) label='Total Sleep Time (person-mean centered)' LABELATTRS=(size=14 weight=bold) VALUEATTRS=(size=12 weight=bold); yaxis grid values = (.6 to .85 by .05) label='Breakfast Consumption' LABELATTRS=(size=14 weight=bold) VALUEATTRS=(size=12 weight=bold) GRIDATTRS=(color=GrayCA thickness=.1mm);run; The below graph was produced:     The only thing I could not do was to make the grid lines transparent, which is not an option in GRIDATTRS. -confooseddesi89 ... View more