A wonderful comic recently pointed out that with so few presidential elections in U.S. history, something new always happens. But apparently not everyone got the message, so National Review’s Dan Foster writes:

Is there precedent for an incumbent president being down three in the major nat'l tracking polls 10 days out and winning?

Here’s the thing: In the polling era, at best the only precedents for an incumbent president having a close reelection contest are Harry Truman in 1948, Gerald Ford in 1976 and George W. Bush in 2004. Of those, Truman really precedes the polling era, so we’re really only talking about two cases. Every other president since Truman who made it to the general election either won easily (Eisenhower, Johnson, Nixon, Reagan, Clinton) or lost decisively (Carter, George H.W. Bush). 

In other words, there are presumably all sorts of unprecedented things that will happen to Barack Obama this time around, because there are so few relevant precedents.

By the way, the group with closely contested reelections are easily distinguished by their approval ratings, which are right around 50 percent.

Now, in the event, Foster is misreading the polls, with the eight (!) national tracking polls showing a close split, and the overall polling averages also showing what amounts to a dead heat: As I write this HuffPollster has Obama up by 0.1% while the poll-of-polls at Real Clear Politics says Mitt Romney leads by 0.9%.  And it appears that Obama has an Electoral College advantage, with (per Pollster) a solid lead despite the overall tied contest; if this apparent split turns out to be accurate, Obama could lose the national vote by somewhere between one and three percentage points and still win the election.   

But the main point here is: These kinds of questions just don’t make sense. Anything premised on a close race for reelection to the presidency just doesn’t have enough past cases to draw any conclusions at all. Similarly, if anyone asks whether a president can be reelected with a certain level, say, of unemployment, the most likely answer is: We can’t know! There aren’t enough examples!

That doesn’t mean we can’t extrapolate from the current numbers, of course; careful observers can learn from various other elections and find patterns, for example, of how undecided voters are likely to split. But be very wary of people who assert that something in a presidential election is unlikely to happen because it hasn’t happened before. If there’s one thing we know for certain, it’s that every single presidential election breaks dozens of “never been done befores.”