Sister Kissing Update

In recent days, the probability of a 269-269 electoral tie has increased slightly in our simulations. This is mainly because Ohio has shifted from a lean Obama state to a toss-up, and the way that the math works out, it is very difficult to come up with plausible tie scenarios if Obama wins Ohio.

In this afternoon's simulation run, a tie occurred 66 times out of 10,000 simulations, involving 10 distinct scenarios to produce such an outcome. A lot of the more far-fetched scenarios that the model had imagined before (scenarios involving McCain winning Massachusetts or Obama winning Texas, for instance) have come off the board as we draw closer to the election and as the huge demographic earthquakes these things would require become less likely. The flip side of this, however, is that with a smaller conceivable set of battleground states, some specific, tie-producing scenarios come up more frequently. The two most plausible scenarios for producing a tie are as follows:

(22 times out of 10,000 simulations): Obama wins Kerry States + Iowa + New Mexico + Nevada. I somewhat dismissed this scenario in the polling write-up today, arguing that Obama was unlikely to win Nevada if he lost Colorado, but in fact such possibilities occur reasonably often.

(22 times): Obama wins Kerry States + Iowa + New Mexico + Colorado - New Hampshire. This is simply Al Gore's map, plus Colorado.

The only other scenario to come up in the double digits was this one:

(10 times) Kerry States + Iowa + New Mexico + Colorado + Virginia - Michigan. The five Bush states that Obama is most likely to flip are Iowa, New Mexico, Colorado, Ohio, and Virginia. If Obama succeeds in flipping four of the five, but loses Ohio and the companion state of Michigan, we'll have a tie.

From here onward, we get pretty obscure:

(5 times) Kerry States + Iowa + New Mexico + West Virginia. Obama has not made much of an effort to contest West Virginia, but we know very little about it -- it hasn't been polled much and its demographics are very unique -- so our model has it coming into play occasionally.

(2 times) Kerry States + Iowa + Colorado + Nevada - New Hampshire. This is mathematically equivalent to the first scenario we discussed, but with New Mexico replaced by Nevada. Since Obama's polling considerably better in New Mexico than Nevada, it is unlikely.

Here are the remaining five scenarios:

(1 time) Kerry States + Florida - Minnesota
(1 time) Kerry States + Colorado + New Mexico + Montana
(1 time) Kerry States + Colorado + New Mexico + North Dakota
(1 time) Kerry States + Ohio + Colorado + New Mexico - Michigan
(1 time) Kerry States + Ohio + Iowa - Minnesota

*-*

I am also going to introduce a slight change to our methodology. Previously, for the reasons explained here, we had all ties going to Obama. I am going to continue to assign all ties to Obama in instances where he wins the popular vote. However, in cases where the candidates tie and McCain wins the popular vote, I will split the ties 50:50.