Wednesday, March 2, 2016

Super Tuesday Polling: How Did Monmouth Do?

To say the 2016 primary season has been surprising would be an understatement.  A presumed Democratic nominee who was supposed to coast to victory has faced a tough challenge.  The GOP race is now coalescing around a front runner who practically everyone would have laughed off less than a year ago.

This topsy-turvy situation has amplified the already significant challenges that pollsters face when trying to take the pulse of voters in presidential primaries and caucuses.  Super Tuesday presented the first large-scale polling test of the nomination contest.  Interestingly, only a few national polling organizations devoted significant resources to polling these races (more on that below).

Monmouth University conducted polls in four of the nine primary states in the weeks leading up to Super Tuesday – Alabama, Oklahoma, Texas, and Virginia.  We had a significant miss in the Oklahoma Republican primary where, like every other poll in the state, we forecast a Donald Trump victory when Ted Cruz was the actual winner.  On the other hand, Monmouth was the only poll to accurately forecast a Bernie Sanders victory in the Democratic contest in that very same state.  Monmouth’s forecasts were accurate in the other three states we polled.

Monmouth’s Democratic primary polls were closest – or second closest in the case of Texas – to the actual margin of victory of all the polls taken in those four states over the past month.  Our Democratic primary results compared to the actual margin of victory:
   TX :  Vote - Clinton +32  /  Monmouth Poll - Clinton +34
   AL :  Vote - Clinton +59  /  Monmouth Poll - Clinton +48
   VA :  Vote - Clinton +29  /  Monmouth Poll - Clinton +27
   OK :  Vote - Sanders +10  /  Monmouth Poll - Sanders +5

On the GOP side, Monmouth came closest to the actual margin of victory in Texas and Alabama:
   TX :  Vote - Cruz +17  /  Monmouth Poll - Cruz +15
   AL:  Vote - Trump +22  /  Monmouth Poll - Trump +23

Monmouth had the correct winner in the Virginia GOP primary, but the margin was tighter than our poll:  Vote - Trump +3 / Monmouth Poll - Trump +14.  It’s worth noting that our Virginia poll was conducted a week prior to Super Tuesday, specifically before the crucial debate when Rubio decided to take on Trump.  That performance appeared to help Rubio’s performance in Virginia, although not enough to change the overall outcome.

As mentioned, we had the wrong winner in Oklahoma GOP primary:  Vote - Cruz +6 / Monmouth Poll - Trump +12.  Every poll in Oklahoma had Trump ahead by a significant margin.  Examining all the state exit polls from Super Tuesday shows that Oklahoma actually had the highest number of Republican voters (33%) who said they made up their mind in the last few days.  These voters broke significantly for Cruz (37%) and Rubio (38%) over Trump (15%).

Despite the stakes involved in Tuesday’s races, only four national polling organizations polled in at least three of the nine primary states during the month of February.  In addition to Monmouth, CBS/YouGov and NBC/Wall Street Journal/Marist polled in three states and PPP conducted Democratic-only primary polls in all nine states.

In Texas, the one state where all four organizations polled the Democratic primary, Monmouth’s 32 point margin for Hillary Clinton was the closest to the actual result.  CBS/YouGov had the margin at 24 points, PPP had 23 points, and NBC/Wall Street Journal/Marist had 21 points.  In the Texas Republican primary, Monmouth’s 15 point margin for Cruz was also closer to the actual result.  NBC/Wall Street Journal had a 13 point margin and CBS/YouGov had an 11 point margin.

This primary season has been particularly tough to poll, so it is understandable why more national polling operations did not dip their toes in the water.  Despite the one missed call, Monmouth is pleased with its Super Tuesday polling results.

Full information on Super Tuesday polling can be found here:
Republicans:  http://www.realclearpolitics.com/epolls/2016/president/march1GOP.html