There’s a lot of lamentation going on about how Biden had no coattails. That all the talk of progressive priorities is what caused the Democrats to loose seats in the House and prevented them from gaining/flipping seats in the Senate. This was, apparently, part of some screaming at today’s Democratic House caucus Zoom call and some tweets by AOC and the lying shitbird Waleed Shahid. And, of course, how the pollsters all screwed up once again. While I’ve got a couple of specific points on the pollsters, which I’ll save for the end, here’s what I think got missed and ignored. And it got missed and ignored not just by the pollsters, but also by the modelers: Cook, Sabato, etc.
On Sunday night I wrote this:
The political science PhD part of me, in line with Charlie Cook’s projections, looks at the data and information we have and recognizes the high probability that VP Biden wins without much difficult and the Democrats are able to achieve a 52 seat Senate majority. The low intensity warfare professional in me looks at what’s going on and is exceedingly concerned regardless of what happens on Tuesday.
Let’s leave low intensity warfare professional me out of this, as he’s on solid ground, but I clearly got it wrong. And I got it wrong because Cook got it wrong. And Cook got it wrong because the data he had was wrong. And the data he had was wrong because the assumptions underlying the collection of the data were wrong based on a misreading of what happened in 2018.
What everyone missed is that in 2018 if you were pissed at Trump, uncomfortable with Trump, disliked Trump, were concerned about what Trump was doing and you wanted to cast a protest vote against Trump you could only do so by voting for Democrats and against Republicans. And that’s what caused the 2018 blue wave for the Democrats in the House races. And it’s why it didn’t reappear this time.
It didn’t reappear this time because Trump was on the ballot. So if you are pissed at Trump, uncomfortable with Trump, dislike Trump, are concerned about what Trump is doing and wanted to cast a protest vote against Trump you actually could directly vote against Trump and then go and vote how you normally would, how you’re comfortable voting, which was for all the other Republicans just like you always do because you’re a Republican. That’s your team. That’s your tribe. That’s how you identify. And that’s the lens you understand politics and America through. Yes, this means that all these people angry with Trump who voted against him, but voted for all the other Republicans are completely oblivious to the fact that all of those folks enabled Trump, went along with Trump, embraced Trump, refused to stand up to Trump, refused to cross Trump, and are, therefore, just as bad as Trump. They may in fact be worse because they’re not Trump and therefore should both know and be better.
Congressman Max Rose should never have even gotten close to being elected to Congress from his district in Staten Island. That’s not because Max Rose is a bad member of Congress or a sellout or anything else. It is simply that Max Rose, because he’s a Democrat and because he’s Jewish, is a TERRIBLE fit for that district. It shouldn’t be surprising that Staten Islanders, who would normally only vote for a Republican, could actually protest vote against Trump in 2020 by actually voting against Trump in 2020. And that they were then perfectly willing to go back to voting for the Republican running to unseat Rose. Which is what they did.
What the pollsters missed, what the modelers like Cooke missed, and what, as a result, allowed us all to miss because we were working off of their assumptions and data and models is that there does not appear to have been a major shift among suburban white women, and some suburban white men, from the Republicans to the Democrats because of Trump. What there appears to be is that suburban white women and some suburban white men don’t like Trump and will vote against him if he’s on the ballot. When he’s not, they will vote against Republicans as a protest because they can’t directly vote against him.
This is not a fully original theory. My thinking was prompted by this tweet from Dave Wasserman:
A key factor in House Rs' ability to easily hold onto districts that Trump likely lost last night: in 2018, the only way for suburbanites to take out anger at Trump was down-ballot; this year, they could vote against Trump and still vote for down-ballot Rs they like.
— Dave Wasserman (@Redistrict) November 4, 2020
I think there is empirical goodness here that we need to keep in mind going forward.
I said up top that I wanted to make a specific point about the pollsters. I read today, actually reread because I saw someone flag a 2012 article on election technology, polling disparities, and election rigging, and this stood out at me:
The statistically anomalous shifting of votes to the conservative right has become so pervasive in post-HAVA America that it now has a name of its own. Experts call it the “red shift.”
The Election Defense Alliance (EDA) is a nonprofit organization specializing in election forensics—a kind of dusting for the fingerprints of electronic theft. It is joined in this work by a coalition of independent statisticians, who have compared decades of computer-vote results to exit polls, tracking polls, and hand counts. Their findings show that when disparities occur, they benefit Republicans and right-wing issues far beyond the bounds of probability. “We approach electoral integrity with a nonpartisan goal of transparency,” says EDA executive director Jonathan Simon. “But there is nothing nonpartisan about the patterns we keep finding.” Simon’s verdict is confirmed by David Moore, a former vice president and managing editor of Gallup: “What the exit polls have consistently shown is stronger Democratic support than the election results.”
Wouldn’t American voters eventually note the constant disparity between poll numbers and election outcomes, and cry foul? They might—except that polling numbers, too, are being quietly shifted. Exit-poll data is provided by the National Election Pool, a corporate-media consortium consisting of the three major television networks plus CNN, Fox News, and the Associated Press. The NEP relies in turn on two companies, Edison Research and Mitofsky International, to conduct and analyze the actual polling. However, few Americans realize that the final exit polls on Election Day are adjusted by the pollsters—in other words, weighted according to the computerized-voting-machine totals.
When challenged on these disparities, pollsters often point to methodological flaws. Within days of the 2004 election, Warren Mitofsky (who invented exit polls in 1967) appeared on television to unveil what became known as the “reluctant Bush responder” theory: “We suspect that the main reason was that the Kerry voters were more anxious to participate in our exit polls than the Bush voters.” But some analysts and pollsters insist this theory is entirely unproven. “I don’t think the pollsters have really made a convincing case that it’s solely methodological,” Moore told me.
In Moore’s opinion, the NEP could resolve the whole issue by making raw, unadjusted, precinct-level data available to the public. “Our great, free, and open media are concealing data so that it cannot be analyzed,” Moore charges. Their argument that such data is proprietary and would allow analysts to deduce which votes were cast by specific individuals is, Moore insists, “specious at best.” He adds: “They have a communal responsibility to clarify whether there is a vote miscount going on. But so far there’s been no pressure on them to do so.”
Some argue that the Democratic victories in 2006 and 2008 disprove the existence of the red shift. However, this may be a misinterpretation of complex political upheavals that occurred in each of those election years.
While Democrats won a majority in the House of Representatives in 2006, and the White House in 2008, postelection analyses did in fact suggest extensive red-shift rigging. But in both election cycles, these efforts simply failed to overcome eleventh-hour events so negative that they drastically undercut the projected wins for the G.O.P.
In 2006, it was the exposure of Republican representative Mark Foley’s sexual advances toward male congressional pages, and the long-standing cover-up of his behavior by G.O.P. leadership. The scandal swirling around the outwardly homophobic Foley broke in a very ugly and public way, engulfing the entire party and causing a free fall in its polling numbers. The Democratic margin in the Cook Generic Congressional Ballot poll, which had been at 9 percent in early October, jumped to 26 percent by the week of the election.
The collapse of Lehman Brothers months before the 2008 elections had a similar effect on John McCain’s numbers. Pre-election polls showed that the American public blamed the Republicans for the imploding financial markets. “These political sea changes swamped a red shift that turned out to be under-calibrated,” argues Jonathan Simon, who speculates that Barack Obama actually won by a historic landslide, driven by an overwhelming backlash against the policies of the Bush Administration.
There’s a lot more important, solid, and infuriating information at the link! I think the take away is that there has not just been a problem with the professional polling, both in the run up to elections and the exit polling, that is reported in the news media and that is being relied on by everyone to understand what the dynamics are and what is going on, but that the problem is because the pollsters are cooking their own data so it doesn’t make it look like their data is wrong. No one trying to do good faith modeling or projections can do so if the data they are getting is itself misinformation.
A penultimate point about how the reporting of the results is being done. The national news media, especially the campaigning and political reporters have defaulted to their traditional and comfortable framing. If PA, WI, MI, and GA had been allowed to count mail in and early vote ballots like Florida or Ohio are, then by 1 AM at the latest on Wednesday morning we’d be reading about Biden’s historic win, with more of the popular vote than even Obama, that restored the blue wall in the midwest, and that GA had turned purple or was on the cusp of it. But because it’s dragged out, even the best reporting teams that were warning everyone not to be concerned if we didn’t know anything until Thursday or Friday, have now all reverted to horse race reporting, the Democrats are in disarray, why couldn’t Biden do better framing. In reality, Biden did amazing, but the narrative must be observed!!!
The final point I’ll make is that the Senate is still in play, even if the path for Democrats is narrow. Despite everyone being loath to call the Arizona race for Mark Kelly, the margin is too great for McSally to overtake and defeat him. This leaves four Senate seats outstanding: Alaska, both the Georgia seats, and North Carolina. Alaska, even if we didn’t have acknowledged and recognized issues with polling, is notoriously hard to poll. Dr. Gross does have a path to flip that seat, but I have no idea how probable that might be and I don’t think anyone else being honest does. While I’d like to see the Democrats take both the Georgia seats, I think the likelihood is that Warnock defeats Loeffler and Perdue narrowly hangs on, but Ossoff has run better than I think anyone expected, so the seat is in play. As for North Carolina, the state stopped reporting updates because of the Federal court rulings and agreements about how to handle the ballots that are allowed to come in after election day. It is possible that Tillis’s lead has grown, that it has eroded, or that nothing has really changed at all. But we won’t know until next week. The goal for the Democrats is now a 50-50 Senate where they get the majority because a Vice President Harris breaks all the ties. But it is also probable that McConnell maintains a 51 or 52 seat majority. If that happens, expect Schumer to offer Murkowski, Romney, and even Collins whatever they want to not go along with McConnell’s obstructionism.
Open thread!
PS: I told you all for over two months that unless Biden was up by at least 6, if not 10 points in the final Florida polling he was going to lose Florida by 1 to 2 points because there’s something funky going on with the polling in Florida that I’ve noticed as a pattern over the years. I am sorry to write that my concerns were correct.
Marking Beliefs To Market: What I and Everyone Else Missed Or IgnoredPost + Comments (436)