The psychology of bad advice

May 16 2012 by Brian Amble Print This Article

It's one thing advising an individual about an important investment choice. But when you're advising a large number of people Ė if you're an investment analyst, for example - even greater care should be taken to ensure your advice is accurate.

Sadly, however, this is often not the case. In fact according to new research by Sunita Sah from Duke University's Fuqua School of Business and George Loewenstein from Carnegie Mellon, advisors confronting a financial conflict of interest actually give more biased advice to multiple, anonymous recipients than they do to single, identifiable recipients.

The reason for this seems to be driven by increased intensity of feelings toward individuals; advisors have greater sympathy for and more motivation to reduce bias in their recommendations to identifiable individuals.

As the researchers found in their experiment, advisors turned out to be much less biased if they were giving advice to just one person that they knew something about (such as their name and age) than if it was for a group of five advisees or a person they knew nothing about.

According to Sunita Sah , the research can shed light on the behavior of stock analysts who gave recommendations they themselves didn't believe during the dot-com boom, that of auditors during the Enron debacle, and of bond raters during the housing market bubble.

In all of these cases, he said, these advisors were giving biased advice to large numbers of investors who were anonymous to them, so the damage they were causing had little reality for them.

"When advice affects the welfare of a greater number of people, greater attention and care should be taken to ensure its accuracy. However, many advisors face conflicts of interest and next time you need advice on a particular stock, don't read the public recommendations but speak to your advisor one-on-one."

The full report to this research is published the May issue of Social Psychological and Personality Science.