\magnification=1200
\baselineskip=20pt
\nopagenumbers
\font\big=cmr12 scaled \magstep2

\centerline{\bf STANFORD UNIVERSITY}
\centerline{\bf DEPARTMENT OF STATISTICS}
\centerline{\big DEPARTMENTAL SEMINAR}
\bigskip

\baselineskip=12pt
\centerline{4:15 p.m., Tuesday, May 8, 2001}
\centerline{Sequoia Hall Rm. 200}
\centerline{(Cookies at 3:45 in 1st Floor Lounge)}

\bigskip
\baselineskip=15pt
\centerline{\sl Simon Jackman}
\centerline{\sl Department of Political Science}
\centerline{\sl Stanford University}
\centerline{\sl Stanford,  CA 94305}
\bigskip
\centerline{\bf Markov chain Monte Carlo in the Social Sciences}

\bigskip

By any reasonable standard, Markov chain Monte Carlo is one of the
more important recent developments in applied statistics.  After some
initial resistence to MCMC's Bayesian underpinnings, MCMC is now
widely accepted by quantitative social-scientists, providing an
interesting case study in the cross-disciplinary proliferation of
statistical ideas and methods.  My focus is on examples from political
science.  Models and data sets that political scientists had long
consigned to the ``too-hard'' basket are now easily handled via MCMC.
Time permitting, examples to be discussed include (1) estimating the
ideological location of politicians via analysis of their voting
records (roll call data), using an analog of item-response models; (2)
multivariate $t$ regression models for (heavy-tailed) multi-party
electoral data; (3) dynamic hierarchical models for state-level
election forecasting, using the 2000 U.S.~Presidential Election as an
example; (4) mixture models for heterogeneity in survey-based measures
of political attitudes; (5) parameter-driven transitional models of
international conflict (correlated binary data).
\bye