Okay folks. Here it is. The last 14 days. And like it or not, there’s still a chance that Trump could wind up with a big W on November 4th, or whenever they finish counting the votes. But the chance of this happening as of today (but not necessarily tomorrow) are slim to none, and even his own campaign staff are beginning to look for other jobs.
That being said, I always start looking over my shoulder during the last several weeks of any political campaign because no matter how accurate the pollsters all claim to be, the moment that someone starts using regression analysis to predict results, you’re in something of a la-la land, whether you like it or not.
Regression analysis is fine when you want to explain or illustrate how two or more trends moved against each other over time. But it’s not all that accurate or reliable when you use regression methodologies to predict the outcome of any event before the event takes place. And the reason for this lack of reliability is very simple – there’s always a chance that the particular event will turn out to be what Nassim Taleb first called a ‘black swan.’
And no matter how unique and rare the appearance of the black swan happens to be, it happens and there’s no way to predict when it might happen again.
In compiling this week’s report on pre-election polls, I thought I would double down on the possibility that what I have been describing to date might turn out to be totally and completely wrong. So this week I’m going to give you the national and swing-state numbers from two aggregators, Nate Silver’s 538 and the independent political blog, RCP.
It turns out that both aggregate national polls ended up at just about the same number in 2016, with Hillary coming out slightly under 3% more than Trump, which is exactly where things ended up. Both polls also had basically the same odds of a Clinton win on the eve of the election itself, with 538 saying that she was 71% in the bag and RCP outing her number at 69%.
One more point before I give you the numbers. All the pollsters today have Pennsylvania, Michigan and Wisconsin as a Democratic ‘lean.’ All the pollsters had the same three states leaning blue in 2016. All three states ended up going to Trump by a total margin of less than 2/10ths of 1 percent.
Here are the battleground numbers from 538, Joe blue and Trump red:
Here are the numbers from RCP:
With the exception of New Hampshire, the numbers from 538 look better for Joe than the numbers from RCP. The difference isn’t all that great, except that Joe has hit the magic 50%-mark in 6 battleground states according to 538, he’s only at 50% in 3 battleground states according to RCP.
Taken together, Joe’s battleground average in the 538 polls is 49.52%, but in the RCP overall average his number is 48.52% – Trump is 4 points behind (44.8) according to 538, the polls from RCP have him only 3 points behind at 45.08. I know it’s only a nit here and a nit there, but let’s not forget the size of Trump’s winning margin back in 2016.
For me, here’s the most worrisome point in these two aggregate results. The dog shit number from 538 is 5.61, it was 5.89 last week. That’s a good thing to see. But the RCP dog shit number is 6.4, which means that at least half of the dog-shit still haven’t made up their minds. If all those dummies go for Trump, he could win again. At least that’s what we see in the RCP polls, the 538 numbers are somewhat more positive for Joe.
The point is this. Nothing’s done until it’s done. Which means we all have serious work to do for the next two weeks. Money, phone calls, ZOOM, whatever it is. Do it. Just do it.