Confirmatory Factor Analysis

Researchers in the social and behavioral sciences depend on measurement to connect abstract ideas with the data they can gather. Roos and Bauldry present Confirmatory Factor Analysis, a commonly applied method for linking several observed variables to underlying latent constructs that cannot be seen directly. The opening chapters establish key terms and notation, then walk through a range of measurement models that differ in how latent and observed variables relate.

From there the book examines model estimation in depth and explains how to judge model fit. Most chapters work with indicators that approximate continuous variables, while one chapter addresses categorical indicators. Each chapter is built around a worked example, and some chapters work through two. These examples draw largely on national surveys and cover topics including racist attitudes, theological conservatism, leadership qualities, psychological distress, self-efficacy, beliefs about democracy, and Christian nationalism.

Readers can reproduce every example with the datasets provided on an accompanying website, which also supplies code in R, Stata, and Mplus.