Educational Practices to Identify And Support Students Experiencing Homelessness
Category: Student Well-Being
This paper develops new models to evaluate the effects of interventions and intervention-by-site heterogeneity when treatment group assignment is based on a fallible variable and the outcome of interest is determined in part by the corresponding true control variables (measured without error). The specific application concerns a school report card redesign in which school performance is evaluated based on the achievement growth of students in the bottom quartile of prior achievement. We show using Monte Carlo data that the traditional errors-in-variables estimator (EV) produces severely biased estimates of the gap-closing initiative. We develop an augmented EV estimator (AEV) that addresses this bias and is shown to produce highly accurate estimates in Monte Carlo simulations. We also show how AEV can be implemented using regression calibration (RC). Using state data, we find that there are essentially no differences in the average growth in student achievement between students in and not in the lowest quartile. However, the noise-corrected correlation in school growth estimates for the two groups is high (around 0.8), but not perfect. These findings are important given that most (if not all) state accountability systems prioritize reporting of school performance for multiple student sub-groups, including groups with large gaps in prior student achievement.
Keywords
Achievement gap closing, Evaluation methods, Control for test measurement error
Education level
Topics
Document Object Identifier (DOI)
10.26300/h5e1-v925
