Yikes: MSNBC "unskews" the CNN poll showing Trump up by two points

This is unfortunate.


By this same logic, I guess, if you “adjusted” CNN’s result using the 1984 electorate, Trump would be up something like 15 points due to his strength with white voters. Electorates change. They change a lot more over 32 years than they do over four years, but even four years is time enough for decisive shifts to happen. Look no further than Obama’s landslide in 2008. He turned a narrow Bush win in 2004 into a Democratic whupping by mobilizing minorities and young voters like no one before him had done. Trump is looking to do the same thing this year with working-class whites. If he pulls it off, the 2012 electorate will be irrelevant.

I think it can be helpful to look at the 2012 data for clues to see how well a candidate is meeting some of the basic benchmarks he needs to meet. For instance, if we start to see Trump polling consistently at 40 percent among Latinos, we can look back at the fact that Romney pulled just 27 percent within that group four years ago to infer that Trump is resonating with that group and might outperform Romney in states with big Latino populations. It’s another thing, though, to try to predict from that what percentage of Latino voters will turn out, especially in battlegrounds. That’s the trick for pollsters in determining who qualifies as a “likely voter.” Who’s really likely to turn out? How do you go about determining that? One of the great mysteries of the campaign since day one has been whether Obama’s young and minority voters will turn out for Hillary at the same clip they did for him. Todd is making an assumption here, for illustrative purposes, that they will. But based on … what, exactly?

He’s also making a turnout assumption for lower-educated whites. They were a bit more than a third of the electorate in 2012, he notes, but CNN sees them as closer to half of the electorate in 2016. Could that be accurate? Well … maybe! Remember the “missing white voters”? CNN evidently believes they’re going to show up in far greater numbers this time than they did for Romney’s election, which would mean a stronger showing for Trump than everyone currently expects. This is why, in flagging the surprising R+4 sample in this morning’s CNN poll, I noted that it may be that the electorate is trending red and CNN is picking up on it. If you had to bet, you’d bet that it’s an outlier because the weight of other national polls is against it — for the moment — but there’s no way to conclude that it’s “wrong.” Todd is careful not to use that word here, in his defense, but the point of crafting an alternate topline for CNN’s data based on the 2012 electorate is clearly to try to reassure MSNBC’s left-leaning audience that CNN’s result is screwy. He could have just called it a probable outlier and left it at that. To go this far, reworking the actual data, is unskewing.

Funny that we’ve reached this point on the left now after some righties spent August doing the same thing to polls favorable to Hillary. Thank goodness the Trump fans on Fox News will be able to believe in polls again!