Monday, January 30, 2012

Confidence and trust in recommendations

Life is short, so we like to avoid as many bad experiences as we can.  I have seen quite a few movies that are basically two hours of my life that I cannot get back.  I have been to my share of bad restaurants.  So, of course, I’d prefer to avoid more of those experiences in the future.

In some situations, it is easy to find experts that you trust.  For example, for movies, I am a devoted reader of Roger Ebert’s reviews.  When he likes a movie, I generally like it as well, and when he doesn’t, I know that it is safe for me to skip it.  On the other hand, my agreement with Peter Travers of Rolling Stone is lower.  So, I don’t find him to be a reliable guide for movies I should see. 

But, what do people do in cases where they don’t have an expert that they already know they should trust?

A paper by Uma Karmarkar and Zakary Tormala in the April, 2010 issue of the Journal of Consumer Psychology takes up this question.  They were interested in how people decide whether they trust a recommendation based on the combination of the expertise of the judge and the confidence of the judge in his or her recommendation.

They had people read a restaurant review for a new restaurant.  The man making the recommendation was either described as a real expert (a nationally renowned critic published in major newspapers) or as a non-expert (a local administrator who has his own food blog).  The text of the review also mentioned something about the reviewer’s expertise.

In the review itself, the reviewer expressed either a high degree of confidence in his review or a low degree of confidence.  For example, the reviewer might say he had eaten at the restaurant several times and felt very certain about his recommendation, or he might say that he had only eaten there once and felt tentative about making a recommendation.

The researchers found that people were more likely to trust a reviewer when the combination of the level of expertise and the degree of confidence was surprising.  That is, they tended to trust the non-expert reviewer more when he was very confident than when he was not so confident.  In contrast, they tended to trust the expert reviewer more when he was tentative than when he was very confident.

An interesting aspect of this increased trust is that when people were surprised about the combination of expertise and confidence, they read the review more carefully and paid more attention to good arguments than to poor arguments in the recommendation.  

This work suggests that our reaction to recommendations may be strongly swayed by factors that might not be optimal for making good decisions.  That is, experts are often better able than non-experts to isolate factors that affect whether people will find an experience enjoyable.  A confident expert ought to be a good source of information.  But, because we expect experts to be confident, we seem to distrust these recommendations to some degree.  Non-experts may be less able than experts to really determine what makes an experience enjoyable, though it probably does make sense to pay more attention to a very confident non-expert than to one who is tentative.

In the end, it is worth trying to find out more about the people giving us recommendations.  After all, there is a cost to each bad movie and each uninspiring meal.