Mixed-Effects Location Scale Models for Joint Modeling School Value-Added Effects on the Mean and Variance of Student...

Area of Science:

• Educational Measurement and Statistics
• Quantitative Educational Research

Background:

• School value-added models (VAMs) are standard for evaluating school performance based on student learning gains.
• Traditional VAMs use mixed-effects models to estimate average student achievement, controlling for prior achievement and background factors.
• These models provide a 'school value-added score' representing mean student achievement differences.

Purpose of the Study:

• To extend traditional school value-added models by analyzing the variance of student achievement within schools.
• To identify schools with unusually high or low variability in student achievement, beyond average performance.
• To offer new insights for school accountability systems and educational research.

Main Methods:

• Fitting mixed-effects location scale models.
• Extending traditional mixed-effects linear regression VAMs to simultaneously model the mean and variance of student achievement.
• Analyzing student achievement data to estimate both central tendency and dispersion at the school level.

Main Results:

• The proposed method allows for the estimation and study of school-specific achievement variance.
• Identified schools exhibiting statistically significant high or low variability in student achievement, independent of average VAM scores.
• Demonstrated the feasibility of applying location-scale models to VAMs for richer school performance analysis.

Conclusions:

• Analyzing achievement variance in addition to the mean provides a more comprehensive understanding of school effectiveness.
• This approach can identify schools with distinct achievement patterns (e.g., high consistency vs. high variability).
• Findings have significant implications for refining school accountability metrics and guiding educational interventions.