General Discussion
In reply to the discussion: This! This! This all day long. Without comment [View all]Lucky Luciano
(11,284 posts)<nerd>
I think they were just straight up multiplying probabilities of a dump win in each state. This is fine if the overall events are independent.
Correct me if I am wrong below - I use some stats, but I am not a statistician:
If we assume that the polls are from unbiased samples, then the probability of dump winning one state can be calculated based on the assumption that the proportion of pro-dumpers in the sample is asymptotically normally distributed (central limit theorem). The joint distribution of the five states will be independent if the covariance matrix is diagonal since it will be normal and the second moment is sufficient to establish independence. Was it this covariance matrix, which of course implicitly handles fluctuations about the mean the basis for your comments?
I would probably say (though I don't have the data to support it) that the first principal component probably explains 90% or more of the variance in the joint distribution. The second component probably gets 99%. These really are highly correlated states. Maybe NC/FL ate a bit less correlated and would cause us to need the second principal component.
</nerd>