When someone on your team misses a deadline, your brain files it under who they are. When you miss one, it files it under what happened to you. That asymmetry is so automatic that most managers never notice it running, and it quietly shapes who gets coached, who gets written up, who gets the stretch assignment and who gets managed out.

Psychologists call the core of it the fundamental attribution error: the habit of explaining other people’s behavior by their character while underweighting the situation they were in. Lee Ross named it in the 1970s, and it has been poked, replicated and argued over for fifty years since. The arguments are worth knowing, because the version that survives them is more useful to a manager than the tidy textbook version.

The quizmaster who looked like a genius

The cleanest demonstration I know of is a 1977 Stanford experiment by Ross, Teresa Amabile and Julia Steinmetz. Pairs of students were assigned roles by a coin flip. One became the questioner and wrote ten hard trivia questions from their own store of obscure knowledge. The other had to answer them. Everyone knew the roles were random. Everyone knew the questioner got to pick topics they happened to know well.

Afterwards, the contestants and outside observers rated the questioners as more knowledgeable than the contestants. The role gave one person total control over what got tested, and people still read the result as a difference in ability.

An earlier study by Edward Jones and Victor Harris found something similar in 1967. People read essays about Fidel Castro and were told the writers had been assigned which side to argue. They still concluded the writers believed what they wrote. Daniel Gilbert and Patrick Malone later reviewed the pile of follow-up work under the name correspondence bias and argued the error is strongest when the observer is busy or distracted, because correcting for the situation takes effort and the trait judgment arrives for free.

Busy and distracted. If that does not describe a manager reading a missed status update at 4:50 on a Thursday, nothing does.

Supervisors do it too, and harder when the stakes go up

The lab studies are about strangers and trivia. The research that should make managers uncomfortable comes from organizational behavior. Terence Mitchell and Robert Wood ran two studies in 1980 testing how supervisors respond to subordinate poor performance. Supervisors who put the cause inside the employee (effort, ability, attitude) chose punitive responses. And when the consequences of the mistake were serious, supervisors made more internal attributions and more punitive responses, even though the severity of the outcome tells you nothing about why it happened.

Read that last part again. The same mistake gets blamed more on the person when it happens to cost more. That is the outcome bias problem wearing a different hat, and it means your judgment of people is least reliable at precisely the moment it matters most: right after something expensive goes wrong.

The analyst I almost wrote up

Early in my time running an operations team at a large telecom, I had an analyst whose ticket numbers were the worst on the floor. Their average time to close was nearly double the team median, and their breached tickets showed up on my weekly report every single week. I had already formed an opinion. Slow, I thought. Maybe not careful. Possibly not suited to the queue.

I had a performance conversation half drafted when a senior engineer mentioned, almost in passing, that the routing rule for one of our enterprise customers had been set up years earlier to send that customer’s tickets to “whoever is on the late shift.” This analyst had been on the late shift for eleven months. They were getting every ticket from our most complicated customer, the one with three vendors in the path and a change freeze on every other weekend. Nobody else on the team had touched those tickets in almost a year.

When we fixed the routing and spread that customer’s work across the team, the analyst’s numbers moved to the middle of the pack within a quarter. The team median got worse, which told me everything about where the problem had been.

What still bothers me is not that I misread it. It is how certain I felt. I had a dashboard, a trend line and a story, and the story was about a person, because a person is what the dashboard showed me. The routing rule was not on any report I looked at.

The textbook version is too tidy

Here is where it gets more interesting than the poster version. For decades, textbooks also taught a cousin of the fundamental attribution error called the actor and observer asymmetry: you explain your own behavior by circumstances and other people’s behavior by character. It is a lovely, clean idea. When Bertram Malle pooled the published tests of it in a 2006 meta-analysis in Psychological Bulletin, the general effect nearly vanished. Whether it showed up depended on conditions, such as whether the event was good or bad and how well the two people knew each other.

So the honest summary is not “everyone always blames others and excuses themselves.” It is narrower and more practical:

• People reliably underweight constraints they cannot see. That is the part the Ross and Jones studies keep confirming.
• You can always see your own constraints. You lived them. The meeting that ran long, the vendor who didn’t call back, the request from your boss that blew up your Tuesday.
• You almost never see your team’s constraints unless you go looking. They don’t show up in the output. The output is all you get.

The asymmetry between you and your team is not mainly a character flaw in managers. It is an information problem. You have full context on yourself and a summary on everyone else, and your brain fills the gap with the cheapest available explanation, which is personality.

