Using Propensity Score Methods with Statewide Administrative Data to Inform Education and Workforce Policy in Maryland

• In Event: 3-006 - Program Evaluation and Identifying Intervention Targets via Secondary Data

Abstract

Researchers, practitioners, and policy makers are interested in the effects of a program or policy on student outcomes, and large administrative databases are increasingly being made available to researchers in policy contexts. However, these data present unique challenges for the researcher, as data were typically not collected for research purposes and typically lack randomization to groups. Propensity score methods provide a viable solution to mimic randomization for program and policy evaluation. In Maryland, dual enrollment programs, where high school students enroll in college coursework, serve as a developmental bridge to ease the transition from high school to college, and are of particular state policy interest. This paper causally evaluates the long-term college and workforce outcomes of participation in dual enrollment programs in Maryland.
Data were from the Maryland Longitudinal Data System (MLDS), the statewide repository for linked education and workforce data in Maryland. Students were identified as dually enrolled if they had overlapping enrollment dates in a Maryland public high school and a Maryland college. Of the 2009-2010 cohort of 12th grade students (N = 63,000), 7% (N = 4,200; 59% female; 79% not eligible for free/reduced price meals (FARMs); 64% white; 93% non-Hispanic) were dually enrolled in their 12th grade year. Propensity scores were estimated using a logistic regression model predicting dual enrollment program participation by demographic characteristics (e.g., gender, race, ethnicity, FARMs), academic achievement (e.g., high school assessment scores), high school attendance, and school district. Common support and balance on each covariate were good. The outcome model used the matched sample (N ~ 8,000 students) to examine the causal effect of dual enrollment on long-term outcomes.
Dually enrolled students were more likely to enroll in a 2-year college one year after high school (B = 0.86, p < .01) than students who were not dually enrolled. There was no difference in likelihood to enroll in a 4-year college one year after high school (p > .05). Results in years 2-4 suggest that dually enrolled students were beginning their careers at 2-year colleges and transferring to 4-year colleges. Dually enrolled students were also more likely to earn a Bachelor’s degree (B = 0.38, p < .01) and earned an additional $2,000 six years after high school when compared to similar students who were not dually enrolled.
The strengths and limitations of using this approach with administrative data will be discussed.

Author

• Angela Henneberger, University of Maryland at Baltimore
  Presenting Author