2016 Democrats, March 15, Poll Watch, State of the Race, Uncategorized

Last Minute Polling Shifts (Democrats)

March 14, 2016

We might get a couple last minute surveys between now and the time voters start heading to the polls in the morning, but most of the data is in. We can see if Bernie Sanders has made progress since his Michigan victory.

We can see if Hillary Clinton is still well in front in North Carolina and Florida. For a couple weeks, there was a gap between how you might think Bernie would do in Illinois and Ohio and what the polls were showing. A few new surveys give us a different picture.

Let’s do a quick tour of the March 15 map and see what the polls are saying:


Sanders is talking about making this closer than expected, not winning. There is no indication this is anywhere near a toss-up. Michigan is our example of a giant upset. It was the largest average polling deficit for a primary winner since they started measuring this stuff.

He beat his polling average by 23 points, second only to Gary Hart in New Hampshire, back in 1984. The current gap in Florida is 29.6%. Unless you think he’s going to beat the numbers by even more than he did in Michigan, there’s no chance here.

In Michigan, there was dissention between pollsters. Some saw it as a 10 to 12 point race, others well in to the 30s. Here, much more unanimity. The last five surveys have margins of 28, 27, 28, 25, 26.

This type of agreement is very unusual so far. Previously, Quinnipiac, CNN/ORC, and NBC/WSJ/Marist are fairly favorable to Bernie in their measurements. In Michigan, only NBC/WSJ/Marist participated.

They found him trailing there by 17, not 27. CBS/YouGov is another survey that skews towards Bernie. In Michigan they had him down by 11, one of the things that made victory seem almost impossible. In Florida it’s 28.

Quinnipiac is the one pollster to survey Florida in March, both before and after Michigan. Sanders made a little progress, but only from -30 to -26. There is simply nothing to indicate an upset is in the offing.

North Carolina

Clinton is still the clear favorite here based on polling, but it’s marginally more interesting than Florida. The Real Clear Politics average has her leading by 23.7% instead of almost 30.

Instead of no surveys under a 25 point margin, we have a new one from PPP at 56/37. That’s still a big gap, and Hillary is safely over 50%. Even giving the undecided voters to Bernie doesn’t make this super close.

We’re still in a part of the country where Hillary has outperformed her final polling numbers, not Bernie. If you want to find drama, it’s with a few other polls from back in mid-February.

At that time, Elon University saw a 10 point race, and SurveyUSA had it within 15. Conveniently, PPP polled then too, finding a 17 point gap. As mentioned above, they’re showing a 19 point difference now.

So, you can argue Elon and SurveyUSA might get similar results today to what they found a month ago. Another March pollster, Civitas, got results within a couple points of what they found in mid-January.

This indicates things in North Carolina are fairly stable. If the closer race scenario is the more accurate one, more like how things are up North, then Bernie can make this semi-close. If it winds up 55/45 or 56/44, the narrative is bad for Hillary.

That’s probably the most Berners can hope for here.


Finally the polls match underlying conditions. The goal is to use polls to make decent predictions, not to say that someone has a chance in one state because the polls were showing a similar deficit when he won somewhere else.

When we modelled the states before anyone began voting, Ohio looked like a Sanders opportunity, but not quite as strong for him as Michigan. His economic message should work better in the state that is suffering more.

Bernie has another complication that we hadn’t considered back in January. John Kasich is fighting for his political life tomorrow. We figured Sanders would need open primaries to make up for Clinton’s advantage among registered Democrats.

In Michigan, a huge advantage among Independents put him over the top. Underestimating their turnout for Bernie contributed to the polling miss. The same could happen here, but Kasich provides an alternative for voters who reject both Clinton and Trump.

While Kasich did pull a few of these voters in Michigan too, he still wasn’t the home state governor, so the impact is likely greater here. Given that Bernie only won by 1.5%, every little difference matters.

Bernie is making progress here. Quinnipiac had Hillary ahead 52/43 from March 2 to March 7. In the newer poll, taken from March 8 to March 13 it’s 51/46. As you can see, that’s more a matter of undecided voters going to him, than voters leaving her.

The most favorable poll for him is one taken by PPP from 3/11 to 3/13. In this one Hillary leads 46/41. It’s helpful for a couple reasons. One, Quinnipiac usually has a more Sanders favorable model than PPP.

