The Third-Place Mechanism: What Would Actually Test It

Bowling centers in the United States fell by 32% between 2005 and 2023, and league membership is down 89% from its peak, according to reporting by Fortune. Physical visits to public libraries dropped from roughly 1.25 billion in 2019 to about 671 million in 2022. Elks lodges have gone from over two thousand chapters to fewer than eighteen hundred, with membership halved since 1976. These figures, alongside similar counts for diners, movie theaters, and bars, are now routinely cited as evidence for a specific causal story: America has fewer third places, so people have less informal social contact, so loneliness rises, so health outcomes worsen.

Each piece of that chain has published support. None of the sources establish the chain as a whole, and the gap between “plausible mechanism” and “demonstrated mechanism” is worth being precise about, because a policy response built on the wrong link in the chain will not work.

The concept comes from sociology, not epidemiology

Ray Oldenburg coined the term third place in 1989 to describe informal public gathering spots — cafes, barbershops, taverns — distinct from home and work, and argued they were the primary infrastructure of casual civic life. Robert Putnam’s 2000 account of declining civic participation extended the argument into a broader thesis about the erosion of American social capital. Eric Klinenberg’s 2018 book pushed further, arguing that libraries and parks are “social infrastructure” that measurably shapes contact rates, and used the 1995 Chicago heat wave to show that neighborhoods with more of this infrastructure had lower mortality during the crisis.

That is suggestive, not causal in the sense a health researcher would require. The heat wave comparison is a natural experiment of sorts, but it compares neighborhoods that likely differed on many dimensions besides library and church density.

The recent data confirms decline, not mechanism

A 2025 study in Health & Place tracked twelve categories of third place — coffee shops, libraries, museums, recreation centers, restaurants — and found closures across every category between 2019 and 2021, concentrated in census tracts with higher social vulnerability and in rural areas. The University of Colorado Boulder’s Institute of Behavioral Science summarized this work as a warning about health and equity risks, framing the loss of senior centers and churches as a driver of loneliness rather than a symptom of it. Axios reporting in July 2026 drew the same connection between uneven third-place geography and loneliness prevalence data.

These are strong descriptive findings. They establish that closures are real, accelerating, and unevenly distributed. They do not establish that the tracts losing third places are the tracts where loneliness or isolation is rising fastest, and they do not rule out the reverse possibility: that declining social demand caused the closures rather than the other way around. A bowling alley does not close because loneliness increased; it closes because attendance fell. Attendance falling is itself already a measure of reduced informal social contact, which makes the “closure causes isolation” framing partly circular unless the direction of causation is pinned down independently.

The mortality link is real but comes from a different literature entirely

Julianne Holt-Lunstad’s 2010 meta-analysis of 148 studies covering 308,849 participants found that stronger social relationships were associated with a 50% increased likelihood of survival, an effect size comparable to established risk factors like smoking. Her 2015 follow-up separated the constructs, finding isolation carried an odds ratio of 1.29 for early mortality and loneliness 1.26. The American Heart Association’s 2022 scientific statement, led by Crystal W. Cene, put the increased cardiovascular risk from isolation and loneliness at roughly 30%, with 29% higher risk of heart attack or death from heart disease and 32% higher risk of stroke. The Surgeon General’s 2023 advisory compared the mortality effect to smoking up to fifteen cigarettes a day.

These are robust, replicated numbers. But they measure the relationship between a person’s social connection and their health outcome. They say nothing about what causes variation in social connection in the first place. Third-place closures are one plausible upstream cause among several — remote work is another, and a 2026 Science study found that the rise of remote work explains roughly a third of the increase in isolation and mental distress between 2011-2019 and 2022-2024, with remote workers living alone seeing a roughly tenfold increase in days spent alone compared with workers living with others. If both remote work and third-place decline are operating simultaneously, attributing the isolation increase to either one without separating them overstates confidence in the mechanism a reader is being asked to accept.

What Would Actually Test It

The AHA’s own statement is unusually candid on this point, identifying the absence of intervention evidence as the central research gap in the entire field. Applied to third places specifically, a study that tested the mechanism would need three things the current literature lacks.

First, a design that follows the same tracts over time as third places close or are preserved, measuring loneliness and isolation directly — with a validated instrument such as the UCLA Loneliness Scale rather than inferring it from footfall — before and after the change, rather than comparing tract snapshots at a single point.

Second, a natural experiment exploiting variation that is not itself caused by declining demand: a library closing for a funding formula change unrelated to local usage, for instance, rather than a bowling alley closing because people stopped bowling. Without this, the closure and the isolation it is meant to explain share a common cause and the direction of the arrow stays ambiguous.

Third, a way to separate the third-place channel from competing explanations operating on the same population at the same time, chiefly remote work and the broader decline in face-to-face socializing that Axios’s July 2026 analysis of time-use data documents independently — Americans now average about 35 minutes of daily socializing, down from 45 minutes two decades earlier.

Until a study does that work, the honest description of the evidence is that third-place decline, loneliness, and adverse health outcomes are three well-documented trends running in parallel, with a coherent story connecting them and no design yet capable of confirming it.