Author ORCID Identifier
Date of Award
Master of Public Health (MPH)
Dr. Emily Graybill
Dr. Brian Barger
In 2020, the US status dropout rate was 5.3%, or approximately 2 million people. Those who fail to complete high school tend to experience much poorer life outcomes such as worse health, incarceration, homelessness, and shorter lifespans. School administrative detection and intervention strategies surrounding student engagement, a primary driver of academic success, are imperative to mitigating student dropout rates. Analysis of extant student social emotional behavioral screening results through novel processes like structural equation modeling (SEM), in conjunction with early warning systems (EWS), has the potential to improve student outcomes. More than half of all public schools utilize some form of an early warning system (EWS) to pinpoint youths that are at risk of dropping out. Typically, EWSs are triggered through indicators around students’ attendance, behavior, and course performance (or “ABCs”). While, the ABC method provides schools insight, it represents a student’s historical behavior, with intervention occurring after a child’s mental health has manifested into self-destructive behavior.
This study intends to examine whether a well-established behavioral health screener, the Strengths and Difficulties Questionnaire (SDQ), will align with EWS indicator trends. By utilizing a structural equation model (SEM), the relationships between predictors of race and gender, the SDQ factors of externalizing and internalizing, and the outcomes of attendance and behavior. Better understanding these relationships could allow administrators higher levels of policy refinement while also mitigating poor outcomes for at risk students through early intervention.
3,240 high school students (grades 9-12) among six separate schools in the Southern United States were administered the SDQ at the beginning of the Fall 2021 school year and their emotional and behavioral health was rated through a 25 item Likert scale instrument. Analyses consisted of a SEM path analysis examining the exogenous predictor variables of race and gender, the latent subscale constructs of externalizing and internalizing, and the endogenous outcome variables of attendance (total days missed) and behavior. Behavior was derived through a binomial count of office disciplinary referrals (ODRs). The SEM was conducted through R Lavaan.
Through a SEM lens, the significant associations among the sociodemographic exogenous predictors, latent instrument subscales, and the endogenous outcomes, allowed a view into how well the SDQ could be incorporated into EWSs. The externalizing problem subscales exhibited a greater number of significant associations to EWS indicators than did the internalizing subfactors. Particularly, conduct problems showed a strong positive association to behavior as an outcome. Contrary to prior findings, there was little direct association between sociodemographic variables (of race and gender) and being referred for discipline.
This study’s findings suggest that a modified two factor externalizing SDQ could be incorporated into early warning systems and provide administrators enhanced responsiveness to student disengagement while also improving upon the data used to intervene. However, more study should be employed, with greater demographic heterogeneity, to establish greater confidence in the SDQ instrument as an EWS component.
Lewis, R. Scott, "Using Structural Equation Modeling (SEM) To Assess Whether the Subfactor Constructs of the Strengths and Difficulties Questionnaire (SDQ) Can Be Integrated Into an Early Warning System For At Risk High School Students." Thesis, Georgia State University, 2023.
File Upload Confirmation
Available for download on Friday, May 02, 2025