The advice is always the same, and it is always addressed to your character. Know your priorities. Value your time. Learn to be comfortable with other people’s disappointment.
It assumes the problem is that you are bad at refusing. Two decades of research suggest the problem is somewhere else entirely — in two errors that are measurable, ordinary, and almost impossible to notice from the inside. One is about the future. The other is about the person standing in front of you.
The first error: three weeks from now looks empty
Ask someone to do something tomorrow and they will check the day and answer honestly. Ask them to do the same thing in three weeks and the answer changes, because the three-weeks-away version of their calendar is mostly blank.
Gal Zauberman and John Lynch put a name to what happens next. They call it slack — the surplus of a resource you believe you will have available for a given task. Across seven experiments they found that people expect slack to grow in the future, and that the expectation is markedly stronger for time than for money.
That asymmetry is the useful part. Most people are fairly realistic about future money: a bill due in three weeks feels like a bill. Time does not work that way. A commitment three weeks out feels almost free, because the emptiness of a distant calendar reads as availability rather than as a page nobody has filled in yet.
7
experiments showing people expect more spare time later than now — and expect it far more for time than for money, which they forecast comparatively well
So the yes was not weakness. It was a forecast, made in good faith, using a number that is reliably wrong.
The second error: refusing genuinely costs something
The other half is harder to see, because it involves being wrong about someone else.
Francis Flynn and Vanessa Bohns have spent years asking people to make direct requests of strangers — fill in a survey, lend a phone, walk them somewhere — and, first, to predict how many would agree. Across studies involving more than fourteen thousand strangers, the predictions were wrong in the same direction every time. People underestimate compliance by an average of 48%.
48%
the average amount by which people underestimate how likely others are to say yes to a direct request
The explanation Bohns settles on is not that people are unusually generous. It is that saying no is awkward, and the person asking cannot feel it. The asker is running the numbers on the cost of helping: the twenty minutes, the detour, the favour. The person being asked is running a completely different calculation, about the cost of refusing — the pause, the face, the small social damage of being someone who would not.
Both parties are attending to a real cost. They are attending to different ones. And only one of them is in the room where the decision is made.
What this research can and cannot tell you
These are laboratory and field studies of mostly small, low-stakes requests made by strangers. The effect is unusually robust — it survives across samples, request types and settings, and the review lists the conditions that strengthen, weaken and reverse it — but the size of the error in your own workplace, where you know the asker and the request is not small, has not been measured.
What transfers is the direction and the mechanism, not the percentage. Refusing costs something real, the person asking is not weighing that cost, and neither of you is being unreasonable.
Why the usual advice does not touch either one
Now put the standard advice against the two errors it is supposed to fix.
“Know your priorities” addresses neither. You did know your priorities in the moment you agreed; you simply believed the commitment was landing in a week that had room. Clarifying your values does not correct a forecast about slack.
“Get comfortable with disappointing people” is closer, and still wrong in an interesting way. It treats the discomfort as a distortion to be trained away. The research says the discomfort is accurate. Refusing does carry a social cost. Being told to stop feeling it is being told to ignore a signal that is measuring something real.
What the advice never does is the one thing that would work, which is to move the decision away from the moment and the person.
- Convert the date before you answer. Ask whether you would accept the same request for this week. That question is answered against a real calendar rather than an imagined one, and it is the only version of the question with data behind it.
- Take the decision out of the room. “Let me check and come back to you” is not stalling. It removes the exact condition — the other person’s face, in real time — that makes refusing expensive.
- Answer in the channel that costs least. A written no an hour later is the same no, minus the pause. Nobody is owed your discomfort as part of the reply.
- Say no to the calendar, not to the person. “I can’t take that on in October” is a fact about a month. “I don’t want to” invites a negotiation about your reasons, which is a conversation with a much worse structure.
- Treat a fast yes as unreliable data. Your own too — an immediate agreement is the one most likely to be a forecast rather than a decision.
The fifth of those has a second edge, and it points the other way.
Try this
For two weeks, write down every commitment you agree to and how far away it is when you agree. Nothing else — not whether you regretted it, not how it went.
The pattern most people find is not that they say yes too often. It is that almost everything they regret was agreed to more than ten days out, and almost nothing they regret was agreed to for tomorrow.
That is a much easier problem to solve than a personality trait, because it has a specific trigger and the trigger is a date.
The half of this that is uncomfortable
If people say yes to you far more often than you expect, then your own requests carry more weight than you think they do — and the agreement you got may not be the agreement you believe you got.
Bohns is explicit that this is the serious end of the finding. The same underestimation that makes people reluctant to ask for help also makes them blind to the pressure their asking creates, and it does not stay in the territory of small favours. Where there is any difference in seniority, the person being asked is absorbing a cost the person asking has no access to.
The practical version is unglamorous. If you need a real answer rather than a polite one, ask in a way that makes no cheap to say: put it in writing, give it a deadline further out than you need, and say plainly that no is a complete answer. Those are not courtesies. They are how you find out what someone actually thinks.
The bottom line
You are not bad at saying no. You are working with a forecast that overstates how much time you will have, against a cost the other person cannot see and you cannot ignore.
Neither of those is fixed by resolve. Both of them are fixed by changing when you answer, and where.
Sources (4)
Every factual claim in this article comes from one of these sources.
- Resource slack and propensity to discount delayed investments of time versus money
- If you need help, just ask: underestimating compliance with direct requests for help
- (Mis)Understanding our influence over others: a review of the underestimation-of-compliance effect
- With a little help from my friends (and strangers): closeness as a moderator of the underestimation-of-compliance effect
Keep exploring
Continue reading
Deeper into what you just read.
Explore more
Adjacent ground, if this was useful.
One thing worth knowing
A surprising finding. The evidence behind it. And what it means for your life. BetterYou+ sends one useful idea at a time—nothing to keep up with.