Friday, January 22, 2016
Is The Newest CNN New Hampshire Poll Believable ?
The top chart represents the results of the CNN / WMUR Poll of New Hampshire likely Democratic primary voters. They questioned 420 voters between January 13th and 18th, and listed the margin of error at 4.8 points.
The other chart is the Gravis Marketing Poll of New Hampshire likely Democratic primary voters. It was done between January 15th and 18th of 472 voters, and has a 4.5 point margin of error.
You probably noticed that the two polls are vastly different -- with the CNN poll giving Bernie Sanders a huge 27 point edge over Hillary Clinton, and the Gravis poll showing that edge to be only 3 points. Which poll is correct? They both can't be.
For the past couple of days, CNN has been touting the results of their poll (and Sanders supporters have been bragging about it on social media). I think they are making a big mistake. If you do enough surveys, you will eventually get one that is a clear outlier (probably because of a poor sample). That seems to be what has happened here.
The CNN poll is vastly different from all the other recent New Hampshire polls. The average of the five other recent New Hampshire polls (Gravis, American Research, Monmouth, NBC/WSJ, and Fox) only gives Sanders an 8 point edge -- and the RealClearPolitics average (which includes the CNN poll) puts the Sanders edge at 12.6 points.
I do think Sanders has a very good chance of winning the New Hampshire primary, but it will not be by anywhere near 27 points. The CNN poll is simply not believable.
There are also two new polls out concerning the Iowa Democratic caucuses.
The top one represents the CNN / ORC Poll -- done between January 15th and 20th of 280 likely Democratic caucus attendees, and has a 6 point margin of error.
The other is the Monmouth University Poll -- done on January 18th and 19th of a sample of 570 likely Democratic caucus attendees, with a margin of error of 4.1 points.
These two polls also show a big difference -- with CNN giving Sanders an 8 point advantage in Iowa, and Monmouth giving Clinton an 8.4 point advantage. Both can't be right.
Again, it seems the CNN poll could be the outlier here. The RealClearPolitics average of recent Iowa polls shows Clinton with a 3.1 point advantage (very similar to the Monmouth poll).
Iowa is close, and the winner will be the one that gets their supporters to the caucuses -- but I still think Clinton will eke out a win in that state.