Predicting the Election Outcome

This is really a weird year in terms of convergence of views on the outcome of the presidential election.  Polls are all over the place - from the differences between sample skews to the differences between the state polls and national polls, the lack of consistency points to a wide range of potential outcomes.  I've never looked at an election with as much optimism and trepidation at the same time.

James R. Whitson at President Elect had this to say yesterday (emphasis added);

Like everyone else, I've been harping about Ohio, Ohio, Ohio. I mentioned last time that even though Obama doesn't have a large enough lead to move the state to his column using my normal model, that I wanted to do it anyway since his lead had been so consistent. Today I am going ahead and doing that. I didn't follow through in the last update because Romney took a lead in one Ohio poll and if it was a trend I didn't want to go against my numbers too fast. Today the polls still have that slight Obama lead. But my gut tells me it is still way too close to call there...

There has been a lot of talk this week about the disconnect between the national polls and the state polls, with some suggesting one of them has to be wrong. I've noticed the normal trend has been for state polls to lag behind what national polls say, but that is usually because state polls aren't produced as often. That's not the case right now. I agree that something is different this time out. Whether that is a polling methodology issue, too many new and not well established polls poisoning the well, or something else, I don't know.

Others are making some pretty bold predictions  from 315 electoral college votes for Romney (of 538 possible), to almost the same unbalanced numbers in favor of  Obama,  What gives?  It is really hard to filter out the noise and get a good read on things but here's my math.

If we start with the conventional wisdom that Obama has 201 electoral college votes locked down and Romney a similar 191, I'll focus on the swing states and just add up based on my best guess by state to see what the number looks like.  The approach I'm taking in order to avoid over-exuberance on my part is to be a glass-half-empty cautiously pessimistic prognosticator.  It also avoids the information bubble view approach that so many on both the left and right sometimes fall victim. That means I'll give Obama some states that Romney might win.  I'll try to keep the logic succinct.

New Hampshire (4 Electoral College votes): Close but probably sticking with Obama.  If this flips to Romney it could augur a seismic shift in favor of Romney across many of the swing states. 

Pennsylvania (20): While Romney is pushing here late in the game and has reason to be optimistic, failing a seismic shift mentioned above, this state won't flip for Romney.  It's going to be closer than normal but will stick with Obama.  An upset here for Romney and it's game over for Obama.

Virginia (13): The 3 latest polls have it tied or Obama a modest lead, but when you look at the sources, CBS/NYT/Quinnipiac, WeAskAmerica and Gravis, I'm suspicious about their partisan splits.  Rasmussen has Romney up 2 points this week.  Romney will take Virginia.

North Carolina (15):  No contest. Romney.

Florida (28): Over the last week, only 3 polls of 10 have Obama up and those include pollsters I'm highly suspicious about - NBC/WSJ/Marist, The Democratic pollster PPP, and the aforementioned CBS/NYT/Quinnipiac.  The rest have Romney ahead, and he'll win this state.

Michigan (16):  Yes Romney is closing the gap, no he won't win.  There was too much ground to make up for a late charge to close the gap.  Even Rasmussen has Obama ahead up by 5 this week.

Minnesota (10): Same story as Michigan.  People, particularly the low information voters paying more attention now has started the gap closure.  But it happened to late to flip a state like Minnesota.

Wisconsin (10): Wisconsin has a great Republican ground game and wins for the GOP in 2010 and in the Governor Scott Walker recall election.  The poll average has Obama way up, but the usual suspect suspects make up the bulk of the average.  Rasmussen has it a tie.but that's a lot of eggs for one basket.  I'm going to leave it for Obama even though I think it could easily be a Romney state.

Iowa (6): Some too-small-to-count pollsters, Rasmussen (Romney) and WeAskAmerica (Obama) have it close.  NBC/WSJ/Marist with oversampled Democrats, has Obama well ahead.  I'm ignoring them.  It's another 50/50 toss-up. Again, playing cautious ball, I'll leave this one for Obama.

Colorado (9): I'm not convinced, despite the close polling that Obama has a chance of repeating his 2008 win here. Romney will take Colorado.

Nevada (6): Despite the president's frequent slagging of Las Vegas over the last few years, he will win Nevada.

Where does that put the total?  Obama 263, Romney 257.  Which brings us to Ohio.

Ohio (18): Taking out polls that have a D+8 sample and the unknown pollsters there's not a lot left to work with other than Rasmussen, which is showing a tie.  ARG for example is showing Obama leading 49-47.  But the poll D/R/I split is 43/34/23. That shows me that with a more realistic Ohio split, Romney is probably ahead.  The same is true for many other polls that show Obama ahead.  I hate to use the word collusion.  Maybe it's a herd mentality that's driving pollsters to show Obama up by 5.  Maybe they're afraid to be outliers.  I do think Romney will win Ohio, and while it will take a while to report, he should win by a more than one point.

That means I'm looking at a Romney win. 275-263.  But I think Romney could pick up Iowa, Wisconsin and far less likely, Pennsylvania, meaning his possible upside number is 311.  Obama's possible maximum is 290.  He won't make it.

So my final prediction is a Romney win.  We're pretty much down to wait and see.