This streamlit app demonstrates a heuristic for standardized test reliability by allowing the user to explore the relationship between the standard error of measurement of a test like the LSAT and the number of "buckets" that the test can sort students into at the chosen confidence level.
I've striven to make the app self-documenting, but if you want a bit more information on the statistics involved, please see my blog post
If you just want to run this app without cloning the repo, you can do so at http://testbuckets.streamlit.app. Please wait a few moments for the app to wake up.
#First, set up a virtual environment and activate it
python3 -m venv test_buckets
source test_buckets/bin/activate
#Install requirements
pip install -r requirements.txt
#run the app
streamlit run standardized_test_dashboard/main.py