Looking at the RCP average for New Hampshire, you’d think that something is severely wrong with the polling. The last round of polling by Gravis, Monmouth, and Reach all had Trump at between 27 and 35 points, with Cruz and Rubio fighting it out with Kasich for second or third.
Then came American Research Group (ARG) showing that Kasich had soared to 20 points, and Cruz was down to 9. And now CNN/WMUR released a poll taken by the University of New Hampshire Survey Center (disclosure: I’m a UNH alum) showing Kasich at 6 and Cruz at 14.
The ARG and CNN polls cover similar date ranges, 1/13 – 1/18 for CNN and 1/15 – 1/18 for ARG. How in God’s green earth could they come to such completely different results?
So I looked at the ARG data collection method. They appear to have made 600 telephone interviews. UNH called 972 adults. Neither poll disclosed that they used an interactive voice response (IVR) system or “robocaller,” so I have to assume they used live interviewers. ARG wrote that they rotated the names on their list, but so did UNH. The questions appear to have been worded similarly.
The only relevant difference I can find in the polls is that UNH appears to have asked more questions, and asked favorability questions BEFORE asking who likely voters would support. Cruz got 14 percent in that poll, with Trump at 34 percent.
More interesting, in the UNH poll, 20 percent of voters named Ted Cruz as their second choice, if their first choice was not running–the highest by far. Only 8 percent named Donald Trump as second choice. Kasich had 6 percent as second choice and 6 percent as first choice.
Reach and Gravis both use IVR technology to poll voters, and Monmouth interviewed 414 voters using liver interviewers. Monmouth’s results showed Trump at 32 percent and Cruz at 14 percent, with Kasich at 14 percent. Second choice was Cruz at 21 percent and Kasich at 6 percent.
So either asking people to rate the candidates before asking who they’d vote for today changes answers, or something is seriously wrong with ARG’s polling. I’m inclined to believe the latter.
Based on what the polls are saying, N.H. is Trump’s to lose, but Cruz is waiting in the wings. Trump has zero support as a second choice, so any dropouts after Iowa go to Cruz (or don’t count). That could have some effect.
There’s been no new Iowa polling data posted to RCP since last week. It will be very interesting to see what falls out before the actual caucuses.