A private discussion prompts me to say this publicly about how I see the potential for where Tuesday's results will wind up nationally.

Will seem like a fair bit of a cop out, which is fine. I committed not to forecast for this 2020 General, and am damn glad I did. 1/

It's a wild election cycle in so many ways.

If I was forecasting, I'd feel compelled to come up with an exact point I think things will wind up, or at least a reasonably small range of possibilities, as I did in 2016 fairly accurately.

But that seems like fool's game. -cont- 2/

Why?

Here is a simple way of putting it:

I expect that may way of averaging & RCP's way of averaging is going to wind up in the 7.5%-8% range nationally.

538's a point to 1.5% higher.

So then, what's a *normal* national polling error?

A normal national polling error is not what we had in 2016. National polls were pretty good! Off by 1%, on average, unless you average like 538, in which case they were off by 2%. That's a normal polling result that isn't in error.

A normal national polling *error* is ~3.5%. 4/

George H.W. Bush beat his national polling avg in 1992 by 2.6% (avg 8.2%, final 5.6%) but still lost, so no one really remembers.

Bob Dole was down 12.1% in 1996 final polling, but "only" lost by 8.5%, a 3.6% difference.

5/

But, alas, it is not just Republicans who sometimes beat their polling by a substantial amount.

People remember 2012 a bit more. Closer. Involved Obama. Barack beat his RCP avg by 3.2%.

So, that's 3 times 1992 forward that national polls missed by ~3%.

6/

So, a normal national polling error - and boy if it doesn't seem like the conditions are ripe for such an error! -

spreads that 7.5-8.5% range to a 4% to 12% range of final national outcome reasonable possibility.

!!!

But there's more ...

7/

What about the chances of an "abnormal" polling error?

Well, we've had one of those in relatively recent history too, one with four national final polls, average = Carter down just 3.8% to Reagan (1980).

Reagan won by 5.9% more than averages with a 9.7 win.

8/

So now we are into completely obnoxious territory where Biden, a la Rasmussen, could be competing for an actual popular vote win.

*OR* Biden, a la "live caller" Franklin Pierce/Boston Herald and also "J.L. Partners", could win by as much as 14%.

9/

The more reasonable versions of this massive range are Emerson at 4.4% and CNN at 12%.

V glad I do not have to make a *decision* to hang out for the world to see & later, of course, be used against me if, as inevitable for either CNN or Emerson, it's seriously off.

10/

We could land in the sweet spot between CNN and Emerson (around 8-9%) where they have both missed by 3-4%.

Completely reasonable sounding finish, to me. State polls are coming in a little lower than that, so maybe due to Trump's electoral college advantage ...

11/

... Maybe only a regular sized electoral college blowout, rather than a Reagan like maps in Dems favor.

But then, a sea of blue on the electoral map is also quite possible. See @leantossup! (Robert's a damn smart and successful forecaster.)

12/

Tbh, I am rather enjoying keeping things wide open, instead of sweating exactly where I will wind up, knowing I've committed myself w/public criticism to coming in under 538/Econ/LeanTossup range.

I can see the wisdom in @SeanTrende & RCP emphasizing "Toss-up" EC map.

13/

And just to be *abundantly* clear, I am not at all insisting that this thing will come in closer than 538/Econ/LeanTossup range. I clearly be Robert/LeanTossup in a little side-wager by-election deal in the UK once. Ever since, he's consistently waxed me.

14/

All of this is to say that, while I will privately tell a few people where I kind of expect things to land, I really will not be at all surprised at virtually any result next Tuesday or ... when the dust finally settles weeks later in terms of precise national pop. vote. 15/fin