alc<- factor(rep(c("y","n"),c(16,16)))
cig<- factor(rep(c("y","n","y","n"),c(8,8,8,8)))
race<- factor(rep(c("w","o","w","o","w","o","w","o"),rep(4,8)))
gender<-factor(rep(rep(c("f","m"),8),rep(2,16)))
mj<- factor(rep(c("y","n"),16))
counts<-c(405,268,453,228,23,23,30,19,13,218,28,201,2,19,1,18,1,17,1,17,0,1,1,8,1,117,1,133,0,12,0,117)
acm.fit1<-glm(counts~alc+cig+race+gender+mj,family=poisson)
cm.steps<-step(acm.fit1,scope=list(lower=~alc+cig+race+gender+mj,upper=(alc+cig+race+gender+mj)^3))