As I said in my initial post, looking at some of the numbers behind the election results, I've been wrong about a lot this election season. I was wrong that Trump couldn't get the nomination. I was wrong that he would lose in the general election, even against a weak candidate like Hillary Clinton.
I don't think this was entirely because I strongly dislike Trump as a person and as a leader, though I do certainly dislike him. I based my belief that he couldn't win in part on the fact that we've never before elected a president who lacked a track record in political or military leadership. But even more so, I based my assessment on the polls, which had consistently showed him losing to Hillary throughout the last year.
I focused on those polls heavily because in 2012, although I didn't go in for the full "unweight the polls" foolishness, I did allow myself to think that Romney/Ryan had a low but real chance of beating Obama. When they got trounced in exactly the way the polling averages had predicted, I resolved to believe the polls next time. Well, this time the polls were wrong.
We don't really know why the polls were so consistently off this time. I don't think that it's because the media was trying to hurt Trump. There's way too much at stake for polling companies in getting things right. If there were a major polling organization which had nailed this result while everyone else got it wrong, they would be getting massive amounts of credit, and predictive modelers thrive on that kind of thing. However, statistical modeling is hard, and the more we talk about polls in our election media cycle, the more we probably add to effects that throw off the polls. Given that many Trump supporters both distrusted the media and were being labeled as bigots by the media, it is perhaps unsurprising that they became more hesitant to talk to pollsters. Additionally, one of the dynamics during this election is that the demographics of who showed up to vote changed. Urban voters and minorities did not show up in as large numbers to vote for Hillary as they did for Obama, and rural and working class white voters showed up in greater numbers for Trump than they did for Romney or McCain. The polling companies based their demographic modeling on past elections and thus got the demographics of this election wrong.
However, there's another culprit for my blindness about this election, aside from my dislike for Trump and the polls being worng, and that's that although the election was decided in my part of the country (with Trump taking Ohio, Pennsylvania, Wisconsin, and Michigan) it was not necessarily decided by people like me. Take exit polls with a whole shaker of salt, given issues with polling this year, but the swings in support for the GOP presidential candidate this year as compared to 2012 are very interesting. Although Hillary got more votes than Trump among those making less than $50k/yr, the change in that level of support was huge. Trump did 16% better than Romney among those making less than $30k and 6% better among those making $30-50k. White college graduates swung 10% towards the Democrats versus 2012 (even though Trump still won them 49 to 45) and while white non-college graduates swung 15% further to the Republicans. Fewer Republican voted for the GOP nominee than in 2012, and more Democrats did. Married people swung 4% towards the Democrats versus 2012 (though married people still overall supported Trump by 53% to 43%) while unmarried people swung 10% back towards the GOP versus 2012, though a majority still voted for Hillary.)
So in my world, the world of college graduate, moderately high earning, registered Republican, married people, Trump had a pretty lousy year compared to Romney. And yet, in the wider world, the world that contains ALL of the people who actually vote in elections, Trump just surprised the people like me by winning.
It's not just that Trump won by racking up big margins with people who fall outside the 'elite' demographic, his election represents a rejection of the whole elite/professional paradigm in several ways. Think about the "how he won" pieces written about the Obama campaign eight years ago. Obama was credited with having come up with the political campaign of the future. He used social media. He hired a team of data analysts to build statistical models of where people likely to support him lived, send them advertising, contact them, and then call them to make sure they went out and voted. The age gap between McCain and Obama helped in painting the 2008 election as being between the past and the future of campaigning. The GOP tried to catch up, with Romney hiring a data team in 2012, but the Obama team continued to grow and innovate as well and once again their data prowess was credited with winning the 2012 election, despite the unpopularity of some of Obama's key accomplishments. In the world of political operatives, the story of the Obama administration was the story of the ultra-calm, professional candidate, "Mr. Cool", who ran a data driven campaign and thus could get the people to the polls when needed. Hillary Clinton inherited that campaign apparatus and took it even further, (though in the aftermath of her failure some are asking whether it may have misfired significantly.)
It wasn't just that Hillary inherited and expanded on Obama's data driven approach to campaigning. In many ways, Hillary Clinton epitomized the elite "meritocratic" illusion. She was a smart kid who went to the right schools, got her law degree, and was reputed among her intimates to be a hard worker eager to master all the facts about issues she cared about. And yet, Hillary, like many of the aspiring elites cranked out by top schools, does not actually have a very impressive set of personal accomplishments. It's Bill Clinton who was the "Man from Hope" able to win over crowds and convince voters to excuse his dishonesties. Bill won an election against a sitting president and brought the Baby Boomer generation to political power on his own political skills and personal magnetism. Hillary headed up the health care task force, but only because her husband appointed her to it. She held a Senate seat for New York, because her husband's political allies pulled every string to put her into a seat which is a complete shoo-in for whoever gets the Democratic nomination. She performed an undistinguished stint as Secretary of State, appointed to the post by the man who had comfortably defeated her in the Democratic primaries for the presidential nomination.
