Risk comparisons backfire when they equate a risk someone chose with a risk imposed on them. A medical scan is voluntary, a hazard in your own home is not, and flattening that difference reads as dismissive. Use a comparison to give scale, never to tell someone their fear is unreasonable.
In a two-day presentation training session with a large nonprofit in Washington DC this month, the most useful argument of the two days was one nobody won.
The subject was risk comparisons. You know the shape of them. This exposure is about the same as the one you get from a routine medical scan. The level we found is a fraction of what you would encounter doing an ordinary household task. Half the room considered these indispensable. The other half considered them the fastest way to lose an audience. Both sides had evidence, and by the end of day two we had not resolved it.
I am going to lay out both cases honestly, because pretending this is settled would be the wrong kind of confidence. Then I will give you the decision framework I use, which is not a verdict.
Why do risk comparisons backfire?
The core failure is a category error, and once you see it you cannot unsee it.
A medical scan is a voluntary exposure. A hazard in your own home is an involuntary one. You choose the scan, in consultation with a doctor, for a benefit you understand, at a time you agreed to. You did not choose what is in your walls, your water or your air, and nobody asked you.
Decades of risk research point the same direction: people accept risks they choose at rates far above the risks that are imposed on them. That is not innumeracy. It is a reasonable position about consent. When you put a chosen risk and an imposed risk side by side and say the numbers are similar, the listener does not hear reassurance. They hear you telling them their objection does not count.
And then they stop listening, which means the rest of your material, including the parts that would genuinely have helped, does not land either.
What is the boomerang effect of a risk comparison?
This is the failure that should worry you most, because it damages something outside your own story.
Say you tell an audience that an exposure they are frightened of is roughly comparable to a routine medical scan. You intend the comparison to travel in one direction: the scary thing is smaller than you feared. A meaningful number of listeners will run it the other way.
They do not become calmer about the exposure. They become alarmed about the scan. And some of them will decline a scan their doctor has recommended, for a condition that would have been caught.
That is a real cost with a real face on it, and it lands on someone who was never part of your story. A comparison is not a one-way instrument. It transfers meaning in whichever direction the listener is already leaning, and the listener is usually leaning toward worry.
Why does comparing to a catastrophic threshold fail?
The other common move is to reach for the level at which serious harm actually begins. What we found is far below the level associated with acute illness.
Technically that may be exactly right. Rhetorically it is a disaster, because you have just told the audience that the bar for concern is severe illness. Anything short of catastrophe now reads as nothing to see here.
Most people are not asking whether they are about to fall seriously ill. They are asking a much more ordinary question: is this okay for my kids, over years, in the house we already bought. Answering a small, cumulative, personal question with a catastrophic yardstick tells them you were not listening to the question they asked.
Is there a case for using comparisons at all?
Yes, and it is stronger than the critics allow.
A number with no scale is not information. Tell someone a figure in units they have never encountered and you have given them anxiety, not understanding. Their next move is to search for a reference point, and if you did not supply one they will find one, from a stranger, on a platform optimised for alarm.
The scientists in the room made this point with some force, and they were right about it. Comparisons are how technical people establish magnitude. More to the point, comparisons are the only yardstick the public already owns. You cannot hand someone a new unit of measurement in a ninety-second interview.
There is also a fairness argument. Refusing every comparison can leave an audience believing a small risk is a large one, and that has costs too. People move house, spend money they do not have, and carry fear for years over something that did not warrant it. Being unhelpfully vague is not a neutral act. I make a related point about numbers that arrive without meaning in why data alone never convinces.
How do you decide whether to use a comparison?
This is where I landed, and it is a test rather than a rule. Run a proposed comparison through four gates. If it fails any one of them, do not say it out loud.
- Gate 1: consent. Is the thing you are comparing to also involuntary? If one side was chosen and the other was imposed, the comparison fails before you finish the sentence.
- Gate 2: scale, not permission. Is the comparison giving the audience a sense of size, or is it telling them how to feel? Roughly the size of X survives. So you should not worry does not.
- Gate 3: reverse flow. If a listener ran the comparison backwards, would they make a worse decision? If anyone could plausibly conclude that they will now skip a medical procedure, the comparison is unusable no matter how accurate it is.
- Gate 4: the clip test. Would the comparison still be fair if it were quoted with the surrounding sentence removed? It will be. Comparisons are the most clippable thing a technical spokesperson says.
Gate 4 comes from thirteen years of producing at ABC, NBC and Fox. A vivid comparison is exactly what gets pulled for a story, and the qualifier attached to it is exactly what does not. If your comparison only works with its caveat, assume it will run without it.
One more thing that is not a gate but matters as much as all four. Acknowledge the fear first, in one sentence, before any number. It makes complete sense to be concerned about something you did not choose to be exposed to. Then the number. An audience that feels heard will hear a comparison as help. An audience that feels corrected will hear the same words as dismissal, and the words are identical.
What do you say when a reporter offers you a comparison?
This happens constantly and it is the highest-risk moment in the whole interview, because the reporter has done the work of making the comparison memorable and is now inviting you to sign it.
Do not repeat it, including to reject it. Anything you repeat can be quoted as yours, and a transcript does not carry the fact that you were pushing back. That rule and the technique behind it are in bridging does not tell you what to say.
Three usable responses, in ascending order of firmness:
- Substitute. The comparison I would make is this one, and here is why it fits better.
- Redirect to the decision. I would not put those two next to each other. What actually helps your readers is knowing who is most affected and what changes the number.
- Decline cleanly. I do not think that comparison helps anybody. Here is what I can tell you with confidence.
Notice that all three keep talking. A refusal that ends in silence gets edited into a spokesperson who would not answer. A refusal that runs straight into something useful gets edited into a spokesperson who was helpful, which is what you are.
What did the room not resolve?
The disagreement itself, and I think that is the honest state of the field rather than a failure of the session.
The technical staff wanted comparisons kept, because magnitude is meaningless without a reference point and because they watch the public reach for far worse reference points in the absence of a good one. The communicators wanted them dropped, because they have watched the boomerang happen and because a comparison that lands badly costs you the whole interview, not just the sentence.
Nobody moved. What both sides did agree on, in the end, was narrower and more useful than a verdict: never use a comparison as your first sentence, never use one to tell somebody how to feel, and never use one you would not be comfortable seeing quoted on its own. That is the four-gate test in plain language, and it is the most agreement I have ever seen a room reach on this.
If your technical experts have to explain risk in public and the comparison question is live for them, that is exactly the kind of thing a session should settle in advance rather than in front of a camera. Call (646) 233-1424 or tell us what is coming up.
Drawn from a two-day media training session with a federal agency in Washington DC on 13 and 14 August 2026. The agency, the subject matter, the specific substance and all participants are withheld. The disagreement described here was genuine and was not resolved during the session.