I'm updating my prediction for this year's election. Here are my assumptions that drive the outcome:
1. The polls do not accurately reflect the state of the electorate. There is a persistent over-sampling of Democrats as pollsters base their turnout projections on the turnout rates in the 2008 election cycle. The, Democrats were massively energized and Republicans were somewhat demoralized, and so the cycle was something like D+5 or even more. There's simply no way that we'll see that this year, even with what we can expect to be pretty serious voter fraud.
2. There is a real Bradley Effect in play here. The Bradley Effect is almost always misinterpreted to mean that voters are racist; instead, it shows that voters (rationally) do not want to be accused of racism, and so lie to the pollsters. This inflates Obama's numbers by a couple of points over what we will see in the ballot boxes where the secret ballot will allow people to cast their vote without being accused of being a racist. Given the exceptional nastiness of this campaign - the "dog whistle" accusations of the last day are a good example - this reluctance to tell people their real opinion is entirely justified.
3. Perhaps 10% of the voting population has yet to make up their minds, and will not until a couple weeks before election day. These people are simply not political junkies, and while they take their votes seriously, they just aren't particularly interested in tuning in to the give and take until the big day approaches. Given that the economy is on the skids and people rationally vote their pocketbooks, this group will likely break to Romney 2 to 1.
The result of all this is that there is between a 5% and a 10% over-sampling of the polls, and to get an accurate picture of how the election will turn out we need to subtract between 5% and 10% from Obama's poll numbers across the nation.
Election Projection shows the current latest polling on a state by state basis. They currently say that Obama leads Romney by 303 to 235. However, we've just seen that we need to adjust the poll numbers. When you subtract 5% from Obama's numbers, all of the states shown as "Weak Obama" go to Romney. Even more, we're talking more than 5% (but less than 10%). That category is "Moderate Obama", and half of those will go to Romney.
And that will give Romney nearly 350 electoral votes. Here's my map predicting the race (click to embiggen).
Now I may be wrong here, and as I've said before we're probably better off in the long term with a 2012 Obama victory. However, this is how I see this developing. You can create your own map, too. Remember, the interesting discussion is not the result, but your assumptions that drive the result.