Why the manager’s seat makes it worse

Every promotion moves you one step further from the work and one step closer to the numbers about the work. That is the job. But it changes what you see. A frontline person sees the broken tool, the confusing handoff and the customer who changed the requirement twice. Their manager sees a late deliverable.

The quizmaster study has an uncomfortable parallel here. In most teams the manager is the questioner. You decide what gets measured, what gets asked in the one-on-one, what the deadline is and what “done” means. That control makes you look sharper than you are and makes your people look less capable than they are, for the same reason the questioners looked brilliant.

W. Edwards Deming put a number on the stakes years ago. He argued that roughly 94 percent of problems and opportunities for improvement belong to the system, which is management’s responsibility, and about 6 percent come from special causes. You do not have to accept the precise figure to take the point. If most variation comes from the system and your brain defaults to explaining it by the person, you will be wrong in a predictable direction most of the time.

It compounds, too. Once you have decided someone is weak, you watch them more closely, give them less latitude and read every ambiguous signal as confirmation. That loop has a name, set-up-to-fail syndrome, and it is how a manager ends up creating the weak performer they thought they had.

Four questions before you decide it’s the person

None of this means you stop holding people accountable. It means you change the order of operations: rule out the situation first, because the situation is the thing your brain skips. The attribution model Mitchell and Wood tested rests on three cues, and they translate well into questions you can actually ask.

Does anyone else struggle with this same work?

If three people have trouble with the monthly reconciliation, the reconciliation is the problem. If one person struggles where everyone else breezes through, that tells you something different. The trick is making sure “everyone else” is doing the same work. My late-shift analyst failed this test only because nobody else was getting those tickets.

Is this a pattern, or a recent change?

Someone who has been solid for two years and slips for six weeks almost never became a different person. Something changed. A new system, a new client, a sick parent, a reorganization that dumped two people’s work on one desk. Ask what is different about the last six weeks before you ask what is wrong with them.

Is it only this kind of task?

A person who handles customers well but writes weak documentation has a specific gap, which is fixable with specific help. A manager who concludes “they’re sloppy” has turned a narrow skill problem into a character verdict, and character verdicts don’t come with a training plan.

What would I say if it were me?

This one is my favorite because it uses the bias against itself. Write down the explanation you would give if you had missed this deadline. You will produce a list of circumstances in about ten seconds: unclear priority, a dependency that slipped, three competing requests. Then go check whether any of those were true for them. Usually at least one was, and you just couldn’t see it from your chair.

Going and getting the missing context

Questions only help if you get answers, and the answers live with the people doing the work. A few habits that have worked for me and for the operations teams I have worked with since:

Ask “what got in the way?” before “why didn’t you?” The second question puts the person on trial. The first one invites them to show you the constraint. Same meeting, very different information.

Look at the inputs, not just the output. What arrived on their desk, from whom, and when? In a fractional COO engagement with a small service company, the owner was sure their dispatcher was disorganized because jobs kept getting double booked. When we sat with the dispatcher for one morning, it turned out they were keeping the schedule in two systems because the field app didn’t sync with the office calendar. The dispatcher was compensating for a tool, heroically, and getting blamed for the moments the compensation failed.

Do a stretch of the work yourself. Not to prove a point. Take a ticket shift, run the report, process five orders. Thirty minutes inside someone’s workflow surfaces constraints that no status meeting ever will.

Delay the verdict when the outcome is severe. Since the Mitchell and Wood research suggests expensive mistakes push you toward blaming the person, build in a rule: no performance conclusions in the first 48 hours after a costly failure. Look first at what the process allowed to happen, and whether you can design that mistake out entirely. The person will still be there on Monday.

When it really is the person

There is an opposite failure, and I have watched managers fall into it after learning about this bias. They become so committed to “it’s always the system” that they explain away a genuine effort or skill problem for a year, while the rest of the team picks up the slack and quietly loses respect for the manager.

Sometimes it is the person. Someone takes the same work, the same tools and the same support as their peers, and consistently does less with it over months. Someone is careless in ways that are specific to them and persist after the obstacles are cleared. The four questions are not a way to avoid that conclusion. They are a way to earn it. When you have ruled out the routing rule, the broken app and the six-week crunch, and the gap is still there, you can have the hard conversation with evidence instead of a hunch, and the person on the other side of the table will usually know you did your homework.

That is the real payoff. Teams can tell the difference between a manager who has looked at their situation and a manager who has looked at their numbers. The first one gets the benefit of the doubt when they finally do say “this one is on you.” The second one never quite does.

The next time a name on your report looks bad, before you form an opinion, find the one thing about their week that you can’t see from where you sit. There is almost always something.