Forbes1 <- read.table("c:/cjdata/stat401/forbes.txt",header=T) names(Forbes1) Forbes1 <- Forbes1[Forbes1$sector!="Other",] # delete data from 3 sectors Forbes1 <- Forbes1[Forbes1$sector!="Communication",] Forbes1 <- Forbes1[Forbes1$sector!="Medical",] Forbes1$LSales <- log(Forbes1$Sales) Forbes1$LAssets <- log(Forbes1$Assets) plot(Forbes1$LAssets,Forbes1$LSales,pch=as.numeric(Forbes1$sector), xlab="Log of Assets",ylab="Log of Sales",main="ANCOVA on Industry Data") # here we implement models based on factors (like Proc GLM in SAS) Forbes1.LA <- lm(LSales ~ LAssets , data = Forbes1) # Fit all three models Forbes1.sectorLA <- lm(LSales ~ sector + LAssets , data = Forbes1) Forbes1.sectorxLA <- lm(LSales ~ sector/LAssets , data = Forbes1) anova(Forbes1.sectorLA, Forbes1.sectorxLA) # perform F tests for nested models anova(Forbes1.LA, Forbes1.sectorLA) # check residuals for the final model par(mfrow=c(2,2)) plot(Forbes1.sectorLA) par(mfrow=c(1,1)) rm(Forbes1,Forbes1.LA,Forbes1.sectorLA,Forbes1.sectorxLA)