This dissertation’s first two empirical chapters exploit newly available complete count U.S. Census records to explore topics in U.S. historical educational inequality. In the first empirical chapter, I revisit the puzzling finding that U.S. rural youths appeared to participate more in school than urban youths prior to the Great Depression. I link youths aged 6 to 19 appearing in complete count Census records from 1900 to 1920 to their appearance in the 1940 Census, which was the first to report educational attainment. Rural youths ultimately completed fewer grades than urban youths across all subgroups considered in all time periods. The historical U.S. does not comprise as exception to the universal finding of rural educational disadvantage. The second empirical chapter, coauthored with my advisor, explores the impact of the first wave of compulsory schooling policies. Using a novel dataset of father-son linkages spanning 1850 to 1940, we replicate previous findings that compulsory schooling probably closed the gap in school between high and low SES youths in school participation and boosted population-level educational attainment. However, compulsory schooling laws had minimal, and potentially even negative, effects on overall educational inequality. Why? High SES youths responded to the laws by boosting their school participation at older ages in response to the laws relative to low SES youths. Both chapters lead to the conclusion that parity historical age-specific school participation statistics can conceal inequality in educational attainment. In the third empirical chapter, I turn to the question of whether job instability is growing in the 21st century and whether it is stratifying along educational lines. Using the Current Population Survey Job Tenure Supplement, I find that the answer is sensitive to the choice of estimand. An estimand incorporating shorter-term job separation hazard reports a mild increase in job stability and a sharp increase in educational inequality in job stability. An alternative, equally plausible, estimand that utilizes only longer-term job separation hazards yields a decline in job stability and declining educational inequality.