The reason is that the 2012 polls assume the same turnout as for 2008, and that is highly dubious. Morris writes:
All of the polling out there uses some variant of the 2008 election turnout as its model for weighting respondents and this overstates the Democratic vote by a huge margin.The 2008 vote was no ordinary election. Blacks, college kids and Hispanics increased their participation beyond their traditional percentages, and were enthused by the novelty of voting for the first black president. This time around, almost every voting class is less enthusiastic in light of continuing high unemployment, energy and food costs. The 2008 percentages will not be repeated, and polls who rely on them are bound to over represent the Obama vote.
Morris goes on to say that Rasmussen's poll is generally more favorable to Romney because it uses a variant of both the 2008 and 2004 turnout.
The real vote turnout will probably be closer to 2004 than 2008. Morris says:
If you adjust virtually any of the published polls to reflect the 2004 vote, not the 2008 vote, they show the race either tied or Romney ahead, a view much closer to reality.Further, the undecided voters at this point generally break for the challenger, and that means a Romney win.
Read it all here.