treatment and sum contrasts in R

gistfile1.r

#create dummy data

minex <- c()

minex$hand <- c(rep(1,20),rep(0,20))

minex$speed <- c(rep(c(rep(20,10),rep(50,10)),2))

minex$angle <- c(rep(c(rep(45,5),rep(75,5)),4))

minex <- as.data.frame(minex)

beta0 <- 18

beta1 <- -16

beta2 <- -6

beta3 <- 30

beta4 <- 13

sigma2 <- 0.3

minex$distance <- beta0 + beta1*minex$hand + beta2*minex$angle + beta3*minex$speed + beta4*minex$hand*minex$angle + rnorm(dim(minex),0,sqrt(sigma2))

minex$hand <- as.factor(minex$hand)

minex$speed <- as.factor(minex$speed)

minex$angle <- as.factor(minex$angle)



# treatment contrasts

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

m1 <- lm(distance ~ hand + angle + speed + hand:angle + hand:speed + angle:speed, data=minex, x=T)

#m1 <- lm(distance ~ hand + angle + speed, data=minex, x=T)

m1$x -> X

X1 <- X;

d <- dim(X1)

v <- solve(t(X1) %*% X1)

sd <- diag(1/sqrt(diag(v)))

c <- sd %*% v %*% t(sd)

X1.cor.betahat <- round(c[2:d[2], 2:d[2]], 3)

X1.Sigma <- solve(t(X1) %*% X1)

X1.contrast <- solve(t(X1) %*% X1) %*% t(X1)

cbind(X1.contrast[2,],minex$hand,X1[,2])



#repeat with sum contrasts 

options(contrasts = c(unordered = "contr.sum", ordered = "contr.poly"))

m2 <- lm(distance ~ hand + angle + speed + hand:angle + hand:speed + angle:speed, data=minex, x=T)

#m2 <- lm(distance ~ hand + angle + speed, data=minex, x=T)

m2$x -> X

X2 <- X;

d <- dim(X2)

v <- solve(t(X2) %*% X2)

sd <- diag(1/sqrt(diag(v)))

c <- sd %*% v %*% t(sd)

X2.cor.betahat <- round(c[2:d[2], 2:d[2]], 3)

X2.Sigma <- solve(t(X2) %*% X2)

X2.contrast <- solve(t(X2) %*% X2) %*% t(X2)

cbind(X2.contrast[2,],minex$hand,X2[,2])