This article was published by The McAlvany Intelligence Advisor on Wednesday, August 10, 2016:
The difference between pollsters like Rasmussen and prognosticators like Nate Silver of Five Thirty Eight is like the difference between lightning and a lightning bug. Polling is tricky enough. Prognosticating is a whole other matter.
Rasmussen (and its peers) must develop the questions, determine whom to ask, decide when to call (weekdays, weekends, dinner time, etc.), and then confirm that they’re talking to the right person. They have to take into account that sometimes people will give wrong answers, or hide behind their answers because they have opinions that aren’t mainstream. Once the poll is taken, they must determine just how variable – how reliable – the results really are. That explains part of the plus or minus calculation included in them.
Silver (and his peers) build models based on past experience. It’s vastly more complicated