RE: FN-FORUM: security notification

date posted 6th January 2004 16:49


Speaking as a former statistician (it's boring so I started doing
programming instead) this isn't actually true.

Whenever you're doing this kind of work you would generally exclude the
statistical outliers and declare the stats at say 95% confidence, i.e.
exlcude the top and bottom 2.5% who can skew the results.

Just my 2p worth.

Dominic


Actually, from a statistical point of view, all it would take is a small
number of really incredibly smart people to mean that MORE than 50% of
people would be dumber that average.