2020 Commentary #4: Those Elusive Last 12 Electors

.

.

.

Note: This had been titled “2020 Commentary #4: Those Elusive Last 12 Electors,” published on Oct. 26, revised to add a critique of forecasting sites on Oct. 29, and revised to add a call for Maine to pass a new law to prevent stealing of the election at around noon on Nov. 3 and Nov. 5, and revised to add some additional analysis on polls and polling after things finally settled down on Nov. 17.  I restored the original title on December 29.

I’m moving discussion of the presidential election results to here, because this was my most thorough discussion of the election’s endgame — which goes well beyond mere vote-counting.  So everything down to the graphic of Biden as Sisyphus will be new, posted blog style in reverse chronological order  Reading the portion below the graphic  will probably be edifying as well; none of its text and graphics has been changed.

As of the evening of November 5, the remaining votes of six states and two Congressional Districts will decide the outcome of the 2020 Presidential election.  (Most people wrongly believe that the two Congressional districts are already settled; they should be, but aren’t, for reasons to be explained below.)

Joe Biden is generally agreed to have a current Electoral Vote lead of 253 to 214, with 71 votes undecided.  The unsettled states are, in order of the size of electoral vote:

• Pennsylvania (20 EVs): Pennsylvania (“PA”) could decide the election for Biden all by itself — and it would be one of the states whose results are most difficult to overturn.
  • 95% of PA’s vote has been counted.
  • As of 9 p.m. PT, Trump’s lead in PA has Trump’s lead in Pennsylvania has fallen to fewer than 24,000 votes.
  • Biden has won these counties, almost all of them urban or suburban:
    • Philadelphia (City of Philadephia and environs)
    • Delaware (near eastern suburbs)
    • Montgomery (near north/northeastern suburbs)
    • Chester (middling north suburbs)
    • Bucks (middling north suburbs)
    • Lehigh (further northwest suburbs)
    • Northampton (even further north exurb)
    • Lackawalla (Scranton and environs)
    • Monroe (between Northampton and Lackawalla)
    • Dauphon (Harrisburg and environs
    • Centre (State College, Penn State and environs)
    • Allegheny (Pittsburgh and environs)
    • Erie (Erie and environs)
  • So new votes coming from those counties and the most pro-Biden the closer they are to an urban center, Erie and Northhamptom being the least so.
  • Trump’s probably temporary lead is presumably why he has been calling — entirely improperly, for the state’s vote count to stop immediately
  • Below find the number of absentee ballots that were in these counties, insofar as it is available, with each county’s partisan margin from the already counted vote (in red or blue).  For example: if Allegheny County is 95% done with 310,796 ballots, it has 5% left to vote — which equals 15,540 ballots.  If we presume that these ballots will be the same as already counted ballots, Biden would get 59.5% of them (adding half of 19 to 50) and Trump would get 40.5% (subtracting it.)  So that would be 9246 for Biden and 6294 for Trump, or a net bump for Biden of 2,952 votes.  These are only the counties that report their number of absentee ballots, but it strongly suggests a Biden victory in this state — and one obtained before the end of the business day on Friday..
    • Allegheny (95% done), +19, 310,796
    • Montgomery, (98%) +26, 238,122
    • Chester, (98%) +17, 117,736
    • Philadelphia, (91%) +62, 75,220
    • Delaware, (91%) +26, 74,110
    • Luzerne, (95%) +15, 55,242
    • Dauphin, (98%) +8,48,638
    • Lackwanna, (98%) +8, 41,168
    • Washington, (98%) +23, 35,519
    • Northampton, (98%) +0.5, 33,265
    • Lebanon, (98%) +32, 19,761
    • Erie, (94%) +1, 14,741
    • Pike, (98%) +19, 13,058
    • Lawrence, (96%) +30, 11,911
    • Bedford, (98%) +68, 5,369
    • Warren, (98%) +39, 5,048
    • Wyoming, (98%) +35, 4,044
    • Jefferson, (98%) +59, 816
    • Perry, (98%) +51, 650
  • How Trump could cheat Biden of a victory:
    • First: I’m not telling them anything here that they don’t already know
    • He’d need help from the Supreme Court, to get them to agree that all counting had to stop on Election Night or at any rate before Friday, when as you can probably tell Biden is likely to pass Trump.  This would be absolutely outrageous — not simply on the grounds that it would deprive voters who had followed all of the rules of the franchise, but (this matters to constitutional scholars) because it would involve a federal court intruding on state voting rights — but it’s not absolutely impossible.  The could also say that all states that arrived after Election Day — even in the mail, as state law allowed, and which Trump’s Postmaster General’s actions slowed down, and even with a timely postmark — which would be similarly heinous, given that Pennsylvania law allowed them to arrive in that time and manner.
      • Adding: but by working all night — so long as it happened everywhere so as to preclude an equal protection argument — the state’s hard work may have precluded all of this.
