The latest CNN poll on South Carolina and Nevada reports this sample information:

A total of 1,009 South Carolina adults were interviewed, including 521 who said they were likely to vote in the Republican presidential primary. In Nevada, interviews were conducted with 1,011 adults, including 285 who said they were likely to participate in the Republican presidential caucus.

The idea that over half of the adults polled at random are going to vote in one political party's primary is ludicrous even in an open primary state like South Carolina.  About 30% of them aren't likely to be registered to vote.  Of those who are registered, probably 40% identify or lean Democrat and are unlikely to cross over even if the voting laws allow them to. Plus there is a primary on the Democrat side, too.   Then there are historic primary voting patterns.  According to the United States Election Project the actual Republican presidential primary turnout in South Carolina in 2012 was 16.7% of eligible voters.    In the 2012 general election, turnout in South Carolina was 56.3%.  Yet CNN is reporting that 51.6% of its sample of adults is likely to vote in the Republican primary. Based on historic information on voter registration, party affiliation and turnout, a sample of a 1000 adults is far more likely to yield a number closer to 50 who will vote in the Republican primary than 521.    

CNN also claimed that 285 of the 1,011 residents of Nevada they polled were likely to participate in the Republican caucus.  Nevada is a closed caucus state, which means that only those registered as Republicans 30 days before the caucus date can participate. According to the most recent statistics from the Nevada Secretary of States office, 34,5 of the voters are registered Republicans. Actual turnout at the Nevada Republican caucus in 2012 was 1.9% of eligible voters.   Thus a random sample of 1000 adults is likely to contain fewer than 5 people who are likely to attend the Nevada Republican caucus.  (1000 x.70 x.34.5 x .019)