https://github.com/sodascience/schools_synth_test

Test code for disaggregated synthetic control with simulated data
https://github.com/sodascience/schools_synth_test

causal-inference simulation statistics synthetic-control

Last synced: about 1 year ago
JSON representation

Test code for disaggregated synthetic control with simulated data

• Host: GitHub
• URL: https://github.com/sodascience/schools_synth_test
• Owner: sodascience
• License: mit
• Created: 2023-07-14T06:51:32.000Z (almost 3 years ago)
• Default Branch: main
• Last Pushed: 2024-01-17T09:28:56.000Z (over 2 years ago)
• Last Synced: 2025-01-25T12:09:36.638Z (over 1 year ago)
• Topics: causal-inference, simulation, statistics, synthetic-control
• Language: R
• Homepage:
• Size: 11 MB
• Stars: 0
• Watchers: 3
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

          
# Disaggregated synthetic control simulation

In this repository, we simulate data on a school intervention and we try to recover the average causal effect using a disaggregated synthetic control method. In this method, we create a synthetic control for each school and then average over schools.

The script [`01_simulate_data.R`](./01_simulate_data.R) contains the code to create the simulated school data, which results in the following simulated effect:

![Simulated causal effect on test scores for a subset of schools](img/simulated_effect.png)

The script [`02_disaggregated_synthetic_control.R`](./02_disaggregated_synthetic_control.R) contains the code to analyse this data and produce school-level effect plots like this:

![The estimated causal effect for a single school](img/trend_plots/trend_ccdf7c91a940.png)

Additionally, there is code to investigate covariate balance for the schools:

![Plot of covariate balance](img/balance_plot.png)

## Contact

This is a project by the [ODISSEI Social Data Science (SoDa)](https://odissei-data.nl/nl/soda/) team.

Do you have questions, suggestions, or remarks on the technical implementation? File an issue in the

issue tracker or feel free to contact [Erik-Jan van Kesteren](https://github.com/vankesteren).