In the wake of the pitched battle here on Saturday night over who won Nevada, and in light of the renewed discussion of the scenario--mentioned here, and favored by most DKos users--in which Edwards will pick up delegates in almost every state from here to the Convention, making him and his supporters the "kingmakers," I thought to write a little diary on how the pledged delegates are chosen in primaries. This diary is based mainly on this source on the S.C. Dem. Party website.
I hope this explanation will help everyone make a little more sense of what happens not only on Saturday but also in their own state on Feb. 5 and beyond.
There are 7 contests taking place on Saturday. The statewide contest and the contest in each Congressional District. Eight distinct sets of delegates will be awarded to the candidates:
Statewide A. Pledged Party Leaders and Elected Officials (PLEO): 6 delegates.
Statewide B. At-Large Delegates: 10 delegates.
CD 1: 4 delegates; CD 2: 5; CD 3: 4; CD 4: 4; CD 5: 6; CD 6: 6.
Each set of delegates is split proportionately among the viable candidates, that is, the candidates getting 15% or more of the vote.
The wikipedia page on all six of South Carolina's CDs, including maps, is here.
No polling has been done with crosstabs on the candidates' support in each Congressional District. And that's a shame. Just as an exercise, I'll go through the delegate calculation on the patently incorrect assumption that each of the three candidates has the same support in each CD.
Using an average of the five most recent polls, if the undecideds break evenly among the 3 candidates, the results will be roughly
Obama 45.53%
Clinton 32.33%
Edwards 22.13%.
I think the methodology of the polls I am using (unlike that of the Clemson poll, which I leave out only because it's too dated) underestimate the number of undecideds. And it could be that all the undecideds will break for Edwards. Who knows? Maybe they will break for Obama or Clinton. But let's just take those numbers above and work with them.
Assuming the counterfactual that the geographical distribution of support is uniform, the rest is just caucus math. Because all three candidates receive greater than 15%, all three will get at least one delegate from each pot.
Statewide A: Obama 3, Clinton 2, Edwards 1.
Statewide B: Obama 5, Clinton 3, Edwards 2.
CD 1: Obama 2, Clinton 1, Edwards 1.
CD 2: Obama 2, Clinton 2, Edwards 1.
CD 3: Obama 2, Clinton 1, Edwards 1.
CD 4: Obama 2, Clinton 1, Edwards 1.
CD 5: Obama 3, Clinton 2, Edwards 1.
CD 6: Obama 3, Clinton 2, Edwards 1.
Obama would get 22 out of the 45 delegates at stake (49%).
Clinton would get 14 (31%).
Edwards would get 9 (20%).
As you can see, there would be some discrepancy between delegate results and the popular vote, as there was in Nevada (although there the numbers labeled by the Clinton campaign as the "popular vote" were already removed from the popular will by one layer of caucus math).
How does the caucus math affect things?
First, in each district with just 4 delegates to award, provided that all three candidates take more than 15% of the vote, the winner of a plurality of the vote automatically gets 2 delegates. So in a three-way race, those 4 delegate districts become more important, in a sense, for the frontrunners--a one-vote victory in the primary means a one delegate advantage, and these delegates can add up.
Second, in any district with 5 delegates to award, the presence of a viable third candidate makes it easier for a trailing candidate to win just as many delegates as the 1st place winner. In our example, South Carolina's 2d CD, Clinton gets her 2d delegate the following way.
Delegate scores: Obama 45.53/20 = 2.28; Clinton = 1.62; Edwards = 1.11. First, truncate those scores to award as many delegates as possible (Obama 2, Clinton 1, Edwards 1). Then, the fifth and final delegate is awarded to the candidate with the highest remainder, i.e., Clinton because Clinton's .62 is greater than Obama's .28 and Edwards' .11.
I note that nationally, in CDs with 5 delegates, if Edwards can pull exactly 15%, Obama does not need to be all that close to Clinton in order to split the 4 remaining delegates 2 and 2 with Clinton. Remarkably, in a 5-delegate CD, if the results are Clinton 52.4%, Obama 32.6% and Edwards 15%, the delegate split will be 2-2-1.
The bottom line, which I won't belabor, is that luck plays a significant role in who gets how many delegates. But delegates are the name of the game, particularly if Edwards remains viable (i.e. above 15%). If the results modeled above were replicated across the country (but with Obama in Clinton's shoes, and vice versa), Clinton would be just short of a majority of pledged delegates. It would then be up to Edwards and Obama to convince enough superdelegates to join with them to keep Clinton from winning at the Convention.