Documentation. Witnesses. Facts. Truth. That's what they're afraid of.

Monday, December 26, 2016

Post-Mortem Roundup, Week of 12/25/16

Lots of stuff to go through this week, though I'm guessing it'll be pretty soon that post-mortems will start to dwindle. At least Ramona Grigg hopes so:

All I ask is, no more postmortems. I don't want to rehash how or why Trump won. I don't want to hear that it was all Hillary's and the DNC's fault, or that anti-government voters wanted massive change, or that the racists won the day, or that Vladimir Putin and the Koch Brothers caused enough of a sneaky upheaval to cause half of America to go crazy and vote for an unqualified, ruthless carnival barker who lies with every breath he takes. I've read and heard it all.
We're not there yet. There's still new information coming out, and Ari and I will be working on ours for another month or two. Then we'll stop (mostly).

I was going to start by dissecting Nate Cohn's piece on the dissolution of the "Obama Coalition," particularly white working class voters (boy, am I sick of that term) in the northern Midwest. However, Ari beat me to it:

Nate Cohn's December 23rd column points to the cause of Clinton's loss as her inability to hold Obama's coalition together. His two pieces of evidence are:
1. Undereducated whites in Northern states voted for Trump in greater numbers than they did Romney.
2. Black voters in key Southern states did not vote at the same level as in 2008 or 2012.
The first point is wrong and the second completely misses the reason why.
Read more here.

Ari's basic assertion is that the 2008 "Obama Coalition" was more of a fluke caused by both the Bush economic collapse (which resulted in scenarios like this, which could only hold for so long), and because heck, Obama has the kind of charisma we really haven't seen in American politics in ages. If we're depending on that, we're in trouble... or not. The demographic progression John Judis and Ruy Texeira wrote about years ago appears to be real; it's just that as the groups that make up the Democratic coalition grow, Nixon's Southern Strategy, which won over the Southern working class whites from the Democratic Party to the Republicans, may taken hold among rural and exurban white in the North, or as Booman puts it, the Southification of the North. If that's the case, it ain't economic anxiety causing Rust Belt whites to vote more Republican. If it were, you'd expect nonwhite working class voters to trend in that direction, too. But it's not new for American white people in the lower-to-middle classes, who have been told for centuries that minorities are coming to take their stuff, to respond to that by digging in further.. In any case, 2016 was not the year the demographic bubble burst for the Republicans.

On Twitter, Elliott Lusztig's tells us why he believes that bubble is getting ready to burst soon:

Kurt Eichenwald disagrees; he thinks the Dems need to not lose the white working-class vote in order to win:

Hot off the presses, Ari points out that the "white working class" is, for the most part, too small to swing a presidential election based on addressing their issues alone. It is especially important to note that whites make up a small portion of the overall working class.
Of the 43.1 million people living under the poverty line, white people account for 28.6 million or 66%.  On the other hand, whites account for 79% of the people living over the poverty line. The CPS data does not breakdown age-race demographic combinations but we do also know that of the 43.1 million poor, 66% are above the age of 18.  If those people voted at the same rate as wealthier Americans (our first assumption), then we would be looking at about 15.7 million people under the poverty line that voted.  Since the CPS does not offer data for race-age subdivisions, we have to make a second assumption that the distribution of the poor over the age of 18 is the same as the distribution of the poor in general, such that 66% are white.  Multiplying that 15.7 million by 0.66 we find a total of 10.4 million voters here that make up the "White Working Class" - in effect, 7.6% of all voters.
Hillary turning Obama's 28-point margin of victory into a 12-point margin of victory, therefore means she lost 1.2% net votes to Trump. Compare that to the 2-3% that she lost because of Comey's interference and another 2-3% that she lost because of Putin's interference. And don't forget that a good sized chunk of that 1.2% may have flipped because of Comey or Putin. 
Booman doesn't quite agree with Ari (we're back and forth on this today), but regardless of whether the WWC vote is necessary to bring back to 2008 levels to win the Presidency back, there are other considerations:

