Individual and School Correlates of Attendance and School Performance Among Homeless Students

Background:   Research consistently documents the relatively poor academic performance among children who are homeless relative to their low income and housed peers.  Current literature focuses on student-level academic risk factors and points to important individual differences within homeless youth.  There is surprisingly limited consideration of the school-level factors that may play a role in shaping academic performance among homeless youth, given that schools are often implicated as a key intervention lever.  This study thus explores relationships between both individual- and school-level factors and a rich set of student-level academic performance outcomes including standardized achievement scores, grade point average (GPA) and attendance.  Given high rates of mobility among homeless students, the relationship between attendance and standardized tests scores and GPA is also examined.

Methods:  This study draws on four years of archival student and school data from a large urban district (school years 2006-2007 through 2010-2011) to estimate relationships between individual, family, and school factors, and student performance.  This longitudinal sample encompasses 2659 students in 119 schools.

Students’ daily attendance, end of year GPA, and standardized reading and mathematics achievement test scores are the key outcome variables.  In addition, a rich set of student-level characteristics are included as controls (e.g. student age, grade level, racial and ethnic background, home language and English literacy characteristics, economic background  and prior school performance (e.g.,  prior history of grade retention, special education participation). School characteristics include rates of free and reduced lunch participation, minority enrollment, English language learners, credentialed teachers, and daily attendance as well as school size and performance characteristics such as overall achievement proficiency rates and an indicator of whether the school met performance targets relates as specified by No Child Left Behind (NCLB). 

The nested and longitudinal structure of these observational data is exploited to implement fixed-effects regression modeling techniques—at student and school levels-- to control for the influence of unobservable, time-invariant student and school characteristics.  Standard errors in regression models are adjusted for the clustering of students within schools.

 Results:  Prior student performance (special education and grade retention) were consistently and negatively related to reading and mathematics achievement, GPA and attendance.  School performance characteristics, but not demographic and structural characteristics, were also independently related to student attendance and performance.  That is, homeless students in schools with higher performing children and adolescents, over and above student and school demographic characteristics, showed better performance over time.  Moreover, student attendance (both in terms of days enrolled and daily attendance) was a robust correlate of GPA and achievement. 

Conclusions:   This study provides robust empirical support for the contention that variation in school characteristics—over and beyond student characteristics-- shape homeless student academic performance.  These findings underscore that attributes of school settings are a factor that should be considered in intervention planning.  Moreover, some of the negative effects of homelessness on student achievement outcomes could be mitigated by interventions focusing on student placement and attendance.  This may be especially relevant in larger urban districts where provisions related to NCLB provide school choice options.