Telling It
Like It Is

The Confession That Costs Nothing

Advisors may be tempted to dress up guesses as forecasts because candor feels expensive. But research across twenty-eight experiments found its price: zero.

Dries, C., Schneider, C. R., Oeberst, A., McDowell, M., Egharevba, G., & Rebitschek, F. G. (2026). The impact of uncertainty communication on trust in its sources: A systematic review and meta-analysis. Royal Society Open Science, 13(4), 251896. Open access: doi:10.1098/rsos.251896

Climb high enough and you can see the whole valley: the river, the village, the switchback road you came up by. What you still cannot see is what the weather will do tomorrow. Every expert lives on that ridge.

A scientist asked about a new vaccine, a forecaster asked about a storm track, an economist asked about next year's rates: each can describe the terrain in front of them, and each knows exactly where their sight runs out. The question is whether to admit it.

For years, two camps have argued about that admission. One holds that saying 'we are not sure' is honesty, and honesty reads as trustworthiness. The other holds that audiences hear doubt as weakness, that confidence is the coin of expertise, and that every hedge spends some of it.

The argument turned urgent during the pandemic, when guidance shifted week to week and officials had to decide, live on camera, how much of their own uncertainty to show. Health communication guidelines said be transparent. A well-known line of psychology, the confidence heuristic, warned that transparency would read as incompetence and cost the messenger.

Twenty-Eight Experiments, Weighed Together

Charlotte Dries of the Harding Center for Risk Literacy in Potsdam and five colleagues across Germany, England and New Zealand decided the argument had run long enough on anecdote. They registered their plan publicly in advance, then combed three research databases for every experiment ever run in which a source either did or did not admit uncertainty and the audience then rated its trust in that source.

The sweep started with 3,909 candidate papers and ended, after screening, with 28 qualifying experiments: studies about vaccines, climate change, tax projections, unemployment forecasts, even a museum exhibit. In one, the BBC ran the test live on its own news site.

The manipulations were small and honest. In one study, a fictional 'Dr. Anglada' either promised that a new vaccine would work or said she was hopeful it would. One group got the confident version, one got the caveats, and both then rated how much they trusted the messenger.

What They Found

Pooled together, the effect of admitting uncertainty on trust was Hedges' g equals 0.00. Zero, to two decimal places, with the confidence interval hugging it from both sides. Twenty-three of the 28 experiments individually found no effect at all.

The five exceptions were small and pointed both ways, and the one clear outlier ran in candor's favor. Twenty-eight experiments in a row failed to catch audiences punishing experts for admitting the limits of their knowledge.

Then the team attacked its own result. Many original studies had let words like 'may' and 'around' leak uncertainty into the supposedly confident version, so they re-ran the analysis on the seven cleanest studies. Still nothing. Restricting to studies whose participants verifiably perceived the difference? Still nothing. Publication bias? None to be found.

Even the audience did not matter. Education, political leaning, health literacy, tolerance for ambiguity, optimism: none of it changed the verdict.

One more finding earned its own sentence: in two of the studies, sources that flagged their uncertainty up front lost far less trust later, when the evidence changed and their guidance had to move. The hedge, it turns out, is also armor for the day you have to update.

What This Means at Your Desk

For a profession coached to sound certain, a fear this thoroughly tested and this absent deserves to lose its grip on how you talk to clients about the future. The studies were about scientists and forecasters; the moves below are ours, translated for the desk. Five places to start.

Consider what the swap sounds like. Old: 'We see the S&P ending the year around 6,900.' New: 'Most outcomes we model land between down five and up twelve percent, and the plan works across that band.' The second version is longer, cheaper, and true, and it matches what the plan actually models. Nothing in twenty-eight experiments says it costs you the room.

Give the range you actually believe. If the analysis supports a band of outcomes, present the band, and put the confidence interval on the plan projection instead of the single line down the middle. The fear that a range reads as weakness has now been tested twenty-eight times without materializing. Retire it, and take the pressure off your forecasts while you are at it.

Say you don't know about the things nobody knows. Where rates go next year, what the election does to markets, which quarter the correction lands. The admission costs you nothing with the client, and it buys you the buffer the research hinted at: the advisor who flagged the uncertainty loses less trust on the day the picture changes.

Expect no credit for it, either. Hedging is not a trust strategy, and clients will not lean in because you gave a range. The case for candor is narrower and better: it is free, it is accurate, and it leaves you owning fewer false promises when a point estimate misses.

Audit your materials for false precision. Projections quoted to a decimal place, single-number retirement dates, target prices with no band around them. Each manufactures a certainty the analysis does not contain, and this research removes the commercial excuse for the habit. Round to the precision you can defend.

