The return-to-office memo has a standard paragraph. The wording varies, but the claim doesn't: we're better together. Collaboration happens in hallways. Remote employees drift, lose touch with the culture, and eventually leave. The office isn't a building, it's a community.
It's a testable claim. A team of researchers recently tested it on 7,704 people, and the community argument did not survive contact with the data.
A design that's harder to wave away
Most of the evidence in the remote work debate comes from asking people how they feel and stopping there. A cross-sectional survey is cheap and easy to dismiss: moods drift, questions lead, and nobody follows up to see whether feelings turned into resignations.
This study, published in July in the peer-reviewed journal Frontiers in Psychology and getting fresh attention this week after a University of Colorado Boulder write-up, was built to follow up. Alyssa Lezcano, Stephanie Zajac, and Courtney Holladay of the MD Anderson Leadership Institute, with Stefanie Johnson of CU Boulder's Leeds School of Business, started from a 2023 engagement survey at one large healthcare organization in the southwestern United States: 1,869 fully remote employees, 2,099 hybrid, and 3,736 onsite. Then they waited a year and went back to the organization's actual turnover records to check who had left.
The lag is the point. It can't fix everything. Nobody was randomly assigned to work from home, and the survey is still a snapshot of how people felt in 2023. What the lag does is convert the softest question in this debate, "who says they're doing well?", into a harder one: does any of this predict who stays?
The well-being ladder
The survey asked employees, among other things, whether they feel energized by their work, supported by their manager in decisions about their health and well-being, and able to find balance between work and life. The researchers rolled four such items into a well-being score on a 1-to-5 scale.
Fully remote employees averaged 4.22. Hybrid employees averaged 4.12. Onsite employees averaged 3.89. Every gap between rungs was statistically significant: remote above hybrid (p = .0003), hybrid above onsite (p < .001), remote above onsite by the widest margin (p < .001). Well-being tracked remoteness step for step, with no sweet spot in the middle. Top rung to bottom is a third of a point on a five-point scale, a consistent lean across thousands of people rather than a chasm.
This is a healthcare system, though, and in a healthcare system "fully remote" skews toward administrative, billing, and analytics roles while "onsite" includes bedside clinical work. Some unknown share of the well-being gap belongs to the jobs, not the locations. Concede that upfront, and then notice how little of the office-mandate argument it rescues: job mix might explain who finds the work draining, but the mandate case rests on connection, and job mix gives you no reason to expect a remote connection deficit. The connection comparison is where the argument gets decided.
The ordering also surprised the researchers themselves. Johnson expected the middle rung to win. "I actually thought you would be happiest if you were part time out of the office," she said in the CU Boulder release. "Then every once in a while you get to see people, get that human connection."
Looking for the disconnection
The well-being result is the headline. The connection result is the story.
The connection measure was simple. An open-ended survey item asked employees for 3 to 5 words describing the organization's culture. Three of the authors rated every word employees offered for whether it signaled positive connection at work (words reflecting inclusion, unity, care, and collaboration made the cut), then each response was coded for whether it contained any such word.
Remote workers' answers did 61% of the time, against 58% for hybrid and 55% for onsite. The group that never sets foot in the building described the culture in connection language at the highest rate of the three. How firm that difference is depends on the test: a simple chi-squared comparison found it significant, the regression weighted by group size did not, and the authors call the differences minimal, which is why the paper's abstract reports the connection results as nonsignificant. Read conservatively, remote employees feel about as connected as everyone else. Read generously, slightly more.
Neither reading contains a penalty. Word-coding culture descriptions is a blunter instrument than a validated connection scale, and no null result proves an effect is zero. But the memo's version of the claim is that the disconnection is large, obvious, and general. In a sample of 7,704 people, damage that size should have left a mark, and the mark runs the other way.
What the study can't prove, and says so
Remote employees also had the lowest turnover a year later: 7.8% of them left, against 8.8% of onsite employees. If you've seen this study shared this week, you may have seen that spread quoted as proof that remote work retains people.
The authors don't claim that, and neither should anyone else. The turnover difference was not statistically significant. The number leans the right way, which is not the same as evidence.
