February 25, 2016
Apparently South Carolina has either extinguished the Bern or never felt it in the first place. Other SEC Super Tuesday states are looking extremely unhospitable. A recent Georgia poll has Hillary up 50 plus. Of the next 20 states to vote, Bernie is leading in 2.
Some of that is old data. A few states were last surveyed in 2015. Others not since people actually started voting. Still, aside from an epic lead in Vermont and pretty decent polling in Massachusetts, he’s not ahead anywhere that votes soon.
If you scan the Real Clear Politics compilation of recent polls, he has a pile of 20 to 30 point deficits. Outside of Vermont, only West Virginia shows a similar deficit for Hillary. A 10 to 15 point shortfall is a cause for celebration.
If the optimistic Texas polls are correct, he’s trailing like that, if the majority are correct, he’s down 60/30 or so, with the chance of getting obliterated on delegates next Tuesday. It’s not just the Old Confederacy.
Polls have him down 20+ in places like Pennsylvania, Maryland and New Jersey. You might wonder what the big news is. Hillary started as a clear favorite and looked on the ropes after New Hampshire, but revived herself in Nevada.
There is some indication Sanders is doing better with Latino voters, but African Americans are still largely rejecting him. Beyond being a polling matter, it appears this actually happened in Nevada and was the largest contributor to his defeat.
Seniors are almost as intractably negative on Bernie. On top of all this bad news, a few recent articles have pointed out several key March states have students on Spring Break during the primary or caucus. There’s no way that helps.
Betting markets have reacted to all of the above, slicing his chances of getting the nomination in half since Saturday. Hillary is back to being an 8 to 1 or 9 to 1 favorite. All of this seems reasonable.
One catch. National polls are showing a completely different picture. Based on the state polls, she should have a double digit lead in national polls. He’s ahead by large margins in Vermont and West Virginia. Neither are populous. She’s ahead by a lot in places with a lot of people.
Instead, a new poll released today from IBD/TIPP has Hillary up two, 45/43. The majority of the survey happened after Nevada, so any bubble deflation is taken into account here. It’s not an outlier. Fox News had Bernie up 47/44 a few days ago.
Overall, Hillary leads the Real Clear Politics average by 5. How is this possible? The math simply doesn’t work. The states without recent polls are smaller than those where he’s clearly losing. Either Bernie is holding a stealthy 80/20 lead in California, or something is off.
Actual voting results match the national polls. FiveThirtyEight is doing calculations to show how Bernie would need to do in each state based on being even nationally. They did one calculation based on what was known ahead of any results and just updated it based on Nevada.
Their latest estimates say Bernie should have won Iowa by 19 points instead of tying. He should have won New Hampshire by 32 points instead of 22. He should have tied in Nevada instead of losing by 5.
So, if we believe this (and I do), he should have trailed by almost 20 points nationally at the time of Iowa, 10 points nationally when New Hampshire voted, and 5 points now.
The RCP average isn’t a perfect proxy, but they had Hillary up by 15 when Iowa voted, 13 when New Hampshire voted and 5 when Nevada voted. Close enough. As mentioned above, she’s still at +5.
If you go down the FiveThirtyEight chart, Bernie should be down 4 in Ohio. He trails by 13. He should trail by 32 in Georgia, the average is 40. Michigan adjusts to a dead heat, instead Hillary is up 16.
We know the national margin is approximately 5. We know that FiveThirtyEight managed to calculate things so that the Nevada result is in line with the current national polls. They’ve given us a chart so that we can take the number for a tied race and subtract 5 from Bernie’s number to show where he should sit in state polls.
Only using polls taken in the past 7 to 10 days, far enough after New Hampshire for Bernie’s latest progress to register, we find an average gap between state and national results of 8 to 10 points.
Some are very close. He should lead in Vermont by 78 and leads by 76. He should trail in Texas by 18, he’s behind by 17. But he should have a 3 point deficit in Tennessee instead of 25. Curious, eh?
I have a possible explanation. Different pollsters. Most of the organizations contributing to the national average have not polled the vast majority of the individual states. The pollster most sympathetic to Bernie is CNN/ORC. Quinnipiac also gives him a boost.
With the exception of Quinnipiac in Ohio, none of the above mentioned states have surveys from either. There are wide gaps in results. Though Bernie is down by an average of 17 in Texas, some surveys have him within 10, others down by 30 plus.
For the entire contest, the disparity between polls is greater on the Democratic side than GOP. Sometimes it’s a good Trump state, sometimes not as much, but the polls are generally in the same range, especially on him. Not with Hillary and Bernie.
Just a couple weeks ahead of the Iowa vote, several pollsters still had Bernie down by 20-25 points. He tied. While they were reaching that conclusion, CNN/ORC and Qunnipiac had him leading.
I think the state polls are selling Bernie a bit short. This doesn’t mean he’s competitive in Georgia. It doesn’t mean he’s going to hit his March 1 targets. Being 5 points down nationally is not good enough on Tuesday. It will leave him with a large delegate deficit and no more than 5 victories out of 12.
But the polls are showing an even worse outcome. If it’s just the matter of Berniephile pollsters not balancing out Berniephobes, he’s likely to outperform expectations and have a good narrative.
If voters get influenced by what they’re hearing and stay home, it may become a self-fulfilling prophecy. We don’t have a good way of knowing if Bernie supporters are voting because they want to do their part for the cause, regardless of result, or if they think he can win.
Unless CNN/ORC and Quinnipiac begin weighing in to most of the Super Tuesday states in the next couple days, we may never know what might have happened if the state and national numbers were already in line.
Assuming I’m right, and assuming his voters aren’t demoralized (two major assumptions), Bernie will win Vermont like we think, win Massachusetts like most think he should, win Minnesota and Colorado, which both lack recent polling, and make Oklahoma very close.
The rest all seem out of reach. If he survives March 1, we’ll get up on the 2nd and take a look at applying this theory forward to Michigan on the 8th, and the big March 15 states.