The Question Changes the Answer

Why asking for information is also giving it away

By Glenn C Nunes

I. The polite question

Due diligence has a favorite question:

“Is there anything I should know?”

It is broad enough to sound thorough and polite enough not to sound suspicious. It seems to put the whole transaction on the table.

It does something narrower.

It asks the other side to decide what belongs on the table.

The question appears everywhere one person knows more than another: used cars, houses, businesses, employment interviews, settlement discussions, commercial contracts. The subject changes. The structure does not.

One side possesses a fact that matters. The other side does not know the fact—and may not know enough to ask for it by name.

So the less-informed party asks for everything.

The better-informed party supplies something.

That something may be true. It may be detailed. It may even be responsive in the ordinary conversational sense. The seller can describe the maintenance, the battery, the paint, the employees, the customers, the condition of the roof, or the history of the relationship. Nothing said has to be false.

The important fact can simply fail to make the list.

This is not an argument against open questions. When you do not know what problems exist, a broad question may uncover something you did not know enough to name.

But an open question is an invitation to begin the inquiry.

It is not the inquiry completed.

II. Three questions

In 2018, Julia Minson, Eric VanEpps, Jeremy Yip, and Maurice Schweitzer published an experiment involving the sale of a used iPod.1

The sellers knew something the buyers did not. The device had frozen and crashed twice before, for reasons that remained unknown. Sellers also had an incentive to obtain a higher price.

The researchers assigned sellers to receive one of three questions:

“What can you tell me about it?”

Or:

“It doesn’t have any problems, does it?”

Or:

“What problems does it have?”

The first question was general.

The second identified the relevant subject but assumed a reassuring answer.

The third assumed a problem existed and asked the seller to describe it.

The results were not close.

After the general question, approximately 8 percent of sellers disclosed the freezing problem.

After the reassuring question, approximately 61 percent disclosed it.

After the question that assumed a problem, approximately 89 percent disclosed it.

The study did not compare three kinds of sellers. Different participants were randomly assigned to the three conditions, but they received the same role, the same information, and the same basic incentives. What changed was the question.

Disclosure moved from roughly one seller in twelve to nearly nine in ten.

The questions also defined three different tasks.

“What can you tell me?” asked the seller to choose the subject.

“It doesn’t have any problems, does it?” identified the subject but supplied the preferred answer.

“What problems does it have?” treated the existence of a problem as settled and asked for an inventory.

All three questions requested information.

Only one made the damaging information the unavoidable subject of the answer.

III. What the question tells

The question changed more than the range of available responses.

It changed what the seller thought about the buyer.

Participants who received specific questions tended to perceive the buyer as more knowledgeable. Those who received the negative-assumption question also perceived the buyer as more assertive. Those impressions statistically accounted for part of the difference in disclosure.

The experiment did not establish every step in the seller’s internal reasoning. It did not ask whether sellers believed this buyer would investigate further, discover the problem independently, or punish an evasion.

But it established the more important point.

A question is a two-way transmission.

It requests information from the respondent while supplying information about the asker.

A general question may communicate that the asker does not know which subjects matter and has left the respondent free to choose among them.

A specific question communicates that the asker has identified the subject.

An assumptive question communicates something further: the asker already regards the problem as plausible enough to place it inside the question.

The wording therefore changes the informational position of both parties.

The buyer learns from the answer.

The seller learns from the question.

That helps explain why the general question can fail without producing a lie. It does not require the seller to deny anything. It permits selection.

The seller can answer with the favorable facts and omit the unfavorable one. Every sentence can be true. The buyer can nevertheless leave with a seriously false impression.

Truthfulness and disclosure are different things.

A general question can receive a truthful answer and still fail as an inquiry.

IV. The missing fourth question

The experiment supports a powerful result. It does not support every possible lesson one might draw from it.

There is a fourth question the researchers did not test:

“Has the iPod ever frozen or crashed?”

That question is specific, but it does not presume either answer.

Its absence matters.

The experiment shows that the negative-assumption question performed dramatically better than the general question and substantially better than the reassuring question. It does not tell us how the negative-assumption question would have performed against a neutral question naming the precise defect.

The authors identify that comparison as a question for future research. They also acknowledge that general questions can be useful when the asker does not yet know which problems to investigate.2

The lesson, then, is not that every inquiry should begin with an accusation.

It is that broad and specific questions perform different jobs.

A broad question searches for subjects you have not anticipated.

A specific question tests the subjects you have.

That suggests a sequence.

Begin broadly enough to allow the unexpected fact to surface. Then ask separately about each important problem that is reasonably foreseeable.

“What should I know?” may belong at the beginning.

It should not be mistaken for the end.

Open questions are not the problem.

Unfinished inquiry is.

V. The price of suspicion

The question that assumes a problem does something useful.

It also spends something.

An assumptive question can signal knowledge and attention when the assumed problem is plausible. The same question can signal poor judgment or hostility when the assumption is not.

Minson and her colleagues tested that boundary in a later experiment involving job-interview questions. Some participants were asked about relatively common workplace misconduct, such as using work time for personal email or social media. Others were asked about much less common conduct, such as online gaming at work.

