Identifying bias. A Bayesian analysis of suspicious agreement between beliefs and values.

Stefan_Schubert

Here is a new paper of mine (12 pages) on suspicious agreement between belief and values. The idea is that if your empirical beliefs systematically support your values, then that is evidence that you arrived at those beliefs through a biased belief-forming process. This is especially so if those beliefs concern propositions which aren’t probabilistically correlated with each other, I argue.

I have previously written several LW posts on these kinds of arguments (here and here; see also mine and ClearerThinking’s political bias test) but here the analysis is more thorough. See also Thrasymachus' recent post on the same theme.

How many beliefs concern propositions which aren't probabilistically correlated with each other, though?

Also, this seems to ignore the possibility of deriving one's values from one's beliefs.

Also, this seems to ignore the possibility that people cannot research everything themselves, and therefore have to trust others to get some of their beliefs, and that if they have reason to distrust someone who happens to be using a true belief, they can justifiably distrust true beliefs. Climate change is extremely convenient for the left to promote leftist policies, to the point where anyone on the right who doesn't know a lot of science would justifiably think "the people who believe in climate change are probably engaged in motivated reasoning".

this seems to ignore the possibility of deriving one's values from one's beliefs

The possibility (or something closely related) is raised in the section headed "Causes of agreement between political views and factual beliefs" -- it's option (b) in the list near the start of that section. But that option gets dismissed rather rapidly. (Too rapidly, I think.)

0gjm10y

In so far as this is true, it is also true that climate change denial[1] is extremely convenient for the right. (And vice versa.) [1] I intend this here to mean simply "denying" rather than "denying in the face of what ought to be overwhelming evidence"; there doesn't seem to be a neutral way of putting it. This is a general phenomenon: if there are rival positions X and Y on a factual matter, which if true would support rival positions P and Q on a matter of policy, then you may suspect partisans of P of bias when they assert X, but you may equally suspect partisans of Q of bias when they assert Y. So if you are not yourself very partisan, what difference should this make to your opinions about X versus Y? It should make you treat someone's opinion about X/Y as less informative in so far as they have a partisan position on P/Q that would explain it. (But bear in mind that the causation may go X/Y -> P/Q rather than the other way around, so the appropriate discounting is less than you might naively think.) So, in this case, you might reasonably be very suspicious about American politicians' statements about climate change, because in the US the issue is very politicized. So, where else should you look? * Outside the US. It's really only in the US that such a tight relationship between politics and opinion on climate change is present. Outside the US, even right-wing politicians (whom you might, in terms of the model here, expect to have a strong bias against belief in anthropogenic climate change) generally agree that the climate is warming because of human industrial activity, and they generally favour taking measures to reduce the warming. (But "conservative parties do not challenge coal or petroleum in countries with large reserves of these resources".) * The opinions of actual climate scientists. Being a scientist is of course no guarantee of being unbiased, and there are ways for scientific institutions to be biased even when individual scientists are not

14

Identifying bias. A Bayesian analysis of suspicious agreement between beliefs and values.

14

14

14

Identifying bias. A Bayesian analysis of suspicious agreement between beliefs and values.

14

14