Replication Material for "Understanding Majoritarian Violence"

Description

This repository contains the replication materials for the paper:

Milliff, Aidan and Blair Read. 2024. "Understanding Majoritarian Violence: Evidence from Vigilante Violence in India."

The repository provides all the necessary resources and instructions to reproduce the results presented in the paper.

Table of Contents

• Introduction
• Installation
• Replication
• License
• Contact
• Citation

Introduction

Extralegal violence threatens human rights and public order in many societies. These threats are especially acute in India, where mobs often punish transgressions of majority-community norms---like eating beef or pursuing inter-religious or inter-caste romantic relationships---through targeted vigilante violence. We develop a new theory explaining vigilante violence in India, an important case that differs from the patterns of vigilantism observed in other contexts. We use newly-available data to show that spatial variation in vigilante attacks in India is strongly associated with the strength of majoritarian institutions, measured via the presence of majoritarian-ideology schools. We further show that the relationship between majoritarian institutions and vigilante violence is driven by communities where majoritarian institutions have persisted longer, not areas of new expansion. We contextualize these findings by examining three high-profile attacks in India and identifying shared patterns: Though attacks appear spontaneous and mass-driven, they are usually coordinated by a core group of perpetrators affiliated with a durable organization which uses the spectacle of violence to generate a larger group of peripheral participants. While most recent literature on vigilantism finds that mob justice substitutes for state law enforcement, we find that vigilantism in India is associated with the rise of para-statal organizations operating even where there are no state capacity vacuums to fill. Because vigilante violence in India is not a spontaneous, grass-roots response to state weakness, we caution that efforts to increase the capacity of local law enforcement or the legal system are unlikely to curb majoritarian violence.

Installation

We use the renv package to create an "environment" that makes it easy to install (correct versions of) all the software necessary to reproduce our analysis.

To set up the correct environment, simply:

1. Clone this repository

usethis::create_from_github(
  'milliff/milliff-read-replication',
  destdir = "[YOUR TARGET DIRECTORY]",
  fork = F,
  protocol = 'https'
)

2. Run the following commands to load all the required packages (listed in /milliff-read-repository/renv.lock) into your environment

renv::init()
renv::restore()

Note: We used R 4.3.1 (Beagle Scouts) on an aarch64-apple-darwin20 (64-bit) platform running macOS Sonoma 14.5 for all the results in the paper.

Replication

Each numbered item in the list below refers to an individual source code file in /milliff-read-replication/code/. For each file, we provide a brief description of the analyses, list the data sources it calls from /milliff-read-replication/data/, and the outputs it creates in /milliff-read-replication/results/.

1. 01-hcw-descriptive-statistics.R

   • Description: Summarizes hate crime data from India Religious Hate Crime Watch
   • Data: hc_locs.csv; subdistricts.json
   • Outputs: Figure 1; Figure 2B; Figure A.1; Table A.1; Supplementary Figures

2. 02-vb-descriptive-statistics.R

   • Description: Summarizes Vidhya Bharati schools data
   • Data: vb_locs.csv; subdistricts.json
   • Outputs: Figure 2a; Figure 5; Figure A.2; Table A.2

3. 03-main-estimation.R

   • Description: Estimates primary results (relationship between VB schools and hate crimes)
   • Data: main_df.RData
   • Outputs: Figure 3; Figure 4; Table A.3; Table A.4; Table A.5

4. 04-treatment-duration-estimation.R

   • Description: Estimates VB-hate crime relationship as a function of VB duration in community
   • Data: main_df.RData
   • Outputs: Figure 6; Figure 7; Figure 8; Figure A.3; Table A.8

5. 05-additional-estimation.R

   • Description: Re-estimates main results, showing robustness to different functional forms and model families
   • Data: main_df.RData
   • Outputs: Table A.6; Table A.7

6. 06-state-weakness.R

   • Description: Builds/compares measures of state capacity, a key alternative explanation
   • Data: main_df.RData
   • Outputs: Figure 9; Table A.10; Table A.11; Table A.12

7. 07-vb-placement.R

   • Description: Models placement of VB schools, the main IV
   • Data: main_df.RData
   • Outputs: Table A.13; Table A.14

8. 08-diff-in-diff-estimation.R

   • Description: Estimates VB-Hate Crime association as staggered-treatment difference-in-differences
   • Data: did_df.RData
   • Outputs: Figure A.6; Figure A.7; Table A.21

9. 09-matching-estimation.R

   • Description: Estimates VB-Hate Crime association using matching and balancing
   • Data: main_df.RData; genmatch.RData; nnmatch.RData
   • Outputs: Figure A.8; Figure A.9; Table A.22

10. 10-alternative-explanations.R

    • Description: Addresses additional alternative explanations including school expansion (non VB), perpetrator impunity (via governing party), schools following HCs
    • Data: main_df.RData
    • Outputs: Figure A.4; Table A.15; Table A.16, Table A.17; Table A.18; Table A.19

11. 11-effects-pre-2014.R

    • Description: Separates estimation into pre- and post-2014 election
    • Data: main_df.RData
    • Outputs: Figure A.5; Table A.20

12. 12-scaling-estimation.R

    • Description: Replicates 3 (main-estimation.R) with scaled variables to ease effect interpretation
    • Data: main_df.RData
    • Outputs: Table A.9

13. __data-creation-pseudocode.R

    • Description: Provides instructions for re-creating main datasets
    • Data: Does not include data; Many constituent data-sets cannot be reshared in original format
    • Outputs: main_df.RData; did_df.RData

License

The software in this replication package is licensed under CC BY-NC 4.0. To view a copy of this license, visit https://creativecommons.org/licenses/by-nc/4.0/.

This license requires that reusers give credit to the creator. It allows reusers to distribute, remix, adapt, and build upon the material in any medium or format, for noncommercial purposes only.

Contact

Direct questions, complaints, and compliments to Aidan Milliff

Citation

If you use this replication package or data in your research, please cite the original paper:

Milliff, Aidan and Blair Read. 2024. "Understanding Majoritarian Violence: Evidence from Vigilante Violence in India."