How Do You Know if Someone Is Lonely? |
Understanding Loneliness
Take our Loneliness Test
Find a therapist near me
No loneliness measure scores well on all three quality dimensions. That's a problem for policy.
The same population can look 20 percentage points more or less lonely depending on the measure.
A traffic light system rates loneliness measures on comparability, validity, and coverage.
Two reports on U.S. loneliness came out in 2025. One said about 41 percent of adults were lonely. The other said about 61 percent. Same country, same year—a 20-percentage-point gap.
These weren't fringe studies. These were experts using established measurement tools. And yet they painted fundamentally different pictures of the same population. If the numbers can swing that much depending on which questionnaire you pick up, what are we actually tracking?
My colleagues and I have been digging into this through the LONELY-EU project, a European Union-funded effort to evaluate how loneliness is measured across all 27 EU member states. We recently published two policy briefs—one on Europe's measurement challenge and one on a quality rating system we developed to address it—and I want to share what we found, because it has implications well beyond the EU.
Loneliness is linked to a 26 percent increased risk of early death. In Spain alone, it's estimated to cost more than €14 billion a year. The UK and Japan have appointed loneliness ministers. The former U.S. Surgeon General has declared it a public health crisis (but see this post as a reminder about it not being an epidemic). The World Health Organization has launched a Commission on Social Connection. There's real political momentum here. But as we lay out in our measurement challenges policy brief, the tools we use to measure loneliness weren't built for what we're now asking them to do.
When our team evaluated the major loneliness questionnaires across all 27 EU countries—surveying more than 25,000 people—we ran into a paradox. The one measure that can be meaningfully compared across all countries (the Three-Item UCLA Loneliness Scale) mostly captures the painful feeling of being lonely. It's weaker at picking up whether someone actually lacks social relationships or faces structural barriers to connection. So a program that successfully connects isolated people might look like it failed—not because it didn't work, but because the tool can't see what changed.
The measure that does capture the full picture—emotional distress and social support (the De Jong Gierveld Scale)—can only be reliably compared across 5 of 27 EU countries. Use it to compare loneliness in France versus Poland, and you may be comparing apples to something that isn't even fruit. And then there's the single-question approach ("How often do you feel lonely?"), which many surveys still rely on. Our data show it fails validity checks in eight countries and can't capture the multidimensional nature of loneliness at all.
It's Not Just a Methods Problem
An elderly widow in rural Greece experiences loneliness differently from a young professional in Amsterdam or a refugee in Berlin. Current instruments—developed primarily in the United States and the Netherlands—may not capture these variations. When a measure can't "see" certain forms of loneliness, whole populations drop out of the data. And populations that don't show up in the data don't get resources.
Within the same EU country, prevalence estimates from different surveys in the same year can differ by 8 percentage points or more. Try evaluating whether a national loneliness strategy is working when the numbers shift that much depending on the questionnaire.
A Traffic Light System
In our measurement quality policy brief, we introduce a traffic light rating system to help navigate this problem. It evaluates every loneliness measure on three dimensions: Can you compare scores across countries? Does it actually measure loneliness (and not depression or general dissatisfaction)? And does it capture the full range of loneliness experiences?
Understanding Loneliness
Take our Loneliness Test
Find a therapist near me
Green means strong evidence. Yellow means caution. Red means don't use this for the decision you're trying to make.
No existing measure gets green across all three. The UCLA scale scores green on comparability and validity, but yellow on content coverage. The De Jong Gierveld scale scores green on validity and content, but red on comparability. The single-item measure: red, yellow, red.
The picture is sobering, but at least it's honest. If you're a policymaker deciding how to track loneliness across a continent, you need to know what your tools can and can't do.
We're developing the EU SIL Index—a new measure designed to achieve what no current tool does: reliable cross-country comparability and comprehensive content coverage across the broader domains of social connection, calibrated over multiple EU countries and translated into multiple EU languages.
But a better questionnaire alone won't get us there. Europe also needs the infrastructure to use it well: coordinated data collection, quality assurance protocols, and integration with statistical agencies. You don't just need a good test—you need a system for administering it, collecting the results, and acting on them.
We know that social connection matters. Our ability to track it, compare it across populations, and figure out whether our interventions actually work depends on getting measurement right. It's not the most glamorous part of this conversation, but I'd argue it's the most consequential.
Hans Rocha IJzerman, Ph.D., is Director of the Annecy Behavioral Science Lab and Associate Researcher at the University of Oxford. He is a PI on the LONELY-EU project, a Horizon Europe initiative to improve loneliness monitoring and policymaking across Europe.
Conflict of Interest Disclosure: Dr. IJzerman is the founder and CEO of Entrelacs, a company developing AI-powered personalized mental health assessment. The measurement findings discussed here are based on independent psychometric research (Paris et al., 2025).
There was a problem adding your email address. Please try again.
By submitting your information you agree to the Psychology Today Terms & Conditions and Privacy Policy