BART Fare Gate Installation and Ridership Analysis

Overview

This repository contains the full replication materials for an empirical analysis of BART fare gate installations and their effects on station-level ridership. It is structured to support end-to-end reproducibility, including raw data, analysis scripts/notebooks, and the final paper.

The repository is intended for academic replication, evaluation, and review. After unpacking the data archive, all scripts in scripts/notebooks/ should run without modification, subject to the software requirements listed below.

---

Repository Structure

BART_Fare_Gates/
├── data/
│   └── bart_data.zip              # Replication dataset (Git LFS tracked)
├── scripts/
│   └── notebooks/
│       ├── Regression_Analysis.ipynb
│       ├── Regression_Analysis_monthly.ipynb
│       ├── bart_data_cleaning.do
│       ├── bart_descriptive_stats.do
│       └── ...
├── AustinCoffelt_TermBARTPaper.pdf # Final paper
├── .gitignore
└── README.md

Notes:

• Generated figures, tables, and logs are intentionally excluded from version control.
• The data archive is tracked using Git LFS.

---

Data Access and Setup

Step 1: Clone the repository

git clone https://github.com/austin7384/BART_Fare_Gates.git
cd BART_Fare_Gates

If using Git LFS for the first time:

git lfs install
git lfs pull

Step 2: Unzip the data

Unzip the replication data in place so that the directory structure matches what the scripts expect:

cd data
unzip bart_data.zip
cd ..

After unzipping, the data/ directory should contain the raw and intermediate data files referenced by the notebooks and Stata scripts.

---

Running the Analysis

All analysis code is located in:

scripts/notebooks/

The workflow proceeds in three logical stages:

1. Data cleaning and construction

   • bart_data_cleaning.do

2. Descriptive statistics

   • bart_descriptive_stats.do

3. Regression and event-study analysis

   • Regression_Analysis.ipynb
   • Regression_Analysis_monthly.ipynb

The notebooks assume that the data has already been prepared by the Stata scripts. Output paths are relative and will be created automatically if they do not exist.

---

Software Requirements

The analysis was developed and run using the following environment:

• Stata (for .do files)

• Python 3.9+

  • pandas
  • numpy
  • matplotlib
  • statsmodels
  • jupyter

A standard scientific Python stack is sufficient. No proprietary Python packages are required.

---

Replication Notes

• All file paths are relative; the project should run as-is once the data archive is unzipped.
• Randomness is not used in estimation; results should be exactly reproducible.
• Figures and tables in the paper can be regenerated from the scripts, though generated outputs are not tracked in Git.

---

Paper

The final paper is available in the repository root:

AustinCoffelt_TermBARTPaper.pdf

---

Contact

For questions regarding replication or code, please contact:

Austin Coffelt

---

License

This repository is provided for academic and research purposes. Please cite appropriately if using or extending this work.