Rehearse the sentence. 'The honest answer is a range, and here is the range' lands differently when it does not sound like an apology. Deliver it with the same energy you would give a conviction pick. The disclosure is safe; the delivery is the part you still control.

The Bottom Line

An industry that sells confident numbers is betting that clients punish doubt. Researchers pooled every experiment ever run on the question and found the flattest result statistics can produce: a zero, checked and re-checked. Keep dressing up guesses as forecasts if you like, but the evidence says nobody was paying you for the costume.

The Open-to-Being-Wrong Advantage

People trust experts who seem open to being wrong, believe them more, and seek out their advice. Faking it is the one thing that does not work.

Koetke, J., Schumann, K., Bowes, S. M., & Vaupotic, N. (2025). The effect of seeing scientists as intellectually humble on trust in scientists and their research. Nature Human Behaviour, 9(2), 331-344. doi:10.1038/s41562-024-02060-x

You have met the expert who is never wrong. He holds court at the end of every dinner table, and a question about his confidence lands as a question about your manners.

His instinct runs deep in every profession that sells advice: certainty is what expertise sounds like, and doubt is what it costs. Four psychologists have now spent five preregistered studies finding out whether audiences actually agree.

The Experiments

Jonah Koetke and Karina Schumann of the University of Pittsburgh, with colleagues at Vanderbilt and Vienna, study a trait called intellectual humility: the awareness that what you know might be limited or wrong. A first survey of 298 Americans pointed one way: the more humble people perceived scientists to be, the more they trusted them.

Surveys cannot say which way the arrow runs, so the team built an experiment. Participants read a news article about a fictional scientist, Dr. Susan Moore, and her research on long COVID treatments. One detail varied. In one version, a colleague remarks that Dr. Moore 'is not afraid to assert what she knows.' In the other, the same colleague says she 'is not afraid to admit when she doesn't yet know something.'

One sentence of character, and the ratings moved. The humble Dr. Moore scored 6.4 out of 7 on trustworthiness; the assertive version scored 5.6, below even a plain control that said nothing about her character. The pattern held as the scientist became a psychologist, then a climate researcher: 2,034 people across five preregistered studies.

The behavior followed the ratings. Offered the chance to actually read the climate scientist's advice on plant-rich diets, 36 percent of participants took it when she had been described as intellectually humble. For the version who admitted nothing, 21 percent did.

The scientist's identity kept not mattering. In one study the same article introduced either Dr. Sandra Wilson or Dr. Robert Wilson; in another, the climate researcher's name signaled her race and ethnicity. Nothing changed. Trust followed the humility, not the person.

Advice-seeking is the measure worth pausing on, because it is behavior rather than sentiment. Participants did not simply say kinder things about the humble scientist; more of them spent their own time reading her work. At your desk, that is the difference between a client who nods politely and a client who calls you first.

Why the Doubter Wins

The authors trace the effect to what a conclusion signals. An expert who names what could prove her wrong is telling you her claims track the evidence rather than her ego. When that expert then states something firmly, the firmness carries information: you already know she would have flagged it if the picture were murkier. The braggart's confidence tells you nothing, because he sounds identical whether he is right or guessing.

Note where the damage concentrated. In the head-to-head studies, the never-wrong version did not just miss the humility bonus; it fell below the neutral control. Reading like someone who cannot be wrong actively cost the scientist trust. Arrogance, in these experiments, was not a style choice; it was a tax.

The Catch

So an expert should simply perform some humility? The fifth study is the cautionary tale. The team asked researchers in the field for the recommended ways to convey intellectual humility, and tested the three leading approaches on a census-matched sample of 679 Americans: talking about updating your beliefs and sharing credit, describing the limits of your methods, and describing the limits of your results.

Two of the three backfired. Readers who saw the scientist catalogue her study's limitations believed the research less than readers who got no humility talk at all. Only the personal kind, the scientist who talked about revising her views, avoided the damage, and even it lifted trust only indirectly.

Audiences reward the trait and punish the performance of it. Flagging your own study's weaknesses read as doubt about the work rather than honesty about knowledge. The signal has to leak out of how a person operates over time; a manufactured version reads as either weakness or theater.

What This Means at Your Desk

The mapping to your practice is close: same credentials, same recommendation, fifteen points of advice-seeking between the expert who names what could change her mind and the one who admits nothing. The backfire finding sets the constraint. Every move below is a demonstration, something a client watches you do, never a line you ask them to believe.

