Middle Managers Report the Lowest Psychological Safety of Anyone at Work

A study published by Harvard Business Review in October 2025 put a number on something most people who've ever managed a team already knew in their body: middle managers score 68 out of 100 on psychological safety — lower than individual contributors, and lower than the executives sitting above them. They sit in the one position that's supposed to translate strategy into reality and reality back into strategy, and it's the least safe seat in the building.

That's not a complaint about middle managers being soft. It's a structural finding, and structural findings are the ones worth taking seriously, because they don't go away when you hire nicer people or run a better onboarding. The job itself is built to produce this.

What the Numbers Actually Say About Middle Managers

The psychological safety concept, in its rigorous form, comes from Amy Edmondson's research at Harvard Business School — a team's shared belief that it's safe to take interpersonal risks, admit mistakes, or say "I don't know" without being punished for it. Most psychological safety research over the past decade has focused on teams: does an individual contributor feel safe speaking up in a meeting, does an engineer feel safe flagging a bug that's their own fault. The October 2025 HBR data did something less common — it broke the numbers out by organizational layer, and the layer that came out worst wasn't the bottom. It was the middle.

The explanation isn't mysterious once you look at what the role actually requires. An individual contributor's honesty travels one direction: up, to a manager who has some responsibility to receive it well. An executive's honesty travels one direction too: down, from a position where very few people are positioned to punish them for it. A middle manager's honesty has to travel both directions, constantly, to two different audiences with two different incentives — and there's rarely a peer group positioned similarly enough to absorb what doesn't fit in either direction.

Why the Bridge Position Is Structurally the Loneliest One

Picture the actual daily mechanics. Leadership hands down a target that isn't going to be hit, or a reorg nobody explained, or a policy the manager privately disagrees with. The manager's job is to translate that into something a team can actually work with — without lying to the team, and without undermining leadership. Then the reverse: the team is exhausted, or scared about the reorg, or has legitimate concerns about the target — and the manager has to translate that upward, in language leadership will actually hear, without looking like they can't control their own team.

Every one of those translations requires the manager to privately hold a version of the truth that's rougher than either version they're allowed to say out loud. And there's rarely a natural audience for the rough version. Peers in other departments are managing their own translation problem and don't have bandwidth to hold yours. The manager above them is often the person whose message needs softening in the first place. The hierarchy itself becomes a stress cascade — pressure compounding as it moves through each layer, and the middle is where the compounding is heaviest because it's absorbing pressure from both ends at once.

The Vacuum: What Happens When the People Absorbing Bad News Can't Share It

This is where it stops being a wellness problem and becomes an information problem. When the layer responsible for translating reality upward doesn't feel safe delivering unwelcome truths, the truths don't stop existing — they just stop traveling. A manager who doesn't feel safe telling leadership "this deadline isn't real" will instead tell the team to work weekends and tell leadership the deadline is on track, because both of those feel less risky in the moment than the honest version. Multiply that by every manager in a company simultaneously making the same calculation, and you get an organization where the people at the top are making decisions on information that's been quietly rounded toward optimism at every layer it passed through on the way up.

Fewer than one in five employees, according to the same research cycle, report that their manager has actually explained how AI is going to affect their specific job — not because managers don't know, often, but because "I genuinely don't know yet and neither does leadership" is exactly the kind of rough, un-translatable truth the middle layer has the least safety to say out loud. The vacuum doesn't produce silence. It produces confident-sounding half-answers, which are worse, because they look like information and aren't.

What This Costs the People Below Them, Not Just Above

The instinct is to treat this as a manager-wellbeing story — give middle managers more support, more training, a better Slack channel to vent in. That's not nothing, but it mistakes the mechanism. Psychological safety theater — the surveys, the values posters, the "safe space" language layered on top of a structure that hasn't actually changed — doesn't fix a structural bottleneck, it just gives the bottleneck better PR.

The actual cost lands on the team below the manager, not just the manager themselves. A team led by someone who's translating leadership's anxiety without a safe place to process their own gets that anxiety passed straight through, usually amplified, because unprocessed pressure doesn't evaporate on the way down a hierarchy — it just finds the next available outlet. The manager isn't the end of the chain. They're a relay that's currently overloaded, and an overloaded relay degrades everything downstream of it.

What Actually Fixes This — Not Another Wellness Webinar

The fix that has a shot at working isn't therapeutic, it's structural: build an actual peer channel for managers at the same level, insulated from the reporting relationships above and below them, where the rough, untranslated version of reality is allowed to exist out loud. Not a support group — a working group, with enough shared context that "this target isn't real" or "I don't know how to answer the AI question either" can be said plainly, tested against other managers' read of the same situation, and only then translated into whatever gets said upward or downward.

Companies that skip this and go straight to "trust your manager, they've got this" are asking the loneliest layer in the organization to keep absorbing pressure from both directions with nowhere to set it down. The data says they're already buckling. The question worth asking isn't whether your middle managers seem fine. It's who they're allowed to tell when they aren't.