Monday, 24 October 2005

Only The Stupid Get Caught

Alt.folklore.urban currently has an amusing/scary thread on the perils of certain sets of self-selecting data.
During the infancy of what is know called 'Operations Research' a study was performed in during to examine fighter aircraft damaged by enemy ammunition, in order to determine which parts of the fuselage should receive additional or improved shielding.
A then-Junior Researcher pointed out that they were only able to examine the aircraft that had been able to RETURN from battle.

Then there's this:
This reminds me of the tale of the researcher who subjected some prison inmates to an intelligence test and concluded that crime was caused by low intelligence. It was eventually pointed out that he had documented that criminals of low intelligence spend time in prison.

It turns out that this might be wrong, and that the more accurate way of presenting the above data is this:
The right conclusion is that people of low intelligence spend time in prison. This could occur just because they're not able to effectively defend themselves when the charges are false.

The more you think about this the more you realise it applies to anything where failure is catastrophic, but success is hardly noticed.

No comments: