Re: roadmap status update/update request

[Thomas Vander Stichele <thomas apestaart org>]

Hi,

> I think that's a way better number of people 
> participating than MOST commercial companies would ever dream of to make a 
> market research on. They usually do their researches based on a sample of 
> 2-3,000 people. Gnome would have 100 times that, and so I do take it that it 
> would be accurate "enough".

I'm sorry, but I have to step in here and defend the basic principles of
statistics being violated here.  Sample size is not the only thing that
makes the outcome of a statistical "polling" accurate.  Your sample also
needs to be representative of whatever group you're sampling from.

Ie, if you ask 1 million rabbits if a Macintosh should have three mouse
buttons, you'd probably get a very reliable answer.  But it would be an
answer representative of what a rabbit is likely to think on the
subject.  It doesn't say *anything* about what a human would think.

As such, these companies doing their research on 2000-3000 people
probably have a much better selection process and take care to make sure
the sample chosen is representative of their target audience.  Getting
100 times that amount of people in your poll is likely to completely
mess up your poll instead of making it more statistically sound.

Defend mathematics ! Don't make statistics say something that you try to
make it say - do the research properly.


I now go back to taking my daily cough medicine.

Thomas


Dave/Dina : future TV today ! - http://www.davedina.org/
<-*- thomas (dot) apestaart (dot) org -*->
- It's late - can we finish this some other time ?
- Oh - so you want to jump right to the kissing then. 
<-*- thomas (at) apestaart (dot) org -*->
URGent, best radio on the net - 24/7 ! - http://urgent.fm/