Censoring specified directly...
model {
for (i in 1:6) {y[i] ~ dnorm(mu, 1)} # uncensored data
for (i in 7:9) {y[i] ~ dnorm(mu, 1)I(8,)} # censored data
mu ~ dunif(0, 100)
}
Data:
list(y = c(6,6,6,7,7,7,NA,NA,NA))
node mean sd MC error 2.5% median 97.5% start sample
mu 7.193 0.3478 0.003604 6.515 7.19 7.875 1001 10000
y[7] 8.571 0.4809 0.005356 8.018 8.446 9.805 1001 10000
y[8] 8.575 0.49 0.005317 8.018 8.448 9.802 1001 10000
y[9] 8.578 0.487 0.005635 8.018 8.451 9.804 1001 10000
Censoring specified via zeros trick...
model {
for (i in 1:6) {y[i] ~ dnorm(mu, 1)}
for (i in 1:3) {
zeros[i] <- 0
zeros[i] ~ dpois(p[i])
p[i] <- -log(phi(mu-8))
}
mu ~ dunif(0, 100)
}
Data:
list(y = c(6,6,6,7,7,7,NA,NA,NA))
node mean sd MC error 2.5% median 97.5% start sample
mu 7.192 0.348 0.003687 6.519 7.185 7.882 1001 10000