    • Trump could try to convince the state legislatures in the state to decide that voting was so messed up and so close a margin that they should just send a delegation of their own choosing.  They can do this under the Twelfth Amendment, but no one has though that anyone in any race for the last century or so would ever do this. Even if they wanted to do so, though, Pennsylvania has a Democratic Governor, and would not sign such a law.  But technically, the constitution gave this role just to the legislatures, so some suggest that they do so by joint resolution.  Supreme Court precedent makes clear (in a case called Smiley) that constitutional delegations to the legislature were intended to apply to their power to make a law, and if the state procedure included a requirement that the law had to be signed by the Governor, then that’s what they’d have to do.  Would the current Supreme Court toss out that requirement and allow a purely legislative stripping away of a states’ citizens rights to have their votes determine the state’s electors?  Uh — I hope not!
      • Trump’s Congressional delegation has been tied: but Rep. Connor Lamb, in one of conservative counties surrounding Pittsburgh’s Allegheny County, is in a very tight race this year, leading by about a point.  If he loses, Trump has another Congressional state delegation to support him.
    • The legislature might simply refuse to certify the election — through means that could conceivably include protracted absence of a quorum — which could means that Pennsylvania would send no electors to the Electoral College.  That would make it more likely (or perhaps even assure) that no candidate would reach the 270 required for an Electoral College majority, which would mean a “contingent election” — which means that Trump wins.
    • Violence, kidnapping, extortion, assassination, bribery, sabotage, arson, torture, ballot fraud (either stuffing or ballot’s physical or electronic destruction) — you know, the usual fare in an election in a failed state.
      • Now: how could democrats possibly fight back against any of these?
      • You can’t do much about the Supreme Court at this point except to put on your best case and speak to conscience and to history.
      • The Twelfth Amendment option could be countered by street protests, petitions, and generally telling any Republican who voted for such a proposal would be politically dead forever.  The problem is: it’s probably not true — and who knows if we’d even be having popular elections anymore.
      • For the contingent election: the only thing that comes to mind is for the Democrats to challenge the credentials of enough people who want to take their seats that Biden would win the contingent election.  This would be a horrible precedent, but possibly justified in response to a horrible situation.  Does Nancy Pelosi have the guts to try this with maximum force?  I don’t know, what do you think?
      • Against violence, etc. — it’s certainly a disadvantage when one side has most of the arms and lives in more remote areas.  I got nothing, and if I had anything, obviously I would not be publishing it here!
• Georgia (16 electoral votes)
  • Trump is literally up by only 665 votes here at just after midnight PT, and the presence of so many uncounted votes in Atlanta and other areas of large Black minorities make it unlikely that he will be in the lead when dawn breaks.
  • Georgia (“GA”) could create an electoral college tie at 269, which — absent any other wins, would lead to a “contingent election” that Trump would probably win.  Such an election would involve votes by state delegations rather than individuals — and going into the election a majority of 26 states — the minimum needed — have delegations with Republican majorities.
  • Georgia is one of the two states — Arizona being the other — where Republicans have the Governorship and control of both legislative chambers.  That means that, if they wanted to, they could steal the votes from their voters without creating a federal constitutional question.  The hope here is that enough of its Republicans simply would not want to do it — and would be willing to take the heat from people who may be able to kill them for not doing it.
  • The likelihood that Biden will not end up with more votes than Trump the close of business there today is minimal. The question is what they will do about it.
  • Beyond the above, the things that Trump could do here are similar to those already discussed — failing to certify the election results is probably the most possible, though I’ve never heard of this being done.
• North Carolina has 15 electoral votes — and Trump will probably win it, so I’ll skip it for now
• Arizona has 11 electoral votes — enough to give Biden the narrowest win in Trump keeps Nevada (and it’s still useful otherwise, especially if Trump wins the absurd recount he has filed in Wisconsin, which I’m not otherwise addressing.
  • While Arizona has a Republican trifecta, it is also the state of John McCain, Jeff Flake, and Cindy McCain, and I just don’t think that a majority of its legislature would favor a society-exploding Twelfth Amendment “solution.”
  • It seems quite likely that Phoenix area will come through tomorrow — despite armed hooligans threatening the ballot count
    .
• Nevada has 5 Electoral Votes.  No time left to discuss it!

