Further thoughts from Science Online 2013
Two of the sessions I attended dealt with probability and statistics in some fashion, and how to deal with them. The really big-picture take-away from the discussions was that people don’t understand statistics and related subjects, so one needs to be extra careful in conveying this information. It’s necessary to provide context when citing any numbers in order to minimize the potential to misunderstand them.
One need look no further than the hubbub raised by Nate Silver’s predictions and how they were misinterpreted— the “80% probability that he will get 51% of the vote” and similar predictions. Once you get past “4 out of 5 dentists recommend Crusty toothpaste” heads start swimming. One can go to the next step and look at situations where you have low probability but devastating consequences — examples were things like an asteroid impact or the possibility that the Large Hadron Collider would destroy the earth. Some people get very upset over topics like these when they adopt the “I could be the one” thinking that prompts people to buy lottery tickets every week.
This leads into issues of not properly understanding risk, something humans are really bad at assessing. Fear of issues that are not under our control can evoke strong responses, even when the odds are slim. We see reports of every plane crash and they tend to affect many people, so one might get the impression that plane travel is unsafe, even though it’s safer than driving.
Related to this is the fact that science is inherently tied in with uncertainty, and that can be innocently misunderstood or cynically exploited. The LHC example, in which one of the opponents said that since it either would blow up the world or wouldn’t, the odds were 50/50. That’s such a laughably horrible misunderstanding of probability that it was really effective when The Daily Show responded by mocking it. Unfortunately, not all misuse of probability is so egregious that people can see how ridiculous it is.
Beyond this, there are those who are quite happy to undermine the credibility of science by exploiting statistics and the uncertainty. Science is always subject to change when more and better data are obtained, but even though this is a strength of the system, there are those who use it to imply that we can’t trust it — peddling the idea that the only choice is between knowing everything and knowing nothing.
It’s a difficult problem. Unfortunately, with such a wide spectrum of issues, we’re short on answers on the best way to deal with them.
Edit: I like this commentary regarding the recent Jared Diamond blurb I had linked to, and rare+spectacular risk vs everyday risks, and how we react. (Though I think shark attack might work better than tiger attack as an example.)
I’d go one further and say that many scientists (including physicists) don’t have a good grasp of statistics — because beyond adding the occasional error bar to a result, many don’t really do in-depth statistical analyses of their work.
Perhaps things have changed, but when I was a physics undergraduate in the 1980s we were not required to take a formal course in statistics (ie taught by a statistician) and the assumption seemed to be that we would “pick it up” if we needed it.
I agree. There are numerous example of scientists getting error bars, and other uses of stats, wrong. It’s not an example of us doing it right and everyone else not understanding. But the lay person is probably not going to understand better than scientists.