Are you unhappy with the election results?

If you are thinking of leaving the country, and happen to be a college senior majoring in computer science who is interested in doing an English-language masters at one of the top universities in Europe (in beautiful and exciting Amsterdam), then click here.

Democrats Lost Because Democrats Didn't Vote

Many stories have been written (and are yet to be written) about why Hillary Clinton lost. Economists like the idea of "workers are hurting badly" (although see below). Social scientists would probably prefer the "there are a lot of racists out there" line. We're data nerds, so we have a different angle: Democrats didn't bother to vote this time. We even have a downloadable Excel spreadsheet to help make our case. Here are the data from the spreadsheet:

It's a big chart with lots of little numbers, so where do we start? The second and third columns are the Clinton and Trump raw vote in 2016. The fourth column is Clinton minus Trump, so positive means Clinton won the state and negative means Trump won it. From the bottom line we see that Hillary Clinton won the national popular vote in 2016 by 282,546 votes out of a total of 119.8 million votes or 0.2%. Trump's margins in Michigan, Wisconsin, and Pennsylvania were 11,837 and 27,357, and 68,236, respectively. If Clinton had gotten 107,430 more votes in these three states in the right proportion, she would have won the election. In other words, a change of less than 0.1% of the vote, properly placed, would have flipped the presidency. A Trump surprise it was, but a landslide it was not.

Columns five through seven are the analogous data for 2012. Column 8 is how many more Democrats voted in 2016 compared to 2012 as a percentage of the 2012 vote. So for example, in Washington 30.42% fewer Democrats voted for Clinton than voted for Obama in 2012, despite the U.S. population being over 3% larger in 2016. Washington Democrats, you are not doing your job. Column nine is the same thing for the Republicans. Here we see that North Dakota Republican turnout was 14.86% higher in 2016 than in 2012. North Dakota Republicans, you are performing your civic duty very well, congratulations.

Now, let's go back to the bottom row again. In 2016, 60.1 million Democrats voted, compared to 65.9 million in 2012, even though the population was 3.5% larger this year. That's a huge drop-off. In contrast, Republican turnout was down only slightly, from 60.9 million to 59.8 million this year. So our first conclusion is that the Democrats lost because 6 million fewer of them voted this year than last time.

The next question is, where did the dropoff occur? Was it, for example, largely in red states that have adopted stricter requirements to vote? The first thing we note is the champions in not-voting were Washington and California, two very blue states. The table is sorted on column 8, so we see the state where Democratic turnout improved the most is—Texas. What about the states that Obama won but Clinton lost? These are marked in orange in the last two columns. In Iowa and Ohio, Democratic turnout was way down and Republican turnout was somewhat up. Neither state had serious voting restrictions. The Democrats have only themselves to blame. If as many Democrats had voted in Ohio as in 2012, Clinton would have carried the state by over 100,000 votes. Wisconsin and Michigan have similar stories. If Clinton's turnout had matched Obama's, she would have won both states easily.

Pennsylvania is different. There, Democratic turnout didn't fall too much, but Republican turnout was up almost 9%. Finally, in Florida, we have one of only five states where Democratic turnout was better than in 2012 (and only three where it was higher when you correct for population growth). Unfortunately for the Democrats, Republican turnout improved by even more than Democratic turnout.

So, our conclusion is that Democrats lost because their turnout was down, rather than because millions of new Republicans suddenly decided to vote. What we can't see is why. In some cases, it could be due to restrictive state or local laws, in others it could be because supporters of Sen. Bernie Sanders (I-VT) decided both Clinton and Trump were corporate stooges and there was no difference between them. Maybe other reasons. We'll leave that for others to figure out. (V)

Was the Trump Voter Motivated by Economics or by Racism

One of the big questions of 2016 was what motivated the people who voted for Donald Trump, besides the little (R) after his name. One theory is that economically hard-hit workers liked his promises to fix the economy and create new jobs. Another is that many voters gravitated to his attacks on blacks, Mexicans, Muslims, and other minorities. The exit polls shed some light on the issue. One of the questions asked of voters was their annual income in six broad ranges as shown below. The numbers in the boxes are Trump's share of the vote minus Clinton's:

