• Choose a category
• All
• Classic-Fiction

Our experiment presented participants with a series of vignettes in three different institutional contexts: politics, medicine, and consumer products. In each context, the cases differed only by the extent to which an actor’s financial incentive was described to be dependent upon a particular outcome.

Across all three of the domains we tested, the mere suggestion of a link between financial incentives and a particular outcome significantly influenced the participants’ trust and confidence in the underlying actor or institution. Doctors’ advice was judged to be less trustworthy if the procedure they recommended was tied to a financial incentive. Politicians were judged to be less trustworthy if they supported a policy consistent with the agenda of contributing lobbyists. Researchers for consumer products were judged less trustworthy if their work was funded by an agency that had a financial stake in the outcome. And most surprisingly to us, these variations in the hypotheticals we presented also significantly influenced the participants’ judgments of their own doctors, politicians, and consumer goods. Even the suggestion of one bad apple was enough to spoil the barrel.

In each of these contexts, of course, we might well say that the participants made a logical mistake. In none of the cases did we prove that the money was affecting the results. In none of the cases did we even suggest that it was. But logic notwithstanding, trust was affected merely because money was present in a way that could have biased the results. We infer bias from the structure of the case. Rightly or wrongly, this is how we read.22

4.

The field of “conflicts of interest” focuses on the question of when we should be concerned about dueling loyalties within a single decision maker or single institution. If, for example, you’re a judge deciding a billion-dollar lawsuit brought against Exxon, the fact that you’ve got any financial connection to Exxon, however small, is enough to disqualify you from that suit. Your decision should depend upon the law alone. And one fear addressed by “conflicts” rules is that your loyalty might be split between the law and your own personal gain.

But come on—a single share of Exxon stock is enough to get a judge kicked from the case? Does anyone actually believe that a judge would throw a case because her stock might move from sixty dollars to sixty-one? Why does the law worry about such tiny things? Or, more sharply, why would it require a judge to step aside merely because, as the law states, her “impartiality might reasonably be questioned”? Shouldn’t the test be whether the judge is partial? And if she is not partial, then shouldn’t the question of whether people “might reasonably question her impartiality” be irrelevant? We don’t lock people up in jail merely because other people “might reasonably” believe they’re guilty. Why do we kick a judge from the bench?

Imagine a judge we know is impartial. Put aside how we know that; just assume that we do. If we know the judge is impartial, why should the fact that others might “reasonably” think otherwise matter? Sure, if we don’t know, what others might “reasonably” think might be important. But what if we do know?

The answer to these questions is that uncertainty has its own effect. The law might say someone is innocent until proven guilty. But law be damned, if you learn that a school bus driver has been charged with drunk driving, you’re going to think twice before you put your child on his bus. Indeed, even if you think the charge is likely false, the mere chance that it is true may well be enough (and rationally so) for you to decide to drive your kid rather than risk his life on the bus. The charge doesn’t make the driver “guilty” in your head; but it certainly will affect whether you think it makes sense to let him drive your kid.

That’s the same (Bayesian) principle that guides conflict-of-interest analysis.23 The legal system doesn’t assume that a judge is partial merely because her “impartiality might reasonably be