Nate Silver has made a science of reading and interpreting the collective work of pollsters. His blog, fivethirtyeight.com, is a must read for anyone who wants to understand what the polls are saying, especially when they can seem to be pointing in different directions.
Case in point: the highly regarded Gallup tracking poll had shown a dead heat for several weeks until most recently, while other polls that do the same thing were showing President Obama with a four to six point lead over Mr. Romney in the upcoming presidential election.
Second case in point: when the Gallup poll finally “caught up” to the others last week in the results it was announcing, the Rasmussen poll showed Romney with a two-point lead.
Third case in point: earlier this week, an ABC-Washington Post national poll showed Obama leading Romney by only two points (well within the poll’s margin of error), when most of the others released at the same time showed Obama with a much more significant lead (i.e., outside of the margin of error).
Through it all, Mr. Silver, using a highly sophisticated “model” that consists of so many factors and measuring devices that just looking at his chart of the methodology he uses is enough to bring on a massive headache, daily posts his model’s prediction of the likely election result. At the start of this week, he had Obama with an 85.7 percent chance of winning, thereby leaving Romney with only a 14.3 percent chance.
If he’s correct, the activity and events that will unfold in the next month (including the debates that started this week; the continuing unfolding of the story in Libya, where U.S. Ambassador Chris Stevens was killed last month; the state of the economy, with a final unemployment report due out at week’s end; and any other “October surprise” or other event that might cause an upheaval in the perception of voters) would all seem to be preemptively irrelevant. Or, to put it another way, it’s over.
Except, of course, that it isn’t over, as Mr. Silver would be quick to point out, because all of the things that will happen (or might happen) over the next month can affect voters’ perceptions and decisions, as can totally fortuitous events like weather conditions on election day in various parts of the country and potential court decisions that can limit voter eligibility in certain states.
And the other reality is that, even with the aggregate of the current polls seeming to weigh so heavily in Obama’s favor, the fact is that the election isn’t decided on the basis of the overall popular vote. Just recall the 2000 election, when Al Gore won the popular vote (by over a half a million votes) but lost the election, if you need a reminder of that fact.
No, despite the fact that we call ourselves a democracy (democratic republic would be the more accurate description), it’s that weird entity that no one really understands called the Electoral College that will decide the election, and the math of that vote isn’t nearly as lopsided, even now, as that 85.7 percent probability figure would lead you to believe.
To be specific, at the same time that Mr. Silver posted that probability figure earlier this week, he predicted the following Electoral College result: Obama 320.1; Romney 217.9, which would hardly be the kind of margin you might intuitively expect to see with the President a prohibitive 6 to 1 favorite to win re-election.
It doesn’t make much sense until you realize how the Electoral College math works. Most of the nation’s 50 states will vote for either candidate by decisive margins (in the range of 60-40 splits). It’s just the way the country breaks politically these days. In some parts of the country (principally the nation’s west coast and its northeast) the tilt is toward the Democrats. In much of the rest of the country, it’s toward the Republicans. It hasn’t always been thus, but it is now.
As a result, that Electoral College math starts with Obama looking at somewhere in the neighborhood of 225 Electoral Votes and Romney securely holding about 190. The remaining 120 or so come from those nine or ten states that are closely divided, i.e., highly competitive. And in many of those states, the electorate is still unsettled. Or at least enough of the electorate in those states is unsettled that a swing of four or five percentage points could tip their Electoral College votes for Mr. Romney instead of Mr. Obama.
To be clear, that kind of a swing isn’t all that likely to occur with only a month left until the votes are cast. (Actually, a significant number of votes will be cast in the weeks leading up to November 6 via the early voting that many states allow and by the casting of absentee ballots that all states allow. But, for the sake of simplicity, let’s set that fact aside for now.)
But if, as I expect, the support for President Obama is broad but not particularly deep, then a relatively minor change in the electorate’s attitude could cause as much as a three or four point flip in the popular vote. And that kind of a change could easily put enough of the swing states back into play to allow Mr. Romney to become much more competitive.
So the caveat that needs to be attached to the very rosy-sounding predictions for Obama is that those predictions are based on maintaining that slightly above-the-margin-of-error lead in the national polls.
Thus can these pre-election polls be understood to be accurate in terms of what they reflect of the current mood of the electorate but not necessarily predictive of the actual Election Day results.
That great philosopher, Yogi Berra, summed it up pretty well when he said (when discussing a team’s seemingly dismal prospects in a baseball game), “It ain’t over until it’s over.”
The cautionary version of that thought, one that over-confident Democrats would do well to ponder, is “don’t count your chickens before they’re hatched.”