What we see is that voters making less than $50,000 per year strongly went for Clinton. Those voters making less than $30,000 per year went for Clinton by 12 points and those making $30,000 to $50,000 went for her by 9 points. Voters making more than $50,000 went for Trump. If Trump's base was primarily voters who are hurting economically (making less than $50K), we would have expected the poorer voters to support Trump. In fact, the reverse is true. This suggests that his attacks on minorities played a bigger role in rounding up voters than economic hardship. (V)

Other Key Findings from the Exit Polls

Chris Cillizza of the Washington Post has dug into the exit poll data and come to a number of conclusions as follows:

• Trump won the white vote by a record margin
• There was no surge of female voters
• There was no surge of Latino voters
• Education mattered yugely
• Trump did better with white evangelicals than Romney
• Trump didn't bring lots of new voters to the process
• The economy was a big issue—and Clinton won it
• This was a change election and Trump was the change candidate
• Obamacare was a wind beneath Trump's wings
• Trump's personal image was and is horrible
• Clinton's e-mail hurt her
• This was a deeply pessimistic electorate
• People didn't think Trump lost the debates as badly as I (Cillizza) did

In short, quite a few predictions (female surge, Latino surge) didn't pan out. (V)

Things Are Turning Ugly

Donald Trump is not actually the president-elect yet (he presumably will be on December 19, although maybe not—see below). He is, however, the president-elect in waiting. And in the 48 hours since he assumed that status, things have not been pretty. There was always a fear that November 9, 10, etc. would bring some very bad behavior, and that has come to pass.

To start, on the anti-Trump side, there have been massive protests across the nation. It is the right of citizens to gather peacefully, but in many places the protests have turned to violence or destruction. Buildings owned by, or associated with, Trump have been popular targets for vandalism, as have police cars. If history is any guide, such protests will peter out eventually, but they could easily resume on key dates in the future, like Inauguration Day.

Meanwhile, despite Trump's victory speech—in which he called for unity—many of the pro-Trump folks have, not unexpectedly, taken his success as license to act on their baser instincts. There is, for example, the middle school in Michigan where white students greeted their Latino classmates on Wednesday with chants of "Build the wall!" At another school in Pennsylvania, black students were subjected to the chant, "Cotton Picker, You're a N***er, Heil Hitler." There have been dozens of reported incidents of racial slurs, threats of violence, acts of sexual assault, and the like. One Muslim woman had a knife pulled on her because she was wearing a hijab, several other Muslim women have decided to discard the garments out of fear. One Latina was approached by an older, white man and advised, "I can't wait until Trump asks us to rape your people and send you back over the biggest damn wall we're going to build. Go back to hell, wetback." A woman in Pittsburgh was approached by three men in Pittsburgh who declared they were going to, "grab her by the p***y," and then proceeded to do so. There are many more examples at the "reported" link.

There is no question that Donald Trump has let this particular cat out of the bag by endorsing, legitimizing, and engaging in such behavior. But can he put the cat back in the bag? And will he work hard to try to do so? Those are questions we cannot answer right now. We can only hope that we get an affirmative answer in both cases, and sooner rather than later. (Z)

Classes, Exams Canceled on Wednesday

A number of professors, including ones at Yale, UConn, UCLA, Stanford, and other institutions, made last-second changes to their courses in view of Tuesday's results, given how upset their students were. Some classes were canceled, a few exams were postponed or otherwise reconfigured, and a handful of other concessions were made to relieve distraught young scholars of their obligations, as needed, for a day or two. Fox News picked up on the story, which has become—to many on the right—emblematic of the mamby-pamby individuals who run the academy, and their shameful coddling of a generation of young people.

