The Atlanta Loneliness & Third-Place Index
We pulled every census tract in the five-county Atlanta metro, joined the CDC's loneliness estimates to an independent map of where community gathering spaces actually exist, and looked at what happens where a "third place" is missing.
The headline number
Across the 1,006 census tracts that make up Fulton, DeKalb, Cobb, Gwinnett, and Clayton counties, the population-weighted average share of adults reporting feelings of loneliness is 38.27%, based on the CDC's most recent PLACES estimates. That means well over a third of the metro's adult population lives in a tract where loneliness is the modeled norm, not the exception.
How does that stack up against the rest of the country? We ran the same population-weighted method on 16 major U.S. metros: Atlanta lands squarely in the middle of the pack at #7, well behind the loneliest major metro (Baltimore, at 41.2%) but still ahead of ten others. See the full Loneliest Big Cities in America ranking.
Which counties have it worst
Loneliness isn't evenly spread across the metro. Clayton County's tracts run more than seven points higher than Cobb's, a gap wide enough to matter for anyone deciding where to put outreach money or a new community center.
| County | Tracts | Avg. loneliness (per tract) |
|---|---|---|
| Clayton | 70 | 43.57% |
| DeKalb | 203 | 39.01% |
| Gwinnett | 220 | 38.94% |
| Fulton | 327 | 37.38% |
| Cobb | 186 | 35.89% |
The third-place connection
We joined the loneliness numbers to the Third Place Index, an independent project that scores every U.S. census tract on how many community gathering spaces it has, parks, cafes, libraries, faith spaces, and similar low-cost places to be around other people without an agenda. Across all 1,006 Atlanta-metro tracts, third-place access and loneliness move in opposite directions: the correlation is a moderate r = −0.23.
Split the metro into quartiles by third-place access and the gap is concrete. Tracts in the bottom quartile for third-place access average 39.91% loneliness. Tracts in the top quartile average 36.97%, a gap of 2.9 percentage points between the tracts with the least community-space access and the ones with the most.
Explore the data yourself
Every one of the 1,006 tracts below, mapped and charted. Switch what the map colors by, see how third-place access and loneliness actually relate tract-by-tract, and search or sort the full table. This loads on scroll and pulls from the same joined dataset linked above.
Bubble size is population. Click a tract for details, or click the map once to enable scroll-to-zoom.
Third-place access vs. loneliness
Every tract, one dot each, colored by county. Downward slope = the r = −0.23 correlation, visible.
Average loneliness by county
Population-weighted per-tract average, five counties.
Calculating…
| Tract (GEOID) | County | Loneliness | Third-place %ile | Population | Median income | % Bachelor's+ |
|---|
What's in this data, and what isn't
Loneliness figures come from the CDC's PLACES project (2025 release, using 2023 Behavioral Risk Factor Surveillance System data), which models tract-level prevalence from a national health survey rather than a direct census count, the standard, peer-reviewed way small-area health estimates are produced when a full census isn't possible. Third-place scores come from the Third Place Index, built by data communicator Evan O'Neil from OpenStreetMap location data, scored on a 0–100 percentile scale within comparable community types. Neither dataset originates from NexSpark, a dating app, or a friendship app; both are public and independently produced.
What we did: pulled the loneliness estimate and total population for every tract in the five counties from CDC PLACES, pulled the matching third-place percentile, place count, and other tract attributes from the Third Place Index bulk data, joined the two on census tract GEOID, and computed the population-weighted averages and correlation above. All 1,006 tracts in the five counties matched on both sides, no tracts were dropped or filtered out.
For journalists and researchers
The full joined dataset is public and free to query, no key required: the county-level summary and the tract-level rows (add ?county=Fulton, DeKalb, Cobb, Gwinnett, or Clayton to filter). Both source datasets are public: CDC PLACES on data.cdc.gov and the Third Place Index. We're glad to answer questions about the methodology, drop us a line via nexspark.org.
Common questions
Where does this data come from?
Two public sources, joined at the census-tract level. Loneliness prevalence is from the CDC's PLACES project, which models self-reported loneliness from the Behavioral Risk Factor Surveillance System (2025 release, 2023 data). Third-place access is from the Third Place Index, an independent project by Evan O’Neil that scores every U.S. census tract on the number of community gathering spaces (parks, cafes, libraries, faith spaces, and similar) drawn from OpenStreetMap. Neither dataset comes from NexSpark or any dating or friendship app.
Does low third-place access cause higher loneliness in Atlanta?
This data shows a correlation, not a causal link. Tracts in the bottom quartile for third-place access average 2.9 percentage points higher loneliness than tracts in the top quartile, and the overall correlation across all 1,006 tracts is a moderate negative r = −0.23. That is consistent with the third-place theory of loneliness but doesn’t prove it: income, age, commute patterns, and dozens of other factors vary across the same tracts and aren’t controlled for here.
Can I download or verify the underlying data myself?
Yes. The full joined dataset, all 1,006 tracts, is available as JSON at api.nexspark.com/api/atlanta-data/rows, and the county-level summary is at api.nexspark.com/api/atlanta-data/summary. Both are public, unauthenticated, and free to query. The source data itself is public too: CDC PLACES at data.cdc.gov and the Third Place Index at thirdplaceindex.org.
Why did NexSpark build this?
NexSpark Friends helps Atlanta residents find a small, matched Circle of friends, and the third-place shortage is part of why that’s gotten harder to do on your own. We wanted a real, tract-level picture of where that shortage is worst before writing about it, so we pulled the public data and built one.
Keep reading
The data
The Loneliest Big Cities in America
The same population-weighted method, run on 16 major U.S. metros. Atlanta ranks #7; see who's #1.
The data
Run clubs are the new third place
Why a 6am jog is doing the work a barstool used to.
The data
The friendship recession, explained
12% of adults now have no close friends. What it takes to opt out.
Loneliness doesn't stop at friendship.
NexSpark Matchmaking uses the same honest, in-person-first approach for dating. See how matchmaking compares to dating apps.