Skip to content

README.md

Latest commit

 

History

History
151 lines (102 loc) · 4.41 KB

README.md

File metadata and controls

151 lines (102 loc) · 4.41 KB
Raw

eicompare_stata

Stata wrapper for the eiCompare R package via rcall. Provides Stata commands for ecological inference (EI) analysis commonly used in racially polarized voting (RPV) litigation and research.

Overview

eicompare_stata brings the full power of the eiCompare R package to Stata users without requiring any R programming. It supports three standard EI methods:

  • Goodman's ecological regression — fast, linear, can produce out-of-bounds estimates
  • Iterative EI (King's method) — bounded 2×2 EI with uncertainty measures
  • RxC Bayesian EI — joint multi-candidate, multi-group MCMC estimation

The package also provides commands for side-by-side method comparison and visualization.

Requirements

Requirement Minimum Version
Stata 14.0+
R 4.0+
rcall (Stata) latest
eiCompare (R) 3.0.6+

Installation

Step 1: Install rcall

github install haghish/rcall

If you don't have the github command:

net install github, from("https://haghish.github.io/github/")
github install haghish/rcall

Step 2: Install eicompare_stata

github install RPVote/eicompare_stata

Or from a local copy:

net install eicompare_stata, from("/path/to/eicompare_stata")

Step 3: Verify setup and install R dependencies

ei_setup, install

This checks that R is accessible, verifies rcall works, and installs the eiCompare, ei, and eiPack R packages if they are missing.

Commands

Command Description
ei_setup Check and install dependencies
ei_good Goodman's ecological regression
ei_iter Iterative EI (King's method)
ei_rxc RxC Bayesian EI (MCMC)
ei_compare Compare results across methods
ei_plot Visualize EI results

Quick Start

* Verify dependencies
ei_setup

* Load your election data (precinct-level with vote shares and demographic shares)
use my_election_data, clear

* Run Goodman's regression
ei_good cand_a cand_b, races(pct_white pct_black pct_hisp) totals(total_votes)

* Run iterative EI
ei_iter cand_a cand_b, races(pct_white pct_black pct_hisp) totals(total_votes) seed(12345)

* Run RxC Bayesian EI
ei_rxc cand_a cand_b, races(pct_white pct_black pct_hisp) totals(total_votes) seed(12345)

* Compare all three methods
ei_compare cand_a cand_b, races(pct_white pct_black pct_hisp) totals(total_votes) seed(12345)

* Generate a comparison plot
ei_plot cand_a cand_b, races(pct_white pct_black pct_hisp) totals(total_votes) ///
    saving(ei_results.pdf) seed(12345)

See the full vignette for a detailed walkthrough with real-world guidance.

Vignette

The package includes a comprehensive vignette covering:

  • Data preparation and variable requirements
  • Running each EI method with explanation of parameters
  • Interpreting results
  • Comparing methods for RPV analysis
  • Visualization options
  • Practical tips for Voting Rights Act Section 2 litigation

Read the full vignette: eicompare_stata_vignette.md

How It Works

Each Stata command:

  1. Exports the current dataset to a temporary CSV
  2. Calls the corresponding eiCompare R function via rcall
  3. Imports the results back into Stata
  4. Displays a formatted results table
  5. Stores key values in r() for programmatic use

No R coding is required — the .ado files handle all R code generation internally.

Data Requirements

Your dataset should contain precinct-level data with:

  • Candidate variables: Vote shares as proportions (0 to 1) for each candidate. Must sum to 1 within each precinct.
  • Race variables: Demographic group shares as proportions (0 to 1). Must sum to 1 within each precinct.
  • Totals variable: Total number of votes (or voting-age population) per precinct.

Citation

If you use this package, please cite both eicompare_stata and the underlying R package:

Collingwood, L. (2026). eicompare_stata: Stata wrapper for the eiCompare R package.
  https://github.com/RPVote/eicompare_stata

Collingwood, L., Oskooii, K., Garcia, S., & Barreto, M. (2020). eiCompare: Comparing
  ecological inference estimates across EI and EI:RC. R package.
  https://github.com/RPVote/eiCompare

Author

Loren Collingwood, University of New Mexico Email: lcollingwood@unm.edu

License

MIT