|The good news is that the election will be over in just nine days.
The bad news is that between today and Tuesday, November 8th, most of us may suffocate from the endless stupidities being shouted by Trump and his surrogates, and repeated by the dumbed-down media. Trump is after voters, and the media is after audience and the advertising dollars that come with viewers and readers. Many have told me over the decades that I’m going straight to Hell. I almost wish such a place existed, for it would be a great pleasure to see Kellyanne Conway and the other Trump surrogates along with various TV “journalists” on fire for real.
This election has been full of information, 99% of which is not informative. Those looking for the truth are pretty much on their own. Lost in the mind-numbing endless repetitive chatter, lies, misleading claims, bitter accusations, and Trump insults are these truths that are among the more important ones just now:
Not bad, fellow citizens; not bad at all.
Truthteller “Late decider prediction tool” update
We’ve continued daily tests of our new model, and it is still performing as we’d hoped. Meaning it’s quietly tracking the steady, unspectacular drift of the remaining third party and undecided supporters to Clinton and Trump. As hypothesized back in September before the first debate, the great majority of these late deciders are indeed falling to Clinton.
Readers will recall we’re testing the new tool using the main twelve Battleground state daily poll averaged results. These are the most hotly contested states, the ones that will decide the election. As of last night, October 29th, since September 26th in these key states:
Just to remind you, the primary purpose of the new tool is to predict the final national winning margin for the victorious candidate. This percentage is the key predictor in guessing how many House seats, if any, might flip to Democrat occupants. More on this below, but for now we can say that the tool predicts that Clinton will win with at least a 15% margin over Trump. To be blunt, this projection is regarded by most experts as wishful, possibly insane, thinking.
The main analytical problem in 2016
As noted here and in many other election analysis sites, the undecided likely voters are taking their time to make up their minds. Similarly, the greater number of third partiers than has been the case in prior elections are hanging on to their candidates right down to the wire. We cannot really blame these late deciders, given the tendency in this ugly contest for a new revelation to appear every ten days or so. These last roughly fifteen percent of the likely voters have made life miserable for the polling organizations. Many of their most revered “rules” and past voter behavior patterns simply don’t apply this year, with its two intensely disliked main candidates, and its two third parties.
Seeing this relatively huge pool of late-deciders on the horizon is the reason we invented a special-purpose, applies-to-2016-only predictive model in September. It’s experimental, and hard to test, since there are no fat files full of past data to use, simply because this type of election has never occurred before.
What we expect to see
With eight more testing days to go, we may find a flaw in the math, but based on the trial so far, we’re confident that the model will come reasonably close to predicting Clinton’s final victory margin. With each passing day, the shift of the majority of late deciders to Clinton has been rock-steady.
Then came a stunning confirmation of our tentative projection from a quality polling service. Over nineteen million voters have already voted, so Reuters/IPSOS finally has enough of a pool to build a sample of these early deciders. Yesterday they announced that Clinton’s reported winning margin among these folks was, indeed, 15%! We cannot say for certain if that margin will hold up through November 8th, but if it should be close to that, then we’ll feel our effort was vindicated.
Much more importantly, a huge winning margin of twelve percent or more will give the Dems a good chance of adding 30 or more House seats to their present total, and taking control of the Congress.
We’ll publish our final projection next weekend, before Election Day.
|…The Clinton Coalition is on the verge of swamping the Republicans in a wave not seen since Lyndon Johnson crushed them in 1964. Many pundits are finally awakening to this possibility; we intimated as much months ago in our first Battleground map.
Could something suddenly throw up a roadblock to the looming progressive victory? Well, sure; in an era of Wikileaks and sensationalistic revelations it would be daft to assume that the contest is over before it’s truly over: we’ve only to remember how Trump threw himself under his own fancy bus just ten days ago.
But time is short, and it’s unlikely in our judgment Clinton’s juggernaut can be slowed, much lest stopped. The notion that some massive mob of mystery Trump supporters will emerge from their shadows of shame, vote for him, and pull off a surprise win is, sorry, absurd.