If this is still true, the PPP survey is better for him and would show further momentum. It was taken over the last 3 days, while Qunnipiac’s survey is for 6. Also, PPP has Hillary below 50%. It’s much easier to win undecided voters than those leaning to the other candidate.

The above isn’t a complete picture. Monmouth polled from 3/11 to 3/13 and found Hillary ahead 54/40. They found her ahead by a very similar margin in Michigan in a poll taken equal distance from the actual vote.

If you want to play the Michigan game, this shows a tied race. They were off by 14 points there. CBS/YouGov has Clinton up 11 here, just like in Michigan.

PPP has a mild house effect toward Clinton, and Bernie does better with undecided voters in Northern states, so you can adjust that one to roughly even.

Quinnipiac adjusts to being almost even if you account for having the poll go back a few days further than the others. NBC/WSJ/Marist had Hillary ahead 58/38, and normally does not favor her, but most of the poll was taken before the Michigan results.

I can take all the data and argue it’s almost a dead heat. I just can’t make a case Bernie is ahead. He’s not. If you had no poll information and only knew he’d narrowly taken Michigan, you would calculate an even race to a very mild Clinton advantage.

This one comes down to which voters show up. However, if the contest is decided by more than 3 to 5 points, there’s far more chance of a solid win for Hillary than Bernie. A win for him is likely incredibly narrow.


A week ago, this looked like an easy victory for Hillary. Two pre-Michigan surveys had her leading by 37 and 42 points. Both were off, even for then. One had Hillary head by almost 30 points in the Downstate area, which was a virtual impossibility. Bernie will win that area tomorrow.

Still, it’s really hard to adjust 37 points down to a close race. The ground has shifted noticeably in the past few days.

NBC/WSJ/Marist showed a 51/45 lead for Hillary. They took the survey from 3/4 to 3/10, meaning they’ve missed some of the post-Michigan momentum. It also doesn’t take into account anything that may have ensued from the Trump rally cancellation in Chicago on Friday.

We don’t know if the turmoil, which included many Berners as protestors, even if not at all sanctioned by the campaign, will help or hurt Sanders in the state. My gut reaction is it will help significantly with voters of color in the Chicago area.

As we know, Hillary has done far better among non-white voters so far, in part based on the perception that she’s done more to fight for them over the years than Bernie. Having a full squad of people shouting down Trump promotes the idea Team Sanders is doing something to stop a candidate many Democrats regard as a bigot.

That’s all speculation. So is figuring having thousands of young voters in the streets on Friday increases the chances they turn out to vote on Tuesday. Likely, but not provable until it happens.

What we do know is Bernie has at least a shot in Ohio. During the same period NBC/WSJ/Marist had him trailing by 6 in Illinois, he was down 20 in the Buckeye State. So, by the transitive property, he’s got a very good shot here.

The two most recent surveys look even better. CBS/YouGov has Bernie ahead 48/46 and PPP has Clinton ahead 48/45. Since CBS/YouGov (an Internet poll) is normally more favorable to Sanders than PPP, consider both surveys a similar result.

It leaves them effectively tied. Sanders clearly has the momentum. While Ohio shows him making progress, Illinois indicates a flood. Hillary is below 50% in the two newest polls, barely over it in the other.

The state is definitely in reach for Bernie. It’s not a done deal just yet. The NBC/WSJ/Marist poll is adjustable upward to account for time and momentum. However, the sample breakdowns are very favorable to him.

They indicate almost as many voters between 18 and 29 as 65 and over. That’s not normal. It could happen. The younger voters of Chicago seem very energized, but if he was trailing in a survey with those data sets, it’s too early for Sanders to plan his Illinois victory speech.


Not much data here. One mostly pre-Michigan poll with Hillary ahead 47/40, and a new PPP survey with Bernie ahead 47/46. Based on the results of neighboring states, I’d assumed Sanders was ahead even after the first poll.

The second gives me no reason to change that assessment. A Clinton win tomorrow would surprise me a bit. Bernie is relying fairly heavily on the type of low-to-middle income white voters that propelled him in Oklahoma and Kansas.

There’s a slight chance of a backlash against Berners gone wild in Chicago the other day, but presumably most voters who were pro-Trump were already planning on voting for him in the open primary. It’s not like people have mixed feelings about The Donald.

Unless you see a late poll putting Hillary clearly ahead, assume Bernie will win here, if perhaps not by that much. Until results prove otherwise, I’m going to continue to pay at least as much attention to results in neighboring states as polls in those which haven’t voted yet.






Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s