For the problem with the single-district votes, see the “MAINE NEEDS A NEW LAW” section below, which addresses the prospect that Nebraska’s unicameral legislature could just snatch back Omaha’s single electoral vote if it was the margin of victory, and suggests that Maine (a blue trifecta) should be prepared to retaliate in kind with respect to Northern Maine’s single red electoral vote if Nebraska does so.

• Actions not tied to individual states
  • Faithless electors
    • The Supreme Court ruled that states could prohibit electors from voting for anyone other than for whom they were instructed by passing laws preventing it.  How many states have done so?  My guess is: a lot fewer than should have!  In a very tight race, it could seriously come down to this.  Parties have to vet electors perfectly to make sure that they’re reliable — or else perhaps suffer the consequences
  • Crazy executive orders
    • Trump has been getting away with crazy violations of settled law for years now, because it’s clear that Congress (mostly because of the Senate majority) would not stop him.  Well, what is to prevent him from ordering that there can be no federal money spent to collect electoral ballots — and that no one can do it on a volunteer basis?  It would be insane —     but who with the power to stop him would do so?
  • Crazy judicial orders
    • I’m not even going to speculate here, but with Trump controlling the appellate judiciary, who knows what could be in store?

ORIGINAL STORY APPEARS BELOW!

“Sisyphus Joe” Biden can pretty easily roll that boulder most of the way up Electoral Vote Mountain, but that final 12-vote stretch is a real killer.

1. Reliving 2016

(Note: if you like this post, check out our more statistically sophisticated sequel, with updated numbers, at this link!)

Despite the seeming wind in Biden’s sails, Democrats still worry about winning next week’s election — and rightly so.  It’s pretty easy to see where Joe Biden gets the first 220 or so Electoral Votes that he needs to reach 270.  But getting those last 50 seems like a task similar to the one that Hillary Clinton faced when needing only 28 more to win, and — even setting aside the “faithless electors” who didn’t vote as instructed — failed to get them.  Remember that?

We still live in that world — and while various states are closer to flipping blue, each of those swing states could easily stay red, even without the sort of chicanery I discussed last week in the companion post to this one.

It has become common to say, in the wake of that election, that Hillary Clinton lost the election because (out of overweening overconfidence) she did not bother to visit Wisconsin during the  campaign.  Not quite: flipping Wisconsin blue would not have won her the election; it would have left her 18 votes short.  Even flipping both Wisconsin and Michigan would have left her two votes short of 270 on election night, losing 272-268.  (I’m putting aside the “faithless electors,” five of whom deserted Clinton and two of whom deserted Trump, not nearly enough to matter.)

To reach 270, Clinton had to win something like Pennsylvania — not enough by itself — plus one of those two states.  She lost Michigan by only 0.23%, Pennsylvania by 0.72%, and Wisconsin by 0.77%, and Florida (which could have given her victory all by itself) by 1.2%.  (She lost no other state by less than 3%.)  Trump’s narrowest losses were New Hampshire (0.37%), Minnesota, (1.52%), Nevada (2.42%) and Maine, (2.96%).

(If you want to relive and agonize over that night, this section of the Wikipedia page is an excellent resource.)  https://en.wikipedia.org/wiki/2016_United_States_presidential_election#Results

Biden has several advantages over Hillary Clinton:

• he has much higher favorable ratings
• polls are taking educational attainment explicitly into account this year
• subterranean social media isn’t taking us quite as much by surprise
• his ticket isn’t white-on-white
• Trump’s “Hunter’s laptop” October Surprise shattered when it hit the ground
• Voters now know Trump better:
  • they know that he’s a perpetual liar and con man
  • they know that he’s a would-be dictator
  • they know that he’s a gas-lighting, revenge-driven, narcissist
  • they know that he’s a financial basket case and a cheat
  • they know that he’s largely incompetent at governing
  • they know that he violates law and longstanding norms to evade accountability
  • they know he flubbed (and is still flubbing) the response to Covid, at a cost of over 200,000 lives
  • and, most importantly, they dislike him even more now than they did in 2016

But … all of that may not be enough.

