re: The idea of Undecideds breaking for the Challenger

[cthulu2016]

There have been several things written about this the last few days—Nate Silver, Josh Marshall, and others.

The conventional political wisdom is that an incumbent under 50% is in trouble because undecideds break for the challenger. And there is a lot of truth to that... but not in Presidential elections.

In a congressional election the challenger is typically totally unknown except to political junkies so 90% of the campaign process is a referendum on the incumbent. Undecideds are already saying they are not for the incumbent, and they typically couldn't pick the challenger out of a line-up, so they are not for the challenger either.

One they don't like, one they don't know.

But on election day, when forced to chose, they chose against the incumbent even while having almost no idea who the other guy is.

In presidential election we do not go into the last week with undecideds having barely heard of the challenger. Undecideds in election 2012 know who Obama and Romney are, and have for a long time, and are still undecided.

Thus presidential races do not feature the "incumbents under 50% lose" dynamic that local races do.

Also, the length of Presidential races allow the undecideds time to break for the challenger and then break back. Kerry won the debates but lost the post-debate period. Romney won the first debate big and got a big break from independents and undecideds, but unfortunately for him there was a full month and two more debates to go. It appears that in 2012 some people signed onto Romney after the first debate, but provisionally. He was on probation. And he did not wear well.


The late break par excellence was Reagan in 1980 and that was because there was only one debate and it was right at the end of the campaign. That election was so striking that pundits who remember it always expect that dynamic to repeat, but it is actually not at all typical.



Here are the notable late breaks in Presidential elections I recall.

(I don't remember 1960, but if I have my history right, there was no late break. JFK and Nixon were very close throughout the whole campaign, and very close on election night. The debates actually had no dramatic effect on the polls. 1964, 1972 and 1984 were such blow-outs that it would be hard to identify a late break. Undecideds presumably went to the winner, which was the incumbent president each time. I don't recall a late break in 1992. If the race tightened at the end that would be another late break for the incumbent.)

1968 - late move for Humphrey.
1976 - late move for Ford
1980 - late move for Reagan
1988 - late-ish (post convention) move for Bush
1996 - We will never know whether Dole got a late surge. The polling in 1996 was very poor and Dole outperformed his polls, but we don't know exactly why or when or how. Maybe he surged, or maybe the polls had been off all year.
2000 - Gore surged, but that was due to the DWI story so it's a separate category.
2004 - Late break for Bush
2008 - late break for Obama, but he was already well ahead.

So the late movement is 50-50 challengers and incumbents. (5 out of 8 for the incumbent party—Bush in 1988.) Also, the candidate getting the late surge only got inaugurated 50% of the time. (I say inaugurated instead of won because of Bush v. Gore)


And if we count undecideds breaking for the three incumbents LBJ, Nixon and Reagan, which they must have since they were land-slides, the it would be 7-4 for incumbents.

This election is probably going to be quite similar to 2004, though with Obama winning by more EV than Bush.