March 29, 2016
Wisconsin is very important on the GOP side. It’s potentially important for the Democrats. We have somewhat limited data now. We’ll have a lot to work with soon. Our goal today is to take a look at what those surveys might tell us when we see them.
Consider this a guide to upcoming polls.
In a perfect world, we’ll have the following options by the end of the weekend:
Marquette University: Marquette has polled Wisconsin on a quarterly basis for a couple years and monthly for the past few. They’re overdue for March. Their best result for Donald Trump was 30%, so if you see him at or above 35% it’s a very good sign for his chances.
John Kasich never registered well, scoring at 2% in January before moving up to 8% in February. If he’s within a few points of the top, he’s made a considerable amount of progress.
Ted Cruz posted a 19% number in mid-February. With his chances based on consolidating support that would have gone to Marco Rubio and other candidates now out of the race, he should find himself very close to Trump or ahead if he’s going to win.
CNN/ORC: This is the most consistently pro-Trump measurement. With the importance of Wisconsin, I’d expect to see something from them in the next few days. The Donald never beats his final CNN number, but does get very close or equals it in states where he closes well.
It’s a great way to establish his ceiling. Take the CNN number, add 20-25% of the undecided voters (most CNN polls don’t have many) and you have his absolute max. A 35% result isn’t real good for him. It’s the equivalent to 30% in the Marquette poll. He needs 38 to 40 percent here.
CBS/YouGov: This is our Cruz ceiling. While CNN’s methodology of dialing without regard to voting history or registration records is Trump-friendly, this Internet-based survey tends to bring Cruz supporters out of the woodwork.
Normally, Cruz beats his poll averages on election day, but if you take the YouGov result and give Ted 50% of the undecided, it’s about his max. Trump does fairly well in their surveys too, though it doesn’t give him the same boost that it does Cruz.
Kasich tends to suffer. If he’s trailing the two by several points, it doesn’t necessarily mean he has no chance of winning. If for some reason it looks like a legit 3-way race, you should assume this is very good for the governor.
NBC/WSJ/Marist: These guys are the Trump skeptics. A few days before South Carolina, NBC/WSJ had Cruz leading him in a race The Donald won by double digits. They consistently have the two much closer in national polls than others do.
If Ted isn’t leading Trump in either this poll or the CBS/YouGov, it’s likely he won’t defeat him on election day, unless events change between the poll dates and the vote. They’re also fairly conservative on front running numbers. A 38% result for someone here is a big score.
ARG: The Official Pollster of the Kasich Campaign. They are always his best number. If he’s not leading in an ARG poll taken within 2 to 3 days of the vote, he won’t win. Kasich’s ARG number plus 30% of the undecided is his ceiling.
They tend not to like Cruz very much. If he’s even or ahead of Kasich, it’s a very good sign for him. Wisconsin is an open primary, which favors both Trump and Kasich ahead of Cruz. We’ve seen Trump do very well with some types of Independent and even a few crossover Democrats.
In New Hampshire and Ohio, the same was true for Kasich. Cruz does best among solidly conservative Republicans. ARG’s methodology usually allows for more non-GOP voters grabbing a ballot in an open primary than others do.
Tracking: It’s great when a pollster takes multiple readings over the final week of a contest. Daily is fun, but not necessary. We do need multiple readings from at least one pollster to see if there’s any movement. Otherwise different results over the weekend might have more to do with the pollster than movement in the race.
It looks like Optimus Consulting is going to perform this function. They have a readily accessible survey from 3/22 to 3/24, and have polled more recently, though that one is harder to get at. Hopefully, we’ll find a link soon.
They haven’t participated in previous states, so we can’t measure for a house effect favoring a particular candidate, but they will let us see which direction things are moving. So far, their second poll shows no movement from one candidate to another, but the amount of undecided voters grew.
At the moment, Optimus is both the most recent polling, and the one indicator this is a three candidate race, as they have the candidates bunched within 4 points from top to bottom.
It’s far easier to use polling on the GOP side. Every important primary contest with several data points has proven out within a couple points of the polling averages. Not a single major surprise yet. If we get the surveys listed above, we’ll be able to project things very well.
For Bernie and Hillary, not so easy. Let’s leave aside the caucus polls. After Iowa and Nevada, they consistently shorted Bernie by the tune of 40 to 50 points. Let’s leave aside polls from Southern states, which often clipped Hillary’s support by a few to several points.
The most relevant examples are Michigan, Illinois, and Ohio. Iowa is worth a bit of a look, it’s in the general neighborhood and acted more like a primary than caucus in polling and turnout. Minnesota is next door, but we need to remember it was a caucus-like caucus.
Bernie beat the final poll average by more than 20 points in Michigan. Illinois was exactly correct. Iowa was only slightly pessimistic. He ran behind the estimate in Ohio. Even adjusting for caucuses, Minnesota (with extremely limited sample) was the biggest miss so far.
Inconclusive? Not completely. The final Ohio and Illinois polls were from disproportionately friendly sources. As with the GOP, there are some house effects at work. The surveys included in the final Real Clear Politics averages were neutral to favorable for Bernie.
Unpacking this a bit further, he legitimately closed well in each state. However, he was trailing badly in Illinois and Ohio before the final kick. The mix of polls made it appear that he made up more ground than he actually did.
So look for these:
Marquette University: Same logic as the GOP. There’s a history here. The January and February results were very close. If Bernie is going to get anywhere near the margin he needs to start catching up on delegates, this should show him ahead by at least a few points.
CNN/ORC: Bernie normally does well on these. In national polls, they generally give him better results than the average pollster. However, in Michigan, they were just as off as the norm. In Ohio, they had Hillary up by 30.
They didn’t survey in Illinois or Minnesota and were Bernie-favorable in Iowa. For one more round, you can probably figure they are even or have a slight Sanders lean. If he’s going to do well, he should at least find himself ahead in their poll.
CBS/YouGov: They’ve uniformly had a pro-Sanders house effect. In Michigan, everybody was so far off, it wasn’t enough for him to lead their poll, but it was his best showing. Normally, you would take his number and deduct a couple points to get to the actual result.
NBC/WSJ/Marist: They’re generally more accurate than most on the Democratic side. If this is another Michigan, they’ll be wrong too, but otherwise would tend to trust their number as at least being in the ballpark.
PPP: For a while, they were leaning a bit towards Hillary, but in recent contests have actually given Bernie more credit than some of their competitors. In Michigan it made them less inaccurate. In Illinois they were dead on. In Ohio, they shortchanged Hillary.
On the whole, their results are worth taking seriously, though again, the Democratic contest is proving difficult to poll in northern open primaries.
While the Democratic surveys won’t give us the same confidence as we’ll have on the GOP side with a few more polls, it should at least give us a range of possible outcomes. A tracking poll would help too, but Emerson is the only pollster to have produced a single survey so far in March.