Chill

     

       Today, I want to talk to my Democratic friends.  If you’re a Republican and a Trump supporter, you’ve got the day off.  What I’ve got to say today will be of absolutely no interest to you.

            It’s time for the freak-out over Hillary Clinton’s poll numbers to stop.  Put the worry beads away.  And stop thinking about the logistics of moving to Canada.  Donald Trump is not going to win this election.

            I know.  You’re obsessed with the election forecasts on FiveThirtyEight.com and Nate Cohn’s Upshot column in the New York Times.  You probably look at both first thing in the morning, at lunch and just before you go to sleep at night.  And you’ve been watching incredulously as Hillary Clinton’s odds of winning the election have slowly eroded.

            What you need is a reality check and some perspective. 

            Here’s the reality check.  We live in a country that is split pretty evenly between Democrats and independents who lean left on the one hand and Republicans and Independents who lean right on the other.  And, let’s face it, Hillary Clinton’s candidacy comes with a whole lot of baggage.

            Both Donald Trump and Hillary Clinton got bounces after their conventions over the summer.  Hillary got a bigger bounce than Donald Trump did, but that’s neither here nor there.  We call them bounces because public opinion goes up and then it comes down.  The race has settled down to about where it was in the Spring.  Hillary Clinton still enjoys a small but significant advantage over Donald Trump.   FiveThirtyEight.com and the New York Times are still saying that despite Clinton’s slide in the polls, she has a substantially better chance of winning the election than Donald Trump has.

            Here’s the perspective.  Take a look at this graph from RealClearPolitics.com.  This shows you aggregated polling numbers for the 2012 election.  You probably don’t remember this, but President Obama’s reelection was no sure thing.  As of September

21, 2012, Barack Obama held a polling lead of only 3.5 percentage points.  As of the day before the election, the polls were showing an extremely tight race.  Obama led by only 0.7 percentage points. 

            On Election Day, though, Obama beat Romney by a 6 percentage point margin.

            There are a number of lessons to learn from this.  The first is that even if you aggregate polls as RealClearPolitics.com does, the results are still only probabilities.  Polls can only tell us only what particular people said at a particular moment in time.  The math we use to impute what we learn from a relatively small subset of people to the entire population of voters has its limits.  The best we can do is use that math to come up with an estimate of what is really going on in the population at large.  And that estimate is going to be subject to a margin of error that we can never reduce to zero.

            The second is that unless you want to believe that something miraculous happened between November 5 and November 6—and, being liberals, you probably aren’t that much into miracles—you need to look for an explanation for the difference between the predicted margin of 0.7 and the actual margin of 6.  And, there are at least two.

            First, you may have noticed that the polls have seemed to respond to the news.  That’s kind of puzzling, since by now, most people have made up their minds about who they’re going to vote for.  According to FiveThirtyEight.com, there still seems to be an unusually large chunk of voters who haven’t decide yet.  But given the fact that Clinton and Trump are extremely well known, it’s likely that these people will never resolve their ambivalence about these two candidates and that they just won’t vote at all.  The volatility we are seeing in the polls may be coming from this group.

            Political science tells us rather definitively that something like 90% of people who are Democrats or Democratic leaners will eventually vote for their party’s nominee, and the same is true of Republicans and Republican leaners.  Our politics has become increasing tribal.  This election is not a grand policy debate.  It’s a football game, and we just want our side to win.

            If our politics is largely tribal and most people have already decided who their going to vote for, then you’d wouldn’t expect many people to change their minds definitively even after a bad campaign week.  Instead, you’d expect public opinion to be relatively stable.

            What may be going on is known as “non-response bias.”  Writing in The Daily Kos, David Jarman explains that when a candidate has a bad series of days on the campaign trail, his or her supporters may temporarily become somewhat demoralized and not want to talk to pollsters.  When that happens, the polling results reflect an oversample of the other candidate’s supporters.

            You can see how this might work if you look again at the Real Clear Politics graph.  The first presidential debate between President Obama and Mitt Romney took place on October 3, 2012.  As you may recall, the president’s performance was--how to say this--AWFUL.  The day before the debate, the tracking poll had the president leading Romney by 3.3 percentage points.  By October 8, Obama’s polling lead had collapsed to 0.5 percentage points, and by October 15, the day before the second debate, the president’s polling lead was only 0.1 percentage points.  According to Jarman, pollsters who were using more sophisticated techniques “found that there really wasn’t much of a debate effect at all, and the race stayed in pretty much the same narrow band from April on.”

            Clinton got a strong bump after her convention, and its effects were probably prolonged as Donald Trump had a number of unforced errors, including a stupid fight with the Khans, a Gold Star family who had lost a son in the war in Iraq.  You can well imagine that Trump supporters might have been demoralized by all of this and how they might not have wanted to talk to pollsters.  Trump’s mistakes were serious, and that may explain why Hillary Clinton’s probability of winning the election began to flirt with 90%.

            But what goes around comes around.  Trump hired a much more competent campaign staff that obviously managed to impose some discipline on him. He started using a teleprompter, and for a stretch, he stopped making unforced errors.  Far worse, though, more Clinton emails surfaced, Trump’s campaign reared up in righteous indignation over Clinton’s description of a portion of Trump’s supporters as “deplorable” and “irredeemable,” and Clinton was less than forthcoming about the case of walking pneumonia that caused her to collapse in the heat at the September 11 Commemoration in New York.  Clinton’s polling numbers fell back to earth.

            But a second explanation is the fact that modern elections are all about who gets their voters to the polls on Election Day.  Obama, in 2012, knew exactly who his marginal voters—people who generally liked him but might not have prioritized voting--were and had built an organization that could mobilize these voters on Election Day.  The Romney operation was a shadow of the Obama machine, and just didn’t have the wherewithal to do the same thing.  As a practical matter, Democrats tend to live in much more densely packed communities, and so it’s easier to get them to the polls than it is to mobilize Republicans, who like live in less compact geograpies.  The difference between the Real Clear Politics polling average and actual turnout could merely reflect the difference between the marginal Democratic voters who expressed an intention to vote and were mobilize by the Obama machine and the marginal Republican voters who said they were going to vote but never made it to the polls.

            Clinton has spent the better part of the last two years learning from Obama’s experience in 2008 and 2012.  She’s got an army and she’s running it with state of the art analytics.  In comparison, Trump’s team looks like a group of neighborhood kids who have gotten together to play pickup football on a Sunday afternoon.

            Listen, I know you want this to be easy.  Nobody likes to take chances, particularly with something so important.  But even if Clinton’s numbers were still brushing up against 90%, that still wouldn’t be a sure thing.  When her probabilities were that high, Nate Cohn was still comparing her possibility of losing the election to the possibility that a NFL field goal kicker would miss from the 20 yard line.  But as the 2012 Baltimore Ravens will tell you, sometimes field goal kickers do miss.

            So go pour yourself a nice cup of chamomile tea and relax for a few minutes.  Clinton should win this election. My prediction, though, depends on how well Democrats do the nitty-gritty work of getting people to the polls.  I’m assuming that Clinton’s machine will do its job effectively, but if you want to reduce the risk that Clinton will lose, get up off the couch and write a check, or, better yet march yourself down to Clinton headquarters and put in some time canvassing or phone banking.

            Feel better?  Good.  Now be a dear and invite our Republican friends back into the room.