Throughout the 1980s, Republicans and some gullible reporters (and not a few easily-intimidated Democrats) assured us that there was a GOP electoral college lock on the presidency. Just look at all those states Ronald Reagan (and then George H.W. Bush) won! Even if Republicans lost some swing states, all they had to do was to win the states that they always won, and they would win the presidency.
This worked just fine...until an election came along in which the Democratic candidate received more votes than the Republican candidate, and it turned out the "lock" was meaningless.
Well, all it's taken is a grand total of one election for Chris Cillizza to roll out a reverse version of the argument. See, he says, Barack Obama is in great shape for 2012:
[A] detailed examination of the national map heading into 2012 suggests that the president still sits in a strong position for reelection - able to lose half a dozen (or more) swing states he carried in 2008 and still win the 270 electoral votes he needs for a second term.The problem with this kind of analysis is that it ignores the possibility that the Republican candidate might actually do better than the Democratic candidate overall. If that happened, it wouldn't just be the states that Obama won narrowly that would swing to the GOP; it could be lots of marginal states. Cillizza's "detailed examination" is meaningless; all he's telling us is that Obama won big in 2008 and could do a lot worse and still win. That's true -- but Democrats in 1968 and Republicans in 1976 could tell you that large swings from election to election are quite possible. As could George H.W. Bush.
Usually, it's safe to assume that the winner of the popular vote will win the electoral college. In very close races, such as the 2000 contest, that can obviously go wrong...but both parties are capable, over the last few decades, of winning decisively overall, and if so they'll win the electoral college, too.
The correct way to check for an electoral college advantage is to see what would happen in tie races by shifting each state equally to reflect a tie result. Up until recently, it turns out that neither party had any kind of advantage in the electoral college.
It is perhaps worth noting that while the college tilted to the Republican in the only modern election in which it actually mattered, since 2000 the college has tilted dramatically to the Democrats. In 2004, John Kerry lost the popular vote by 3% to George W. Bush, but would have won had Ohio, which Bush won by 2.5%, swung to the Democrats (and a universal swing of that size would have brought New Mexico and Iowa with Ohio, leaving Bush with a narrow popular vote lead but a 284-254 electoral college loss).
Then, in 2008, Barack Obama won the overall vote by a bit over 7 percentage points, but only a few states -- Florida, Ohio, Virginia, and Indiana -- went to Obama by less than that margin; in an evenly shifted 50/50 election, John McCain would have added 72 electoral votes and lost 293-245. Even a ten point shift to McCain only adds another 20 electoral votes and leaves him a solid popular vote winner but an electoral college loser. Only after that does McCain start grabbing a large electoral college lead.
Electoral college advantages are certainly theoretically possible. A party will have an advantage if it does better in small states than large ones, all else equal; or, more likely, it will have an advantage if its votes are distributed more efficiently, with the other party wasting votes in huge landslides in large states. As I said, this doesn't seem to have been the case generally over time. For every Republican Wyoming there's a Democratic Rhode Island, and for every GOP wasted margin in Texas there's a Dem wasted margin in New York. So we'll see if 2004 and especially 2008 is a trend that sticks, giving Democrats a real advantage, but my guess is that it's just a fluke.