I'm striken by the fact that what drives this entire thread is in fact poor math/statistical skills about what is significant sampling. Sampling size determines the margin of error and as long that the results are well outside the margin of error, it's absolutely possible to "project a winner".
Given that the result in South Carolina was very much in favor of Gingrich, a low sampling is required to establish this simple fact.
I stopped looking at the video at that point, the cluelessness having been established right from the start.
There is no substitute for education.