This is a subject on which the authors of this blog are particularly qualified to comment. We are both college professors. One of us (Z) even teaches at UCLA, which is one of the campuses in question. We can understand the challenge these professors faced. Exams and other assignments are an instrument of measurement, intended to gauge students' mastery of course material. If a scientist knew that the thermometer was broken, or the lab was too cold, or a flask was dirty, or the control group had been tainted, he or she would take steps to correct the problem so that the measurements would not be inaccurate. An exam (or essay, or quiz, or other assignment) operates under the same basic logic. A teacher may not be able to concoct the perfect test or testing environment, any more than a scientist can concoct the perfect experiment, but they do the very best they can. And if an externality emerges that could taint the results of a test, that must be corrected for. It could be an outbreak of measles, or a copy of the test falling into the wrong hands, or the emotional impact of a campus shooting (which Z had to adjust for just a few months ago), or any manner of things.

If the inboxes of those professors at Yale, UConn, UCLA, Stanford, etc. were filled with messages of the sort that Z got on Wednesday—from students terrified their families would be deported, or that they would lose their health insurance and thus their cancer treatments, or that they would not be able to visit their families in Indonesia any more because they are Muslim and may not be able to get back into the country—then their decisions are defensible, even if not all professors would have reached the same conclusions. And those who are persuaded that the current generation of young people is somehow deeply flawed would do well to remember the complaint of Socrates nearly 2,500 years ago: "The children now love luxury. They have bad manners, contempt for authority; they show disrespect for elders and love chatter in place of exercise." (Z)

Did Hillary Clinton Have a 98% Chance of Winning?

The senior polling editor at Huffington Post, Natalie Jackson, is a bit embarrassed at having predicted a 98% chance that Hillary Clinton would win the election. She has tried to explain what went wrong. The structure of their model was just to look at the polls, with no corrections or adjustments for anything. The underlying averaging method was a Bayesian Kalman filter, which smooths out the bumps, but nothing else. She admits her problem: She put too much faith in the polls. But isn't that what a polling editor is supposed to do?

She also makes a valid point about the subjective choice of which polls to include. Among other choices they (and we) had to make is whether to include:

• Landline-only live interviewer polls?
• Robopolls?
• Internet polls?
• Partisan polls?

Each of these has a different mix of characteristics. Landline-only polls tend to overweight older, rural voters whose idea of a phone is something that sits on a table, has a large rotating dial on it, and is wired to the wall. But these were precisely the people who voted for Trump. Leaving them out because they didn't include all those hip youngsters with exploding Galaxy Note 7s meant important information was lost. If they had been included and there was a clear difference between landline-only polls and landline + cell phone polls, that would have been a clue that something was going on.

Robopolls are often landline-only because, by law, computers can't dial cell phone numbers. However, a pollster can hire someone to dial the cell phone numbers and then put on the computer. People sometimes respond differently to robopolls than to live interviewer polls. There is some evidence that Donald Trump did better on these polls than live interviewer polls, and Huffington Post's decision not to use these may have been a factor in being so far off.

Internet polling is a different can of worms because (1) it is impossible to randomly select a bunch of Internet users to interview, and (2) the set of people who use the Internet is not much like the electorate, requiring massive corrections to the raw data.

Partisan polls bring up another factor: Is the pollster lying? Maybe some of them do, but maybe some don't, and ignoring them altogether leaves out a potential source of information. (V)

President Ryan?

For the #NeverTrump crowd, it is not too late. There is still one way to stop Donald Trump from becoming president. He won 306 presidential electors on Tuesday, but he doesn't yet have 306 electoral votes. He has zero electoral votes at the moment, and won't have any until the electors vote on Dec. 19. Suppose 37 or more of the electors—many of whom were chosen by the state parties and not by Trump—have a meeting and all agree to cast their electoral votes for someone other than Trump, for example, Speaker Paul Ryan (R-WI) or Mitt Romney, or some combination of them and others. In some cases that might violate state law and require the elector to pay a small fine if prosecuted and convicted (which has never happened).

In this scenario, Trump get 269 electoral votes, Ryan, Romney, and the others gets 37, and Hillary Clinton gets 232. Since no one has a majority, the new House picks the president from the top three and could simply choose its speaker. It probably wouldn't even be that hard to pull off. There would be riots in the streets, though. (V)

Email a link to a friend or share:

---The Votemaster and Zenger