Misquantifying Terrorism Using Statistics

Reader David, who has the blog Photon Courier wrote and linked to this post:

Calculations of probability and cost-effectiveness must be based on the assumption of a trend or lack of a trend. If probabilities of a particular phenomenon are changing over time, you can't just snapshot the phenomenon at one point in time and extrapolate those probabilities into the future. Suppose that for a particular model of car, there are 2 million vehicles on the road. In a particular year, 2 of these cars encounter complete steering failure. How much of an issue is this? A simplistic statistical analysis might conclude that there is only one chance in a million of encountering the failure in any given year--representing a fairly small increase in the risk of driving--and hence, from a cost-effectiveness standpoint, money would be better spent on other aspects of automotive safety.

Read the whole thing.

As an aside, David, I think I read this post somewhere, maybe at your site, maybe somewhere else. It was good then. Thanks for the heads up so I could link to it.