This is an attempt to figure out where Barack Obama fared best among white voters in South Carolina. What I've done is estimate the racial composition of the electorate in each county by assuming (a) more or less all African-Americans are Democrats, and (b) the percentage of white voters that vote in each county will be roughly the same. If we then hold African-American performance constant, we can produce a number that represents the combination of Obama's over/underperformance among black voters and the margin of victory among white voters.
The green areas represent counties where Obama either underperformed among African-Americans, or lost the white vote, or both; the purple areas the reverse. Of the major cities, Obama's best performance was in Charleston, followed by Greenville, Columbia, Spartanburg, Rock Hill, and Myrtle Beach. Outside the cities Obama fared vest in the heavily African-American counties between Columbia and Charleston, fought to a draw in some of the Northwestern counties, and tended to lose the rural counties that were still majority white. His poorest performance was in the "corridor of shame" south of Columbia: Orangeburg, Bamberg, and Allendale counties are also heavily African-American. The light-pink spot in the Northwest is Greenwood, home of the "Fired up! Ready to go!" lady.
Update: I just want to be very clear: this map does not necessarily show where white voters went for Barack Obama. Purple counties might represent a strong white vote for Obama; or they represent above-average turnout from black voters; or they might represent areas where Obama earned 100% of the black vote. Green counties might represent might represent white voters' antipathy towards voting for Obama; or they might be counties where African-American voters were more likely to vote for Hillary Clinton; or they might represent counties where white voters make up a larger share of the Democratic electorate. We just don't know which is the case. Because South Carolina is required to track the race of its voters, we'll eventually have more accurate figures, but that's a month away.
