Using Longitudinal Data to Assess Long-Term Outcomes Associated with Poverty in Maryland Students

Background and Purpose: Poverty is a public health crisis affecting more than 20% of children and youth and places them at risk for developing a serious mental, emotional, or behavioral disorder (Yoshikawa, Aber, & Beardslee, 2012). Additionally, the detrimental impact of poverty is so prominent that achievement gaps are established as early as kindergarten (Williams, Bryan, Morrison, & Scott, 2017), which explains why children persistently living in poverty experience substantial deficits in educational and occupational outcomes (Hair, Hanson, Wolfe, & Pollack, 2015, McLoyd, 1998). In the U.S., low-income students constitute a majority of students enrolled in public schools (Hair et al., 2015). Concentrated poverty at the school level occurring as a result of residential segregation by poverty level (Crosnoe, 2009; Bischoff & Reardon, 2013) may further negatively affect students regardless of their poverty status. Previous studies examining the long-term effects of poverty have been limited due to nonrepresentative samples, nonresponse bias, and attrition (Michelmore & Dynarski, 2017). Using statewide administrative data aids in addressing these limitations. The present study utilized administrative data for a cohort of 6^th grade students enrolled in Maryland public schools to assess the independent effects of student and school poverty associated with long-term educational and workforce outcomes.  

Methods: Data were from the Maryland Longitudinal Data System (MLDS), Maryland’s statewide repository for individual-level education and workforce data. MLDS receives the data from three state agencies and longitudinally links the data to follow individuals over time. The cohort of Maryland public school students who were in 6^th grade (N = 54,465) in 2007-08 (the earliest year of data available in the MLDS) was used for this study. The cohort was predominantly White (45%; 35% Black; 5% Asian; 4% other) and non-Hispanic (89%). Fifty-one percent were never disadvantaged (eligible for free and reduced-price meals, FARMS) throughout middle and high school, 30% were transitorily disadvantaged (sometimes FARMS), and 19% were persistently disadvantaged (always FARMS). Students attended 466 public schools in 6^th grade and 257 public schools in 12^th grade in 2013-14. This study used SAS 9.3 to obtain descriptive statistics and Stata 15 with MlwiN 3.02 to conduct multilevel modeling.

Results: Students who were never disadvantaged were more likely to graduate high school on time (49% versus 15%), enroll in college within 2 years (42% versus 9%), and earn higher quarterly wages ($777 versus $572 among those who never enrolled in postsecondary) when compared to students who were persistently disadvantaged. Results showed school poverty overshadowed student poverty for outcomes like SAT Math scores (school effect size (Cohen’s d): .52; student effect size: .34) and likelihood of graduating on time (school effect: .54; student effect: .37). Student poverty more strongly influenced students graduating despite being retained (school effect: .26; student effect: .36), or dropping out of school altogether (school effect: .23; student effect: .30).

Conclusions and Implications: Concentrated poverty at the school level has significant implications for students’ academic and postsecondary success. A discussion will focus on implications for research, policy, and practice.