Monday, June 2, 2008

Projecting Delegate Allocations in the Final Democratic Primaries: How Many Superdelegates Does Obama Need to Clinch?

First things first. As the Puerto Rico result made very clear, the delegate projections for the last few primaries are going to be pretty rough estimates simply because there are not very many polls in any of these states. In Puerto Rico, I projected a 32-23 delegate split in favor of Clinton, but Clinton won a much bigger margin, taking 38 delegates to Obama's 17.

So, with that said, the projections about delegate allocations from the South Dakota and Montana primaries differ quite a bit depending on which polls you look at. Last week, our projections based on the latest polls had Obama winning 10 delegates in Montana (to 6 for Clinton) and 9 in South Dakota (to 6 for Clinton). Overall, that would give him 19 delegates tomorrow night compared to 12 for Clinton.

However, American Research Group released surveys from Montana and South Dakota today suggesting a very different result. These surveys give Obama just a 48-44% edge in Montana and put Clinton ahead by a 60-34% margin in South Dakota. If these polls are correct, then there would be an 8-8 split in Montana and South Dakota's delegates would be allocated 10 to 5 for Clinton. Overall, that would give Clinton 18 delegates tomorrow night compared to just 13 for Obama.

Why does this matter? The Obama campaign would love to line up enough superdelegate endorsements tomorrow to allow them to pass the 2,118 mark and have a victory party tomorrow night. But they might need a few more than they had initially thought they did if the American Research Group polls are accurate. Here are some graphics showing how close Obama will be to 2,118 after either scenario (assuming no more superdelegates endorse before tomorrow night):

Depending on which polls you believe, Obama will either end up 20.5 or 26.5 votes short of the nomination. So, if Obama wants to go over the top tomorrow night based on the results of the primaries, he needs at least 21 superdelegate endorsements between now and tomorrow night and to be safe he'd like to have 26 or 27 endorsements to bank. It will be interesting to see if he can pull together two dozen endorsements in the next 24 hours.


Ok, so which projections am I going with. Well, frankly, I don't trust these American Research Group polls for a couple of reasons. First, in Montana, we had a poll conducted last week by Mason-Dixon that showed a very different picture. This survey put Obama up 52% to 35% over Clinton. In the American Research Group survey, Obama was winning 18-49 year olds by a margin of just 48-43% while the Mason-Dixon poll had him winning that age group 56-30%. Second, there seems to be little reason to think that South Dakota would be a huge blowout for Clinton. As Nate Silver at notes:

"But South Dakota isn't all that idiosyncratic a state, and 26-point win just doesn't make any sense in the context of what we know about the demographics of this race. The national tracking polls are fairly stable, and the campaigns aren't behaving like South Dakota is a 20-30 point Clinton win."

Nate predicts a 5% edge for Obama in South Dakota.

Of course, one problem in pegging the expectations in South Dakota is that every state that South Dakota borders held a caucus. The only primary we have for comparison is the beauty contest held by Nebraska (on South Dakota's border) which Obama narrowly won, 49-46%. Nevertheless, the 60-34% margin just seems way off base. So I'm sticking with the projections from last week.

No comments: