He assumes that Obama will just lose three states from 2008 – Indiana, North Carolina, and Florida.
Every other swing state will go to Obama – Virginia, Ohio, Colorado, Iowa, Wisconsin, New Hampshire, Nevada, Pennsylvania, Michigan, Minnesota…
So Silver isn’t just using the RCP averages to count Obama states; he’s then screwing with those averages to get the results he wants.
In other states, the latest polls show the race tied – Ohio, Wisconsin, and Michigan (where a Democrat pollster shows a 47-47 tie.) Unless undecideds suddenly swing towards the incumbent, these states would be Romney’s, no?
Not to mention the surprisingly strong showing for Romney in states like Minnesota. And New Hampshire, where an Obama lead of 1.4 percent is almost entirely due to a week-old outlier poll showing him ahead by 9.
Against this backdrop, it’s reasonable to question Silver’s “model.” What secret sauce is he using to entirely discount the tied polling in Virginia and Colorado, the latest tied polling in Ohio, Wisconsin, and Michigan, and Romney’s surprising strength in other states? Or is he, like the paper that employs him, simply ignoring inconvenient data like the Libya disaster and Rasmussen polling? He owes it to his readers to tell them.