|
|
|
|
|
Jens 'n' Frens
Idle thoughts of a relatively libertarian Republican in Cambridge, MA, and whomever he invites. Mostly political.
"A strong conviction that something must be done is the parent of many bad measures." -- Daniel Webster
|
|
|
|
|
Saturday, April 12, 2014 :::
There has been some recent discussion in the blogosphere of confirmation bias. The best I've seen is by Yuval Levin, at that link, and he links to several others (including Paul Krugman, whose photograph appears next to the dictionary entry for "confirmation bias," claiming to be immune).
What I haven't seen pointed out, perhaps because it's not especially important, is that confirmation bias can be rational - if you get new information which may or may not be reliable, less skepticism is warranted if it conforms to your priors than if it challenges them. If I believe that (probably) most cows are blue and you believe that (probably) most cows are green and we both learn that a random sample of cows is overwhelmingly green, it would be rational for me to think it more likely that the sample was badly formed than if the sample were overwhelmingly blue. After all, if most cows actually are blue - as I believe - most good samples will be dominated by blue cows.
To be sure, I don't believe all, or even most, of people's tendency to prefer information that confirms their preconceptions is simply Bayesian updating of different priors. But I think it's worth noting that a small piece of it could be.
::: posted by Steven at 7:31 PM
|
|
|
|