But they haven’t been voting Democratic because they agree with us on pluralism or America’s proper role in the world or how to conduct diplomacy or the importance of science-informed education and policy or the importance of female autonomy and empowerment or the minimum requirements of temperament and experience expected of a president. They are probably more inclined to support the police than Blacks Lives Matter, don’t give a damn about climate change, and are conservative about gay rights. They voted for Obama despite all of these differences from Obama’s metropolitan “coalition of the ascendant.”
What this election did was cleave these voters from the Democratic Party even as the opposite thing was happening in our suburbs.
But the most threatening thing, in my view, is that too many of the suburban voters who abandoned the GOP's presidential candidate wound up sticking with downticket Republicans. In other words, it isn't an even swap.
Worse, even if it were an even swap, it would still be a bad trade due to the demographic dispersion of the vote. The Democrats will continue to lose most districts in this country if all their strength is confined to cities, suburbs, and college towns, and that means Republican dominance in state legislatures and the U.S. House of Representatives for as far as our eyes can see.
Through partisan redistricting, like it or not, Republicans have handed the keys to Congress to rural and exurban white voters, and we can't change that until 2021 at the earliest. So in the meantime, what can we understand about their needs?

Ian Reifowitz says there's a kernel of truth this time to jobs moving from whites to minorities, but that's based on location:

Let’s start with the topline stuff. This data, produced by the independent Economic Cycle Research Institute, covers the past nine years since the peak in employment we hit in November 2007, i.e. before the Great Recession. Before I dive in deeper, this sentence from the ECRI’s report sums it up:
Whites actually have fewer jobs than nine years ago, while Hispanics, Blacks and Asians together gained all of the net jobs added, and more.
ECRI explained the importance of place, and how it affected the data on race and employment:
Part of the reason may be that these jobs, predominantly in services, were created in metropolitan areas, rather than in rural areas and small towns where factories were shuttered as the manufacturing jobs disappeared. There is little reason to expect that those jobs are coming back to those areas away from the urban centers.
Looking at the numbers, metropolitan areas gained jobs over the past nine years, 5 percent more than they had in 2007, while the rest of the country shed 2 percent. In that sense it has been a lost decade for large geographic swaths of the country, in particular rural America—which has seen death rates increasing among rural whites, especially women, with suicide and drug overdose causing just about all the spike among younger people.
All these things connect. They help explain why Trump won enough votes in non-urban parts of Pennsylvania, Michigan, and Wisconsin to put him over the top. Overall, among whites without a college degree, Trump won by 39 points. More importantly, he outperformed Mitt Romney by 14 points among that group, while Hillary gained 10 points among whites with a degree compared to President Obama four years ago. Even if the exit polls are off, they’re not off by enough to undercut what those figures tell us about the white working class vote.
Whites are losing jobs while others are gaining them because the jobs have been moving to the cities and out of the rural and exurban areas that used to be  "company towns."

"Icallbs" explains why, despite the fact that Hillary would likely help those voters more (to the extent they can be helped), they voted for Trump:

In the end it is not about the solutions that Trump is or is not providing, it is about his attitude toward minorities and anger at the “elite” that attracted them. The black President oversaw an economic recovery that benefited urban areas and minorities. As they see it, he helped “his kind”. Hillary is an educated woman who they felt would do the same, no matter what she said. They want someone instead who gives them permission to lash out at these people and provides them empty promises that the white rural working class will rise again without them having to adapt to accommodate technology, diversity and a new economy.

But according to Kevin Drum, we wouldn't even be talking about any of this if it weren't for the Comey letter:

Let's add this up:
  • Trump gained 0.9 + 3.1 - 1.7 = +2.3 percent
  • Clinton gained -0.9 + 2.3 - 3.1 = -1.7 percent
The October poll ended on the 24th. FBI Director James Comey released his infamous letter on the 28th. The November poll then showed Hillary Clinton with a net loss of 4 percent compared to Trump. This compares to net movement of only a few tenths of a point in the final days of the 2012 election. 

Nate Silver agrees:

No automatic alt text available.

