* Chapter 6 Model Diagnostics in SPSS . COMPUTE age_2 = age - 2 . EXECUTE . COMPUTE age_2sq = age_2*age_2 . EXECUTE . * Fit Model 6.3 and save conditional predicted values and residuals . MIXED vsae WITH age_2 age_2sq BY sicdegp /CRITERIA = CIN(95) MXITER(100) MXSTEP(5) SCORING(1) SINGULAR(0.000000000001) HCONVERGE(0, ABSOLUTE) LCONVERGE(0, ABSOLUTE) PCONVERGE(0.000001, ABSOLUTE) /FIXED = age_2 age_2sq sicdegp age_2*sicdegp | SSTYPE(3) /METHOD = REML /PRINT = G SOLUTION /SAVE = PRED RESID /RANDOM age_2 age_2sq | SUBJECT(CHILDID) COVTYPE(UN) . * Generate Figure 6-6 (note: standardized residuals are not automatically available) . GRAPH /SCATTERPLOT(BIVAR)=PRED_1 WITH RESID_1 /PANEL COLVAR=sicdegp COLOP=CROSS /MISSING=LISTWISE . * Note: reference lines at zero can be added to these plots manually by double-clicking on the plot . * Generate Figure 6-7 . GRAPH /SCATTERPLOT(BIVAR)=age_2 WITH RESID_1 /MISSING=LISTWISE . * Generate Q-Q plots for Figure 6-8 . SORT CASES BY sicdegp . SPLIT FILE LAYERED BY sicdegp . PPLOT /VARIABLES=RESID_1 /NOLOG /NOSTANDARDIZE /TYPE=Q-Q /FRACTION=BLOM /TIES=MEAN /DIST=NORMAL. SPLIT FILE OFF . * NOTE: EBLUPs of multiple random effects cannot currently be saved in new variables in SPSS . * Generate Figure 6-11 . GRAPH /SCATTERPLOT(BIVAR)=PRED_1 WITH vsae /PANEL COLVAR=sicdegp COLOP=CROSS /MISSING=LISTWISE .