ABC tracking poll "mysteriously" shows 10 point shift toward Trump in one week

I recorded an interview with News Talk 103.7FM in Pennsylvania earlier this week which is airing today and I really wished I’d known about the new ABC tracking poll which just came out when I did it. During the interview I was asked how much faith I was putting in the polls at this late stage of the race and the fact that a few days ago most of them were already refocusing election coverage on what color drapes Hillary Clinton would be installing in the Oval Office. The question seemed designed to see if I would bite on the current conspiracy theories which hold that major polling outlets are stacking the deck with the intent of depressing the turnout of Donald Trump voters.

That wasn’t my take at all, but I did question how the polls were being structured in terms of their sample makeup and predicted that we might be seeing a significant shift in the numbers over the final days. That ABC tracking poll I linked to seems to bear it out because in a matter of only days they’ve taken the race from a blowout in Hillary’s favor to a two point horse race inside the margin of error.

It’s a tale of two electorates in the ABC News/Washington Post tracking poll, with shifts in intended turnout moving a large advantage for Hillary Clinton a week ago to a far tighter 2016 presidential race today.

From a 50-38 percent Clinton lead over Donald Trump in the tracking poll’s first four days, Oct. 20-23, it’s a 47-45 percent contest in the latest results. The movement has been in Trump’s favor, +7, while the -3 in Clinton’s support is not significant, given the sample size.

Changes in the poll’s latest four nights compared with the previous four are not mainly about people shifting in their candidate preference, but about changes in who’s intending to vote.

Keep in mind that this latest sample was taken entirely before the news of the FBI announcement broke. So what could possibly explain a shift from a twelve point lead to a two point margin almost overnight? I would hope that any serious observer can’t honestly believe that tens of millions of Americans are changing their minds back and forth, over and over, based on nothing more than the latest inflammatory comment made by Trump on the stump or some witticism spouted by an MSNBC spokesmodel.

As I discussed in the aforementioned interview, I don’t buy the conspiracy theory and I said so then. That’s not to say that we shouldn’t cast a jaundiced eye on polls which provide a built-in Democratic advantage of as high as fourteen points in some cases, but I’m far more concerned with the likely voter model. Any professional pollster who is honest will tell you that there is as much art as science involved in predicting the future in this fashion.

It all comes down to the likely voters. The problem this year, unlike the past several presidential elections, is that I don’t believe the pollsters have nailed down who exactly is likely to show up. We frequently rely on voters who are four in four the most, while assigning a bit less value to those who generally only show up for presidential elections. But we’re in a season of weirdness right now. On the Democratic side we saw massive unrest among supporters of Bernie Sanders. He was bringing out a large army of relatively unknown and predominantly younger potential voters. Will they show up a week from Tuesday? And if so, will they still vote the D line without him on the ticket? As for the GOP, Trump is at least rumored to be bringing out disaffected voters who have long since been too disgusted with the system to bother going to the polls. On the flip side of that coin, he’s turned off enough regular conservative voters that they might just stay home or vote third party.

The point here is, how confident are we that we know who will actually vote on election day? If we can’t answer that question with at least a moderate degree of accuracy then the margin of error could be significantly higher than we normally expect. The likely voter model may be getting shoved in one direction or the other by the unlikely voter model. ABC tossing out a number showing a two point race less than a week after declaring the spread to be twelve seems to bear this out. I still maintain that we’re in uncharted territory here thanks to nominating two candidates with the lowest approval and trust ratings in modern history.

So do you still think the race is over and Clinton has already won? I wouldn’t bet the entire farm on it.