* read in antisocial data; use stata_sem_data, clear *provide summary statistics; summarize * random intercept only model. Stata does not estimate * saturated model with empty covariance coverage cells * so there are no omnibus measures of fit. You can conduct * LRTs of nested models as is done here; sem (anti6-anti14 <- Int@1), /// noconstant mean(Int) method(mlmv) /// satopts(iter(0)) estat ic estimates store intercept *expand random intercept to include random slope; sem (anti6 <- Int@1 Slope@0) /// (anti7 <- Int@1 Slope@1) /// (anti8 <- Int@1 Slope@2) /// (anti9 <- Int@1 Slope@3) /// (anti10 <- Int@1 Slope@4) /// (anti11 <- Int@1 Slope@5) /// (anti12 <- Int@1 Slope@6) /// (anti13 <- Int@1 Slope@7) /// (anti14 <- Int@1 Slope@8), /// noconstant mean(Int Slope) method(mlmv) /// satopts(iter(0)) estat ic estimates store slope *compare intercept to slope model; lrtest intercept slope *regress intercept and slope on gender; sem (anti6 <- Int@1 Slope@0) /// (anti7 <- Int@1 Slope@1) /// (anti8 <- Int@1 Slope@2) /// (anti9 <- Int@1 Slope@3) /// (anti10 <- Int@1 Slope@4) /// (anti11 <- Int@1 Slope@5) /// (anti12 <- Int@1 Slope@6) /// (anti13 <- Int@1 Slope@7) /// (anti14 <- Int@1 Slope@8) /// (Int Slope <- male _cons), /// noconstant method(mlmv) /// cov(e.Int*e.Slope) /// satopts(iter(0)) estat ic