Demonstrate, never announce. A pitch-deck slide about your firm's intellectual humility is an announcement, and announcements failed. Changing your model when the data change, and telling the client why, is a demonstration. The first costs a designer an afternoon; the second is visible evidence of the trait working in support of their money.

Answer 'how sure are you?' with calibration. Give the honest odds, then name the development that would change your view. That answer shows the client your conclusion is attached to evidence, which is the exact trait these studies rewarded, and it costs you nothing on the days you turn out to be right.

Volunteer a past reversal. Somewhere in your record you exited a position or retired a strategy because the facts moved. Tell that story in prospect meetings, with the original reasoning and what changed. In study five, the only humility that survived scrutiny was a person describing how she updates her beliefs. This is your version of it.

Let clients watch you update. When a thesis weakens, say so before you act. 'I am less confident in this position than I was in March, and here is what I am watching' is humility operating in the open, on a live holding, where nobody could mistake it for theater. It is also a rehearsal for the harder call if the exit ever comes.

The Bottom Line

Five studies put the never-wrong expert against the expert who owns her limits, and the never-wrong expert lost on trust, belief, and follow-through, scoring below even the versions that said nothing at all. Clients want conclusions that visibly track the evidence. The only humility that fails is the kind you have to announce, so stop announcing and start showing.

Say It Before You're Asked

The same apology restores more trust when it arrives before anyone demands it. A new experiment suggests why the unprompted version wins.

Langlinais, L. A., & Tomlinson, E. C. (2026). Forced versus voluntary apologies: Trust repair as a matter of perceived values. Negotiation and Conflict Management Research, 19(1), 30-45. Open access: doi:10.34891/0kq7-ec36

Somewhere in the research literature there is a coworker named J.B., and everyone who meets him recognizes him. He is friendly, well liked, and qualified on paper. He has also just cost you a paycheck.

That is the scenario 156 working adults stepped into: J.B. owed you a report you needed to meet a deadline, and on the Zoom call he admits he forgot, got confused, and never finished it. You do not get paid.

The Experiment

Luke Langlinais of Belmont University and Edward Tomlinson of the University of South Florida ran the study in two parts, days apart. Part one delivered the violation. Part two reminded participants what J.B. had cost them, then offered a choice: demand that J.B. apologize, or wait and see what happens.

Then the apology came, or did not, at random. Everyone who received it got the same words, ending 'I hope you can forgive me though, I won't surprise you by failing like that again.' The only thing that differed was the road it arrived by.

Days later, everyone rated how much they trusted J.B. now. Same colleague, same failure, same sentences of remorse.

What They Found

The apology that had to be demanded repaired less. Victims who waited and received the words unprompted put their trust in J.B. at 2.46 on the study's scale; those who had to extract the same words landed at 2.02, a moderate gap (d=0.48) in favor of the volunteer.

The researchers traced the gap to a simple perception: shared values. Participants had rated themselves on five values, honesty, sincerity, empathy, respectfulness, and capability, and later rated J.B. on the same five. An unprompted apology closed the distance: J.B. began to look like someone with values like yours. A demanded apology left the distance wide. He read as someone going through motions you had scripted.

The authors frame it as two kinds of justice. An apology volunteered is restorative, a gift that humbles the giver. An extracted apology is retributive, punishment that puts the offender in his place. Punishment can feel deserved, but in these data it did not rebuild the relationship. Trust scholars have long stressed that the offender taking the initiative spares the victim the discomfort of raising the issue; these results agree.

One more surprise: people who score high on what psychologists call the tendency for interpersonal victimhood, a standing sense of being the wronged party, were less likely to demand an apology, not more. The authors' guess: demanding requires feeling you have the standing to, and for some, staying the injured party is worth more than the repair.

What This Could Mean at Your Desk

Beat the client to the phone. The moment you find the mistake, the clock on 'voluntary' starts running. An apology delivered before the client knows anything is wrong is the purest version of the signal this study measured: nobody asked, and you said it anyway.

Treat their discovery as a missed deadline. Once a client spots the error and has to raise it, whatever you say next repairs from the weaker position, and the study suggests you are working against the discount that comes with being asked.

Watch the instinct to stay quiet. The silent fix feels safe because it avoids an awkward conversation. This research prices that comfort: you are trading away the one version of the apology that carries the shared-values signal, honesty offered when silence was available.

The Takeaway

Last issue's apology research found that repair runs on the promise of reform. This study adds who has to move first: the same words rebuilt more trust when nobody had to ask for them. If there is bad news in your book right now, the most valuable apology is still the one the client never has to request.