On Thu, 24 Feb 2005, Bill Bartlett wrote:
> At 11:53 AM -0500 23/2/05, Yoshie Furuhashi wrote:
>
>> That means that there is a one-in-twenty chance that the alleged
>> association between "consulting a lawyer" and "less improvement in NPOS"
>> was wholly accidental.
>
> So, a nineteen in 20 chance that it was deliberate? I prefer short odds
> myself. To be honest though, I'm not a gambling man, don't see the point.
> Don't have much faith in statistics either, frankly I distrust the motives of
> most statistical analysis. I always want to know what the agenda of the
> people behind them are. "Lies, damn lies and statistics," as the saying goes.
As somebody who regularly teaches stats, I think the biggest problem is not willful manipulation of statistics; it's misunderstanding stats. If you understand the principles of inferential stats--and you can, it's not that tricky!--people will have a hard time using stats to lie to you.
(In passing, Bill's way off base on his interpretation of the statistics Yoshie reported. Many probably already see the flaw, so I won't waste bandwidth.)
Miles