Monday, April 17, 2017

Listen: Trump's approval and reversion to the mean.

Trump's numbers will probably rebound somewhat, and you shouldn't be surprised when they do.

In statistics, there's a phenomenon known as reversion (or regression) to the mean.  The principle is simple: generally speaking and absent any continuous trend, any extreme metric will tend to become less extreme.  Another way of putting this is that there's a sort of center of gravity for most statistics, and they tend to drift towards that point over time.

To put this more concretely, imagine a tenth-grade teacher who has been working for ten years.  On average, that teacher's students tend to score slightly above average on the SAT.  If this teacher has a year when their students do exceptionally well, through whatever quirk of luck or fate, then it is likely that this teacher's students will return back to the average the next year.

I know it sounds very obvious when phrased that way, but it's an idea that it's easy to forget and hard to implement.  Take the recent United Airlines controversy.  You probably are familiar with it -- the New York Times summarizes it thus: "The video of Dr. David Dao, 69, of Kentucky, being bloodied as he was pulled off the flight in order to make room for four United employees has ignited conversation and outrage around the world."

Now, you'd be right to guess that this had an impact on the stock price for the company.  United shares plummeted the day of the incident, dropping from $72 a share to $67 a share.  Then it stayed around this level, dropping down a bit more to $65 within a few days.  It was tempting for people to write about how the incident had lowered the value of the company by $1.4 billion (the drop in price multiplied by number of shares, roughly).  But this was exactly the moment to look at the long-term average share price.  While technically true that the company had suffered a pretty severe hit, was it really likely that the price would stay at this point for the long term?

The long-term average for the company had hovered a bit above $70 a share since last year, and this is presumably driven by larger market forces such as labor, cost of fuel, etc.  Without a change in any of these essentials, and given the extremity of the drop, then, we should expect a reversion to the mean.  And sure enough, the fundamentals being basically the same, the price has now rebounded almost all the way back to where it was: $70 a share.

So what does this have to do with politics?  Well, this tendency of different sorts of trends dictates that extreme swings will tend to revert back to the mean in the long-term.  And in politics, partisanship is the average tendency of today's Americans.

President Trump started out his presidency with an aggregate approval/disapproval rating of 47.8%/42.5% at his high point, according to 538.  Considering he got 46% of the popular vote, that makes sense -- but it's actually a little unusual, since most presidents begin with approval ratings that extend past their own voters.  Obama's approval rating when he began, for instance, was 78%.  There's a widespread respect for the process and for democracy that combines with the very visible "fresh start" dynamic of a new presidency to boost the new chief executive during their "honeymoon" period.

Within two weeks, however, these numbers plunged to 44.8%/44.7%.  And by the end of the first month, Trump's aggregate approval/disapproval was 50.6%/44.1%.  And after two months, he was sitting at 53.6%/40.6%.

Of course, it's not unusual for presidents to experience a steady decline in support.  This is "political gravity" -- some voters become disillusioned by inevitable realities at the same time that the ineluctable hiccups of any major government erode the good-will of marginal supporters.

But political gravity usually takes time to assert itself.  Obama hovered around 60% approval for about 135 days, Bush stayed at 55% approval for about the same amount of time, and Clinton lurched erratically around between 50-60% for his first 110 days.

Trump's slumping poll numbers have been unusually terrible.  Presumably this is a reflection of his dramatic reversals on a host of key campaign promises, a demonstrable lack of results for many of the others, and a public and crushing failure on his first major legislative initiative.  And of course, it doesn't help that he began with historically low numbers -- he didn't have much of a cushion of soft support to lose, since he actively antagonized the press and soft Democrats.   But if we take a step back, we have to remember that not much time has passed.  It might seem like years, but he's only been president for 88 days.  And his most ardent supporters, just like the overt partisans in every presidency, have kept he faith.

As far as I see it, then, the bottom line is this: the partisan reflex and cognitive dissonance in Trump's initial supporters is strong enough to suggest that his numbers will bounce back a little.  Particularly at a time when he's taking highly visible and Extremely Presidential actions such as bombing the Middle East (a perennial favorite past-time of American presidents), there's a lot of potential for President Trump to bolster his support among marginal supporters and the disillusioned.  His current levels of support are at extreme lows for this time in a presidency, and so we should expect some reversion to the mean of partisanship -- particularly when he still has some cards to play.

So when Trump's approval rating increases somewhat, don't be surprised or dismayed.  We should expect it and plan for it, and none of the underlying dynamics are likely to change.

No comments:

Post a Comment