If you look at the other tabs, the results would be different though.
Perferences by Party:
RP: 14*.60 + 20*.40 = 16.4
by Tea Party support:
RP: 15*.27 + 17*.73 = 16.46
So something is fishy. The results for all candidates is about the average of the likely to vote splits. Did they have a typo for the likely to vote %? I don't know.
No, that's the thing. In their South Carolina poll, the split is 90-10 definite/probably voters, and the numbers add up correctly: http://americanresearchgroup.com/pres2012/primary/rep/sc/