Forecasting New Hampshire: Iowa Fallout Favors Rubio

January 26, 2016

It’s never too early to predict how results from a caucus that hasn’t happened yet will impact a primary that takes place still later.  The last few days in Iowa have helped Marco Rubio in New Hampshire.  Here’s how:

Trump moves ahead of Cruz

It’s still a contest, but Trump is now clearly ahead in polls.  FiveThirtyEight’s polls-only measure has Trump at a 57% chance of winning Iowa, Cruz 36%.  Overall, they have the two mostly tied (Cruz 47%/Trump 44%).

Things are close enough for a ground advantage to grind out a Cruz victory.  A high percentage of Trump supporters have never caucused before.  The debate on Thursday can make a big difference at the margins.

Forecasting Iowa: Update Two (Which Pollster Set is Dead Wrong?)

October 27, 2015

Polling has some variance, that whole margin of error thing.  It’s a snapshot in time, in this case months before anyone votes.  At this point the majority of voters say they haven’t made their final decision.  Even some who think they have might think differently later as events change.  Sampling methods vary.  Live phone interview is considered more credible, but many voters won’t answer calls from unrecognized numbers.

Still, with all those caveats, there’s still some overall consistency.  Donald Trump led everywhere until Ben Carson passed him in Iowa and now in one national poll.  You aren’t going to see a Huckabee Leads in Iowa headline tomorrow.  Speaking of the Hawkeye State, reading polls of their voters over the past few months was usually very straightforward, and remains so on the GOP side.

Though the margins may have differed slightly, they moved in the same direction.  Constructing a narrative was easy.  Ben Carson catching and passing Trump was visible weeks ago.  The underlying data showed him with more upside, higher approval rating, high amount of second-choice mentions, etc.  Eventually the inevitable happened.  Whether Carson can retain the lead is another matter, but he was going to have his moment in the sun.

Democratic voters are a bit harder to read.  The trend line is fairly consistent.  Between May and Labor Day, Bernie Sanders made up a lot of ground.  Polls from the spring showed him trailing by 40 to 50 points, with Hillary having quadruple his support (think 60% to 15% for example).  He spent part of September in a virtual tie with Clinton in the Real Clear Politics average for Iowa.

Buried within that time of September stasis was some inconsistency.  Within a week of each other, a CBS/YouGov poll put Bernie up 10, while a PPP survey had Hillary ahead by 21.  Huh?  The same thing happened in late August.  The Des Moines Register, via Ann Selzer had Clinton +7, sandwiched between three other polls with Hillary +25, +28 and +34.

By any standard Bernie was in better shape in September than August and has regressed a bit in October.  That much we know.  But he’s either still very much in contention (CBS/YouGov -3, Des Moines Register/Bloomberg -7) or in the process of being obliterated (Loras College -38, Monmouth -41).  It’s a huge difference.