Foundations in Causal Analysis in the Social Sciences

Catalog description

Methods for estimating causal effects in observational data, including matching, counterfactual, and related approaches. SOC 572 Foundations in Causal Analysis in the Social Sciences (3) This course investigates methods for estimating causal effects in observational (non-experimental) data. There are three common strategies for estimating causal effects in observational data. One strategy is to condition on variables to block all the back-door paths from the causal variable to the outcome variable. Examples: matching (including propensity scores) and regression with control variables. A second strategy is to use exogenous variation to isolate the "unconfounded" covariation in the causal and outcome variables. Examples: instrumental variables (IVs), natural experiments, fixed effects. A third strategy is to isolate the mechanisms through which the causal variable operates to affect the outcome variable. These methods are becoming more standard in sociology and related fields. The course will cover the following methods: standard regression and its limitations; natural experiments and instrumental variables; counterfactual or potential outcomes methods; matching methods, including propensity scores; fixed-effects and random-effects; structural equation methods; mixed methods. The course is distinguished from other courses in the university by its focus on the use of statistical methods for causal inference. Whereas other courses describe regression or structural equation models as general methods for one's statistical toolkit, this course concentrates on when, and how, regression or structural equation approaches can be used to draw causal inferences from non-experimental data. Examples will be drawn from a variety of social science disciplines, including economics, political science, and criminology, as well as from sociology.