active=rep(c("r","s","t","u","v","w"),c(6,6,6,6,6,6)) passive=rep(c("r","s","t","u","v","w"),6) wt=c(0,1,5,8,9,0,29,0,14,46,4,0,0,0,0,0,0,0,2,3,1,0,38,2,0,0,0,0,0,1,9,25,4,6,13,0) display=data.frame(active,passive,wt)[which(active!="t"),] ### Fit Log-linear model wt=active+passive: Test Quasi-independence for the incomplete table library(MASS) active=display[,1] passive=display[,2] freq=display[,3] displaytable=table(active,passive) displaytable[cbind(active,passive)]=freq LoglinModel=loglm(freq~active+passive,param=T,fit=T) loglin(fitted(LoglinModel),margin=list(1,2),param=T,fit=T) glm(freq~active+passive,family=poisson(log)) summary(LoglinModel) anova(LoglinModel) fitted(LoglinModel) #Parameter Estimates LoglinModel$param