Synthetic National Longitudinal Education Data: A Pilot Study of K–16 Education + Career Pathways

• In Event: Poster Session for Quantitative Methods and Statistical Theory

Abstract

This pilot study develops and tests synthetic national longitudinal education data that track Kindergarten students’ college and career pathways by combining nationally-representative datasets: ECLS-K 1999-2007 (Kindergarten to 8th grade) and NELS 1988-2000 (8th grade to 8 years after high school graduation). The merge of these existing national datasets, linked together via statistical matching and imputation techniques, can help bridge the gap between elementary and secondary/postsecondary education data silos and supplement state/local longitudinal education databases. By imputing missing data for (unobserved) education/job attainment based on matching variables, this study explores the early indicators of Kindergarten students’ readiness for college and career. With synthetic national longitudinal data, machine learning methods are applied for training and testing logistic regression models of educational attainment.

Authors

• Jaekyung Lee, University at Buffalo - SUNY

• Joseph Aaron Jaeger, University at Buffalo - SUNY