[kesg]

47% to 46%, with 7%"undecided". But the "undecideds" usually break towards the incumbent. Figure 2/3. So 47% + (2/3 x 7) = 51.67%. And the IEM vote share was last quoted as 51.8%!

Urban legend. So is the opposite claim that undecided voters usually break for the challenger. If you really want to know how undecideds will break, look at the President's job approval rating. It is the single best historical predictor of how he will actually poll on Election day.

[Doctor Stochastic]

I've seen three claims (four if you count this thread.)

One: Undecided break for the challenger.
Two: Undecided break for the incumbent.
Three: Undecided break proportionally to the decided.

Four: (On this thread) Undecided don't vote.
Five: (My observations) Undecided break for the cutest canditate.
Six: (My observations) Undecided break for the taller candidate.
Seven: (My observations) Undecided break for the candidate listed first on the ballot.