Replication Archive: India Public Opinion Toward the Major Powers

Cambridge Elements in Indo-Pacific Security

Authors: Aidan Milliff & Paul Staniland

Contact: [email protected]

Last updated: 2026

---

Overview

This archive contains all data and code necessary to reproduce the numbered figures in the Element. Running main.R from the project root will reproduce all 28 figures and write them to results/figs/.

Extras

In addition to replication data/code, the repository contains two resources for other researchers interested in historical surveys from the India Institute of Public Opinion.

1. roper_ascii_read.R contains code that can be adapted to read IIOPO data, which comes in ASCII/punchcard format from the Roper Center.
2. IIOPO_FP_Categories.docx and IIOPO_FP_Questions.xlsx contain an index of ALL foreign policy questions, by topic, asked on IIOPO surveys between 1956 and 1988.

---

Requirements

Software:

• R 4.5.3 (see renv.lock for exact package versions)

Package management: This archive uses renv to lock all package versions. To restore the exact environment used to produce the figures, run renv::restore() before sourcing main.R. Packages will also auto-install on first run if missing.

---

Data

Pre-processed data files are in data/. The underlying surveys require institutional access and cannot be redistributed; only the variables necessary to reproduce the figures are included. See source notes below.

File	Description	Source
IND_ADM1.geojson	Indian state boundaries shapefile (for survey location maps)	datameet/maps
gallup_indf.RData	Gallup World Poll, India, 12 waves (2006–2019) — analysis variables only. Cannot be redistributed. To rebuild from raw data, see code/00_build-gallup-indf.R and Obtaining Restricted Data below.	Gallup Organization
pew_clean.RData	Pew Global Attitudes, India, 13 waves (2002–2019). Cannot be redistributed. To rebuild from raw data, see code/00_clean-pew.R and Obtaining Restricted Data below.	Pew Research Center
roper_allyears.RData	Indian Institute of Public Opinion (IIPO/IIOPO) microdata, 40 waves (1974–2001)	Roper Center for Public Opinion Research
dk_rates.csv	Pre-computed "Don't Know" response rates per survey question (summary only)	Derived from Gallup
iiopo-59-88.xlsx	IIOPO topline aggregates, 1959–1988 (pre-microdata)	Roper Center
pew_over_time.xlsx	Pew longitudinal favorability aggregates	Pew Research Center
pew_1618_states.RData	Pew 2016–2018 state identifiers (for maps)	Pew Research Center
pew_0709_cities.RData	Pew 2007–2009 city identifiers (for maps)	Pew Research Center

---

Obtaining Restricted Data

Two datasets require institutional access and cannot be included in this archive. Construction scripts are provided so that researchers who obtain the raw data can rebuild the .RData files used by the analysis scripts.

Gallup World Poll (gallup_indf.RData)

Source: Gallup Organization
What to request: Gallup World Poll microdata for India, waves 2006–2019 (March 2019 release: The_Gallup_032219.dta). Academic data-sharing arrangements are available through the Gallup Analytics portal or via a direct data-sharing agreement with Gallup.

To rebuild:

1. Obtain The_Gallup_032219.dta (or the South Asia subset South Asia Subdata.dta)
2. Open code/00_build-gallup-indf.R and set raw_data_path to your file's location
3. Un-comment and run the corresponding line in main.R

Pew Global Attitudes (pew_clean.RData)

Source: Pew Research Center
What to request: Pew Global Attitudes Survey SPSS (.sav) files for India-inclusive waves: 2002, 2004, 2005, 2007, 2008, 2009, 2010, 2011, 2012, 2014, 2015, 2016, 2017, 2018. Files are freely available from the Pew Research Center data archive after registration.

