/* Binary Regression Examples FILE: Binary-Reg-Examples.doc Examples (modified) from Piegorsch and Bailer (2005) ANALYZING ENVIRONMENTAL DATA Wiley*/*====================================================================;* SAS code to find LC50;* logistic regression specified by d=logistic option;data copper; input conc ydead @@; N = 20; logconc = log(conc); obs_prob = ydead/N;datalines; 0.1 2 0.2 2 0.3 5 0.5 8 1.0 15 2.0 15 2.5 17proc probit data=copper; model ydead/N = logconc / d=logistic inversecl; output out=copper_out p=pred_prob; run;proc gplot data=copper_out; plot obs_prob*logconc pred_prob*logconc/overlay; run;*====================================================================;* SAS code to fit logit, probit and CLL links; data grndwtr; input landuse $ sewer $ nitrate chloride ytce N @@; datalines; Unde yes 1.0 10 0 17 Unde no 0.8 17 0 59 Agrc yes 0.9 9 0 7 Agrc no 7.0 20 2 48 LowR yes 1.6 6 0 5 LowR no 3.5 14 2 21 MedR yes 3.9 18 12 43 MedR no 2.6 15 5 86 HigR yes 5.2 33 33 76 HigR no 4.2 18 1 17 Recr yes 2.3 16 3 26 Recr no 1.7 11 4 38 Inst yes 3.3 36 8 32 Inst no 0.9 11 0 22 Tran yes 4.0 66 7 29 Tran no 1.3 12 1 30 Comm yes 4.3 24 20 42 Comm no 1.1 30 0 7 Indu yes 3.7 23 17 33 Indu no 2.2 19 3 12* fit logit link via PROC GENMOD; proc genmod; title "*********************logit link*********************"; class landuse sewer; model ytce/N = landuse sewer nitrate chloride / dist=binomial link=logit expected;* fit probit link via PROC GENMOD; proc genmod; title "*********************probit link*********************";class landuse sewer; model ytce/N = landuse sewer nitrate chloride / dist=binomial link=probit expected;* fit CLL link via PROC GENMOD; proc genmod; title "**********************CLL link**********************"; class landuse sewer; model ytce/N = landuse sewer nitrate chloride / dist=binomial link=cll expected;
View Full Document