At the end of the day, Hillary's accomplishments are those shared by many in the 'meritocracy': she's smart and a hard worker, but even if her hard work is her own, the positions she has been given to do that hard work are more the result of her connections than of her own virtues.
But in the world of scientific campaigning, this was supposed to be enough. Hillary conquered her instinct throw fits and fire subordinates at signs of trouble and ran a no drama, Obama style operation. She hired all the best data analysts and ran a scientific campaign. The political pundit class all sagely opined that Trump would lose because he lacked the analytics to run a modern get-out-the-vote operation. I agreed with them. It seemed like Trump had learned none of the lessons of the last eight years.
What is more, Trump offended against another tenet of elite culture: expertise. He had never held a government position, and he showed none of the ability to study and grasp complex subjects which elites value in themselves. Clinton may have had bad policies, but she at least seemed to go about policy making the right way. We might not be able to count on her to make the right decision, but we could at least count on her to be wrong within normal parameters.
Frankly, I think that the elites are right on this, and it's one of the things that worries me hugely about the impending Trump administration. Clinton and her supporters made idiots of themselves running around claiming that she was "the most qualified qualified presidential candidate in history", as if spending eight years in the senate and four years as secretary of state is some kind of record breaking level of qualification. But she was at least somewhat qualified for the office and Trump is not.
This sort of argument plays well with professionals who are used to the idea that most people could not simply step in and do their jobs without lots of training. Most people would agree that being a doctor or an engineer requires some degree of training, but to distant eyes being a "boss" looks like it's pretty easy. Indeed, Dilbert, whose author became one of the more frustratingly vocal Trump boosters during the campaign, probably sums up pretty well what the average American thinks is involved in management. And yet, for those who actually work closely with upper executives (and I would surmise similarly politicians) there is a huge amount of skill involved in high level leadership. Not everyone can do it, and many who do do it do it badly. One of the several things that solidified me against Trump is that since I work with a number of high level executives in my company and others, I see many of Trump's personality and leadership traits as being those of the most unpredictable and frustrating type of executive.
However, most voters are not drawing on their personal experiences with upper management in evaluating presidential candidates. Trump told everyone he was a successful business owner, and after all, hadn't all of America seen him hiring and firing people on The Apprentice for over a decade? Obviously he had all sorts of leadership ability, and he wouldn't be like these dysfunctional Washington insiders who made all sorts of promises to the voters and then somehow failed to deliver on them once they arrived in DC. For voters who wanted to see things shaken up by a someone capable of getting things done, Trump looked like the ultimate candidate.
And why is it that the professional, expertise-based approach to leadership which is shared by policy and business elites with candidates like Clinton and Romney has so little appeal in the wider country? Perhaps because while the last twenty years have been very good for the professional elite, they have not been good for many middle class and lower middle class Americans. In a gradual shift which has passed some significant tipping points since the year 2000, automation and outsourcing have done away with a lot of "lower skill" jobs in the US. Professionals have continued to do the financial analysis, the marketing, the sales and the management, and they've done well in the process as companies have become flush with cash. But for those outside the growth industries and professions, the fact that consumer goods are cheaper and better than twenty years ago does not make up for the lowered career prospects.
In conditions such as this, to many voters "leave it to the experts" does not seem to be working very well, whichever party those experts belong to. Instead, people are looking for some kind of transformative change. Obama seemed to offer that, particularly to those who already leaned Democratic by reason of ethnic, family, or regional affinity. Sure, to other elites he may have looked like a cool and competent manager who used data in innovative ways. But perhaps that actually had little to do with why he won. Perhaps the vaunted data operation was not actually the key to his success. Perhaps the core reason that Obama won is that he deeply inspired ethnic minorities, and also gave hope to a lot of core Democratic voters, including working class voters who have traditionally leaned Democratic in their voting, particularly in states like Pennsylvania, Wisconsin and Michigan. That passion, rather than his data driven operation, brought Obama to the White House.
