Thursday, October 30, 2008

A November Surprise Scenario

Over this past weekend we had a discussion going about the effect the presence of early voting in a state would have on the McCain campaign's home-stretch strategy. The basic hypothesis was that states with early voting -- where Barack Obama would have some votes in the bank heading into election day based on the advantage the Illinois senator has enjoyed in early voting -- would not be focused on as heavily as states where early voting and absentee voting were minimized. In other words, if something were to happen to change people's minds in some way, you'd get more bang for your buck if you were campaigning the hardest in states with as little early voting as possible. States like Pennsylvania and Virginia.

But we can push this concept a little further. And, in fact, SarahLawrenceScott has done just that. If we were to look at the early voting information we have access to so far in this cycle and gauge whether that looks like it is on track to surpass the numbers from 2004, we can then make some basic assumptions that may give us a better idea of why the McCain folks are focusing where they are focusing.

First, let's look at the assumptions Scott has put together:
  1. Assume something triggers a 4% loss for Obama (this was roughly the size of the Rev. Wright/McCain clinches drop, the Palin drop, and Obama's "soft" support in current polls) whether an Obama revelation or some geopolitical. We'll call it a November surprise.

  2. Assume a 2.5% overperformance for Obama in early voting. In other words it is not that a greater proportion of his supporters vote early, but rather that GOTV would push his numbers up in a state that had 100% early voting.

  3. Assume 50% of undecideds who vote early go for Obama.

  4. Assume that after the surprise, just 20% of election day undecideds go for Obama.
If you start off with the Pollster averages for each state and combine that with the above assumptions you end up with something like this:

If we take these new, post-November surprise state margins and apply the three-category thresholds that FHQ employs -- >7% = strong, 3-7% = lean, <3% = toss up -- we end up with a map that has toss ups that look an awful lot like the targets the McCain campaign has.

[Click Map to Enlarge]

Virginia would flip back to McCain. Pennsylvania would be much more competitive than what the polls show, as would Ohio. And Florida and Wisconsin would look like dead heats. It could be argued that, well, those are the toss ups anyway. Well yeah, except Minnesota, Pennsylvania and Wisconsin don't really fit the bill according to recent polling in each of those states. However, the goal is to consider not only the potential effect early voting has, but to ascertain whether strategically, the McCain campaign is hedging its bets, hoping for something akin to the George W. Bush drunk-driving revelation (Though, perhaps it would have to be bigger) that broke during the final weekend of the 2000 race, happening in this race. Maybe, maybe not, but the decision to focus resources on the three states I just brought up looks more rational in this light than it does on the surface.

NOTE: I want to thank Scott for putting this material together and sharing it will me. It really adds to the earlier early voting discussion we had.

Recent Posts:
National and State-Level Factors in US Presidential Election Outcomes: An Electoral College Forecast Model

The Electoral College Map (10/30/08)

Liveblog: The Obama Infomercial


Unknown said...

I'll also add that McCain deemphasizing Michigan and Colorado looks more rational. Colorado's a bad bet for McCain because it's literally "done"--at this point the number of votes already cast in Colorado is more than half of the total number that was cast in 2004. That means Obama's lead there is pretty much locked in.

Jack said...

Why do we have to discuss these depressing scenarios?

In the interests of political science, I know.

Odd that super-blue Rhode Island is on that map as a tossup.

If we assume that the only way, or by far the most likely way, for McCain to win is a November surprise, then his decisions make sense. Except for Iowa.

Now, was this the reasoning that the McCain campaign went through when picking which states to compete in? We have no way of knowing.

Unknown said...

Jack--Rhode Island is an anomaly because the handful of polls that have been conducted there lately have a ridiculous number of undecideds. That's probably just wrong--the polls were local, and probably did not push people hard enough to state a preference. So when my scenario gives 80% of the late undecideds to McCain it flips Rhode Island to toss-up.

Jack said...

Yeah, that's right, I remember you pointing that out before.

It's really because Rhode Island is tired of being overlooked, and the voters claim they're undecided in an effort to get attention focused on the tiny state. Never works, though.

Unknown said...

Would a 4% drop be a little extreme with only approximately 5 days left in the election?

Unknown said...

Anthony--A 4% drop is not extreme if there's an "event." Drops of approximately that magnitude in that amount of time occurred twice so far this cycle: mid-March, and mid-September.

Another way to look at it is to look at measures of "soft" support:

Fox: 8% "somewhat" support Obama
ABC: 3% chance will change their mind, but currently supporting Obama
AP: 3% only "lean" toward Obama

Now "somewhat" support isn't the same as "could change my mind," so the 8% is high for our purposes. But 3% of the electorate are currently supporting Obama, but tell pollsters they're still up for grabs. Under those circumstances, if there's a substantial event does 4% seem unreasonable?

The event or revelations has to be quite significant, but it's all McCain has left.

Anonymous said...

I've got to say, this early voting in Colorado is borderline amazing. The implications of so many people voting early are fascinating.

What effect does that have on election day turnout?

1) It depresses turnout.
2) a) People say, "Hey, the lines will be short(er), why not go vote!"
b) "Wow! I want to be a part of this!"


Jack said...

I'd say that those who normally vote on Election Day will vote if they haven't already. It might encourage some who don't normally vote because of long lines on Election Day, but it won't do anything for those who don't vote because of apathy.