When Is a Loneliness Trend Actually a Trend

The American Enterprise Institute’s Survey Center on American Life reported in June 2021 that 12% of Americans have no close friends, up from 3% in 1990. Among men, the share with no close friends rose fivefold over the same period, from 3% to 15%, while the share with six or more close friends fell from 55% to 27%. Those numbers get quoted often, and they should be, because they represent something rarer than they appear: a genuine trend line, not a snapshot dressed up as one.

The distinction matters because most of what circulates as evidence that loneliness among young adults is worsening is not actually a trend at all. It is two different surveys, asking two different questions, of two different populations, placed next to each other and described as before-and-after.

The comparison that holds up

The AEI figures hold up because the 1990 and 2021 numbers came from comparable survey instruments asking about the same underlying construct — number of close friends, self-reported — administered to the same country’s adult population. That is what makes a 31-year gap usable. The survey is not claiming to have tracked loneliness continuously since 1990; it is comparing two points using compatible wording, which is a narrower and more defensible claim.

This is a rarer condition than it sounds. Robert Putnam’s Bowling Alone, published in 2000, was able to document decades of civic decline in the United States largely because it drew on survey series — General Social Survey items, membership records, time-diary studies — that had asked the same questions repeatedly for years. The strength of that book’s central claim rests on that repetition, not on any single striking statistic.

Cigna’s loneliness index offers a smaller-scale version of the same discipline. Its 2020 U.S. report found 61% of adults sometimes or always feeling lonely, a rise of 7 percentage points from the prior year’s administration of the same index. Because Cigna used its own consistent instrument in both years, that 7-point figure is a real within-series movement, not an artifact of switching measures. It is a short series — two years, not three decades — so it says less than the AEI comparison does, but what it says is trustworthy on its own terms.

The comparison that does not

Contrast this with how “loneliness has risen among young adults” claims are typically assembled. The AARP Foundation’s 2018 survey of adults 45 and older used the 20-item UCLA Loneliness Scale, a validated academic instrument, and found one in three older Americans lonely. The Harvard Graduate School of Education’s Making Caring Common project reported in February 2021 that 36% of Americans overall, and 61% of young adults aged 18 to 25, met its own criterion for serious loneliness — a different instrument, a different age framing, a different year, collected under pandemic conditions that the AARP survey did not experience.

Lay the AARP one-in-three figure next to the Harvard 61% figure and the temptation is to read a generational gap, or a pandemic effect, or both. Neither reading is supportable from the numbers alone. The two surveys did not ask the same question. One used a 20-item scale built and validated over decades of psychometric work; the other used a study-specific measure of “serious loneliness” with its own threshold. A difference of that magnitude between two unrelated instruments tells you the instruments differ. It does not tell you the population changed.

The Harvard report compounds this by leaning on retrospective self-report for its trend claim: 43% of young adults said their loneliness had increased since the pandemic began. That is a person’s memory of a change in their own state over roughly a year, collected at a single point, not two independent measurements of the same state taken a year apart. Recalled change and measured change are not interchangeable. People are not reliable narrators of their own trajectories, particularly for a mood state as variable as loneliness, and a single cross-sectional survey asking “has this gotten worse for you” cannot substitute for two comparable readings taken at different times.

What young-adult loneliness research is missing

The practical consequence for anyone trying to understand whether young adults are lonelier now than five or ten years ago is that the data mostly cannot answer the question, because almost none of the instruments being used were designed to be repeated. The AARP survey exists to characterize midlife and older adults at one moment. The Harvard survey exists to characterize the pandemic’s early effects. The Cigna index is the exception that shows what a repeatable design looks like, and even it has only run long enough to establish a two-year movement, not a durable trend.

This is not a small technical quibble. Instrument mismatch produces headline numbers that move by 20 or 30 percentage points between “surveys,” and those swings get reported as if they were social change when they may be entirely attributable to which 20 items were asked, in what order, with what threshold for counting someone as lonely. A researcher who wants to know whether loneliness among 18-to-25-year-olds is rising needs the same instrument administered to comparable samples of that age group at multiple points — ideally the UCLA scale or an equivalent validated tool, not a bespoke corporate or foundation index retired after one or two cycles.

The AEI series on friendship is useful precisely because it did this, even if imperfectly and over a long gap. It asked substantially the same question in 1990 and in 2021 and can therefore say something about direction, if not about what happened in between. A study with five-year intervals using the same wording would be more useful still, because it would show whether the decline was gradual, concentrated in a particular period, or reversed and returned.

Until that kind of repeated, comparable measurement exists specifically for young adults, claims about whether their loneliness has been rising for a decade should be treated as plausible hypotheses supported by adjacent evidence — declining friendship counts, rising self-reported loneliness in single surveys, worse outcomes on measures like missed work reported by Cigna’s youngest respondents — rather than as an established trend. The pieces are suggestive. They are not, yet, a time series.