| FinSaint |
10-13-2011 05:19 PM |
Quote:
Originally Posted by saintfan
(Post 339329)
If I break my arm, and you break your arm, statistics can be used to tell us when our arms might heal.
Yours may heal faster. Mine slower, other the other way around. The mountain of data you would need to make the claim that my arm will heal x days faster than yours boggles the mind.
Stats can give you part of the story. You can organize what data you have and you can come to some conclusions. Those conclusions are dependent on the data. How can you prove the data you have is complete? You'd also have to account somehow for assumptions.
9 out of 10 dentists say my toothpaste is better than yours. A different set of dentists are one shy of unanimous about the superiority of yours. Hmmmm....
Don't get me wrong. I'm not arguing that stats have no value. I'm rebutting Cantons claim that I must present some statistical evidence to indicate that Drew does, in fact, improve our o-line's statistics. Of course I could poll every NFL player and to a man they would agree with me, but is that enough? A reasonable person would likely think so, but can we ever prove that, statistically? Good luck with that...
Oh, and no worries about quoting wikipedia. I do it all the time. We know that, statistically speaking of course, everything on the internet is true. :-)
|
I don't know what to say to the first part about the broken hand...? Statistics could be used to give a broad idea of how long a recuperation from a similar type of injury usually takes. But it would only be a guesstimate, because there would be so many variables involved, for which statistics couldn't account for... so why would we even try?
To the second part I would say that if the method of collecting the data is scientifically constructed... the data sample has to be representative and, therefore, you can make valid statements based on it. I've never heard of a complete data set in any type of large scale research involving humans... something like a census would probably come the closest, but, alas, that would also fall short of a complete data set. You can have a data set collected from a representative sample on which to try and prove or disprove a hypothesis, but in most cases you just end up either corroborating or contradicting the initial hypothesis, which leads to further research.
As for the dentists... they are bought and payed for by the company making the tooth paste. :D
But, I need to add a disclosure, because I haven't really studied statistics, so I'm not a hundred percent on all that is involved. But my second college major was sociology, which is as close to statistics as you can get methodology-wise. |