When the presumed behavior was rare, the negative-assumption question did not produce the same disclosure advantage. It also imposed an interpersonal cost: the interviewer was judged less favorably, including on warmth and competence. The cost was especially pronounced when the assumption appeared implausible.

This is not merely a warning to be nicer.

It is a problem of informational credibility.

“What problems has this used device had?” can sound like the question of an experienced buyer.

“How much money have you stolen from your employer?” can sound like the question of someone who has confused suspicion with evidence.

The grammar is similar.

The signal is not.

The practical rule is therefore narrower than assume the worst.

When a problem is reasonably plausible, ask a question that requires the other side to engage with it. When the premise is speculative, the question may spend trust and credibility without buying useful information.

Suspicion is not free.

The question should not be more confident than the evidence behind it.

VI. The exit that remains open

Even a well-framed question has another limitation.

It cannot make the answer true.

An earlier study by Maurice Schweitzer and Rachel Croson involved 74 negotiations over the sale of a used computer.3 The sellers knew that its hard drive was unreliable. The buyers did not.

In the negotiations where buyers did not ask about reliability, no seller disclosed the problem. Seventy-five percent concealed it by omission. The remaining 25 percent made an affirmative false statement.

Where buyers asked directly about reliability, 61 percent of sellers disclosed the problem. None merely omitted it.

But 39 percent gave an affirmative false answer.

The pattern is striking, but it must be described carefully. Whether buyers actually asked the reliability question was not itself randomly assigned, and only twelve buyers failed to ask it. The increase in affirmative lying from 25 to 39 percent was not statistically significant.

The study therefore does not establish that a direct question causes more people to lie.

It establishes something less dramatic and more useful.

A direct question can greatly reduce the room for omission without eliminating deception.

In this simulation, the direct question closed the route marked omission.

It left the route marked falsehood open.

That is why a direct question is not a truth machine.

It forces the respondent to address the subject more clearly. It may convert concealment into disclosure. It may also produce a clear denial from someone prepared to give one.

The clarity of the question sharpens the meaning of the answer.

It does not certify its truth.

VII. The answer to a different question

The next failure often belongs to the listener.

Ask:

“What problems has it had?”

Receive:

“I have taken excellent care of it.”

The answer is favorable.

It is fluent.

It may even be true.

It does not answer the question.

Negotiators routinely evaluate answers by their general impression: Was the person confident? Cooperative? Detailed? Did the explanation sound natural? Did the conversation continue without awkwardness?

Those are social judgments.

The informational judgment is simpler:

What did I ask?

What did they answer?

The difference is easy to lose because an evasive answer often remains near the subject. A description of careful maintenance sounds relevant to mechanical reliability. A statement about a company’s strong revenues sounds relevant to customer concentration. An assurance that litigation is being handled sounds relevant to probable exposure.

Nearness is not responsiveness.

Do not grade an answer for fluency.

Grade it for relevance.

When the answer changes the subject, return to the question. When the answer supplies an adjective, ask for the fact beneath it. When the answer is categorical and the stakes matter, test it against documents, inspection, records, third-party information, or contractual protection as the circumstances permit.

Questions are part of diligence.

They are not diligence itself.

VIII. The second job of every question

Questions are often treated as neutral instruments: insert the right words, extract the hidden fact.

They are not neutral.

A question determines the subject of the answer. It allocates control over what counts as responsive. It communicates what the asker knows, suspects, and intends to pursue. It can make omission easier or harder. It can increase disclosure. It can spend trust. It can also invite a lie.

No single form solves all of those problems.

A general question gives the other side room to identify what you have missed.

A specific question prevents an important subject from disappearing into that room.

An assumptive question signals that the problem is already on the table—but the strength of that signal should depend on the strength of the reason for putting it there.

Then comes the part no phrasing technique can perform for you.

Listen to the answer you actually received.

Ask broadly enough to discover what you did not anticipate.

Ask specifically enough that the important fact cannot hide inside praise.

Ask assertively enough to signal attention, but not beyond what the evidence justifies.

Then verify what matters.

Ask like someone who will notice the dodge.

Then notice it.

Notes

Footnotes

1. Julia A. Minson, Eric M. VanEpps, Jeremy A. Yip & Maurice E. Schweitzer, “Eliciting the Truth, the Whole Truth, and Nothing but the Truth: The Effect of Question Phrasing on Deception,” 147 Organizational Behavior and Human Decision Processes 76, 83–86 (2018), The reported percentages come from Study 3. Participants acted as sellers, knew that the iPod had frozen twice, and had a financial incentive tied to the selling price.

2. Id. at 90–91. The authors note both that general questions may help uncover problems the asker does not know to ask about and that the study did not compare negative-assumption questions with neutral questions identifying the same specific conduct.

3. Maurice E. Schweitzer & Rachel Croson, “Curtailing Deception: The Impact of Direct Questions on Lies and Omissions,” 10 International Journal of Conflict Management 225, 239–42 (1999). The comparison between negotiations in which buyers did and did not ask about reliability was observational rather than a randomized direct-question manipulation. The authors report that the 39-percent and 25-percent rates of affirmative deception were not statistically different.