Setting aside the debate over allocating delegates by proportional representation versus winner-take-all for the primaries (a debate where I think proportional representation wins hands down, as previously argued) and other issues like the impact of the Democrats' schedule of primaries, consider the lessons for how winner-take-all would have played out in the Democratic primary for how best to elect the president in general elections.
In general elections, we currently have winner-take-all, state by state rules (Maine and Nebraska allocate delegates by winner-take-all in congressional district, but have never divided their electoral votes since adopting their approach and are highly unlikely to do so this year either). As we can learn from appplying this rule to the Democratic nomination contests, winner-take-all rules means that:
* The winner in the national popular vote is more likely to be defeated: HIllary Clinton wouldn't need delegates out of Michigan and Florida to win if winner-take-all had been used even though she's behind in the national popular vote and behind by more than 2% in the national vote without Michigan and Florida.
* Big states count more than small states: Barack Obama has won nearly twice as many states as Clinton, but she has won more of the big state. That helps explain why contrary to what some misinformed people contend, the current Electoral College system doesn't help small states. It in fact makes the swing voter in big popular states far more important than anyone else.
* Close states count far more other states: The current Democratic system is essentially a national primary contest unfolding state by state. Getting more votes in every state matters, no matter how close, even if the media likes to obsess over who wins states as if the results were winner-take-all. In contrast, if winner-take-all rules were in place, the candidates would completely ignore states they couldn't win or were sure to win. That's sadly just what they will do in general elections this fall, as revealed over the weekend in the New York Times. Our Presidential Election Inequality report presents powerful data from 2004 about 99% of campaign resources going to 16 states in the campaign's peak season.
* Recounts are a far bigger problem: When Hillary Clinton won Indiana by fewer than 20,000 votes, some in the media started hyperventllating about a recount. That was absurd. If Obama had won Indiana in a recount, he would have gotten only one more delegate. But winner-take-all makes artificial crises out of close results in states like Florida in 2000. The odds of a national recount being impactful with a national popular vote plan are minuscule, as demonstrated in our 2007 report on recounts.
Some might wonder why I don't support proportional allocation of electors in the Electoral College as the Democrats use in their nomination contests. That approach would be better than winner-take-all if applied to all states, but it's not nearly as sensible as every vote being equal in general elections. See chapter four of Every Vote Equal, the book I co-authored about the National Popular Vote plan, and our 2007 report Fuzzy Math.
Fortunately, we should have the National Popular Vote plan in place for general elections in November 2012 to correct all these defects in the current Electoral College system -- the plan has passed in four states and will be debated in all other states in the coming year.