Within-Subject Inference: Fixed-Effects Models for Wage Inequality

This chapter develops fixed-effects models for diagnosing social inequality in wages. It explains when fixed-effects models are preferred over random-effects models, the within-subject transformation, and SAS implementations (e.g., PROC GLM with ABSORB and conditional fixed-effects logit via PROC LOGISTIC with STRATA). A reproducible workflow covers design checks, cluster-robust inference, and interpretation. Using a synthetic wage panel, linear fixed-effects estimates attribute most within-worker wage growth to age, while conditional fixed-effects logit on a Wage Gap Indicator highlights substantial education dividends and muted sector, region, and year effects. This chapter links findings to wage-setting policy, transparency monitoring, and cumulative learning and practice.