Skip navigation
Link to Back issues listing | Back Issue Listing with content Index | Subject Index

On Bertrand Russell

If Bertrand Russell was good for anything, it was for supplying brain dead writers with quotations an inspiration to keep on writing. For those who doubt this, try a little judicious Googling, and enjoy.

Attributed to Russell is the following:
It is undesirable to believe a proposition when there is no ground whatsoever for supposing it to be true.
So here's just one unbelievable thing to believe before breakfast - that astrological signs are associated with illness. Of course they are not, yet if you get a very large mount of data about a large number of people, you can come up with loads of apparently statistically significant associations.

While this might give you pause for a nanosecond, you know it cannot be true. But it is fertile stuff for examining how to avoid the spurious. A huge Canadian study has done the sums, and found 72 statistical associations, but it had to do over 14,000 calculations to find them! The lesson is that we can be caught out by failing to correct for multiple testing, and that failing to do such corrections is common.

Bandolier this month has included some surveys - on what patients want to know about adverse events, on understanding of food labels, and on compulsive buying. Surveys are often given a bad name, dismissed as just being a survey. That may be so if one tries to get too much fine detail.

But fine detail is not what surveys should necessarily be about. They are about headlines. They should knock us in the eye and tell us something we didn't know or hadn't thought about. Our customers (if we can be so bold) tell us that they many of they can't understand labels on food, and that almost all of them (at least in Kansas) want much more information on adverse events than we know, or know how to tell them.

next story