How It Works

Our Methodology Overview

Our approach uses advanced computational methods to generate and evaluate thousands of redistricting plans. Here's how we ensure fairness, transparency, and objectivity in the process.

Data Collection

Census data, demographics, geographic boundaries

Plan Generation

MCMC algorithms create thousands of options

Evaluation

Multiple fairness metrics applied

Selection

Top plans ranked and presented

Step 1: Data Collection

What Data We Use

Comprehensive coverage: We start with publicly available data to ensure our analysis is based on accurate, up-to-date information
Verifiable sources: All data comes from official government sources that anyone can access and verify
Multi-dimensional analysis: We combine population, geographic, and demographic data for complete coverage
Historical context: Include previous election results and demographic trends for comprehensive analysis

Census Data

Population counts, demographic information, and geographic boundaries from the U.S. Census Bureau

Geographic Boundaries

Precise maps of census blocks, precincts, and existing district boundaries

Demographic Information

Race, ethnicity, age, and other demographic data for fair representation analysis

Historical Data

Previous election results and demographic trends for comprehensive analysis

Step 2: Markov Chain Monte Carlo (MCMC)

How Our Algorithm Works

MCMC methodology: We use Markov Chain Monte Carlo methods to generate thousands of possible redistricting plans
Evolutionary approach: Think of it like evolution - the algorithm starts with one map and gradually improves it
Iterative refinement: Makes small changes, keeping good improvements and discarding bad ones
Statistical robustness: Generates a representative sample of all possible fair redistricting plans

1. Start with a valid map

Begin with any legal redistricting plan that meets basic requirements

2. Make small changes

Move a few census blocks from one district to another

3. Evaluate the change

Check if the new map is still legal and potentially better

4. Accept or reject

Keep good changes, discard bad ones, repeat thousands of times

Step 3: Fairness Evaluation

Multiple metrics: Each generated plan is evaluated using multiple established fairness metrics
Weighted combination: We use a weighted combination of these metrics to rank plans objectively
Comprehensive evaluation: No single metric dominates - all aspects of fairness are considered
Scientific approach: Based on established research and statistical methods

Compactness

Measures how geographically compact each district is. Higher scores mean more logical, round shapes that respect natural boundaries and keep communities together.

Polsby-Popper Score: Measures the ratio of district area to perimeter, with higher scores indicating more compact shapes
Reock Score: Compares district shape to a circle of equal area, with higher scores being more compact
Combined Analysis: A combination of both metrics ensures districts have logical, community-friendly boundaries

Demographic Segregation

Measures how fairly different racial and ethnic groups are distributed across districts.

Dissimilarity Index: Measures racial/ethnic segregation between districts
Separation Index: Measures how isolated minority communities are
Isolation Index: Measures how concentrated minority populations are
Fair representation: Ensures all communities have equal voice in the political process

Population Equality

Ensures each district has similar numbers of people, with deviations kept within legal limits.

Generated plans are within 1% of the target
Minimal adjustment needed once community approves a plan
Legal compliance with federal and state requirements
Community input guides final population balancing

Step 4: Plan Scoring and Ranking

How We Combine Metrics

Weighted geometric mean: We use a weighted geometric mean to combine all fairness metrics into a single score
Balanced evaluation: This ensures that no single metric dominates the evaluation
Comprehensive fairness: All aspects of fairness are considered equally
Percentile scoring: For a given map, each metric is assigned a percentile score
Mathematical rigor: The overall score is a weighted geometric mean of those metrics

Ensuring Quality and Validity

Legal Compliance

All plans meet legal requirements including the Voting Rights Act, equal population standards, and contiguity requirements.

Statistical Validation

We use ensemble analysis to ensure our results are statistically robust and not just random chance.

Transparency

Every step of our process is documented and publicly available for review and verification.

Reproducibility

Our methodology can be independently verified and reproduced by other researchers and organizations.

Technical Implementation

Software and Tools

Open-source foundation: Our analysis is built on established, open-source tools and libraries
Proven reliability: All tools are widely used in academic and professional settings
Community support: Active development and maintenance by the open-source community
Transparency: Anyone can examine and verify our methodology

GerryChain: Python library for redistricting analysis

GeoPandas: Geographic data processing

NumPy/SciPy: Mathematical computations

NetworkX: Graph algorithms for district connectivity

View Our Code on GitHub

Our complete codebase and documentation are available on GitHub for transparency and collaboration.

Navigation