Time Series Analysis: An Introduction for Social Scientists
Statistical models can be applied to time series data-- chronological sequences of observations-- to examine the movement of social science variables over time (e.g., public opinion, government policy, judicial decisions, socioeconomic measures), allowing analysts to estimate relationships between variables and test hypotheses. This course introduces time series methods and their application in social science research. We will focus on conceptualization and practice, with some attention to the underlying statistical theory.
The course will provide a comprehensive discussion of the core concepts in time series analysis, including autoregressive, moving-average and unit-root processes, and the two most important assumptions underlying time series models: stationarity and exogeneity. The lab component of the course will demonstrate the use of, and assumptions underlying, common models of time series data: autoregressive distributive-lag, moving average (autocorrelated error), differenced data, ARMA, equilibrium (error) correction, and vector autoregression models. These models will be demonstrated with the goal of giving participants the tools necessary to apply them to their own research.
Prerequisites: A sound background in linear regression models is assumed but prior training in time series analysis is not required. The course will make use of basic algebra. The lab component of this course will employ Stata. Some familiarity with Stata would be helpful but for those without that familiarity, a crash course on Stata will be provided during the first class session.
Interested participants should be clear that this is a course on analyzing time series data. While a course on time series is a very useful precursor to learning panel data analysis, this course does not itself cover panel data analysis techniques.
Fee: Members = $1500; Non-members = $3000
Tags: time series