Predictive model of near-term academic risk combining school and child welfare data¶

Element · All elements
Evidence · 5 claims (5 for) · 2 studies (2 associational), q2 · 0 of 2 report an effect size · 5 claims rest on one study

Description¶

The article describes a predictive model that uses both school data and child welfare data to flag students at risk of near-term academic problems. The fact sheet states it "describes predictors related to near-term academic problems, meaning academic problems in the next quarter or semester," and that the companion report "details the approach used to develop a predictive model." The model's purpose is early identification of at-risk students so educators can intervene within the coming academic term.

Design Implications¶

Context¶

Requirements¶

• Access to linked school and child welfare data sources to compute the predictors

Constraints¶

• The model targets near-term academic problems, defined as problems in the next quarter or semester, not longer-horizon outcomes

Target Learners¶

• K-12 students whose records appear in school and child welfare data systems

Target Learning Goals¶

• Early identification of students at risk of near-term academic problems so support can be provided within the quarter or semester

Claims¶

• A predictive model built from school and child welfare data identifies students at risk of near-term academic problems [+W]
• A predictive model built on school and child welfare data was assessed for identifying at-risk students in two Allegheny County education agencies [+W]
• A partnership developed an approach for identifying at-risk students using school data linked with child welfare, justice, human services, and public benefits data. [+W]
• Near-term academic problems targeted include absenteeism, suspensions, poor grades, and low state-test performance [+W]
• The report examines how in-school and out-of-school predictors relate individually to each type of near-term academic problem [+W]

Related Elements¶

• Linked school and human services data approach for predicting near-term academic risk
• Machine learning early warning algorithm using in- and out-of-school data to estimate per-student, per-problem, per-quarter risk scores
• Predictive-model approach using school and child welfare agency data to flag near-term academic risk
• Arizona linked education–child welfare student database

Examples¶

• Use linked school and child welfare data to flag students at academic risk in the coming quarter or semester

Key Sources¶

• Regional Educational Laboratory Mid-Atlantic. (2020). Using School and Child Welfare Data to Predict Near-Term Academic Risks: Fact Sheet. Regional Educational Laboratory Mid-Atlantic. https://www.mathematica.org/publications/using-school-and-child-welfare-data-to-predict-near-term-academic-risks-fact-sheet