Although Obama leads in the polls, the problem is the polls.
As Jay Cost has shown, most polls are currently polling registered voters (which are usually 1-2% more favorable to Democrats than likely voters) and they are using models predicting an electorate of D+4 to D+7:
http://www.weeklystandard.com/blogs/morning-jay-are-polls-skewed-toward-obama_648770.html?page=2But historically elections average only D+3 (2004 was D+0 and 2000 was D+4). In fact in the June total polling by Rasmussen found unweighted, respondents were a R+1.4%. So most polls are likely favoring Obama by at least another 2%. This means a likely shift of 3-4% to Romney when the votes are tallied, hence the election may well result in a Romney win.
However, I do agree that if Romney wins the popular vote by less than 2% there is a very good chance Obama wins in the Electoral College. In 2008 Obama won the popular vote by 7%, but it would have taken a 2.5% popular vote win by McCain for him to have won in the Electoral College.