Nate Cohn at the New York Times explains why the poll results from Monmouth University might be skewed toward Hillary Clinton a bit too much.
Most other polls have tended to show a tight race, but Monmouth shows a blowout. Why?
One possibility is that Mrs. Clinton has made gains over the last few weeks, thanks to Joe Biden’s withdrawal from the race or her performance in the Benghazi hearing. But another possibility is the sampling frame of the survey. There are good reasons to believe that the Monmouth poll excludes many voters who are supporting Mr. Sanders.
What’s a sampling frame? Basically, it’s the people who could be selected to participate in the survey. According to the Monmouth poll’s methodology description, the poll’s sample was drawn “from a list of registered Democratic voters who voted in at least one of the last two state primary elections."
These two conditions — being a registered Democrat and recent primary participation — exclude many of Mr. Sanders’s supporters.
It would have been more accurate for Monmouth to report that the poll results among registered Democrats skewed heavily toward Hillary Clinton, rather than suggesting the broader frame of Iowa caucusgoers. This kind of poll was what suggested Clinton's inevitability in 2007. There was, therefore, great surprise to the mainstream press and many casual observers, when Barack Obama won the upset that night.
Polls are of limited usefulness, and they certainly aren't suggestive of the broader picture if the frame is zeroed in on a tight group.