One in five high-school students in Passaic, NJ, did not graduate on time. To design interventions to support them, the district needed to know which students were most at risk of dropping out.
SDP Fellows Tara Chiatovich and Elizabeth Rivera Rodas dove deep into district data and built a model to identify the characteristics that predict which Passaic students are less likely to graduate, as early as 4th grade.
The SDP fellows unearthed local trends that diverge from national data. In various grades, students’ math scores, attendance records, and special-education status were predictive, but their proficiency in English and whether or not they received subsidized school meals were not. The analysis underscored the importance of collecting data locally to craft the right interventions for the right students, rather than relying on more general analyses at the state or national level.
Too Many Dropouts
After years of school-quality reforms, public schools in the city of Passaic, NJ, had boosted the on-time graduation rate from 60 percent to 79 percent. But one in five students dropping out was still far too many. The district needed to create targeted programs to support young students most likely to leave high school. First, it needed to find out who they were.
There is no shortage of research and strategies aimed at keeping high-school students on track. But the accuracy of common dropout flags, such as those from Early Warning Indicator Systems, is poor. And looking at students based on what factors make them generally less likely to graduate on time leaves districts prone to providing the wrong sorts of supports to the wrong students.
Those general improvement strategies had already helped the district improve. But to elevate everyone’s chances of getting to graduation, Passaic needed to analyze its own data to discover which factors among its own students were tied to dropping out.
Power of Prediction
SDP Fellows Tara Chiatovich and Elizabeth Rivera Rodas used a combination of state and district data to track students in the classes of 2011 through 2015 backwards throughout their years in school. They wanted to know: which student characteristics are associated with a greater likelihood of on-time graduation, and when in a student’s career do those factors have the greatest potential impact on his or her success in high school?
Chaitovich and Rivera Rodas pored over state and local data for 3,478 Passaic students, starting in the 4th grade. They looked for trends in attendance records, report cards, and scores on standardized tests. They analyzed their findings by characteristics like gender, race and ethnicity, whether students qualified for subsidized school meals, and if they received services like special education or English as a Second Language classes.
Through rich rounds of analysis, the SDP fellows found common characteristics that were associated with a greater likelihood of students graduating on time. Their unique look across many years of many students’ careers in Passaic schools provided a detailed picture of not only who the district needed to help, but how and when.
Take limited English proficiency (LEP) students—a common characteristic in the city, where 92 percent of students are Hispanic and 23 percent are LEP. Broadly speaking, LEP students nationwide are less likely to graduate high school on time. But in Passaic, Chaitovich and Rivera Rodas found, that’s only true if students are considered LEP in high school. Being LEP in the elementary or middle-school grades, but not in high school, didn’t make students any more or less likely to graduate on time.
Chaitovich and Rivera Rodas found other nuances by parsing data just for the class of 2015. While a student’s overall grade-point average was linked to his or her likelihood of graduating on time, grades for just math and English were not. And attendance was also related to on-time graduation, but specifically, attendance during a student’s freshman and sophomore years of high school.
While the data was not complete—the fellows did not have disciplinary data, for example—it served as a starting point to tailor dropout prevention services to the students who could benefit most. And it also showed the importance of district-level data.
As the fellows said, “The focus of Passaic’s effort to boost on-time graduation should rest on what they can change, and they should use their data to monitor efforts to gauge whether their interventions have the intended effect.”
Read Tara and Elizabeth's full capstone report.