First, a categorical binomial process x = {0,1} of either success (alive) or failure (death) was introduced, in which x = 0 if a piglet was still alive at weaning and x = 1 if the piglet died before weaning. Second, for each piglet,i, an underlying liability of survival and the probit link
function f(yi) = Φ–1(yi) was introduced, in which yi is the probability of success [alive; i.e., yi = P(xi = 0)] and Ф is the cumulative Gaussian distribution.