Public repo to demonstrate using Stata's IRT functionalities to explore item response theory on a real dataset.
The data in the repo is a partial subset of the following dataset: Daniel O. Gilligan; Naureen Karachiwalla; Ibrahim Kasirye; Adrienne M. Lucas; Derek Neal, 2021, "Replication Data for: Educator Incentives and Educational Triage in Rural Primary Schools", https://doi.org/10.7910/DVN/FJOL7N, Harvard Dataverse, V1, UNF:6:9T9h8aWC5AXitExRjEjZPg== [fileUNF]
We thank the authors for making their data public.
The main script is "IRT_demo.do" -- this contains the exercise accompanying the coding lunch presentation.
"make_data.do" creates the subsetted data used in the exercise
This is meant as an exploration of Stat's IRT capabilities, NOT a best practices for calculating test scores based on IRT OR a real analysis of the underlying data.