There’s one big disadvantage for Biden that isn’t a surprise: the judiciary.  In some critical swing states and in the federal trial courts, appellate districts, and (as of the confirmation of Amy Coney Barrett) the Supreme Court — is both hostile to him and of doubtful scruples.  Frankly, Biden needs to get to 270 decisively — either by margins in swing states too large to challenge, or by enough of an Electoral Vote lead that even a hyper-conservative majority Supreme Court won’t want to brazenly twist the results.

Side note: The Court will surely at least hesitate to follow the lead of the 1876 Election Commission that resolved each any every one of 20 disputed Electoral Votes in favor of Republican Rutherford Hayes over Democrat Samuel Tilden (who had landed one vote short of victory) by an 8-7 party line vote.  (The results was ratified by Congress by an agreement negotiated by Southern Democrats and Northern Republicans — the “Compromise of 1877” — in exchange for consideration including withdrawal of the few remaining Northern troops from the South and the de facto end to Reconstruction — at least as it applied to treatment of Blacks.)

History has not treated the framers of that bargain kindly.  Yes, given that “winners write the history,” the Court could go totally full-on authoritarian — but upending the results in one state is a probably far more palatable than having to do so in three or four.

Putting all that aside for a moment, let’s look at who we would expect to win in the absence of [insert euphemism for cheating].

2. Where Biden Stands

Biden’s base in this election is actually far larger than Trump’s.  Biden starts with virtual assurance of winning the electoral votes from Hawaii, Washington, Oregon, California, Illinois, Vermont, one of the four votes from Maine, New York, Massachusetts, Connecticut, Rhode Island, New Jersey, Delaware, Maryland, and DC — a base of 183 votes without him having to break a sweat.

Trump, by contrast, can only feel an “I don’t even need to contact voters there” level of confidence — which I define as at least a 15% lead in reliable polls — in states with 70 Electoral votes.  These fall into two contiguous areas:

• The northern plains and mountain region includes Idaho, Wyoming, North Dakota, South Dakota, and three out of five votes from Nebraska
• the latter includes Oklahoma, Arkansas, Louisiana, Alabama, Tennessee, Kentucky, and West Virginia.

So that’s Biden leading 183-70, for now.

The candidates break even when we examine states where he’s moderately likely to win — roughly between 10 and 15 points.  Biden adds 33 more votes (for a total of 216) from Colorado, New Mexico, Virginia, New Hampshire, and Maine‘s two statewide votes, while Trump adds 24 (for a total of 94) from Utah, Indiana, Mississippi, and one more vote from Nebraska.

Biden leading 216-94!  Seems pretty good, right?  Keep reading.

Now we’ll look at states that merely lean their way — meaning leads of between 5 and 10 points.  Biden picks up 42 more votes (bringing him to 258) with Nevada, Minnesota, Wisconsin, and Michigan.  Trump adds 37 from Alaska, Montana, Kansas, Missouri, and South Carolina.  In states where the lead in polling is 5% or more, Biden leads by 258 to 125!

It’s got to be over, right?  RIGHT?  Wrong.

The remaining states include:

• two big ones: Texas (38 EVs) and Florida (29 EVs)
• four medium pretty large ones: Pennsylvania (20), Ohio (18), Georgia (16), and North Carolina (15)
• one medium one: Arizona (11)
• one medium-small one: Iowa (6)
• and two single vote districts:
• Omaha, Nebraska and all-but-Southern Maine.
  • (Note: Nebraska is a red state and Maine is a blue state, with legislatures and Governors to match, so neither can change the rules on allowing this separate district vote without inviting retaliation from the other.  But only Biden needs that EV to win if he reaches 269, so Maine could conceivable change its law him its northern district’s vote after he lost.  However, this would likely lead to a U.S. Supreme Court case, which would find a way to slap it down.)

Arizona’s 11 votes would give Biden 269 votes and a tie (which Trump would likely win in a vote of House delegations); and Arizona plus either of the single-state districts would give him the win.  (The only state that doesn’t give Biden an outright win is Iowa — but winning it is still useful to ward off faithless electors.)

