See: <http://virtuelvis.com/archives/2005/05/statistics-nonsense> ...for more on "statistics".
The guy is using maybe interesting arguments, but it is far from general true, as he is referring exclusively to log-analysis with wrong settings and/or obsolete stats packages.
If you have a methodology that works, is almost accurate and you know there are no quirks in how any agent could misbehave, then you're fine. If you're not interested in absolute numbers (that may wary due to particular implementation of methodology), but only in trends and ratios, you should be fine.
-- Jan Brasna :: www.alphanumeric.cz | www.janbrasna.com | www.wdnews.net ****************************************************** The discussion list for http://webstandardsgroup.org/ See http://webstandardsgroup.org/mail/guidelines.cfm for some hints on posting to the list & getting help ******************************************************
