Heh: MSNBC 'Unskews' CNN Poll Showing Trump in the Lead

Posted: Sep 07, 2016 10:20 AM

Yesterday's eye-popping poll showing Donald Trump overtaking Hillary Clinton in CNN's four-way nationwide survey touched off an amusing round of "unskewing" from political observers -- many of them nervous Democrats and liberals. The R+4 sample is unrealistic! The national top line result doesn't match key swing states! The margin among independents is way off!  Granted, some of that skepticism may be well-founded; there hasn't been a presidential year general electorate that has tilted that heavily Republican in decades, for instance.  Allahpundit reminds us that even in 2004 -- the only election in which the GOP has carried the popular vote since 1988 -- the D/R turnout split was dead even at 37 percent.  Nevertheless, what's so entertaining about the feverish spin surrounding CNN's new numbers is that many of the doubters have gotten a lot of mileage out of ridiculing conservatives for grasping at very similar arguments in the recent past.  

Some right-leaning analysts, including yours truly, regularly entertained the notion of adjusting the 2012 polling, arguing that major pollsters' samples were unrealistically slanted toward Democrats.  It made sense on paper, but it was wrong.  The polls were by and large proven correct when Barack Obama's re-election team defied expectations and turned out a D+6 electorate.  FiveThirtyEight also wrote up a "no, the polls aren't skewed" explainer a few weeks back, smacking down Trump supporters who were coping with his then-disastrous standing by insisting that the surveys were rigged or wrong.  But now that the shoe is on the other foot, some on the Left have reflexively resorted to the same self-soothing tactic.  Such reassuring scoffery was showcased on MSNBC late Tuesday, when a full-screen graphic unskewed CNN's poll results, altering the numbers to reflect 2012 turnout numbers. Fun:

There's a crisp $20 bill waiting for the first person who can find that network casting similar suspicion on pro-Obama polling results in 2008, based on the 2004 exit polls.  I'll wait.  Meanwhile, a number of left-leaning polling gurus are mocking MSNBC and similarly-minded lefties on social media for embracing poll denialism:

Silver correctly notes that even if you don't want to believe CNN's latest data set, there has been a very clear anti-Hillary trend in the national numbers over the last two weeks, and excuse-making about lagging swing state indicators is weak sauce. I made a similar argument as well:

In fact, based on Silver's closely-watched model, Trump's chances have inched up to a one-in-three shot.  Given how things looked in early August, Team Trump will happily take that improvement, even if they're still necessarily indulging some less-than-realistic electoral math fantasies.  I'll leave you with this, which could deliver some genuine comfort for those aboard the Trump train:

It's an established fact that he's struggling mightily with nonwhite voters and younger people. But is this year's electorate going to be older, whiter, and less educated than even some of the most solid data would project?  Even if that's the case, what about his unusual problems among college educated whites?

UPDATE -- Here's the video: