The partisan ratio in the methodology (at PDF) is 34% Dem. 29% Rep. 32% Ind.
They say according to their exit poll the 2008 ratio was 39% Dem. 32% Rep. 29% Ind. So they are predicting the Republican turnout in 2012 to be lower than 2008 and basing their formula on that.
Its BS - maybe a 3+ Democratic advantage if you want to be generous but up 5+? This election is nowhere like 2008 even if the media would like to pretend there’s a huge groundswell of support for Obama waiting in the wings.