Experts from REL Mid-Atlantic developed this R package to help state education agencies stabilize school- and subgroup-level accountability measures by reducing the influence of random error. The goal is to produce estimates that are more reliable, especially when results are based on small groups of students.
This package is intended for state education agency analysts and other users who already work in R, an open-source software environment for statistical computing. If you can run R scripts and install packages locally, you can use this code to run stabilizations within your own data systems.
The code is provided as a source package named a4etool that you can download to your local machine and install using the install_local function from the remotes package. The package uses cmdstanr for model fitting. Instructions for installing and configuring cmdstanr are available at: https://mc-stan.org/cmdstanr/.
For guidance on using the a4etool package, please refer to the package’s vignette by installing the package and running this code:
> library(a4etool)
> vignettes(“a4etool”)
To run a stabilization, you will need:
- Values for each measure, subgroup, school, and year you wish to stabilize. These values might include metrics such as the percentage of students achieving proficiency, the median student growth percentile, mean scale scores, or the graduation rate.
- The subgroup size, or number of students, included in each value for each subgroup, school, and year.
- Unique identifiers for each school in your dataset.
