Gigamantis wrote...
Online polls are meaningless. They always have been and they always will be. An online reaction like this may be unprecedented but it doesn't mean anything. We don't know the actual size of these polls person by person and if they're even remotely legitimate.
The sales estimates are the only legitimate information we have and they're promising. If someone wants to conduct a scientific poll and draw conclusions from it they should. Passing around an internet poll is beyond poinltess, though.
A random sample of only about 750 people can predict the a population of 10 million with a margin of error of 3-5%.
We know the BSN poll of 71,500 people shows 2% liked the ending as-is.
Let's say BSN poll is a random sample (we know it's not). A sample of 71,500 would have a margin of error of less than 1%, but let's call it 1% for now.
Under a normal distribution, the entire population people that liked the ending as-is would be 2% +/- 1%. 68% of the time it would be between 1% and 3%, 95% of the time it would be between 0% and 4%, and 99% of the time it would be between 0% and 5%.
A self-selecting sample has predictive power if the sample is large enough and the results are lopsided enough. Both criteria exist in the BSN poll. It is almost 100 times larger than what is required for a random sample. It is so skewed that it really can't be any more skewed. I mean, what is lower than 2%???
So, even if we use a 20% margin of error, 68% of the time, the percentage would be between 0% and 22% and 95% of the time, the percentage would be between 0% and 42%. That's still not a majority.
The thing is with a sample size so large, the margin of error won't be 20%. It will be less. So we have even higher confidence that the people that liked the ending as-is is not a majority.
Nothing is certain unless we ask every man, woman, and child what they thought of the ending, but we can state with a high degree of confidence that the majority didn't like the ending as-is.
Modifié par kbct, 12 avril 2012 - 06:04 .