Confirmatory Factor Structure and Predictive Validity of the Early Identification System—Student Report in a Community Sample of High School Students

The severity and prevalence of youth mental health struggles continue to increase. Many of those youth will not access effective mental health services. In addition, the stress of the COVID-19 health crisis has required students to attend school remotely, leaving less face-to-face interactions with struggling students. Thus, regular screening practices for mental health-related concerns for youths in schools should take priority. If youths are not screened properly, social, emotional, and behavioral health needs are likely to remain unaddressed, leading to significant challenges for educational professionals. For example, unattended mental health problems contribute to behavioral disruptions, and in turn, contribute to declines in teacher self-efficacy, emotional exhaustion, and burnout. To adopt effective school-based tiered models of prevention requires universal screening for social, emotional, and behavioral health as a first step. The purpose of the present study is to confirm the factor structure, examine the invariance, and investigate the predictive validity using disciplinary data for 5,262 high school students who completed the Early Identification System—Student Response (EIS-SR). The EIS provides a dashboard that permits school personnel to observe students who are in the top tier of risk (5%), students who are at moderate levels of risk (20%), and patterns of risk by grade level and at the universal school level. The present study uses a large sample of public high school students (grades 9 – 12) in a Midwestern state (n = 5,262). There are three research questions for this study: (1) using a split sample confirmatory analysis approach, would the EIS-SR replicate earlier observations of the factor structure in a pure high school sample? (2) would the EIS-SR have any measurement invariance by grade level and gender within the sample? (3) would EIS-SR scores predict later student-level discipline reports, attendance, bullying, and suspension data? Based upon earlier analyses with separate samples, we hypothesized that the EIS-SR to factor similarly as observed, to be invariant with regard to grade and gender, and to be predictive of student outcomes. Analyses were conducted over three phases. First, we confirmed the factor structure of the EIS-SR. Second, we investigated the degree of measurement invariance by gender and grade level. Lastly, we examined the predictive validity of the EIS-SR on outcomes. Our analyses indicated the EIS-SR did fit the previously observed factor structure with the items loading on seven distinct scales. Tests for measurement invariance indicated support that the EIS-SR measured the seven factors equally well regardless of both gender and grade level. EIS-SR subscale scores predicted spring office disciplinary referrals, both in and out of school suspensions, and attendance.