********************************** * Import data and only keep 1995 * ********************************** clear matrix set memory 1g use "http://biostat.mc.vanderbilt.edu/wiki/pub/Main/CourseBios312/salary.dta" keep if year==95 ************************************************** * Convert stings variables to numeric indicators * ************************************************** * Make a male indicator variable; Female is reference group gen male=. replace male=1 if sex=="M" replace male=0 if sex=="F" table sex male regress salary male yrdeg * Residuals versus fitted values rvfplot, yline(0) * Residuals versus predictor yrdeg rvpplot yrdeg * Studentized residuals predict sresid, rstu histogram sresid qnorm sresid lowess sresid yrdeg dfbeta scatter _dfbeta_1 id summarize _dfbeta_* gen yrdegsq = yrdeg^2 regress salary male yrdeg yrdegsq predict sresid2, rstu histogram sresid2 *95% confidence interval for the expected salary of a male subject who obtained a degree in 1990 predict yhat predict stderr, stdp gen cil_yhat = yhat - invttail(1597-3-1, .025)*stderr gen ciu_yhat = yhat + invttail(1597-3-1, .025)*stderr list id male yrdeg stderr yhat cil_yhat ciu_yhat if male==1 & yrdeg==90 *95% prediction interval for the individual salary of a female subject who obtained a degree in 1980 predict std_f, stdf gen cil_f = yhat - invttail(1597-3-1, .025)*std_f gen ciu_f = yhat + invttail(1597-3-1, .025)*std_f list id male yrdeg stderr yhat cil_f ciu_f if male==0 & yrdeg==80 lowess sresid2 yrdeg sort sresid twoway (lowess sresid yrdeg, mean) (lowess sresid2 yrdeg, mean) replace yrdeg=110 if id==23 replace salary=25000 if id==8 scatter salary yrdeg regress salary male yrdeg dfbeta list id _dfbeta_3 if id==8 list id _dfbeta_3 if id==23 scatter _dfbeta_3 id predict sresid3, rstu hist sresid3 qnorm sresid3 * Other diagnostics predict cookvalue, cook predict hatvalue, hat scatter cookvalue id scatter hatvalue id list id _dfbeta_3 _dfbeta_4 cookvalue hatvalue if id==8 list id _dfbeta_3 _dfbeta_4 cookvalue hatvalue if id==23