The TruthTeller swing vote predictor tool
We’re still testing the new model with each day’s results, but the trend is clear: third party and undecided voters are making up their minds for Clinton, not Trump.
Long-demonstrated political experience says that most third party and undecided voters will opt to support one or the other of the two major party candidates as the contest nears decision day. Conventional wisdom has it that these last-minute deciders will tend to split roughly fifty-fifty.
Seeing the very substantial number of third party and undecideds this year, we wondered if that might foretell a different split. Our theory was, loosely, that the late-deciding voters would disproportionately decide for Clinton. Our reasoning was simple:
We therefore concluded that
As announced to our readers, I proceeded to construct a last-minute swing vote model that could predict such a result in credible numbers.
The express objective? To test my hypothesis that in this specific and uniquely messy election, as the three categories of “swing” voters gradually opt for one or the other major parties, the disproportional majority of them will go for Clinton.
How it works
Unlike current models, ours contrasts a consolidation of polling in the twelve main Battleground states with the averaged consensus national poll. We know the results in the BG states will tend to be tighter, since that is where the contest is being waged most intensely. But the national trend is also important, since that correlates better with how each party will do in the Congressional races.
The model employs only recent state polls in the more or less “traditional” Battleground states, namely: Virginia, New Hampshire, Pennsylvania, Michigan, Wisconsin, Colorado, Florida, North Carolina, Nevada, Ohio, and Iowa, plus a twelfth I added last Spring, Arizona, after I realized the Latino vote was swelling there along with Colorado and Nevada. (Georgia, which I saw as another potential new swing state, has so far this cycle failed to develop sufficiently to make the Battleground cut, but should be there by 2020.)
Basically, our BG sub-model allocates 1200 percentage points — twelve states times 100 — across the five distinct voting categories:
plus what I define as the three categories of “swing” voters
We were ready to begin test runs on September 26th, the day of the first debate.
Complications — what else is new?
After a week or so of initial runs, we were gratified to see just the sort of trending our thought-model had suggested. In the two weeks between the first and second debates, we saw a steady increase in Clinton’s percentage of the vote across the BG states, excepting Iowa.
In the same period, we saw Trump’s share of the vote holding more or less steady. Meaning her gains had to be coming from the third parties and/or the undecideds. But the undecideds were also holding at the level of September 26th, so her gains were in fact coming from the two third party candidates. These voters were NOT going to Trump, just as our hypothesis had predicted. Satisfactory, dear Watson.
Then, on Friday October 7th, we all saw the Access Hollywood tape. Almost overnight the numbers began to roil, mainly due to millions of women, and men too, rejecting Trump’s boorish behavior, along with his insincere apology for it. In the week since, the polling has been all over the NBC Studios lot, as women come forth to accuse Trump of being a sexist liar and molester.
How a trend becomes a wave
Just before the 2nd debate, we were ready to share the preliminary results of our testing with you readers. The results at that point projected a strong Dem victory in every one of the twelve key Battleground states, even including Iowa. More importantly, the new tool forecasted a national win for Clinton of 12.2 points over Trump.
This margin is terribly important, since anything above 6 or 7 points nationally can spell a major win in the Senate and House of Representatives contests. In 2008, for example, the Obama team flipped 21 House seats after beating McCain by 7 points. If our estimate of 12% turned out to be correct, the 30 seats needed to return the House to Democratic control would likely be within reach.
But then, two immature boors on a bus happens, and the polling numbers go bananas.
They’re still unsettled, but our latest run, Saturday the 15th, suggests a winning Clinton margin above 14%. Gulp. We’ll keep running the tests, and hope that we see a final trend emerge about five days after the upcoming Wednesday third debate. That should let us make a prediction for the Democratic prospects in the House, with, still, nearly two weeks to go before voting day.
Watch this space!