When Clinton tried to run the same operation, but with none of the passion, the votes did not turn out. Instead, Trump, with a message of "Make America Great Again" is the one who inspired passion. Some of that passion was in the same people who had voted for Obama. Union workers voted for Trump at higher rates than they had voted for any GOP candidate since 1988. After eight years during which it looked like politics was just another information age industry, something the smart kids could move into, master with data, and run as technocratic experts, perhaps it turns out that what matters is old fashioned inspiration. Reagan had it. Bill Clinton had it. 2000 was a toss up, but after 9-11, George W Bush had it. Obama had it. And this year, Trump had it. The appeal was only to certain classes and regions, while in other parts of the nation his election has instigated near panic, but it was there, and it was the reason why Trump ran instead of Clinton. As we move forward, the elites had better realize: expertise is not enough to win elections. Being smart, hard working, and checking all the right boxes along the career path will not hand you the keys to power. What makes or breaks a presidential campaign is the ability to deeply inspire a constituency to go out to the polls and vote. If neither side has a candidate like that, the best data operation may win. But if someone has "it", not matter how professional the other side is, the candidate with "it" is out ahead.
Thursday Random
13 hours ago
7 comments:
Brendan, I warned you and a lot of other people of this possibility for the last year. Because of your dislike of the man, you didn't listen. Now we're stuck with him. I preferred Newt or Carson, but I saw that the media preferred Trump. I tried to point out to you (and others) that the Wikileaks revelations indicated that Hillary's people, the DNC, and the press were working to favor Trump at the expense of other candidates. Again, you didn't listen, or seem to.
So, I switched parties to attempt to nominate Sanders, after more than two score years as a Pub. I warned you (and others) that the Wikileaks revelations indicated that Hillary's people and the DNC were working with the press to favor Hillary at Sanders' expense. Again, you didn't seem to listen. She secured the nomination.
And it seems that you got played by the dirt that Hillary appears to have dug up about Trump. I tried to warn you (and others) that the Wikileaks Podesta revelations indicated that Mr. P. had put out the call years before for any dirt on Trump. It appears that he obtained it, in the form of a taped conversation conducted by one of the execrable Bushes. You were more upset about the words used by Trump than the calumny created by Hillary against Trump.
Finally, I tried to tell you (and others) that the Podesta revelations also indicated that Mr. P. was suggesting ways that the polls themselves be gamed by the polltakers. There also appeared to be a massive cooperation by the press, who in the main were the people creating the polls.
Fortunately, enough other people listened to me and others like me so that Hillary did not get the election. I wish, however, that you had listened to me. While it gives me little satisfaction to say it, as I never wanted the man to be elected, I must nonetheless say it:
I. Told. You. So. Cassandra probably felt much the same way that I do just now.
Bernard,
You were definitely right that it was not impossible for Trump to win the nomination and again that it was not impossible for him to win the election. And I was wrong.
I do agree that the Dems (Hillary included) tried to give Trump a push in the primaries, and indeed found it quite plausible then too, I just didn't think that GOP electorate would allow themselves to be influenced. (And I think over time we'll gradually hear more about the extent to which WikiLeaks and it's push for the Donald were a Russian front operation. Whether that will work out well for them remains to be seen, but as with Brexit their side won.)
I have to admit, Gingrich and Carson were pretty much at the bottom of the heap for me in terms of GOP candidates. I suppose I would have voted for Gingrich is nominated. Carson I think is even more unqualified than Trump, except that he's not such a nasty personality. I suppose that counts for something, though. He probably could have been well managed by party handlers.
It would take a huge amount to convince me that the polling errors were primarily the result of Democrat manipulation. There are so many polls run by so many organizations with so many agendas that I just don't think it would have been possible to buy them all off for a year. I think some combination of a shift in who turned out to vote (which both determined the election and threw off the pollsters' likely voter filters) and the inherent difficulties of something like polling in our increasingly self conscious media age contributed to the miss. We're now dealing with a sort of Heisenberg Uncertainty Voter, where once you poll him he changes because you're talking to him (or decides not to pick up the phone because he doesn't want to talk to pollsters.)
But as you say, you did tell me that Trump could win, and I was wrong.
I'm not sure that we can conclude yet that the data-driven idea was wrong. One of the things that has become much more clear in hindsight is that Trump's campaign actually had excellent analytics -- far superior to anything the Clinton campaign had. Contrary to common wisdom at the time, Trump was spending millions on state-of-the-art analytics, and Trump internal polling showed him exactly where to campaign with most effect in the final weeks, while everyone else was baffled as to what he was doing.
