We told you yesterday about a CNN/ORC poll that showed Trump holding a national lead over Clinton for the first time since July, and that was not fun. Sure, it was only a two-point lead, well within the margin of error, but coming off a month when Clinton posted several double-digit leads in national polls, it still represented and alarming reversal for Democrats. When confronted with the poll results, though, MSNBC said, “not so fast.”
Chuck Todd, backed by MSNBC’s statisticians, discovered that the poll had a fatal flaw that skewed the results in Trump’s favor:
“Whites without a college degree appear to make up nearly half of their sample. In 2012, by the way, whites without a college degree was slightly more than a third of all voters…The point is, your numbers may not be wrong but your weighting may be, your assumptions. So the CNN folks assumed an electorate that is not an impossible scenario for Trump, but it would be an historic shift if it occurred.”
When they “unskewed” the poll to be representative of how the electorate looked in 2012, Clinton had emerged with a four-point lead—a six-point swing from what CNN/ORC reported:
As Todd points out, it’s not impossible that the electorate will be chock full of whites without college degrees—even more so than 2012. Then again, there’s no real evidence for that, and making the assumption is quite a leap.