Disclosure isn't all it's cracked up to be; psychological factors can easily undermine its effectiveness
August 1, 2004
For more information, contact: Benjamin Haimowitz, HHaimowitz@aol.com
A favorite answer to a much-fretted-about problem in business and politics may not be the solution it is widely thought to be, if a new study is to be believed.
In fact, it may make matters worse.
The problem is conflict of interest, increasingly a matter of concern in the wake of business scandals of recent years. And a favorite answer, almost invariably incorporated into responses to the problem, is disclosure.
Thus, medical journals now ask researchers to disclose the sources of their research funding. TV financial-news shows require stock analysts to reveal any conflict of interest. McCain-Feingold mandates public disclosure of political contributions. And an entire section of the Sarbanes-Oxley Act of 2002 is dedicated to enhanced disclosure by corporations and their auditors.
Now a new study, presented on August 10th at the annual meeting of the Academy of Management in New Orleans, uncovers "psychological factors that may undermine disclosure's effectiveness as a remedy...and may even cause disclosure to backfire, harming rather than helping the recipients of advice."The authors of the study, Daylian M. Cain, George Loewenstein, and Don A. Moore of Carnegie Mellon University, undertook the research to test their suspicion that disclosure was more problematic than widely appreciated.
For one thing, they surmised, disclosure might actually tempt advisors to provide more biased advice than they might otherwise offer, a strategy the authors compare to "expecting one's audience to cover its ears and thus compensating for this by yelling even louder." Moreover, the advisors would feel justified in doing so, because disclosure would reduce any feelings of guilt they might otherwise have.
Further pitfalls might arise, they suspected, from the inherent difficulty in "estimating the impact of conflict of interest on an advice-giver." As the authors explain, "While most people think conflicts of interest are a problem of overt corruption ...considerable research suggests that bias is more frequently the result of motivational processes that are unintentional and unconscious. Failing to appreciate the role of unconscious bias will cause estimators to underestimate the extent to which advice might be distorted."
In addition, research has revealed that "even when estimators realize they should make some adjustment for the conflict of interest that is disclosed, such adjustments are likely to be insufficient. As a rule, people have trouble unlearning, ignoring, or suppressing the use of knowledge, even if they are aware it is inaccurate."
With these considerations in mind, Mr. Cain and his colleagues put disclosure to the test via an elaborate experiment they carried out with 146 students at Carnegie Mellon University.
Participants stationed in individual cubicles were divided randomly into advisors and estimators and were asked to perform separate tasks relating to jars filled with coins -- six jars containing amounts ranging from $10.01 to $27.06. Advisors were permitted to handle, but not open, each jar and to study it for as long as they liked. For each jar, they then wrote down a number on a piece of paper, which was then delivered as a suggestion to an estimator. The estimators then got about a 10-second look at the jar in question as an experimenter carried it by at a distance of about three feet, and then provided an estimate of how much money the jar contained.
Thus, advisors had a considerably better opportunity than estimators to calculate the value of the coins in each jar.
Advisors and estimators functioned in three quite different conditions -- the accurate condition, the high-undisclosed condition, and the high-disclosed condition.
--In the accurate condition, advisors were instructed to write down their best estimate of the amount of money in each jar, and estimators were aware of this instruction. Both advisors and estimators in this condition were paid based on the accuracy of estimators' assessments.
-- In the high-undisclosed condition, advisors still wrote down a number that was passed on to an estimator. But now the advisors were paid based on how high estimators' evaluations were, while estimators were still paid based on their accuracy. Advisors were fully aware of this conflict of interest, but it was not disclosed to estimators.
-- In the high-disclosed condition, advisors and estimators were paid on the same basis as in the second condition, but now both estimators and advisers were fully aware of this.
Counter to what common sense might suggest, disclosure encouraged distortion of advice. The amount that advisors exaggerated was significantly greater in the high-disclosed condition than in either of the other two conditions. In the accurate condition, advisors gave estimators suggestions that were, on average, within one dollar of their own personal estimates. In the high-undisclosed condition, their suggestions were more than $3 greater than their own personal estimates, and in the high-disclosed condition they were more than $7 higher.
Also counter to what common sense might suggest, estimators did worst in the high-disclosed condition: the average error of the estimates in that condition was significantly greater than the average error in either of the two other conditions. The authors attribute this disheartening result to two factors -- the greater bias in the advice the high-disclosed estimators received and the greater dispersion in their estimates.
Errors, the study also finds, translated into less pay. Estimators earned less money when conflicts of interest were disclosed than when they were not, and advisors made more money with disclosure than without disclosure. Estimators made the most money in the accurate condition, in which there was no conflict of interest.
"To the extent that a similar effect occurs outside of the experimental laboratory, disclosure would supplement existing benefits already skewed toward information providers," Cain and his colleagues write.
The authors acknowledge that "disclosure might be more effective when recipients of advice have extensive professional experience" -- as, for example, would be the case with government regulators or judges. Still, "the general conclusion that disclosure is most likely to help the sophisticated estimator is somewhat dismaying, since unsophisticated estimators are exactly the ones who are most likely to need protection from exploitation...Many of the most significant decisions people make in their lifetimes are made only once or but a small number of times."
The Academy of Management, founded in 1936, is the largest organization in the world devoted to management research and teaching. It has about 15,000 members in more than 90 countries, including some 10,000 in the United States. The academy's 2004 annual meeting drew more than 7,000 scholars and practitioners to New Orleans for more than 1,000 sessions on a host of issues relating to corporate organization, the workplace, technology development, and other management-related subjects.
- Media Coverage:
- The Washington Post. Disclosing the dirt about full disclosure. (Sunday, August 22, 2004).