Political Science | Data Analysis: Topics in dynamic Analysis
Y577 | 28481 | Hellwig

Much that we do in political science involves change over time.
What explains whether party systems will be stable or susceptible to
dealignment? Do politicians respond to public opinion or are publics
led by elite cues? Does the transition to a social democratic
government cause unemployment levels to trend downward? Why does
popular support for presidents to move over the course of the
election cycle? How does the duration of wars affect the eventual
prospects for peace? These are just a few of the questions which
imply a dynamic understanding of social and political
relationships.  The objective of this course is to introduce
students to a variety of techniques for specifying and estimating
dynamic models.  After introducing some of the concepts essential to
dynamic modeling, the course begins by examining Box-Jenkins, or
ARIMA modeling.  We then turn to times series regression and
consider how the properties of time-series modeling involve
additional (diagnostic) challenges but also additional (substantive)
benefits compared to cross-sectional analyses.  We then will devote
consecutive weeks considering vector autoregression, unit-roots and
cointegration, time series models for heteroskedasticity, and pooled
cross-sectional time series, duration/event-history analysis, and
panel designs.  While we’ll spend considerable time on the technical
problems involved in time-series, the primary emphasis is on applied
work. It is recommended that students complete the first two courses
in the department statistics sequence before enrolling in this
course.  Student evaluations will be based on class participation
and data analysis assignments.