Abstract
Bayesian statistics have made great strides in recent
years, developing a class of methods for estimation and
inference via stochastic simulation known as Markov Chain
Monte Carlo (MCMC) methods. MCMC constitutes a revolution
in statistical practice with effects beginning to be felt
in the social sciences: models long consigned to the "too
hard" basket are now within reach of quantitative
researchers. I review the statistical pedigree of MCMC and
the underlying statistical concepts. I demonstrate some of
the strengths and weaknesses of MCMC and offer practical
suggestions for using MCMC in social-science settings.
Simple, illustrative examples include a probit model of
voter turnout and a linear regression for time-series data
with autoregressive disturbances. I conclude with a more
challenging application, a multinomial probit model, to
showcase the power of MCMC methods.
Users
Please
log in to take part in the discussion (add own reviews or comments).