# How to Say No, and Why It Is So Hard
> Refusing is harder than anyone asking you realises, and there is research showing exactly how much. What that means for how to say no, and what silence costs.
Written by Ahmet Abiç (Founding editor). Reviewed by Noah Cortez (Stanford Journalism Program). Published August 18, 2026.
Source: https://gritwright.com/social/how-to-say-no/
---
import Figure from '../../../components/Figure.astro';
import ResearchNote from '../../../components/ResearchNote.astro';
import noConversation from '../../../assets/guides/social/no-conversation.webp';
import noDeclining from '../../../assets/guides/social/no-declining.webp';
You said yes to the thing on Saturday. You did not want to go, you knew that while you were agreeing, and you agreed anyway because in the moment the refusal seemed to cost more than the evening.
The standard treatment of this is that you lack backbone. That is not what the research says. There is a specific, measured reason refusing is hard, the person asking you has no access to it, and understanding the mechanism is more useful than being told to grow a spine.
## The short answer
The problem is not that you are weak. It is that you and the person asking are experiencing two different conversations.
| What is happening | The measured version |
| --- | --- |
| Why refusing is hard | The awkwardness is real, immediate, and yours alone |
| What the asker sees | Almost none of it. They think you are freely choosing |
| How wrong they are | They underestimate agreement by about half |
| The two failure modes | Never refusing, and refusing badly. Both are penalised |
| What silence costs | Measurable, in one long study alarmingly so |
| What works | Short, clear, repeated unchanged if pushed |
## Why it is genuinely hard
The best evidence on this comes from [a research programme](https://ecommons.cornell.edu/server/api/core/bitstreams/445e3b34-ce08-4f5b-8671-466f49963d40/content) that approached this from the other side: not why people cannot refuse, but why people are so bad at predicting whether their requests will be agreed to.
Across a programme of studies in which participants **asked more than 14,000 strangers** for a variety of things, people were consistently **too pessimistic** about getting a yes.
The error is described as **robust**, persisting across samples and request types, and **substantial: requesters underestimated compliance by an average of 48%.**
The mechanism identified is the important part. The error results from **requesters' failure to appreciate the awkwardness of saying no to a request.** People regularly agree to things, including things they would rather not do, **to avoid the discomfort of refusing.**
Bohns notes the corollary: because people do not themselves face the immediate prospect of saying no, they discount how powerful that concern is, and so **view compliance as more voluntary than the person complying experiences it.**
Take a moment with the size of that. In one of the studies described, participants in a charity run predicted they would need to ask an average of 210 people to hit their fundraising target. They needed 122.
The finding people usually take from this is a cheerful one about asking for things more often, and that is a fair reading. The reading that matters here is the mechanism, and it is not cheerful at all.
You are not imagining the difficulty. The pull towards yes is a documented force strong enough to bend the behaviour of thousands of people across dozens of studies. It is the reason you agreed on Saturday, and it operates at the moment of asking, in person, when the other person is looking at you.
## The person asking has no idea
The second half of the finding is the one that should change how you feel about refusing, and it rarely gets quoted.
Because the asker never experiences the awkwardness of saying no to themselves, they systematically read your yes as a free choice. Which means the social debt you are carefully protecting, the sense that refusing would damage something, exists almost entirely on your side of the table.
This is worth stating plainly: they asked you a question. In their model of the situation, a no is one of the two available answers, and they will have moved on from it considerably faster than you will.
It also gives you a usable check when you are wavering. Ask what you would think if they declined the same request from you. Almost nobody answers that with anything worse than mild disappointment, and that answer is a better estimate of the real cost than the feeling in the room.
## Two ways to get this wrong
Once people decide to start refusing more, they often overshoot, and there is [research showing that both directions carry a penalty](http://www.columbia.edu/~da358/publications/ames_flynn_assertiveness.pdf) rather than one.
Across three studies, using both colleagues' open comments and quantitative ratings, the authors found a **curvilinear** relation: people seen as **markedly low in assertiveness, and people seen as high in assertiveness, were both appraised as less effective.**
They locate the reason in a tradeoff between two kinds of outcome. **A high level of assertiveness worsens relationships**, while **a low level limits goal achievement.**
They also report that perceptions of a person as having too much or too little assertiveness are **widespread**, showing up constantly in how colleagues describe each other's weaknesses.
That study is about workplace leadership and the transfer to friendships and relationships is an inference rather than a finding. It is a reasonable one, and the two shapes are recognisable anywhere.
**Too little** is the version this article is mostly about. You keep agreeing, resentment accumulates quietly, and eventually it comes out somewhere unrelated and disproportionate. Nobody around you understands where it came from, because from their side nothing was ever refused.
**Too much** is the correction people make after reading an article like this one. Boundaries announced rather than held, a refusal delivered with more force than the request warranted, a general air of being difficult on principle. It solves the resentment problem and creates a relationship problem.