So that’s great news, right?  Well, maybe … but maybe not enough of it.  Let’s leave Biden with that 258-125 lead for now and look at the ten swing contests (eight states, two districts) to see where they each stands.

This is a good time to tell you about the rule of thumb that polls have often underestimated Republican support by about 3%, because of facts like:

• voters who vote by mail, but their vote won’t count, because:
  • a mail-in ballot doesn’t get to the Registrar
  • a mail-in ballot somehow gets lost or discarded
  • they make an error in marking their ballots
  • they err in the marking or use of envelopes
  • they spoil their ballot by, for example, signing it
  • they lost their ballot
  • they gave it to someone else who didn’t deliver it
• voters who intend to go to vote in person, but can’t because:
  • they have an emergency (often child care)
  • they are called into work
  • they can’t find transportation
  • the polling station runs out of ballots
  • the polling station runs out of the right kind of ballots
  • the voting equipment breaks
  • the polling station closes down for lack of staffing
  • the polling station has problems because of Covid-19
  • and before you ask why they didn’t vote early, or vote absentee, many states and counties aren’t as well-served as Orange County, and can’t do these things — and, in fact, the wealthier and more conservative counties tend to be better served
• voters who go to the polls but don’t end up voting because:
  • the lines are too long — and they can be very long
  • poll workers won’t let them go the the bathroom and return to the line
  • poll workers won’t let them get food and return to the line
  • a home emergency crops up (often related to child care) and they have to leave the line
  • they face intimidation or violence
  • they didn’t realize that their old precinct was closed
  • they didn’t realize that they didn’t have necessary ID
  • they didn’t realize that they had to re-register when they moved
  • they didn’t realize how far away the closest polling station was
  • they had a sudden medical issue
  • they lost their ballot and couldn’t get a new one
  • they forgot their glasses and are denied help at the polls
  • the polling site ran out of ballots
  • or God knows what else

These sorts of things happen disproportionately to less well-maintained, less well-staffed, less-quickly fixed proportionately poor and Democratic precincts, especially ones that the state government wants to suppress.  So, considering a 3% Democratic polling advantage to be an even race is a pretty good guide.

Now let’s go take a look at those swing contests, starting with the most pro-Trump down into the most pro-Biden!

[Note: alas, this section is not entirely complete, because there is so much to know, and new court decisions are still spewing out.  So it will be updated when research and time allow.]

The Ten Swing Contests

(1) Texas

There’s a big jump between Trumps worst “leaning” state — Alaska, where he leads by 7.7% — and his best swing state, Texas.  Trump’s lead in Texas is just 2.4%.  Even with the “3% rule,” this is not an insurmountable margin.  It’s just not particularly likely this year — especially with Texas’s hard core Republican Governor — who has limited drop-off boxes to 1 per county, applying both to the smallest county and to Houston’s Harris County’s 4.1 million population —  and legislature.  If Biden wins here, of course, it’s all over.  But that’s not the way to bet.

(2) Ohio

Trump has a 1.0% lead here.  But there are problems for Biden here, to be added.

(3) Georgia

Trump has a 0.4% lead here.  But there are problems for Biden here, to be added.  (In the interim, just on’t count on this one.)

(4) Iowa

Trump has a 0.1% lead here!  But there are problems for Biden here, to be added.  (In the interim, just on’t count on this one.)

(5) Northern Maine

Biden leads by 0.2% here.  Hard to believe, but this could literally be decisive.  (In the interim, just on’t count on this one.)

(6) North Carolina

Biden leads by 2.1% here.  It starts to look a little safer.

(7) Florida

Biden leads by 2.3% — yet somehow Republicans almost always seem to win the close ones here.

(8) Arizona

Biden leads by 2.5% — tantalizingly close to the 3% rule of thumb.

(9) Omaha, Nebraska

Biden leads by 4.7%.  This is the most likely swing contest to go Biden’s way — and Maine should pass a law saying that if Nebraska denies Omaha its Electoral Vote then it will do the same with Northern Maine.

(10) Pennsylvania

Biden leads by 5.1% — but voting is very complicated here and the 3% rule is very likely a significant underestimate.  The Supreme Court has gotten Pennsylvania to agree to segregate those ballots arriving after Election Day so that it can decide — with Justice Barrett involved — whether to throw them out.  This is a complete mockery of justice, give voters’ reasonable reliance and settled expectation in having the extra time to vote.

