Tuesday, August 26, 2008

The power of "anecdotal evidence"

We have a bike rack on our 12-year-old Isuzu Trooper that fits into the trailer hitch. I mentioned to my wife the other day that perhaps next year we should add a hitch to our 6-year-old Acura MDX, so we can use the bike rack on the MDX after the Trooper gives out.

She laughed. "What if the Acura dies first?" she said.

My wife holds the perception very firmly that the Isuzu is a highly-reliable, trouble-free car, and that the Acura is a fragile thing, constantly in need of expensive maintenance.

Statistics say otherwise. JD Power gives Acura four stars for reliability (out of five), and Izusu only two stars--tied for the worst rating.

But looking beyond statistics, at the stories, the Trooper has a bunch of ardent fans. Some people have had terrible problems and hate the car, yet many others, a larger number, love it. (See this group of epinions posts for an example.)

And the MDX's reliability has some detractors as well (see reviews from Edmunds.com), with many complaints about transmission problems (writer crosses fingers).

What's most amazing is the firmness with which the reviewers--positive and negative--hold their opinions. It demonstrates the "tyranny of the mean," in which a lot is lost by averaging ratings together. A car is an emotion-laden product--expensive, used daily, inconvenience-causing when broken. Opinions then vary dramatically based on personal experience. The stories are arguably more important than the statistics when evaluating this type of product. And product managers, as well, should be especially attuned to the stories customers are telling about their products.

The lesson here is that "anecdotal evidence" should not so easily be dismissed, and that statistics can be useful but, just as when buying a car, you should look under the hood for yourself.

It might help you figure out why your wife thinks the old Izusu kicks the newer Acura's butt.

, , , ,