| Subject: |
|
Re: Re: Re: yet more bad analysis with good math |
| Name: |
|
Ron Langeveld |
| Date Posted: |
|
Jun 3, 07 - 3:36 PM |
| Message: |
|
Hi Wim,
What Ken may be refering to is a common problem with statistical analysis : you cannot prove that all relevant factors have been inluded into the data to establish a relation between two variables beyond a doubt.
For example : a clear correlation can be found between the consumption of alcohol and the likelihood to break a leg during skying. The conclusion with these two variables considered would be : drinking a couple of beers while skying will help you losen up and not break a leg when you fall. The relation is plausible to some extend and the correlation really exists. If you take into account a third variable, in this case the level of skying experience, you would have found out that the people most likely to break a leg are the inexperienced skiers and that this group is less inclined to drink and ski :-) If you want to predict the likelihood of breaking a leg while skying then the determining variable would have to be the level of skying experience, maybe in combination with one or more less determining factors.
It would have been nice if Ken had mentioned one or more other variables that can better explain the length of a tournament than the ones you did. I think you did a good job in using the available data to predict the tournament length. That is what counts. We do not have to understand all relations and reasons to make a balanced decision. Your "theory" does make sense, and luckily it can still be "falsified" according to Popper ;-)
Ron |
|
Replies:
|
|
|
|
|
Pending TCCMB Exhibition Matches!
| 
TCCMB FAQ/Terms of Use |
Visit CC.COM's Online Store!
TCCMB
Video (Windows Media Player)
Reply
Home
bravenet.com
