Update your priors
According to Bayesians, rational people should update their beliefs about the world as new information becomes available. But do people actually do this? In a recent post I discussed how the Ukraine conflict provides new information about the costs and benefits of engaging in trade with our adversaries.
The New York Times has another good example:
Anne Fuqua keeps a list of suicide deaths. She’s chronicled hundreds of cases of chronic pain sufferers who have killed themselves after losing access to opioid medication since 2014. . . .
Between the mid-1990s and the early 2010s, the number of opioid prescriptions written for Americans roughly doubled, driven by dishonest pharmaceutical marketing campaigns and unscrupulous entrepreneurs who opened so-called pill mills to sell drugs. Medical guidelines, legislation, law enforcement and other measures have since returned painkiller prescribing to pre-crisis levels. But because people who lose access to medical opioids are rarely provided with immediate treatment (whether they are experiencing pain or addiction or both), the result has been more overdose and suicide deaths, not fewer.
Despite these dismal facts, American medicine and law enforcement continue to fight the last war. Policymakers still operate under the assumption that too many opioids are being prescribed. Overdose deaths — including those among adolescents — are now overwhelmingly caused by street fentanyl, not prescription medications. And fatalities have nearly doubled since 2012, in concert with the decline of the medical supply.
So why don’t policymakers update their priors? The federal government tried a new policy, hoping it would make things better. Instead things got much worse. Why not go back to the old policy?
Or take pot legalization. Before some states began legalizing marijuana, drug warriors warned about specific consequences from this policy change. For the most part, those consequences did not pan out. (A study by provides another example.) And yet I see very few pundits change their mind on policy questions, as new information comes in. What’s going on?
One possibility is that the actual reasons for policy views are not the same as the stated reasons. In the 1990s, anti-immigrant people warned that immigrant communities would stick with their home language and fail to learn English. We now know that the children of these immigrants do learn English, and that the language problem is temporary, just as with European immigrants in the 1800s. But the anti-immigration people typically did not change their minds with confronted with this new information, rather they looked for other reasons to oppose immigration—say crime. And if it turns out that immigrants have a lower crime rate than the native born, another reason will be found.
I suspect that in many cases, there are unstated reasons for policy views, which remain in the background. Perhaps opponents of pot legalization worry that it would “send the wrong message”. They might realize that this reason is unpersuasive to many people, and thus seek more “consequentialist” reasons, say increased pot addiction among teenagers. When pot use among teenagers doesn’t rise, they don’t change their views, as the anti-pot view was actually based on a different factor.
Perhaps some anti-immigration people are simply uncomfortable with lots of new people who look and act differently, but are embarrassed by their concern. The stated concern about language ghettos then becomes a sort of respectable reason to oppose high levels of immigration.
PS. This graph shows how the crackdown on legal painkillers led to a much greater loss of life from illegal opioids such as fentanyl, where the dose is hard to ascertain and thus overdoses are common: