David Leonhardt's "Economic Scene" column in yesterday's New York Times is about the risks of trusting experts. Leonhardt discusses the research of Cornell University economist Henry Schneider, in which Schneider conducted an experiment to see how accurately auto repair shops diagnosed problems with customers' cars. Only 20% of repair shops warranted a passing grade, according to Professor Schneider.

This calls to mind data points from the Pew Internet Project's March 2005 survey which asked internet users about the online resources used in making major decisions. Our survey involved several questions. First, we asked online users whether the internet played a crucial or important role across five specific topics: buying a car, making a major investment, getting additional career training, choosing a school for self or child, or helping someone with a major illness or health condition. Some 39% of internet users said the internet played a role in one of these five decision points in the prior two years.

Next, we asked this group of online users what specific role the internet played in helping them with the decision. One third (34%) said they found advice and support from other people, 30% said they did basic information searches or compared options, and 28% found professional or expert services to aid them in the decision. Experts clearly played a role for online users, but so did many other kinds of resources, notably just using the internet to ask other people for advice

What do our findings mean in light of Professor Schneider's research? If only 20% of auto shops earn passing grades, a lot of people may understand that, for cars and other matters, another opinion is well worth seeking. The internet gives people access many voices for advice, and the chance to participate in conversations with a lot of people. They might value expert advice (although we don't know, from our survey, whether people consulted an expert first and then turned to the internet), but want the perspectives of others. This may be an online manifestation of using "word of mouth" in decision-making process.

Our research suggests that some internet users have a "trust but verify" posture toward experts. That said, none of this means that there should be widespread adoption of the mantra "Don't trust an expert, use the internet instead." We don't know, from our surveys, whether those who used non-expert online advice got accurate information. Perhaps that's a new research path for experimental economists like Professor Schneider.