We will be discussing public opinion and polling at some point in the future, and here's a controversy getting to the heart of how pollsters try to measure future events. Polls which ask respondents who they will vote for have to determine the likelihood that that respondent will in fact vote. More broadly, it means that they have to determine what the breakdown or Democratic and Republican voters will be. It varies from election to election, but generally more people identify themselves as Democrat than Republican. Pollsters are estimating that the electorate will skew slightly more (4% or so) Democrat than Republican. Some Republicans are questioning whether this assumption is warranted and whether Romney is performing better than he actually is.
Here is one of many articles written about this point. Good stuff for figuring out electoral polls actually work.
