Time-Invariant Variables' Time-Varying Effects : Misinterpretations of the Fixed-Effects Model in Ascriptive Inequality Research

A popular statistical approach, the fixed-effects regression model is known for its ability to produce unbiased coefficients by adjusting for unobserved time-invariant individual heterogeneity. This ability is contingent on the assumption that time-invariant variables must not have time-varying effects, which, otherwise, would interfere in the process of coefficient estimation. In ascriptive inequality research that focuses on relatively fixed identities such as race and gender, this assumption is often overlooked, leading to misinterpretations of the findings. Demonstrating with case studies, we intend to explain and clarify two types of such misinterpretations: (1) time-invariant variables' time-varying effects, when measured in the model, are mistaken as time-invariant variables' unbiased coefficient estimates; (2) time-invariant variables' time-varying effects, when unmeasured in the model, confound coefficient estimates of time-varying variables.