Threat levels: doing the numbers
Being interested in world event, and also a student of statistics, I'm always intrigued by the language of politicians when it comes to threat or risk.
For instance, this week we reached the dizzying heights of a 'CRITICAL' terrorist threat level, which is distinguishable from 'SEVERE' simply by the imminent nature of the threat. However, when we were downgraded to 'SEVERE' this morning, the questions started to surface. Recently politicians haven't said much of the terror threat level, but in the last few days I'm constantly hearing about how a threat is 'highly likely' now we are at the SEVERE level.
What does that mean? It seems it can be misleading in two distinct ways:
- How likely is highly likely? It suggest a very high probability that an attack will happen, but what is that? 0.5? 0.7? 0.99? More than likely it is an extremely low probability figure if we talk of an attack in the near future, maybe less than 0.01. Certainly, the impact of such an event could be catastrophic, which may justify using such language in a relative sense.
- Once we retreat from the immediacy of an event (ie the CRITICAL level), we have no reference to the timespan of these threats. We are expected to assume the level is describing the 'climate' in which we live. But does this mean an attack is actually highly likely? The language used does not make sense if we consider the intended message..
Perhaps I shouldn't think so much about it. Maybe they should publish estimated probability figures with 95% confidence intervals.. That would satisfy pedants like me.