The Path to 270: McCain's Michigan-Ohio Strategy

Electoral vote analysis in a time when one candidate is ahead outside the margin is massively uninstructive. It mostly consists of teaser-style pieces like, "Obama is within single digits in Alaska!" and "look at those North Dakota numbers!"

If Obama is playing in either of those two states in October, it's game over and no amount of analysis will matter. I won't be wasting my time focusing on what someone needs to do to get from 300 to 350, because all you need is 270. 

So I'm going to approach this differently. I am going to this as though the election is tied, and give you scenarios for how either candidate claws above 270 for the win based on the current configuration of state polls. Obama may look good now, but what if his support recedes to back a near-tie? Which states does John McCain pick off if that happens? What's the exact order of the target lists on both sides?

For this analysis, I'm shamelessly relying on FiveThirtyEight.com's model since I find it to be the most compelling model out there, both in taking into account non-polling inputs (important in blowout states), and weighting polls based on their recency and track record. Then I mash them up with 2004 data to show the projected swing (mostly to Obama) to get hints at the likely configuration of the 2008 map under different scenarios.

The data underlying this analysis is published here.

Based on 538's projection for each state, Obama is leading in the popular vote by 49.7% to 46.4%. This 3.3% margin amounts to a 5.8% swing to Obama off Bush's 2.5% 2004 victory. Any state with greater than a 5.8% is an Obama-outperform state. Any state under is a McCain-outperform state.

As I've said, I'm not interested in what happens if Obama wins by 3 or 4%, other than to say that in that case, It's over. Despite the hype of a split popular/electoral vote winner, probability says that it's highly unlikely unless the popular vote margin is under 1% (0.5% turned out to be the breaking point in 2000). I will explore how McCain can maneuver in a 1-2% popular vote loss scenario, however.

So I am rolling the margin back to a McCain-Obama popular vote tie, and readjusting state victory margins accordingly. I'm also taking the state needed to tip the election off the map. Interestingly enough, it's Ohio, the median state in 2004. And the electoral map looks like this:

This is an Obama 264, McCain 254 map.

For McCain to win, he needs this map, plus Ohio. Subsequent McCain offense states are: 1) New Mexico (0.7% Obama margin in a tie popular vote), 2) Michigan (-0.9%), 3) Pennsylvania (-2.7%), 4) Iowa (-3.2%), 5) New Hampshire (-4.1%) and 6) Wisconsin (-4.4%).

Obama's offense states beyond Ohio are 1) Colorado (0.3% less pro-Obama than the nation), 2) Virginia (3%), 3) Florida (5%), 4) Missouri (5.2%), 5) Nevada (5.3%), and 6) Indiana (5.4%).

Under the polls as they currently stand, it's Obama 306, McCain 232.

What's striking about this map is that Obama has relatively little margin for error in the 0-3% victory margin zone. He has just Ohio and Colorado as padding through 3% (taking him to 293), when he adds Virginia. Beyond that, he doesn't add another state until Florida at the 5% mark. And he has to defend McCain picking off Michigan, which is within 1% of his national average performance.

This suggests a McCain firewall in Virginia and Colorado, and to a lesser extent Florida, but that won't win it. Rather, the key to victory may lie in targeting Ohio-Michigan as a megastate and trying to shift both states 1-2 points in his direction by brute force. If he does this to tune of just 1%, and nothing else changes, the electoral vote is McCain 291, Obama 247 under a tied popular vote.

The intense targeting behind Florida (2000) and Ohio (2004) shows that it's possible to generate one-state swings that pull "ground zero" states towards the national median. The trick this time will be to spin the Buckeye-Wolverine axis as the new Florida/Ohio, because McCain needs both. This also means we can expect to see lots of Dearborn-to-Toledo bus tours for the Straight Talk Express. 

To some extent, McCain should not get too distracted with defending his home turf. If this election becomes about Florida, we lose. The same is also true of swing states du jour Virginia and Colorado, which are still slightly to the right of the country, though they used to be 3-5 points moreso. If the narrative is about those states exclusively, the media will force McCain to play defense the next four months. McCain can't let this happen.

In a state-by-state situation, it's hopeless to try and compete as though you're 5 to 10 points behind, trying to rebuild strength in states you'd easily win in a close race. Instead, you compete as though you're 1-3 points behind, and seek to throw the opponent off balance in some big states he needs to win. The logical conclusion here is a maniacal focus on Ohichigan, heart of Bitter America. 

In an upcoming post I'll analyze how the state-by-state landscape has changed nationwide and how state polling trends jive with the results of the Democratic primaries.