Previous Chapter Chapter 15: Fixed-Effects Panel Regression Next Chapter
Chapter 15: Fixed-Effects Panel Regression
Fixed-effects (FE) regression is a method that is especially useful in the context of causal inference (Gangl, 2010). While standard regression models provide biased estimates of causal effects if there are unobserved confounders, FE regression is a method that can (if certain assumptions are valid) provide unbiased estimates in this situation (other methods are instrumental variables regression and regression discontinuity; see Chapters 13 and 14 in this volume). Since unobserved confounders are ubiquitous in social science applications, FE regression should be standard in the toolkit of modern social research.
FE regression is most often used with panel data, and therefore the focus of this chapter will be on FE regression with panel data. However, before we begin, we want ...