Abstract: Using Big Data to Inform Prevention Science in Maryland (Society for Prevention Research 27th Annual Meeting)

177 Using Big Data to Inform Prevention Science in Maryland

Wednesday, May 29, 2019
Seacliff B (Hyatt Regency San Francisco)
* noted as presenting author
Angela Henneberger, PhD, Research Assistant Professor, University of Maryland at Baltimore, Baltimore, MD
Bess A. Rose, Ed.D, Statistician, University of Maryland at Baltimore, Baltimore, MD
Boyoung Nam, BA, Doctoral Candidate, University of Maryland at Baltimore, Baltimore, MD
Dawnsha R. Mushonga, PhD, Post-doctoral Fellow, University of Maryland at Baltimore, Baltimore, MD
Introduction: Prevention science is increasingly focusing on the use of big data, including administrative data, to inform policy. The current funding formula for education in Maryland provides additional funds for higher-poverty schools in a linear fashion, where an additional dollar amount is received for each low income student. The Kirwan Commission on Innovation and Excellence in Education was established under Maryland State law in 2016 and is charged with making policy recommendations that would enable Maryland’s preK-12 education system to perform at the level of the best-performing systems in the world. Under consideration by the Commission is the possibility of exponentially increasing the dollar amount for low-income students as the concentration of school poverty increases. Researchers at the University of Maryland were called upon to help inform this policy decision.

Method: Data were from the Maryland Longitudinal Data System (MLDS), Maryland’s statewide repository for individual-level education and workforce data that are longitudinally linked across three state agencies. The cohort of Maryland public school students who were in 6th grade (N = 63,427) in 2007-08 was used for this study (50% white; 36% Black; 89% non-Hispanic; 64% eligible for free/reduced price meals). Poverty was measured using student eligibility for FARMS in 6th-12th grade, and school level poverty was measured by aggregating individual student poverty to the school level. High school dropout, assessment scores, and college enrollment were measured using administrative records.

Results: Forty-six percent (N = 29,189) of students were never eligible for FARMS throughout middle and high school, and 18% (N = 11,313) of students were eligible for FARMS every year. School level poverty was categorized into 10 percentage point buckets, and dummy codes were created for each bucket. After controlling for the role of student-level poverty, race, ethnicity, and school-level racial composition, results indicate that the steepest increase in dropout and decline in postsecondary enrollment occurred around deciles 2-4, and peak at about decile 6. For high school assessment scores, significant declining thresholds were found across the spectrum of school poverty. Additionally, students who were usually in poverty experienced the worst outcomes, experiencing even worse outcomes than students who were always in poverty.

Conclusions: Commissioners used these results to offer a preliminary recommendation of an exponentiated per pupil dollar amount for schools with high concentrations of students living in poverty (>40%). A discussion will focus on the challenges associated with using administrative data for prevention science.