It's possible, though, that a weaker conclusion might still be possible. The Clinton campaign seems to have been governed by analytics almost exclusively, to the point of ignoring advice that turned out to be correct -- local Democrats in Florida, Wisconsin, and Michigan kept insisting that there were serious problems with building turnout that needed to be addressed, and Bill Clinton kept insisting that the campaign should focus a strategy that might well have won Clinton the election, and both were ignored because of the analytics. Numbers give the illusion of being solid information, but they are really only as solid as the causal reasoning underlying them; because of that, they can't always outdo direct observation or solid experience when the latter are closely linked to the actual causal situation the numbers are about.
Darwin,
Thank you for this discussion. I confess to a little bit of Schaedenfreude as I have been enjoying the analysis, everywhere, of why she lost.
I do have to add to your excellent work two data points.
While I don't think there was any one thing that caused the election to go the way it did, I do think the overall gist of what you are saying is that Clinton listened to the statistics and Trump listened to the people. I did vote for him, but not enthusiastically.
Also, I live in a swing state, and stay home with my children, so I can say that we were getting AT LEAST 5 surveys, every day, up until the day of the election. And that started, modestly, around Nov. 2015. There are three eligible voters in my house and we consistently refused all surveys. It started out as a matter of principle ("my voice will be heard at the polls, not in surveys") and then finally because we were just sick of answering the phones. As I was watching the coverage on Election Night, it was funny to watch the pundits talk about people such as me who didn't participate in the surveys (I never lied, just refused.), complete with a facial expression as if they had just smelled something bad.
Maybe one lesson for future candidates is to rely less on polling, and more on what people are actually saying. You know, treat people as human and not as statistics. Kinda revolutionary, no?
Darwin,
Sorry for taking your chosen name in vain. I'll try not to do that again. Thank you also for the acknowledgement of my attempts to warn you and others of the possibility of Trump winning both the Pub nomination and the presidency. I never wanted to be right in either of those ways.
That said, I think that if you were to look at the evidence fairly, you would see, if not a conspiracy, a certain confluence among most journalists. The old saying was 'birds of a feather flock together'. The new statistical mathematicians have shown it only takes four rules for individual birds to follow in order to coordinate in that flock. And, as Mencius Moldbug has remarked, 'Cthulhu always swims to the left.'
Nonetheless, I would suggest that you read Le Wik's article on 'Journolist', which I think you will find to be most illuminating: https://en.wikipedia.org/wiki/JournoList. I have no reason to doubt that another such 'list' currently exists, which has not yet been 'outed'. At any rate, if you would also read the Podesta Wikileaks, or some of the summaries done of them, I believe that you would find a substantial cooperation between many members of the media and the Clinton campaign team.
I must also point out that one of the Podesta Wikileaks e-mails, dated January of '08, has a 32 page attachment detailing just how to game the polls. You can find that e-mail here: https://wikileaks.org/podesta-emails/emailid/26551 I have no doubt but that Podesta and his crew worked further with the media to fine-tune that game plan in the following eight years.
Bernard,
I don't have any problem believing that many reporters willingly connived with the Clinton campaign to slant their coverage. (It wouldn't surprise me if a few at laces like Fox and BrietBart worked similarly with Trump's campaign.) However, I don't think that the Clinton campaign would have been able to systematically bias polls. Again, polling organizations have too much stake in getting things right, and after all, there are GOP leaning organizations running polls too.
I think you are perhaps misinterpreting the WikiLeaks documents that you link to. The attached memo is a proposal for internal Democratic polling. The organization involved Atlas is a polling/consulting firm which specializes in doing research for Democratic campaigns and organizations in order to help them with their targeting, etc. Reading through, I don't see anything about manipulating public polls, just about how to use polling to target the persuadable voters in each state.
These are not the same polls which get reported publicly. The ones that go into RealClearPolitics, 538, etc are conducted by media organizations and research organizations. Some have a partisan lean and some don't. However, campaigns tend to be pretty secretive about the polls which they conduct internally -- often running them more frequently than public polls and weighting them based on what they believe the turnout demographic will be.
Why exactly would any campaign want to bias public polls in their favour? It seems to me that it would be much better to be very slightly behind in the polls rather than slightly ahead. If you're ahead, your supporters can take it easy, your donors can save their money, and those on the fence can think about third party candidates to send a message without actually letting the other guy win. If no one does anything, you should win anyway, so the temptation has to be to do nothing. If you're behind, your supporters all need to show up, your doners will recognize the need to give more now, and those on the fence will need to hold their nose and vote anyway because OTHERWISE THE OTHER GUYS WIN. The default becomes bad, so there's more incentive to try to change it.
I would bet that if the election were run again tomorrow, the proportion of Clinton supporters would increase. So, if there were a concerted conspiracy to make misleading public polls in their favour... why?
Post a Comment