Does Conflict Disclosure Build Trust

Answer to the question

Does disclosing a conflict make advice trustworthy?

Sometimes, and not for the reason the theory gives. In experiments, choosers trusted a conflicted adviser less after disclosure yet complied more, 81 per cent against 53. Disclosure helped most when it led advisers to decline the conflict, as 63 per cent accepted one without disclosure and 33 per cent with it.

Key takeaways

• Conflict disclosure has two supports: listeners discount, and advisers decline conflicts they would have to declare. In one study, 63 per cent of advisers accepted a conflicted payment without disclosure and 33 per cent with it.
• Cain, Loewenstein and Moore (2005) found that listeners under-discount biased advice even when told, and that disclosure can license advisers to exaggerate further.
• Sah, Loewenstein and Cain (2013): where advisers recommended the worse option, 53 per cent of choosers took it without disclosure and 81 per cent with it, because refusing a disclosed request felt uncomfortable.
• Where the conflict cannot be avoided, the label has to work through the listener, and that is where it fails most often.

In the Final Edition of Nudge, Richard Thaler and Cass Sunstein recommend a debt app founded by a former student of Thaler's, and then stop to say so: "We have no financial stake in his company."

It is one plain sentence, and it is conflict disclosure doing what the theory asks of it. The reader learns what the authors stand to gain, which is nothing, and sets the weight of the recommendation accordingly. The experiments that tested the theory have been less kind.

What the label is supposed to do

The logic has two halves. The listener discounts: told that an adviser is paid by the outcome, they trim the advice and decide for themselves. The adviser is restrained: knowing the interest will be seen, they temper what they say, or step away from the interest altogether. Medicine, finance and accounting all require [conflict of interest disclosure](https://thewhyweproject.com/articles/native-advertising-and-the-inversion) of one kind or another, which is why it is [the default remedy in so many professions](https://thewhyweproject.com/articles/why-we-trust-online-ratings-anyway).

The strongest evidence for the first half comes from medicine. In a randomised study published in the New England Journal of Medicine in 2012, 269 board-certified internists read abstracts of hypothetical drug trials that differed in rigour and carried one of three funding statements: industry, the US National Institutes of Health, or none. Physicians rated rigour accurately. An industry statement still lowered their ratings of rigour, their confidence in the results and their willingness to prescribe, compared with no statement at all, and they were about half as willing to prescribe on an industry-funded trial as on an NIH-funded one. Readers do something with the label.

The second half has support too. In three experiments with real money, advisers could choose whether to accept a payment that created a conflict. Without disclosure, 63 per cent took it. With disclosure, 33 per cent did (Sah and Loewenstein, 2014). Required to declare, many advisers preferred to have nothing to declare.

Thaler and Sunstein's sentence, read this way, is the good outcome: the interest was absent, and the page said so.

A confession that works as a licence

Against that sits the paper that started the argument. Daylian Cain, George Loewenstein and Don A. Moore reported in 2005 that conflict disclosure can leave advice worse. In one design, advisers who could inspect a jar of coins up close were paid more when the estimators relying on them guessed high, while the estimators saw the jar only briefly and from a distance. Across the experiments, listeners did not discount biased advice as much as they should have, even when the conflict was disclosed.

The authors add a second failure on the adviser's side. An adviser who has disclosed can feel morally licensed, [the pattern known as moral licensing](https://thewhyweproject.com/articles/what-recycling-was-asked-to-do), and strategically encouraged, to exaggerate further, on the grounds that the listener has been warned. The licence is a separate mechanism from the discount and it points the other way. A warning that lowers trust on one side can raise the permitted bias on the other, and the net effect depends on which moves more.

The 2014 paper that produced the encouraging figure above also contains the unwelcome one. In its first experiment the conflict could not be avoided, and disclosure produced higher, more biased recommendations than silence did. The remedy that thins conflicts out when advisers can leave them inflates advice when they cannot.

The listener complies anyway

The sharpest result concerns the listener, because conflict disclosure also changes what the listener feels they owe. In the first of six experiments, 240 people recruited in Pittsburgh played in pairs. Advisers recommended one of two die-roll gambles to a chooser, and almost everyone preferred the better gamble in a pretest. Some advisers were paid only if the chooser took the worse one. Where those advisers wrote out their interest, choosers trusted them less. They also followed the bad advice more. Among choosers whose adviser recommended the worse gamble, 53 per cent took it without disclosure and 81 per cent took it with disclosure.

