Metropolis-Hastings-Algorithm

For implementing Metropolis-Hastings algorithm simulated data were used with k=3 covariates and a constant term. A multivariate normal distribution was taken as a prior for , N(0,) where is a diagonal matrix with on the diagonal. After considering a number of values for , = 1 was chosen. The posterior distribution was simulated using as proposal a multivariate normal distribution centered at the current update of and with a covariance matrix given by the inverse of Fisher information evaluated at the current update.

Gibbs Sampler for binary data

For this analyse, we simulated N independent binary random variables where each yi comes from a Bernoulli distribution with probability of success .