The target is not maximum assertiveness. It is a no that is unmistakable and unremarkable at the same time.
## What never saying it costs
There is [one finding](https://pubmed.ncbi.nlm.nih.gov/17634565/) on this that is more serious than the rest of the literature, and it deserves to be handled carefully rather than as a headline.
**3,682 participants** were examined between 1984 and 1987 and followed for **ten years** for coronary heart disease and mortality.
After adjusting for age, blood pressure, body mass index, smoking, diabetes and cholesterol: **women who self-silenced during conflict with their spouse had four times the risk of dying** compared with women who did not, a hazard ratio of 4.01.
The comparison points matter as much as the finding. **Marital happiness, satisfaction and disagreements were not related to the development of heart disease or death in either men or women.**
For men, the associations that emerged were different: married men were roughly half as likely to die during follow-up as unmarried men, and men whose wives were upset by work were 2.7 times more likely to develop heart disease.
The headline result is in women, and I am not going to quietly generalise it to the men reading this, because that is exactly how a good finding turns into a bad statistic. What was measured in men in this study was something else.
What does transfer is the structure of the result, and it is genuinely surprising. How happy the marriage was did not predict outcomes. How much the couple disagreed did not predict outcomes. What predicted them was a specific behaviour during conflict: holding it in.
That is the same conclusion arrived at from a completely different direction in [what makes a relationship work](/social/what-makes-a-relationship-work/) and in [how to stop fighting](/social/how-to-stop-fighting/). Arguing is not the risk factor. What you do instead of arguing might be.
## Saying nothing is not the same as saying no
The most common alternative to refusing is not agreeing, it is going quiet, and there is [a substantial literature](https://magazine.tcu.edu/winter-2016/understanding-the-silent-treatment/) on what that does inside a relationship.
A meta-analysis of **74 studies covering more than 14,000 participants** examined the demand and withdraw pattern, in which one person presses for a discussion and the other retreats from it.
The pattern was **associated with lower relational satisfaction, lower intimacy and poorer communication**, and the association held regardless of which partner was doing which.
The researchers describe it as one of the most consistently damaging patterns identified in relationship research.
A silent no is not a diplomatic no. It is the withdraw half of that pattern, and the other person experiences it as an unanswered question, which is what produces the pressing that produces more withdrawal.
The practical upshot is that avoiding the conversation is not the low-cost option it appears to be at the moment of choosing it. It defers the cost and adds interest.
## How to actually say it
The mechanics are simple and almost everyone does the same three things wrong.
**Say it early.** The longer the gap between the question and your answer, the more the answer looks like a verdict. A no in the first ten seconds is a preference. A no after four days of consideration is a rejection.
**Keep it short.** One sentence of refusal, at most one of reason. The instinct is to soften it with explanation, and explanation is counterproductive, because a reason is a thing that can be solved. Give three reasons and you have handed over a list of problems to fix.
**Do not apologise more than once.** A refusal wrapped in repeated apology signals that you think you have done something wrong, which invites the other person to agree.
**Offer the alternative only if it is real.** "Not this weekend, but I would like to do this" is useful when true and corrosive when it is a way of avoiding the no. If there is no alternative, do not invent one.
**Repeat rather than improve.** If they push, the temptation is to produce a better reason. Do not. Say the same sentence again, in the same words. A refusal that keeps changing shape looks like one that is still being negotiated.
**Match the volume.** This is the Ames and Flynn point in practice. A no delivered with more force than the request needed is the overcorrection, and it costs you on the relationship side without buying anything on the outcome side.
## The medium is doing half the work
Where a request reaches you determines how much pressure you are under when you answer it, and [the same research](https://ecommons.cornell.edu/server/api/core/bitstreams/445e3b34-ce08-4f5b-8671-466f49963d40/content) measured this directly.
Bohns states the finding plainly: **it is much more awkward, and therefore less likely, for someone to say no to a request asked directly and face-to-face than to one asked indirectly or over email.**
And requesters are **largely oblivious to this**. When the same request was posed either as a dropped hint or as a direct question, targets said they would be more likely to comply with the direct one, while requesters expected the indirect version to work better.
The same pattern held for the medium the request travelled through.
Turn that around and it is a map of your own vulnerability. The evening invitation delivered in person, in front of other people, at the end of a good conversation, is the highest-pressure configuration that exists. A message is the lowest.
Which gives you the single most effective tactical move in this whole article: **do not answer a face-to-face request in the room if you do not want to.** Not as avoidance, and not for days. "Let me check and come back to you tonight" moves the decision out of the setting where the awkwardness is maximal and into one where you can answer the actual question, which is whether you want to do the thing.
The rule for using this honestly is that you have to come back. Buying time and then going quiet is the withdraw pattern again, and it costs more than the refusal would have.
