Perspectives are a funny thing. Here’s one you may have noticed.
Managers and the staff they manage don’t rate safety the same.
In the three studies here that compared their ratings, the managerial group rated safety more favourably than the staff group it was compared with, although each study drew the line between seniority levels in a different place.
What that means in real life is this: If your survey results arrive as an organisation-wide figure, that (seemingly consistent) material difference disappears into the average.
You’ll never see it, and you’ll never know.
I think that’s worth a little digging.
The gap
The clearest data come from hospitals.
In US Veterans Affairs (VA) hospitals, 4,547 staff responded to a safety climate survey and results from 29 hospitals were analysed. The survey counts “problematic” answers, meaning answers that suggest safety climate is absent. Senior managers gave a problematic answer about 10% of the time. All other staff, supervisors and front line together, did so about 18% of the time, nearly twice as often.
Among nurses the gap was wider. Senior nurse managers gave problematic answers about 10% of the time and front-line nurses about 22%, more than twice as often. The authors suggest this may reflect a difference in understanding between the people delivering care and administrators whose work is removed from it.
The gap isn’t confined to just the top though. A 2024 analysis of a 2012 survey in two large Boston hospitals asked 1,492 nurses and nursing assistants and 72 of their unit managers how well their unit put its safety and ergonomic practices into effect. These were the managers running each unit, not senior executives, and on all three measures their average was still higher than their staff’s. The VA study, by contrast, counted supervisors in with the front line.
An exploratory 2026 study of 217 operating-theatre staff in six Norwegian hospitals found the same direction yet again. Staff with leadership responsibilities rated three out of ten safety culture factors more positively than staff without leadership responsibilities. Unit leaders and administrators rated four factors, and the overall patient safety rating, more positively. (Exploratory here means the researchers were looking for patterns to test later rather than testing a prediction, so treat it as consistent with the other work rather than as proof on its own.)
Whose ratings line up with what happens next?
Ok. A gap on its own doesn’t tell you who’s right. So two of these studies went further and compared each group’s ratings with outcomes.
The largest study, of 18,223 staff in 91 US hospitals, asked a much pointier question. Whose view of safety matches what actually happens? The researchers compared each group’s ratings with the hospital’s patient-safety indicators. These are flags in Medicare records that something may have gone wrong. (They’re warning signs, not proof that a patient was harmed.)
The front line’s ratings matched the flags. Hospitals where front-line staff rated safety more highly had fewer of them. By contrast, senior managers’ ratings didn’t match. Actually, the study found no link between how managers rated safety and how many flags their hospital had. To be fair to the managers, that isn’t proof there’s no link, only that this study couldn’t find one. Read into that what you like. The researchers also looked at each group on its own rather than testing the two against each other.
The flags were collected in the year after the survey. But this was an observational study, so it shows that good front-line ratings and fewer flags went together. It doesn’t show that a better safety climate caused there to be fewer.
The Boston study looked at something more concrete. Instead of warning flags, it counted actual staff injuries, and it followed each worker for the 12 months after they filled in the survey.
Staff rated three things about their unit. On two of them, staff who gave their unit better ratings were less likely to be hurt over the following year, by a margin unlikely to be down to chance. The third, a single question about safety leadership, wasn’t linked to injury for anyone.
However, because you knew this was coming, the unit managers’ ratings showed no reliable link with who got hurt, and adding the manager’s ratings to the staff ratings didn’t improve the prediction. On all three measures, the staff ratings on their own gave the best fit.
There’s a caution I should note for transparency. With only 72 managers against nearly 1,500 staff, the managers’ results are less certain. The researchers say the missing link for managers could also come from how they chose people and worded the questions. Fine, I’ll wear that.
Like the US study, this one watched what happened rather than testing a change. It shows what went together, not what caused what.
In both studies, then, the front line’s ratings lined up with outcomes and the managers’ didn’t. It isn’t enough to say the front line is always right. But for me it 100% is enough to say the front line’s view deserves at least as much weight as its managers’. More than equal weight? For me that’s a strong ‘maybe’.
Why the gap
None of these studies tested why managers and staff see it differently. The researchers offer several possibilities:
- Distance from the hazard. Staff are closer to the risks and may judge them more accurately.
- Rating yourself. Safety on your unit reflects on you, and managers may want to see themselves as doing a good job.
- Different information. Leaders see the reporting system and the policies. Staff see how the work actually runs.
- The same policy, seen from two ends. In the Boston authors’ example, managers agree when staff ask for cover to take a break, so they believe breaks happen. Staff often don’t ask when the unit is busy.
- Leaders may be right about some of it. In the Norwegian study, leaders were better placed to judge how often safety events get reported.
What would you add? If you’ve sat on both sides of this, I’d like to hear which explanation rang true.
Outside hospitals the pattern mostly holds, with exceptions
In reviewing earlier work in other settings, the Boston authors found that most studies showed workers’ ratings predicting injuries, but not all. In restaurant workers, staff ratings were linked to injury rates and managers’ weren’t. In a construction study, neither group’s ratings had a reliable link with injuries, but the gap between workers and managers did. A study of blue-collar workers in a university setting also found the gap between workers and managers predicted injuries. A 2017 study found workers’ climate ratings weren’t associated with injury rates, while safety officers’ reports of safety management practices were.
Two of those studies are a reason to pay attention to the gap itself, not only to either group’s score.
So what? A few thoughts
None of these studies tested what follows, but I think there are interesting applications.
- (Hopefully) obviously, survey both groups. The Boston authors caution against manager-only surveys, which they note employers sometimes use for ease, cost or response rate.
- Where surveying both groups isn’t practical, the Boston authors suggest a worker-only survey. Good call. Managers have an oversized influence.
- Ask for the results split by seniority level, and put each level’s scores side by side, item by item. In a council that might mean the executive team, other managers, team leaders and staff. Use at least three levels if your numbers allow, because the studies grouped supervisors and middle managers differently.
- Treat the gap as a finding in its own right that’s worth exploring further. Before reading anything into it, check that each group is answering about comparable work and responsibilities, and report group sizes and response rates alongside the scores. A gap is a prompt to ask why, not proof that seniority caused it.
- For each item where the levels differ, ask two questions. First, what does the front line see that the senior team doesn’t? And second, what does the senior team know that hasn’t reached the front line?
- Settle anonymity before the survey goes out. Agree a minimum group size with your survey provider. In a small council or team, for example, the senior group may be two or three easily identifiable people, so ask the senior team to agree to their own result being reported, or have them predict the front line’s scores before the results are released and compare the two. That tests how well leaders understand staff views, which is useful, but it isn’t the same as comparing their own ratings.
- All of these are hospital studies, and I’m (reasonably, I think) extrapolating.
If you don’t run a safety climate survey as such, apply this to the health and safety items in the staff or engagement survey you already run.
Photo by Joao Viegas on Unsplash