This article compares regression modeling and propensity score analysis as different types of statistical techniques used in addressing selection bias when estimating the impact of two-year versus one-year Head Start on children's school readiness. The analyses were based on the national Head Start secondary dataset. After controlling for covariates, regression modeling showed that program duration (two years vs. one year) was a significant predictor of all six outcome measures, including Peabody Picture Vocabulary Test, Woodcock-Johnson Reading Skills, Woodcock-Johnson Math Reasoning Skills, teacher-reported composite academic skills, preschool learning behaviors, and social skills. When using propensity score analysis that matched children, program duration significantly predicted children's academic outcomes but had limited effects on learning behaviors and social skills. Overall, both methods confirmed the predictive effects of program duration but propensity score analysis offered more conservative findings than regression modeling. Methodological issues and policy implications were discussed based on these findings.