Sunita Sah, with Loewenstein and Cain, called this the burden of disclosure, and their explanation is blunt. A statement meant to say "I gain if you choose this" can be heard as "please choose this because it helps me", and a favour requested face to face, with the outcome visible, is hard to refuse. The measured mechanism was pressure: disclosed choosers reported wanting to help their adviser and finding it more uncomfortable to turn the advice down. They were also less pleased with their choice and liked their adviser less, the signature of reluctant compliance and not of trust.

A label meant to say "I gain if you choose this" can be heard as "please choose this because it helps me".

The effect shrank when the disclosure came from an outside source, when it was not common knowledge between the two, when the chooser could revise later, and when the decision was private.

Medicine shows a version of it. A 2016 study of specialty bias drew on recorded surgeon and patient consultations at Veterans Affairs hospitals and on a randomised laboratory experiment. Disclosing the bias raised patients' trust and raised their likelihood of choosing the treatment that matched the physician's specialty. Nudge makes a neighbouring point about nudges themselves: transparency about a nudge does not reduce its effect and "can easily have the opposite effect", because a disclosed default also conveys a recommendation.

Removing the conflict instead

Set the findings on conflict disclosure side by side and a pattern appears. Conflict disclosure has a clean record in these experiments where it changes who holds the conflict. In the 2014 work, advisers who could decline a conflict did so more often when they knew they would have to declare it. The authors put the motive down to reputation, since people dislike being seen as biased, so the label works on the adviser before it ever reaches the listener. That leaves advisers with nothing to declare but the absence of a conflict, which is what the sentence in Nudge amounts to.

Where the conflict stays in place, conflict disclosure has to work through the listener, and the listener is the weak link. Physicians in the funding study discounted, yet choosers in the die-roll study complied, and one difference is what the listener is asked to do. A reader weighing a published abstract can mark it down in private. A patient or customer facing the person who stands to gain is being asked to refuse them.

The authors of the six burden experiments reached the same place, arguing that policy should focus less on disclosing conflicts and more on decreasing them.

Our reading is that the label earns its place when advisers can walk away from the interest, and costs the most when they cannot. A form that only records the interest leaves it in place. Where the interest stays, the listener must refuse it.

Common questions

Does disclosing a conflict make advice trustworthy?

Sometimes, and not for the reason the theory gives. Conflict disclosure lowered trust in the die-roll experiments yet raised compliance from 53 to 81 per cent, and in jar-of-coins experiments it left advisers freer to exaggerate. It worked best when it led advisers to decline the conflict.

Why do people follow advice more after a disclosure?

Conflict disclosure can read as a request for a favour. In the die-roll experiments, choosers who heard of the adviser's stake reported more pressure to help and more discomfort in refusing, and that pressure accounted for the extra compliance.

When does conflict disclosure actually work?

It works when the adviser can avoid the conflict. In one experiment 63 per cent of advisers accepted a conflicted payment without disclosure and 33 per cent with it, and the advice they gave was less biased.

Sources

Thaler, R. H., & Sunstein, C. R. (2021). Nudge: The Final Edition. New York: Penguin Books

Cain, D. M., Loewenstein, G., & Moore, D. A. (2005). 'The Dirt on Coming Clean: Perverse Effects of Disclosing Conflicts of Interest.' Journal of Legal Studies, 34(1), 1-25. doi:10.1086/426699

Sah, S., Loewenstein, G., & Cain, D. M. (2013). 'The burden of disclosure: Increased compliance with distrusted advice.' Journal of Personality and Social Psychology, 104(2), 289-304. doi:10.1037/a0030527

Sah, S., & Loewenstein, G. (2014). 'Nothing to Declare: Mandatory and Voluntary Disclosure Leads Advisors to Avoid Conflicts of Interest.' Psychological Science, 25(2), 575-584. doi:10.1177/0956797613511824

Sah, S., Fagerlin, A., & Ubel, P. (2016). 'Effect of physician disclosure of specialty bias on patient trust and treatment choice.' Proceedings of the National Academy of Sciences, 113(27), 7465-7469. doi:10.1073/pnas.1604908113

Kesselheim, A. S., Robertson, C. T., Myers, J. A., Rose, S. L., Gillet, V., Ross, K. M., Glynn, R. J., Joffe, S., & Avorn, J. (2012). 'A randomized study of how physicians interpret research funding disclosures.' New England Journal of Medicine, 367(12), 1119-1127. doi:10.1056/NEJMsa1202397

Related concepts

• Moral Licensing
• The Burden of Disclosure
• Conflict of Interest Disclosure

People

• Richard Thaler
• Cass Sunstein