Donald Trump, Marco Rubio Won GOP Debate, Poll Finds
blogs.wsj.comI posted this and wanted to put the word "flawed" before "poll". Mainly because this got front page mention on the WSJ website but if you read how the poll was done it's clearly suspect in it's methodology.
What independent indicators would there be if the majority of national polls had margins of error so far underestimated as to be meaningless? Particularly this far into the future from an election which would provide some assessment of accuracy. Or, put a different way, what published assurances exist that a particular poll has resulted from a properly random sampling of the voting populace? (often: the cell-phone screening problem) At the very least there's a bias to judging a poll as having some accuracy because of the "weather prediction paradox", which is: I say I can predict the weather 23 days from now with perfect accuracy because I can change and refine my prediction as that day approaches and say I was 100% correct in my prediction 22 days ago given my data then as now conditions (or voters' minds) have changed.