Russian interference (covered up by Comey) likely also affected the race negatively for Clinton. Say that's worth another point in either direction, and without two incredibly unlikely occurrences, not only would Hillary have won, her raw numbers could have approached Obama's 2008 numbers and Trump's could have been similar to McCain's -- in other worse, if Russia and the FBI had stayed out of our election, not only would Hillary have won, but she would've won by quite a bit, potentially carrying a couple of the closer Senate seats as well as some additional House races. But it wasn't to be. However, it may indicate that Trump performed near the top of the range of where a Republican can, and Hillary was near the bottom for a Democrat -- in the final weeks, a worst-case scenario happened to her, and yet she lost by 70,000 votes in three states.

But perhaps we should've seen something like that  coming. We know that Republicans play dirty tricks and then use the media to amplify them --- in the last few cycles, we've had "Al Gore Said He Invented the Internet," Swift Boat Veterans for Truth, and Birtherism. In the cases of Gore and Kerry, those were Rovian "attack their strength" approaches (probably the nearest equivalent this time was the Clinton Foundation, which should've been a real boon for her). But this is the first time in decades (though certainly not the first in the last century) that a political party has colluded with a hostile foreign government or one of our intelligence agencies (so far as we know) put its thumb on the scale of our election. And Oliver Chinyere, a former Clinton staffer, said they tried to warn us:

The KKK/Alt-Right/Men’s Rights Activists now see their hero as leader of the Free World. Racism and xenophobia won too, but so did stupidity. Now that Donald Trump is breaking from decades of bipartisan diplomatic foreign policy and US underwater drones are being stolen by the Chinese as we watch, remember there were people telling you this would happen. But back then, everyone seemed more concerned about potentially finding another killer risotto recipe in John Podesta’s emails so…
One thing we haven't considered enough is that, as inconceivable as it is to some of us, Donald Trump actually appealed to some people (scary, I know), or at the very least, could be looked at as a viable President who wouldn't be a major threat to the American way of life. Slate's Michelle Goldberg has a number of quotes obtained from a focus group Planned Parenthood held with Trump supporters.

This leads to an obvious question: If these women think defunding Planned Parenthood is a deal-breaker, why did they vote for a candidate who promised to do exactly that? After all, in a September letter addressed to “Pro-Life Leaders,” Trump pledged to strip Planned Parenthood’s federal funding unless it stops performing abortions. But many of the people in the focus groups didn’t know he’d made this assurance, and those who did didn’t take it seriously.
Goldberg concludes:
If Democrats ever want to regain power, they don’t need to wedge Trump away from the Republican Party. They need to yoke him to it. These voters might be OK with Trump talking about grabbing women by the pussies. What they didn’t know is that they were voting for the federal government to do it.
Yes and no. One of the problems here is that many people just can't believe that certain things can happen -- overturning of Roe v. Wade, elimination/privatization of Medicare and Social Security. Many came of age in the '80s and '90s when those really were third rail issues. Unfortunately, I think those people are wrong. And they voted accordingly -- many even liked him -- his status as a TV celebrity helped.

And finally, I'm not going into detail on this now, but it seem like SurveyMonkey's blog about the election results might be a great resource to tap into for the next few weeks. A taste:

A full accounting for a range of mis-estimates remains months away with the still-to-come release of 2016 voter files and government surveys, but our own, preliminary look at the performance of SurveyMonkey Election Tracking provides an initial take on some of the possible sources of error. Here, we share our initial conclusions about what we did well, what not so well, and why what we’re learning makes us optimistic about the future.
The mission of the SurveyMonkey’s Election Tracking project is provide a platform to collect raw data that can be used by clients in the media and beyond to apply their own weighting, analysis, and interpretation. We also wanted to test whether our approach — recruiting respondents from among the three million people a day who complete a survey on SurveyMonkey — could provide a solution to the challenge of single-digit response rates that relies on traditional polling theory instead of less-tested modeling techniques. 
We went a long way toward accomplishing the goals for 2016 by tracking election preferences at the national level all year and for 97 separate state level contests (including the presidential, senatorial, and gubernatorial races) from Labor Day through Election Day. We offered multiple media partners interactive cross-tabs and the ability to select from multiple weighting schemes and filter the data by registered or likely voters. 
Yes, like almost every other poll out there, our final numbers pointed to a likely Clinton victory, but our data revealed important state-level dynamics.

Mmmm... nerdy.