Using Individual and School-Level Data to Examine School Disengagement and Subsequent Felonies

Background and Purpose: Most of the extant research on factors affecting the School-to-Prison Pipeline examines individual-level, behavioral/legal factors including absences (Fabelo et al., 2011), disengagement (Bender, 2012), school based-offenses (Rocque & Snellings, 2017), and suspensions (Morris & Perry, 2016; Wolf & Kupchik, 2017) with select individual-level, extra-legal factors including biological sex (CRDC, 2014), race/ethnicity (McCarter, 2017), and socioeconomic status (Birkhead, 2012).  Few studies link individual-level variables to school-level variables (Sullivan, Klingbeil, & Van, 2013; Theriot et al., 2010) and to-date, none examine individuals’ school data linked with their justice data. 

This study linked these data sources in order to answer three research questions: 1) Which variables – legal/extra-legal and individual/school-level predict involvement in the justice system?  2) Are there differences between IVs (individual and school level) regarding DVs including unexcused absences, offense commission, suspension rates, and felony bookings? 3) Does school disengagement (operationalized as suspension in the previous school year with unexcused absences in the subsequent year) impact subsequent felony bookings?

Methods: Our 3 research questions are accommodated in a single structural equation mediation model where the relationship between out-of-school suspensions and total felony bookings is mediated by unexcused absences. Using data from a large school district in the southeastern United States, the sample includes all students in the district between the years 2006-2012 who also had justice system contact during the same period (n=27,568).  Uniquely, the school and justice system data were matched using an individual code that links the two data sources but protects the individuals’ identities.

Results: After controlling for individual- and school-level variables, in the lower latent class, an African American student is 1.26 times more likely to have higher count of felonies than a Caucasian student, and in the higher latent class an African American student is 1.65 times more likely to have higher count of felonies than a Caucasian student.  Moreover, in the lower latent class, the odds of having a higher count of unexcused absences is lower (0.92 IRR) for African American students compared to Caucasian students whereas in the higher latent class, there is no difference between these groups in terms of unexcused absences. Finally, testing the school disengagement (the current year association of suspensions and unexcused absences with the lag variable of out of school suspensions) hypothesis, findings suggest that school disengagement does not predict future felony bookings.

Conclusions and Implications: Confirming previous research findings, even after controlling for behavioral/legal predictors (absences, disengagement, school based-offenses, and suspensions), the extra-legal variables of biological sex, race/ethnicity, and disability status were stronger predictors of felony bookings than students’ behavior.  Micro, mezzo, and macro implications for education, justice, mental/physical health, and social work are explored including racial equity impact analyses across systems.