Electoral Map Update

We project no change in the ultimate electoral college tally going into Election Day:
Obama 269, McCain 200, 69 up for grabs.
Those 69 are in only four states: FL, OH, CO, and VA.

163 Electoral Votes for McCain and 264 for Obama project to safe wins of 6 or more points (ranging 53-47 to 51-45 or better, depending on the third party votes). These today have margins a couple points better than the 95% confidence level today, or more, in most of the polls and thus would be difficult to reverse.

We get from those 427 to 538 with 69 tossups in four states, as above, three states with 37 votes leaning to McCain (IN, NC, and MO), and one state with five leaning to Obama (NV).

It looks to be a good night for watching and celebrating. The early focus will be on VA, FL, and OH--if any of them go early for Obama the game will be called soon. More likely, they will be close, and tense, which will bring the Western states' importance forward (as the late-night did in 2004, though they didn't ultimately fall for Kerry). Still, Obama should have a clear lead throughout.

None of the states with the 264 Electoral Votes clearly favoring Obama looks particularly weak. NH, PA, WI, MN, and MI have all firmed up. NM and IA are both on winning arcs for takeovers.

Correcting for DDE and Other Effects

Far as I can tell, the David Duke Effect is different from the alleged Bradley effect, in which white people would tell black pollsters they were undecided when they actually planned to vote against Tom Bradley (LA mayoral election, 198?). What I'm thinking of is people who say they will vote for Obama but will actually vote for McCain, minus those who say they will vote for McCain but will vote for Obama. My reference is to David Duke in the LA governor's race--a primary?

I would argue that the net effect in the national election is not zero, and that it is almost surely positive (DDE>0) in most states. I would suggest it is in the range of 1-3%, and 2% nationally.

3% is big. It turns a 51-47 Obama lead into a 50-48 loss. So states with margins like that and high potential for DDE can not be considered secure for Obama.

Another way to put it is inertia vs. demographic change. The Western states (the clear examples are NM, NV, and CO) are changing toward younger and more inter-racial societies, so we can expect progress over time toward the Democrats. Additionally, African-American populations in these states are relatively small, and not carrying so much historical baggage as in the East. Otherwise, you have to start with the edge to the side defending from the last two elections and make the challenger prove they can win.

Either way, at this point in time (Nixonite Ron Ziegler's memorable "atpit") I choose to disbelieve these small leads for states such as MO, NC, and even VA and FL, the last two of which, if you go purely by the polls, have close-to-statistically significant leads for Obama today.

I believe CO and OH to be states that are winnable for Obama going into Election Day, but he hasn't proved he's won them yet, as he is close to doing for ones like NM, NH and WI.

Ranking Likelihood of Change

The rank order statistic is one of the most stable measures out there. Excuse me, it is the most stable; for example, one can use it to do a much better job than trying to establish the center-point of an unknown distribution. So, it's very clear which states are the most likely to change over.

At one end, this is totally not interesting. Who cares whether Utah, Idaho, Massachusetts, or California is the least likely state to change over (after the territorial entity known as the District of Columbia, of course)?

The key point with regard to this election is not the most likely to change--clearly Iowa. The focus is after Iowa and New Mexico. My argument is that it is Nevada, followed by the quartet of FLOHCOVA (pick your order, but this one reads best). One could argue #8-15 of the Most Likely to Change (some of which are former Kerry states Obama must hold) until Nov. 3, but those are not the ones that will end up making the difference. It's 3-7, and how they fall.

Particularly because McPalin is challenging so little in the blue states, the difference among most electoral estimates for this election--whether the person is objective or partisan, or whether his/her model is sensitive to fluctuations or somehow "sticky"--boils down to how deeply we will go down the ranking of likelihood to change. If Obama only gets his top two, he's five short. His top three, he's got to survive white knuckles. Four or five, a clear win, and six or more, a decisive one (300+). If middle-range North Carolina or Indiana or Missouri go for Obama (I'm expecting each to end up for McCain by 1-5 points), it's a wipeout.