In conclusion:

Biden is likely to win all of the states listed above where he’s leading, getting him to 258 electoral votes.

After that, his best chance is to win Pennsylvania — but the court system there has not been friendly to him and the state has a storied history of fraud from both major parties, possibly justifying the Supreme Court taking some sort of heinous action.  (Note that if the federal court in the Third Appellate Circuit finds for Trump, that decision would stand in the even of a tie Supreme Court vote, and so Justice Barrett would be able to recuse herself from voting without besmirching her reputation or costing Trump the election.

Biden’s second best chance is an exacta: winning Arizona plus the sole electoral vote for Northern Maine — bringing him to 270 and praying for no faithless electors.  (States could still pass laws to prevent them so long as they take effect by December 8.  The Supreme Court might disallow those new laws — but that would be more heinous than even Bush v. Gore.)

Biden’s third-best chance is winning Florida.  (Don’t get me started on that.)

Biden’s fourth-best chance is winning North Carolina.

Biden’s fifth-best chance is winning Georgia, Ohio, or Texas — any one of which would be shocking.

Focus on our neighbor Arizona, readers.  Focus on it hard.

5. MAINE NEEDS A NEW LAW!

I mention it above, but this deserve emphasis.  Maine needs to pass a law saying that if Nebraska eliminates its by district allocation of electors in 2020, and any of its district elections are in the minority, then Maine will do the same.  If Biden wins the Northern Maine elector, Biden wins even without Omaha.  But if Trump wins it — and Biden wins Omaha, and thus the election –what do you expect the bright-red Nebraska legislature to do: just sit there and take it?  No, they’ll take away Omaha’s delegate and cast it for Trump.  And because they’re the only state with one house in its legislature, they can do this and have their Republican Governor sign it within twenty minutes.  Maine, which has two Democratic houses of legislature and a Democratic Governor, should be pass this law right now, as it will take them more time.

6. UPDATE, 10/29

I’m incredulous at what the vote prediction sites are — and aren’t yet — doing.  Fivethityeight.com’s.forecast is the best of a sorry lot, in that (if you scroll down to the snake-like graphic) it shows that one of two scenarios can put Biden over the top (not counting faithless electors): either win Pennsylvania or win Arizona plus Omaha.  Everything else is a bunch of close races that — if you subtract 3% for reducing the Democratic vote — Biden easily loses.

• Pennsylvania (Biden up 5 points — but, you know…)
• Omaha (Biden up 4.5 points)
• Arizona (Biden up 2.5 points)

The site 270towin offers a consensus map showing Biden with 290 Electoral votes — whew, safe margin, right? — but that’s only because they show Biden winning Pennsylvania, Arizona, and Omaha!  If we win Pennsylvania, great — but what we should be guarding against is losing Pennsylvania.

The granddaddy of prediction sites, now called electoralvote.com, predicts a Biden win — granting him Arizona, Florida, Georgia, Iowa, and Pennsylvania — but doesn’t even have a separate prediction for Omaha!  Come on, folks!

I’m hearing pundits on radio TV snickering about Trump going to Omaha — but THAT’S THE RIGHT MOVE, IDIOTS!  Trump almost surely wins if there’s an Electoral College tie — and Arizona alone (where Biden isn’t even far enough ahead for comfort!) only gives Biden a tie!  Trump’s people understand that!

I hate to be a Cassandra, but — with a even a 2% adjustment  Trump is favored to win all of these swing states/contests.

• Ohio — Biden up by less than 0.1%
• Georgia — Biden up 1.8%
• Northern Maine — Biden up 1.2%
• North Carolina — Biden up 1.9%
• Florida — Biden up 2.0%

These electoral vote prediction sites are treating these as independent events and calculating the likelihood of Trump “running the table.” Probability theory tells us that, if they are independent events, he’ll probably win at least one — except that, in 2016, he didn’t!

Right now, I’d predict that:

• Trump wins all five of these nominal “Biden-leaning states”
• Biden does win Omaha
• The Supreme Court disqualifies enough Pennsylvania mail-in ballots to give that state to Trump, and
• Arizona goes down to the wire.

