Their series of studies collected data using a mobile van offering people chances to win prizes - from gift vouchers to chocolate bars - through a dice-roll lottery. They could choose from lotteries A or B, where overall A's prizes were slightly but evidently better. Before comitting, the chooser met another participant, the 'advisor', who handed them a written recommendation of which lottery to pick. In most cases, the advisor had a conflict of interest - they would also get a go on a lottery, but only if the chooser selected the weaker lottery B. The chooser then made their selection, rolled the dice and left; the advisor would then get their turn, if warranted. Participant numbers ranged from 124 to 278 for individual studies.
In the first study, 53% of choosers took lottery B after merely receiving the advice to do so. When the advisor's recommendation also included text revealing their conflict of interest, compliance advice rose to 81%. Yet in both conditions participants rated lottery A as more attractive (this was consistent across studies). A replication examined whether relatively low stakes were driving this abandonment of self interest, by doubling the prizes and recruiting students with presumably lower income as participants. Without disclosure, only 36% took the recommended B, but disclosure took the proportion to 82%.
Was this an altruistic act, choosers electing to be generous and go on with their day? Unlikely: the post-study survey suggested that after compliance, choosers were less happy, sensed more pressure and felt more uncomfortable about the decision, which doesn't suggest general altruism. Instead, the researchers liken this to a 'panhandler effect', where money is passed over due to discomfort over a face to face refusal. A third experiment investigated this: here, when the chooser learned of the conflict from the advisor compliance stood at 90%. When the information came instead from a 3rd party (embedded in the initial instructions) their compliance dropped to 72%; it's less awkward if you're not told directly by the person who hopes to gain, even if they know you know. And if the 3rd party info also stated the advisor was oblivious that you had been made aware of the conflict, the compliance plummet to 47%. This suggests that when choosers comply, it's partly to avoid the perception that they have betrayed the advisor's interests. Without the shared knowledge - I know that you know I stand to benefit - they're happy to disinherit them.
My only quibble with this argument is that in the final, 47% compliance condition, I might personally view the advisor as shiftier. Holding secret information, I may spend the interaction expectantly waiting for them to 'fess up to the conflict of interest - something the experiment actually prohibits. When they don't, I might feel like punishing them by going against their wishes. However, the post-survey scores suggest that there was no significant difference between how much advisors were liked and trusted in this condition and the other disclosure conditions, which goes some way to minimise this concern.
Overall, disclosure leads to more compliance with the advisors interests, especially when disclosure is face to face. This happens even though trust in the advisor drops, and choosers are less happy with the situation. This suggests that the tactic of disclosure practiced simply may be unhelpful for the chooser and ultimately less conducive to the relationship overall. Sah and colleagues agree that disclosure remains important and necessary, but suggest research into smarter ways to deliver it, as well as alternative approaches when conflicts of interest arise.
Paul M Healy, Krishna G Palepu, (2001). Information asymmetry, corporate disclosure, and the capital markets: A review of the empirical disclosure literature, Journal of Accounting and Economics, Volume 31, Issues 1–3, September 2001, 405-440, DOI10.1016/S0165-4101(01)00018-0.
(link to pdf)