h2g2 The Hitchhiker's Guide to the Galaxy: Earth Edition

I wonder if this is more of a "galactica" entry than a guide entry, but it is a complex topic.

I find the main problem with assessing risk in everyday life is lack of information. The media are very good at giving incomplete statistics, either because they don't understand them themselves or because they think that their audience is too dumb. We are frequently subject to some probability assigned to a terrible event (death from vCJD, flooding in a river valley etc.) without the benefit of a confidence interval. I recall the recent figures on vCJD that were the result of a study of samples taken from people who had undergone surgery, these were reported in the media with very scant regard for the fact that with one positive result in a huge number of negative ones, the confidence in the projected number of vCJD suffers was minute.

Weather forecasters are similarly blessed with statistical inadequacy. Although they now see fit to provide a likelyhood of rain figure, they still rarely give any hints about the confidence in the forecast despite the predicatability of the weather being very variable.

I agree, a complex topic. "galactica" entry, eh? Well ... I'd made a few attempts at humor, but in the end the topic seemed to be degraded by my feeble attempts at being clever. It was mutually decided by the subeditor (Bels) and myself to cut most of that out. I don't know if that's what you meant or not, but ... at any rate, it turned out I think!

I must confess I do not know what "vCJD" stands for. At any rate, you're exactlty right in that the public has very little understanding of concepts like "standard deviation" and "confidence interval," concepts that are critical to judging the validity of the statistics presented. I had a professor that liked to say something along the lines of, "statistics is the modern religion - it has become a shadow of its true nature, used and abused for political ends." He might not be far off on that, although I'm sure his choice of metaphor might cause some people to bristle.

I don't know for certain, but I seem to remember being told that weather forecasters are required to meet certain confidence levels in order to report a statistical likelihood for rain, tornados, hurricanes, etc. Maybe the confidence level threshold isn't very good (say, 50% confidence level), but I think there is some rule governing what they can say as a statistic. In the US, I'd check with the NOAA's website - maybe there's something there. (National Oceanic and Atmospheric Administration, that is.)


- JD

I just meant that the H2 article would probably just say that:
"In theory, the risks taken in industrial society are minimised, well quantified and understood by all concerned, but in practise if you pay too little for something (or the supplier demands too much), then someone somewhere is going to take an unquantified and unminimised risk and there is a danger that that person will be you." and it would then go on to list the phone numbers of some dodgy plumbers.