McFatties
Open the map

A county-level data explorer

Do counties with more McDonald's also have higher obesity rates?

Explore county-level adult obesity prevalence, McDonald's location density, and the correlation patterns between them across the United States.

This site explores correlation, not proof of causation. Obesity has many drivers — income, diet, activity, food access, and environment.

Explore the map How the data was built

The map

Every U.S. county, colored by what you choose

Hover a county for details. Click to lock a county for the side panel.

The correlation

McDonald's per 100k vs. adult obesity

Each dot is a county with at least 5,000 residents. The trendline shows a weak positive association — not a causal arrow.

McDonald's per 100,000 residents → ← Adult obesity prevalence (%)

Reading this chart

  • Weak positive trend. Counties with more McDonald's tend to have modestly higher obesity, but the scatter is wide.
  • Outliers are the story. Holmes County, MS has over 50% adult obesity and zero McDonald's. Manhattan has low obesity and high density.
  • Density proxies a lot. McDonald's density also tracks urbanization, commuter traffic, and tourism — not just food environment.
  • Correlation is not causation. Omitted variables like income, activity, and food access likely drive most of the signal.

The rankings

Who sits at the extremes?

Counties are filtered to ≥10,000 residents so per-capita density stays meaningful.

#CountyStateObesity %McD / 100kMcD countPopulation

The insights

Five things the data actually says

Which counties stand out the most?

The extremes are rarely the places you'd guess. Tourist-heavy small counties rack up the highest McDonald's-per-resident numbers because stores serve commuters, not just locals. Meanwhile, the highest-obesity counties cluster in the rural Deep South — often in places with fewer chain restaurants, not more.

Where density and obesity diverge

A meaningful fraction of counties are "mismatches": high obesity with below-median McDonald's density, or vice versa. These outliers are the clearest reason to resist simple causal stories. A restaurant chain is a visible variable, but visible is not the same as causal.

Why correlation is not causation

McDonald's density correlates with urbanization, highway access, tourism, commuter flows, and income. Obesity correlates with income, activity, access to healthy food, and cultural patterns. When two variables share upstream drivers, a positive correlation between them is expected — and explains very little on its own.

The method

How McFatties was built — and what it isn't

Data sources

  • Obesity prevalence: CDC PLACES: Local Data for Better Health, county-level OBESITY measure (model-based adult BMI ≥ 30).
  • McDonald's locations: gavinr/usa-mcdonalds-locations (GeoJSON of U.S. McDonald's store points).
  • County boundaries: topojson/us-atlas counties-10m.json (Census cartographic boundaries, 2017 edition).
  • Population: totalpopulation field from the PLACES county dataset, used for per-capita normalization.

Processing

  1. Deduplicated PLACES records by 5-digit county FIPS, keeping the most populated record per county.
  2. Decoded the TopoJSON county polygons into Shapely geometries and built an STR-tree spatial index.
  3. Performed a point-in-polygon spatial join to assign every McDonald's location to a county (with a short fallback tolerance for coastal stores).
  4. Computed McDonald's per 100,000 residents per county and state-level aggregates.
  5. Computed Pearson correlation (r) and an OLS trendline using counties with ≥ 5,000 residents.
  6. Exported clean JSON for the static frontend.

Caveats and limitations

  • Correlation is not causation. This project does not estimate a causal effect of McDonald's density on obesity.
  • Store density is a proxy for many things — urbanization, commuter flow, tourism, and poverty — not just food environment.
  • County-level data hides neighborhood variation. A county can be "low obesity" overall while containing high-obesity census tracts.
  • Obesity is multi-causal. Diet, physical activity, income, sleep, stress, genetics, and community environment all matter.
  • The McDonald's dataset is community-maintained and may lag recent openings and closures.
  • PLACES estimates are modeled, not direct measurements. They are the best available small-area obesity estimates but carry modeling uncertainty.

What McFatties is

A correlation explorer and an invitation to look at the data honestly. The goal is to make county-level variation tangible, not to argue a causal case.