## Deciding before you are asked
The awkwardness the research describes is a property of the moment, not of the decision. So the most reliable way to beat it is to make the decision somewhere the moment is not.
This is not a trick, it is just where the mechanism points. The reason you agree to things you do not want is that at the instant of asking, the immediate discomfort of refusing outweighs a future evening you can only imagine. Move the choice out of that instant and the weighting changes on its own.
**Decide the category, not the instance.** Not "should I go to this" but "am I doing weeknight things this month." A general position is easier to hold because it is not about the person in front of you, and it stops each request being litigated from scratch.
**Know your actual capacity before the week starts.** Most overcommitment happens because the question "can I do this" gets answered against an empty calendar in your head rather than the real one.
**Have a default.** Not a rude one. A holding sentence you can say without thinking, so that the moment does not have to be won on wit. "Let me look and get back to you" is enough, and it works precisely because it requires no decision at the point of maximum pressure.
**Notice which requests you regret saying yes to.** Almost everybody has a pattern, and it is usually narrower than they expect: one type of event, one type of favour, often one particular person. Refusing well in that one category solves most of the problem without turning you into someone who says no a lot.
## The four situations that actually come up
The general principle is the same everywhere. The friction is different in each.
**A friend, socially.** The easiest case and the one people most overthink. Decline the event, not the person, and be specific about what you would do instead if anything. Almost nobody is keeping score, and the ones who are will reveal it here, which is useful.
**At work.** Say no to the task and yes to the problem. "I cannot take that on this week without moving the other thing. Which do you want?" converts a refusal into a decision for whoever has the authority to make it, which is both more useful and much harder to argue with.
**Family.** The hardest, because the request usually arrives with history attached and the refusal gets heard as a statement about the relationship. This is where brevity matters most. The long version will be about the history within two minutes.
**A partner.** The one with the most at stake and the one where the silent version does the most damage. Everything above applies, plus one thing specific to this context: say what you do want, not only what you do not. A refusal with no alternative in it leaves the other person with nothing to work with, and that is how a small no becomes a subject.
## When they push back
Most pushback is not a challenge to your decision, it is the other person processing an unexpected answer, and it passes if you let it.
Give it a beat. Do not fill the silence, because what you will fill it with is a concession.
Repeat the same sentence if they ask again. Not a longer one.
Do not defend the reason, because defending it makes the reason the subject, and the reason was never the point. The point is the answer.
And if the pushback continues past two rounds, notice what that is. Someone who cannot accept a no is telling you something about how they see the arrangement, and that is worth more than the Saturday you are arguing about. If this is a pattern in a relationship rather than a one-off, that belongs with the harder questions in [how to know when to go](/social/when-to-end-a-relationship/).
## The same finding, pointed at you
Everything so far treats you as the person being asked. The research was designed around the person doing the asking, and read that way it says something worth sitting with.
If people agree to things they do not want in order to avoid the discomfort of refusing, and if the person asking cannot perceive that, then some of the yesses you have received were not the free choices you took them for. Bohns states it directly: requesters **view compliance as more voluntary than targets experience it.**
The same programme demonstrated this with requests that were not trivial. In one study, participants persuaded strangers to commit a small act of vandalism, and the general pattern held: people underestimated how much influence they had, because they could not feel the pressure they were applying.
Two practical things follow, and neither is complicated.
The first is that a yes given in a high-pressure configuration is weaker evidence than it feels like. Asked face-to-face, in front of others, by someone who wants a particular answer, agreement carries a lot less information than the same answer given later by message. If it matters whether someone actually wants to do the thing, ask in a way that makes no cheap to say.
The second is to take a hedge as an answer. "I think so", "maybe", a pause, a change of subject: the research says the awkwardness of refusing is strong enough that many people will route around a direct no rather than deliver one. Treating anything short of a clear yes as a yes is a mistake the person asking is systematically prone to making, because from where they stand it all looks voluntary.
This matters most in the settings where the stakes are highest, and it is the reason the same body of research gets cited in discussions of consent. The standard worth holding is not whether someone said no. It is whether saying no was easy.
## What this does not fix
Saying no well solves a narrow problem, and it is worth being clear about the edges of it.
It does not make you liked by people who liked you for your availability. Some of that will shake out, and the shaking out is the point rather than a side effect.
It does not remove the discomfort. The research says the awkwardness is real and immediate, and nothing in this article makes it stop being real. What changes is that you become willing to pay it, which is different from not feeling it.
It does not work retroactively on the resentment you have already accumulated. That needs its own conversation, and it will go better if it is not delivered all at once.
And it will not always be received well. The measured findings here are about averages, and the average reaction to a clear no is considerably milder than you expect. The exceptions exist, and the useful thing about them is that they identify themselves.