Bias du Jour #7
In one study, subjects read a lengthy description of a week in the life of “Jane,” a fictitious character. The description portrayed her as being extroverted in some situations and introverted in others. Later, subjects were asked to test one of two hypotheses about Jane. Half the subjects were asked to determine whether Jane would be a good real estate agent (i.e., extraverted); the other half were asked to determine whether Jane would be a good librarian (i.e., introverted).
The subjects listed the facts about Jane they considered relevant to their determination. Those determining whether she would be a good real estate agent recalled more facts about her extraversion; those determining whether she would be a good librarian recalled more facts about her introversion. Subjects seemed to form a hypothesis about Jane as a result of the question and then seek evidence for the hypothesis, ignoring the opposing evidence. This is an example of
The Confirmation Bias: We tend to actively seek out and over-weight confirming evidence while we tend to ignore or under-weight disconfirming evidence.
Here’s another example. I tend to think marmalade-colored cats are friendly. Every marmalade cat I’ve met has been that way. So that’s my hypothesis: “Marmalade-colored cats are friendly.” But why do I think this?
I went around the neighborhood looking for friendly cats—the ones that come out to meet me when I walked by. And many of them were marmalade-colored. “See! Here’s another one! I’m right!,” I concluded. After all, every marmalade-colored cat I saw was friendly. Right?
What’s wrong with this picture?
I was unwittingly getting only confirming evidence. Better would have been to look for unfriendly cats, and then look for one of those that was marmalade-colored. This latter cat—if he exists—is inconsistent with my hypothesis: it tends to disconfirm it. Then I would really know if my hypothesis was right.
Cites: Nickerson, R.S. (1998). Confirmation bias: A ubiquitous phenomenon in many guises. Review of General Psychology, 2, 175-220; Snyder, M., and Cantor, N. (1979). Testing hypotheses about other people: The use of historical knowledge. Journal of Experimental Social Psychology, 15, 330-342.