Administrative Algorithmic Dropout Targeting
Observational StudyClinical Trial
Public school systems lack objective, scalable mechanisms to identify students at risk of dropping out before upper secondary enrollment, delaying support until academic failure has already occurred.
Picture this
Instead of waiting for students to start failing classes, administrators analyze prior middle school report cards and standardized admission test scores before the academic year begins to calculate an individual risk score. This functions like a health screening questionnaire at a hospital lobby that flags patients with elevated blood pressure before an acute cardiac event takes place.
What the evidence says
Lower secondary GPA (coefficient = -0.174, p < 0.01) and standardized entrance examination scores (coefficient = -0.003, p < 0.01) accounted for 19.0% of the total variation in upper secondary completion rates.
- Who was studied
- N = 186,293 lower secondary graduates across Mexico City tracked longitudinally from 2008–2009 to 2012.
- How
- Ordinary Least Squares (OLS) regression analysis predicting on-time upper secondary completion using administrative entrance examination and GPA data.
What to do
Deploy automated screening algorithms using baseline administrative GPA and entrance examination scores to identify and rank the top 20% most vulnerable incoming high school students prior to academic year orientation.
From the source
"Once we condition on school fixed effects, GPA and the COMIPEMS admission exams could explain 19% of the total variation in the probability of completing, on-time, upper secondary."
Addressing High School Dropouts with a Scalable Intervention: The Case of PODER