Friday, January 14, 2005

Confirmation bias

I keep talking about confirmation bias. That is the psychological condition whereby people's beliefs about an argument have little to do with the argument, but rather whether they like the results or not. Here is someone who apparently decided to retest these, but using political questions. The results are pretty interesting. The researchers found that they could accurately predict 80% of the time what a response to various questions would be based soley on the subject's political views.

The researchers asked a question about soldiers and torture. In one version of the experiment subjects were presented with strong evidence that the soldier was in the right, and in the other version they were not given the evidence. It turns out that it makes no difference. The subjects' views were split along party lines. Evidence is irrelevant.

During the Clinton-Lewinsky thing they found that regardless of how well or poorly informed one was about the Lewinsky affair, their condemnation or not, was predicted by their party affilation.

The list goes on.

What this shows is that knowledge makes little difference to people who offer their opinions. Most people make decisions based on something that bares little relation to logic or facts or common sense. They decide on their guts and their feelings. Guts and feelings are useless when we are trying to get at good decisions about most things. (Instincts, are by the way, reliable, but only for certain things.)

So next time you hear an opinion being supported by "facts and logic" look carefully. There is usually nothing but some convoluted justification for a gut feeling.

No comments: