Confirmation Bias
From SkepticWiki
Contents |
[edit] Definition
Confirmation Bias, also known as "counting the hits and ignoring the misses," refers to the tendency of people to look for observations that confirm a particular belief, while ignoring, downplaying, or dismissing observations that would serve to contradict it. This is a form of Selection Bias and an invalid form of Inductive Reasoning.
[edit] Examples
Examples of confirmation bias come in many forms and permeate over many different subject areas:
- "I work in a hospital's delivery ward, and I can tell you that it's true that more babies are born during a full moon. I can show you many times where we've had higher admissions than usual, and each of those times there's been a full moon."
This person doesn't tend to notice a rise in admissions other times in the month, or writes them off to just a normal variation in admissions levels. Similarly, low admissions during a full moon will go unnoticed or are considered an exception. Actually looking at the hospital's admissions records over time shows no such tendency of the full moon to increase delivery admissions.
- "I went to see a psychic, and he told me my father's name was Steve and he liked to garden. He's really a psychic!"
This person doesn't really remember the other guesses the psychic made about his father that were wrong. In an analysis of a reading by James Van Praagh, James Randi calculated that, out of 69 names he supplied to his audience, nine of them were hits, and all of them were common names such as Steve or Robert. This 13% hit rate is similar to what has been observed of other mediums such as John Edward and Sylvia Browne. But the person being read is left with an impression of a much greater hit rate.
[edit] References
- "The effect of the lunar cycle on frequency of births and birth complications," The American Journal of Obstetrics and Gynecology, Volume 192, Issue 5, Pages 1462-1464 (May 2005)
- James Randi's commentary including James Van Praagh's hit/miss rate (June 8, 2001)
