Originally Posted by Green Potato
(Post 1449917)
Statistics to the rescue!
We have the data right here, and through a bit of math, we can sort of get an idea of where the real average score will fall according to those polled. I'll use a 95% confidence interval for this, just to keep things standard.
First are the assumptions. We appear to have independence, as I really doubt anyone who voted secretly collaborated with someone else just for a poll, but because opinions were expressed below and that may have affected people's argument, I'll proceed with caution. As for the sample being random, it's far from perfect, but it's close enough. The people voting here are cognizant of what teams will be competing, and are generally good at taking educated guesses. However, this really can't easily extend into the general community's opinion, as people posting on this site are much more likely to be from better, more established teams than the weaker ones. Again, this is a reason to proceed with caution when analyzing the data above.
Also, this poll wasn't purely quantitative, as an potion for more than 140 was given. For this number, I estimated the average to be about 155 based on how normal curves behave, but again, this is a point where the real and hypothetical worlds don't mix well.
Just to give some reference...
The mean is 96.96 points per match.
The sample standard deviation is 30.12 points per match.
Assuming this model is an accurate representation of performance (HUGE leap here, even though we have assumptions to back it up a bit, this will likely be a large source of error), we get a 95% confidence interval of...
(93.59 , 100.33)
And there you have it. I am 95% confident that the true mean expectation of the highest qualifier, according to the Chief Delphi community who bothered to vote on this post and who may or may not have a good grasp on the reality of week 1 competitions, is between 93.59 and 100.3 points.
And that's why the real answer is, "We'll see." :rolleyes:
|
