View Single Post
Old
05-25-2013, 08:41 PM
  #56
MiamiHockey
Registered User
 
Join Date: Sep 2012
Posts: 315
vCash: 500
Quote:
Originally Posted by FreddtFoyle View Post
Miami, why don't you do us all a favour and share your chi-square spreadsheet somewhere like GoogleDocs so we can all learn from your work and discuss it.
Because you gotta admit, when it comes to simple probabilities of winning, when one group wins a championship tournament 4 out of 16 times (25%), and another wins it 7 out of 16 times (44%) in the same period, and the first group, the OUA, has more teams competing for the prize thanks to their two guaranteed spots each tournament, it is pretty difficult to grasp that it "is not significantly different from chance" that the AUS has a more dominant record on the national stage.
The reason I did a proper statistical analysis is because, repeatedly, you have stated that you only deal with facts, and seem to have presumed that you had all of the facts and I had none. Statistics can tell us what the facts are.
The beauty of statistical tests is that they can tell you if what you are seeing is supported by the data. Your hypothesis is that AUS teams are more likely to win the University Cup (because the best AUS teams are better than the best OUA/CW teams). My hypothesis is that is not true, and that the best teams from each league are equally strong - and thus getting into the University Cup is most important thing. The beauty of statistics is that, using the data we have observed, we can test your hypothesis against mine.
To be honest, if you don’t use know how to use Microsoft Excel, posting the file online wouldn’t help, because you wouldn't know what you are looking at. If you do know how to use Microsoft Excel, this is very easy to test yourself - just look online for an example.
To explain this more clearly, imagine that we had a 6-sided die. Your hypothesis is that this die is weighted - that the AUS(s) side is (are) more likely to come up because the best AUS teams are better. The alternative hypothesis is that it is not weighted, and that all 6 sides are equally likely to come up. In other words - that it is a normal die.
Let's imagine that you rolled the die 16 times. How often would you expect each side to come up? Well, roughly 16/6 = 2.7 times for each side. Of course, we've all played board games long enough to know that sh%t happens, and sometimes you end up rolling 4 1's in a row. Does that mean that the die is weighted, or is it just chance?
The chi-square test is a beautiful test because it tests the probability that the results you see are due to chance.
In the 16 University Cups, the OUA has had 41 teams, CW 27, AUS 28. So, the expected number of Cup wins for each conference would be:
OUA = 16 x (41 / 96) = 6.8
CW = 16 x (28 / 96) = 4.7
AUS = 16 x (27 / 96) = 4.5
So, the question for your hypothesis is whether 7 is significantly more than 4.5?
Like all scientists (including the ones you rely on to find cures for cancer and to build your bridges), I use the standard of 95% certainty to say that, yes, the hypothesis is supported. This chi-square test falls well short of that, at around 70%.
Those are the scientific facts.
And, BTW, I worked for CIS teams in various recruiting / coaching capacities from 2001-2010, and I am still in constant contact with my coaching colleagues discussing these matters. Do us all a favour and stop acting like I don’t know CIS hockey. You may not agree with what I have to say, which is fine, but it should be clear to you by now that my opinions are not based on guesswork.


Last edited by MiamiHockey: 05-25-2013 at 09:16 PM.
MiamiHockey is offline   Reply With Quote