What AARP's 45-Plus Surveys Can and Cannot Tell Us About Change Over Time

Most claims that loneliness is “rising” rest on comparing surveys that were never designed to be compared. Different sample frames, different questions, different response scales. AARP’s two national surveys of adults 45 and older are an exception worth taking seriously, precisely because they are not.

The 2018 survey, fielded among 3,020 midlife and older adults, used the 20-item UCLA Loneliness Scale — the same instrument that dominates the academic literature rather than a bespoke set of questions built for a press release. The follow-up survey, published by the AARP Public Policy Institute in September 2025, applied the same instrument to the same age band. That continuity is rare enough in this field to be the actual story.

Why this matters more than the headline number

The 2018 report found that one in three U.S. adults aged 45 and older were lonely. The 2025 follow-up reports a rise in loneliness among this group. Because both surveys used the UCLA scale rather than switching instruments, the difference between the two readings can be reasonably attributed to a change in the underlying phenomenon rather than to a change in how the question was asked. That is a stronger claim than almost anything else in the loneliness literature can support.

Contrast this with the more common situation. Cigna’s 2020 workplace loneliness report found 61% of U.S. adults sometimes or always lonely, using Cigna’s own scale. Cigna’s 2025 report found 57% of Americans lonely, using a survey fielded in mid-2024 but published in June 2025. Both are Cigna products, both are called “Loneliness in America,” and a reader could be forgiven for reading five points of decline into that difference. But without confirmation that item wording, response options, and sampling frame stayed constant across five years, the honest conclusion is that the two numbers are not directly subtractable. They may point in the same direction as the AARP data, or they may not; the instrument does not let us say.

This is the core problem the BMC Public Health review of the field identified in 2023: inconsistent measurement is a structural barrier to comparing findings across studies, not an incidental flaw of a few weak papers. The AARP pairing is useful specifically because it is the exception that shows what a comparable trend line would look like if more survey series maintained it.

What stayed fixed, and what did not

Instrument stability is not the same as complete comparability. Two things can undercut even a same-scale comparison across seven years: sample composition and the world the respondents were living in when they answered.

On sampling, both AARP surveys targeted the 45-plus population nationally, which limits (though does not eliminate) drift from who was asked. On timing, the 2018 fieldwork predates the COVID-19 pandemic by roughly two years; the 2025 report reflects a population that has been through it. A rise in loneliness between the two readings could reflect a durable structural shift, a pandemic-era spike that has only partially receded, or something else entirely. The AARP series, careful as it is on instrumentation, cannot distinguish between these explanations on its own, because it only has two time points. A trend needs at least three.

It is also worth being precise about what “the same instrument” guarantees and what it does not. The UCLA Loneliness Scale measures self-reported subjective loneliness. It was not designed to capture social isolation as a structural property of a person’s network — the size, diversity, and frequency-of-contact measures that the 2018 AARP survey also tracked separately, finding that network size and diversity were the strongest predictors of loneliness scores, alongside physical isolation. Whether isolation itself has grown at the same rate as loneliness among this age group, and whether the two are moving together or diverging, is a distinct question that a subjective-loneliness scale cannot answer regardless of how consistently it is applied.

What would make this comparison stronger

Three additions would turn a useful two-point comparison into an actual evidence base for tracking change.

A third wave, ideally fielded at a fixed interval, would allow AARP to distinguish a trend from a two-point line, which is definitionally incapable of showing anything but direction. A published item-level breakdown — not just the aggregate scale score — would let researchers check whether particular UCLA items are driving the shift, which matters because some items track affect (“I feel left out”) and others track structural contact (“There are people I can talk to”), and a rise concentrated in one cluster implies a different policy response than a rise spread evenly across both. And a matched measure of objective social isolation, run alongside the loneliness scale rather than as a separate module, would make it possible to say whether the 45-plus population is lonelier because its networks have shrunk, or lonelier despite networks that have not changed much — a distinction the 2023 Surgeon General’s advisory treated as central to matching interventions to actual causes rather than to a single undifferentiated “loneliness” problem.

None of this diminishes what the AARP pairing already accomplishes. In a field where most trend claims are built on surveys that changed methodology, sample, or wording between waves, a same-instrument comparison across seven years is close to the best evidence available for this age group. That is a modest claim. It is also, given how rare methodological continuity is in this literature, not a small one.