What the paper does find on retention is more specific, and more useful. Well-being significantly predicted who stayed, in every group. Connection didn't significantly predict it. Work location had no significant direct effect on turnover either; the one significant path from location to retention ran through well-being, an indirect effect the authors themselves call minimal, summarized in their finding that the relationship between work environment and turnover "may be fully mediated by well-being." In plain terms: whatever retention advantage remote work carries here, it ran through people's day-to-day lives being better, and the pathway was real and small.
This is also one organization in one industry. The 7,704 were the survey respondents (out of 20,856) who answered the culture question, though the excluded group didn't meaningfully differ in age, tenure, gender, work arrangement, or turnover. And the well-being measure is a four-item internal scale, statistically reliable (Cronbach's alpha of .84) but not a standard instrument you can compare across studies. A single careful study is a data point. This one is unusually honest about being one.
The hybrid question this reopens
The direction of the finding matches the strongest existing evidence. The best-known randomized trial in this space, run by Stanford's Nicholas Bloom and colleagues at Trip.com, a Chinese travel-technology company, followed 1,612 employees and found hybrid work cut quit rates by a third, with no measurable cost to performance and no difference in promotion rates over two years of reviews. And Pew reported in January 2025 that 46% of people who currently work from home at least some of the time would be unlikely to stay in their job if that ended, rising to 61% among the fully remote. Both point where this study points: flexibility helps, and pulling it back has costs.
Where this study pushes past the consensus is the top rung. The 2024 Trip.com trial tested hybrid against the office and stopped there. This paper put fully remote into the comparison and found another rung above hybrid: the people furthest from the office reported the highest well-being of all, with hybrid significantly behind. One observational study at one healthcare organization doesn't settle that. It does mean the popular compromise position, that hybrid captures remote's benefits while dodging its costs, now has a 7,704-person data point sitting awkwardly against it.
Against all of this, the striking thing about the current wave of office mandates is how little evidence of any kind appears in their reasoning. Johnson, a professor of organizational leadership at CU Boulder's Leeds School of Business, put it bluntly: "Leaders are not making decisions based on data. I think they are just returning to what they are used to."
Questions you can ask a company out loud
If you're weighing a remote role, or deciding whether to wait out a mandate, the study converts into questions concrete enough to use:
- "Do you track engagement or well-being separately by work arrangement, and what did the last read show?" The answer matters less than whether one exists. A location policy backed by no measurement at all is the pattern Johnson describes: leaders returning to what they're used to. The same question works from the inside, too. Well-being was the retention lever in this study, and one of its four items, feeling supported by your manager, is trainable at any distance; a People team that tracks it by arrangement has a lever that doesn't require winning the location fight.
- "What happened to retention and promotions in your remote cohort versus everyone else?" This is the polite version of the real fear: that the next promotion quietly goes to whoever is in the building. The research on that fear is genuinely mixed. The 2024 Trip.com trial found no promotion difference for hybrid employees, but the same company's earlier randomized trial found people working from home four days a week were promoted at roughly half the rate of their office peers despite equal performance, and a 2024 study of fully remote call-center workers found a similar gap. Promotion is the one place the mandate side holds real evidence, which is exactly why a company's own numbers are the question worth asking.
- "When your remote employees describe the culture, does it sound like the same company?" This is the study's own method, and any People team can run it on engagement comments they already collect. You can approximate it from outside before you ever apply: read what remote employees themselves say in public reviews and interviews, and watch for whether the company describes how remote operates day to day or just lists it as a perk.
None of these questions demand a perfect company, just one that has looked. The employees reporting the highest well-being in this study worked somewhere that let them work the way they wanted to; Johnson's own summary of her data was, "This suggests you let people work remotely if they want to work remotely."
And if the mandate on the table rests on the claim that remote employees are disconnected and halfway out the door, remember what this study did. It went looking for exactly that, a year deep, with the turnover records open.
It wasn't there.
Sources: Lezcano, Zajac, Johnson & Holladay, "Rethinking return-to-office: the positive impact of remote work on well-being, connection, and employee retention," Frontiers in Psychology, July 2026 (full text); CU Boulder Today, Aug 12, 2026; Bloom, Han & Liang, Nature, 2024; Bloom, Liang, Roberts & Ying, Quarterly Journal of Economics, 2015; Emanuel & Harrington, American Economic Journal: Applied Economics, 2024; Pew Research Center, Jan 13, 2025.
