options(contrasts=c("contr.treatment", "contr.poly"))

h.glm <- glm(y/n  ~ Rtreat + Blocker + Site * Time, family=binomial, 
weights=n, data=heartattack)


anova(h.glm, test="Chisq")


hglm.final <- glm(y/n  ~ Rtreat + Blocker + Site , family=binomial, weights=n, data=heartattack)

summary(hglm.final)

anova(hglm.final)

u <- resid(hglm.final, type="pearson") # The Pearson residuals are  saved in u
v <- resid(hglm.final, type="deviance") # The deviance residuals are saved in v
sum(u^2) # The result is the Pearson X^2 statistic
sum(v^2) # The result is the scaled deviance


# commands used to prepare data

heartattack$n <- heartattack[,1] + heartattack[,2]

u <- c("active", "placebo")

heartattack$Rtreat <- rep(u, 12)

u <- c("Yes", "Yes", "No", "No")

heartattack$Blocker <- rep(u, 6)

u <- c(rep("Le12H", 4), rep("Mo12H", 4) )
heartattack$Time <- rep(u, 3)

heartattack$Site <- c( rep("Anterior", 8), rep("Inferior", 8), rep("Other", 8))


# finally convert the column types to factors