Happy ending: Biden wins 270 to 268.
Unhappy ending: Trump wins 279 to 259.

If you are young and healthy and free — rather than old, diabetic, and tie to a family like me — go to Maricopa County (Phoenix and surroundings) and help give us the loss-of-Pennsylvania insurance that we need!

(Hey Democrats: do you know who’s reading this and paying attention?  Republicans.)

(7) UPDATE 11/17

I forgot to finish the story last week, though that’s just as well: it has taken until this week for the boisterous clamoring for recounts and ballot disqualification to die down.  Biden won 306 electoral votes to Trump’s 252 (poetically, that’s exactly the same margin by which Trump beat Hillary Clinton in 2016.)  Trump did win all of the five close states listed above except Georgia.  Biden also won both paths I had outlined — winning Pennsylvania and Arizona plus Omaha — in addition to holding on to what turned out to be a swing race in Nevada.

What I published back on October 29, and updated between November 3 and 6, is now pretty much accepted wisdom.  The polls were deeply off — something I’ll eventually cover in the post with a manual stick shift on it — and now there’s a gathering sense that Trump is somehow a “special case” — so polarizing that his presence in a race distorts normal rules of politics, in part by leading Trump voters to simply refuse to engage with pollsters, whom they associate with the Big Biased Media.

In other words, they’re now arguing that there was an unacknowledged systematic error in the polls.  Where have faithful readers read that before?  Oh it might have been in this post, published the day before the election.  This is the one that I was spending time writing when I could have been taking apart the attacks on Gil Cisneros, so it came at a serious price — so you might as well read it!

The anomaly in this list is Georgia — which, given the overall shift due to systematic error, should have been a Biden loss.  That is  a testament to organizing in and around Atlanta — and particularly to the efforts of Stacey Abrams and her team to expand the electorate there.  (It’s also a testament to Georgia’s Secretary of State, a Republican, resisting demands that he throw out votes from Democratic-dominated counties — for which he is now being called a RINO.)

There’s one problem: this is mostly a problem of modeling, not of survey response rate.  Modeling is the “secret sauce” of polling — the art rather than the science — in which pollsters use their best judgment as to how likely various types of people are to vote.  If you’ve established how likely Republicans are to vote, you can make do with fewer Republicans in your sample by simply giving each of them a larger weight in determining the election outcome.  This may result in an “oversampling” of Republicans — sampling 500 to get 300 replies, where one would only would only need to sample 400 Democrats to get that number, but both estimates of how Republicans will vote would be be almost equally good above a certain number.  The problem with such a trick would, again, be systematic bias — that maybe those extra 100 people are different in some fundamental way from the 300 who did response.

Coda

Back when I was in the academic political public opinion community, one researcher — the rare Republican in that field, as I recall (who later co-edited a book to which I contributed an essay) — used to make fun of what was then called the “Shy Republican” hypothesis that Republicans were less willing to admit voting Republican than Democrats were to admit voting Democratic.  His point was that Republicans are perfectly willing to say whom they support and why, because they don’t see anything wrong with it — and indeed, see the earmarks of such views as a positive strength.

Some evidence for “shy Republicans” has been found over the years — people tilt more Republican in responding to Internet polls and to robocall polls than to live interviewer polls — but only in the past few weeks has there been a move towards what we might call the “Contemptuous Republican” hypothesis: that I won’t take your survey because “fuck you, that’s why!”

The contempt applies to both the media and the academics who inform it — which in turn is partly (not wholly) rooted in anti-Jewish animus (given both the actual and exaggerated perceptions that Jews are over-represented in both institutions) and bigotry against women people of color, who are held to be responsible for “political correctness” — both of which afford accusations of socialism, communism, “sexual deviancy,” hostility to religion (by which is meant Christianity, and not even all sects thereof) and more.  Not coincidentally, this is the kind of worldview that both Trump and right-wing talk radio have tried to foster and inflame.  (If you hear the name “George Soros,” you’ll know that you’re talking to one of these.)

How does a survey researcher deal with this?  There’s one way: try to come up with a question that everyone will answer, the answers to which can tell you whether people who are willing to proceed beyond that and people who refuse to do so are much different.

That question might be:

“Which American politician since World War II best represents your views on politics?”

That, I suspect, is the kind of question that will point researchers towards the systematic bias in polling — and, perhaps, towards getting things right.