Four Gallup Loneliness Numbers, Four Different Questions

Between October 2023 and October 2024, Gallup published four separate reports containing a headline loneliness or connectedness figure for the world’s population. Read in sequence, as many outlets have read them, they look like a story: connection is high, loneliness is a quarter of the world, loneliness among workers is one in five, loneliness generally is a bit over one in five. That sequence is not a trend line. It is four different questions, asked of overlapping but not identical populations, over different reference periods, reported as though they measure the same thing.

This matters because the Gallup name carries weight, and because “1 in 5” and “1 in 4” get quoted interchangeably in policy writing as if they described the same underlying rate moving slightly over a year. They do not.

The base survey: connectedness, not loneliness

The foundational dataset is the Gallup-Meta State of Social Connections survey, fielded across 142 countries with roughly 1,000 respondents aged 15 and older per country, between June 2022 and February 2023. Its headline finding, published in October 2023, was that 72% of people worldwide felt very or fairly socially connected — about 3.2 billion people in the countries surveyed, which exclude China and therefore cover roughly 77% of the world’s adult population.

That is a connectedness measure: an affirmative self-report about feeling tied to others. It is not the inverse of a loneliness measure, and Gallup’s own follow-up analysis, published in November 2023, makes the point explicitly — connectedness and loneliness are not simple inverses of one another. A person can report feeling connected in general terms while also reporting loneliness on a specific day or in a specific domain, such as work. The two questions capture different psychological states and should not be subtracted from 100 to reconstruct one from the other.

The same dataset, a different question, a different headline

Two weeks after the connectedness figure, Gallup published a second report from the same underlying survey: 24% of people worldwide report feeling very or fairly lonely. Young adults aged 19 to 29 reported the highest rate, at 27%; adults 65 and older the lowest, at 17%. Forty-nine percent said they did not feel lonely at all.

This is a legitimate and useful number. But it is drawn from a “very or fairly lonely” self-rating item, structurally similar to the connectedness question but asking about a different construct, on the same 2022–2023 fieldwork. The 72% and the 24% are companion figures from one survey wave, not evidence that roughly a quarter of the “unconnected” population is lonely. Any writer who nets these two numbers against each other, or presents them as a before-and-after, is manufacturing a comparison the data does not support.

Workers are not the world, and “a lot” is not “very or fairly”

In June 2024, Gallup’s State of the Global Workplace report introduced a third figure: one in five employees worldwide report experiencing loneliness a lot of the previous day. Fully remote employees reported this at 25%, hybrid workers at 21%, and fully on-site workers at 16%. Loneliness was higher among employees under 35.

Two things changed between this figure and the October 2023 loneliness figure, and both matter. First, the population changed — this is a survey of employed adults, not the general population, and employment status itself is correlated with the social contact that predicts loneliness. Second, and more consequentially, the question changed. “A lot of the previous day” is a day-reconstruction-style item, asking about a specific affect experienced on a specific day, rather than the general “very or fairly lonely” self-rating used in the earlier report. These are not interchangeable instruments. A person might rarely feel lonely “a lot” on any given day while still endorsing a general sense of being lonely, or the reverse.

Then, in October 2024, Gallup published a fourth figure using the same daily-experience item, but for the general population rather than employees specifically: more than one in five adults worldwide report feeling lonely a lot of the previous day. This confirms the pattern of elevated loneliness among younger adults seen elsewhere in Gallup’s work. But it cannot be laid next to the October 2023 figure of 24% “very or fairly lonely” as evidence that global loneliness fell by roughly three points in a year. The instrument changed under the headline.

Why this keeps happening

Part of the difficulty is structural to survey research on loneliness generally. A 2023 review in BMC Public Health identified inconsistent measurement as a persistent barrier to comparing findings across the loneliness literature, and Gallup’s own output across a single year illustrates the problem inside one organization’s work, not just across competing survey houses. Different Gallup teams — global research, workplace research — draw on different item batteries suited to their own reporting needs, and each publishes a standalone headline figure without foregrounding which specific question produced it.

Contrast this with the AARP Foundation’s 2018 survey of adults 45 and older, which used the 20-item UCLA Loneliness Scale rather than a bespoke single item. That choice makes the AARP figure — one in three midlife and older U.S. adults report being lonely — directly comparable to decades of academic research using the same instrument, at some cost in survey brevity. Gallup’s single-item approach is cheaper to field across 142 countries and gives a usable topline number, but it trades away exactly the comparability that would let four Gallup reports in fourteen months be read together as a coherent picture rather than four separate snapshots.

What would actually show a trend

A genuine time series would require the same question, worded identically, asked of the same or a comparable population, at repeated intervals, with the full item text published alongside each topline number. Gallup has the infrastructure to do this — its World Poll already returns to some countries annually — but the loneliness-specific items have so far been deployed opportunistically, attached to whichever report a given item happens to serve. Until Gallup or another survey house commits to fielding one loneliness question, unchanged, on a fixed schedule, headline comparisons across its own releases should be treated as four separate readings of related but distinct constructs, not as a trend.