To rebuild:

1. Download the relevant .sav files and place them in a single directory
2. Open code/00_clean-pew.R and set raw_data_path to that directory
3. Un-comment and run the corresponding line in main.R

---

Reproducing the Figures

Clone or Download the Repository

git clone https://github.com/milliff/lenient-peacock/

Recommended (with renv):

# From the project root
renv::restore()   # installs locked package versions
source("main.R")  # produces all 28 figures in results/figs/

Without renv (packages installed automatically if missing):

source("main.R")

To reproduce a subset of figures, source individual scripts:

library(here)
source(here::here("code/01_data-descriptions.R"))  # fig2-1 through fig2-8
source(here::here("code/02_china-analysis.R"))      # fig3-1 through fig3-9
source(here::here("code/03_us-analysis.R"))         # fig4-1 through fig4-4
source(here::here("code/04_russia-analysis.R"))     # fig5-1 through fig5-4

All scripts use here::here() for file paths and will work correctly as long as R's working directory is set to the project root (the folder containing main.R and .here).

---

Figure Index

Figure	Script	Description
fig2-1	01_data-descriptions.R	Survey respondent counts by year and source
fig2-2	01_data-descriptions.R	IIPO survey locations (1959–2001)
fig2-3	01_data-descriptions.R	Pew and Gallup survey locations (2007–2009, 2016–2018)
fig2-4	01_data-descriptions.R	IIOPO national average approval, 1959–2001 (with pre-microdata)
fig2-5	01_data-descriptions.R	IIOPO city-level approval over time, three countries
fig2-6a	01_data-descriptions.R	Gallup leadership approval over time, by region
fig2-6b	01_data-descriptions.R	Pew favorability over time, by region
fig2-7	01_data-descriptions.R	Distribution of "Don't Know" rates across foreign policy questions
fig2-8	01_data-descriptions.R	Predicted DK count by region and SES
fig3-1	02_china-analysis.R	China approval over time, three sources combined
fig3-2	02_china-analysis.R	Sumdorong Chu crisis: predicted probability (IIOPO)
fig3-3a	02_china-analysis.R	Depsang standoff: predicted probability (Pew)
fig3-3b	02_china-analysis.R	Depsang standoff: predicted probability (Gallup)
fig3-4a	02_china-analysis.R	Doklam crisis: predicted probability (Pew)
fig3-4b	02_china-analysis.R	Doklam crisis: predicted probability (Gallup)
fig3-5	02_china-analysis.R	Heterogeneous treatment effects (interflex), China
fig3-6	02_china-analysis.R	Regional variation in China approval
fig3-7	02_china-analysis.R	Party identification and China approval (Pew)
fig3-8	02_china-analysis.R	Party-in-power effects on China approval
fig3-9	02_china-analysis.R	Interaction effects by party (China)
fig4-1	03_us-analysis.R	U.S. approval over time, three sources combined
fig4-2	03_us-analysis.R	Pew and Gallup U.S. approval combined time series
fig4-3	03_us-analysis.R	Regional variation in U.S. approval
fig4-4	03_us-analysis.R	Interaction effects by region (U.S.)
fig5-1	04_russia-analysis.R	Russia/USSR approval over time, three sources combined
fig5-2	04_russia-analysis.R	Pew and Gallup Russia approval combined time series
fig5-3	04_russia-analysis.R	Regional variation in Russia/USSR approval
fig5-4	04_russia-analysis.R	Interaction effects by party (Russia)

---

Directory Structure

replication-material/
├── README.md                    # This file
├── main.R                       # Master run script
├── code/
│   ├── 00_build-gallup-indf.R   # Rebuilds gallup_indf.RData from raw GWP data
│   ├── 00_clean-pew.R           # Rebuilds pew_clean.RData from raw Pew data
│   ├── 01_data-descriptions.R   # Part 2: survey coverage and descriptive plots
│   ├── 02_china-analysis.R      # Part 3: China approval analysis
│   ├── 03_us-analysis.R         # Part 4: U.S. approval analysis
│   └── 04_russia-analysis.R     # Part 5: Russia/USSR approval analysis
├── data/                        # Pre-processed data files (see Data section)
├── renv/                        # renv package cache
├── renv.lock                    # Locked package versions
└── results/
    └── figs/                    # Output: numbered publication figures (populated by main.R)

---

Why is it called this?

In order to keep project names unique and memorable, names for my github repositories come from a custom generator. You can use it too if you like. See: https://aidanmilliff.com/post/project-name-generator/