March 1, 2016
Welcome to the Land of 10,000 Questions. Democrats have one poll since last August. Republicans have two. None are in the past 6 weeks. The GOP data includes a high percentage of undecided voters, or those committed to candidates who aren’t around.
The most recent look at the Democrats has a result (Hillary +34) completely at odds with the demographics. This isn’t South Carolina. FiveThirtyEight isn’t willing to offer a forecast. I don’t blame them.
For better or worse, I’m not going to let a lack of information get in the way of a flawed prediction. If this turns out wrong, blame the lack of data. If I get close, clearly I’m brilliant.Democrats
Bernie Sanders 51.6%
I decided to ditch the poll from January. It doesn’t fit in. Bernie did worse than a PPP poll from late July/early August. PPP is normally pro-Hillary. Bernie was in better shape mid January than middle of last summer.
PPP surveyed Iowa and Minnesota at the same time last summer. He was a few points better in Minnesota. Since they tied in Iowa, that would give Sanders a small edge for Minnesota if you believe he’s in about the same position nationally.
While Bernie is clearly worse off than he was post-New Hampshire, the pre-Iowa bar is a lower one to match. These caucuses are not as well attended as the Iowa version, so Hillary’s ability to drive regular voters out may not have quite the same impact.
The Sanders campaign has focused heavily here over the last couple weeks. Combined with the support of Congressman Keith Ellison, I’m figuring it’s just enough to get him over the finish line. The Berners haven’t given up yet. He’s raised a ton of money in the past 48 hours.
Hillary Clinton 48.2%
Obama beat Hillary handily in 2008. While she’s reversed her poor South Carolina result, doing better with African American voters than Obama did against her, Iowa and New Hampshire indicate she’s not a sure a bet with the president’s Caucasian supporters.
When in doubt, the FiveThirtyEight numbers are useful. Though they didn’t want to predict the outcome without enough polling data, their latest demographic adjustments have the state as +21 for Bernie in a world where the candidates are tied nationally.
The two new national polls have Hillary ahead by 19 on average. So, if you deduct one from the other, this is about right.
Marco Rubio 28.4%
Rubio led the January poll, but I just concluded we should ignore it completely for the Democrats. Making it the basis for a GOP prediction doesn’t make a ton of sense. It also doesn’t mean we have to completely disregard it.
Especially in Iowa, there were a batch of pollsters who consistently underestimated Bernie’s support. There wasn’t any correlation between that miss and what they found on the GOP side. For the most part, they were about as accurate as the others.
If we go back to the other survey from around Halloween, Rubio ran ahead of his national average in that one too, even if he wasn’t yet leading. Though Rick Santorum won in 2012, Mitt Romney was the 2008 victor.
Rubio is positioned about where Romney was on his first run. On that round, he was the conservative alternative to John McCain and the educated voter-friendly counterpoint to Mike Huckabee. In 2012, when he finished third, he was more establishment center-right.
By Minnesota standards, Rubio is fairly conservative. He got endorsed by Tim Pawlenty, who filled a similar lane when serving as governor. Trump is definitely at a disadvantage compared to his position elsewhere on the map.
Cruz hasn’t spent time here in the past couple weeks like Rubio has, perhaps indicating he doesn’t think he can win. Throw all that together and it points to Marco.
Donald Trump 26.6%
Just because it doesn’t look like a Trump win doesn’t mean I’m brave enough to pick him to finish third. He got as high as 26% in the poll taken around Halloween. Even in the mid-January poll that had him third, it was still close.
Trump has definite momentum compared to where he was a couple weeks ago, and is polling at new highs nationally. There are some white, working-class areas on the St. Paul side of the Twin Cities, plus some rural voters elsewhere.
Wherever Trump lands is his approximate floor anywhere in the country. Don’t be surprised if this is the worst he does for the rest of the contest.
Ted Cruz 23.6%
The Cruz campaign has excellent data management capabilities. Ted desperately wants to finish ahead of Rubio as many places as he can today. He also wants a victory anywhere he can get it.
Despite this, he hasn’t visited the state recently. That’s not a reason to predict him to finish third by itself, but he was at only 4% in the Halloween poll. That was below where he was nationally at the time, meaning his floor here isn’t the highest.
Cruz did wind up second at 21% in the most recent survey and he hasn’t failed to beat his poll numbers on election day. Throw all that together and he’s here.
Ben Carson 10.8%
Carson was at 11% in mid-January. As a rule, he hasn’t lost much from that point anywhere in the country. The new CNN/ORC national poll had Carson back up to 10%. Minnesota looks like a high floor, low ceiling state for him.
He was ahead of his average when he wasn’t doing well nationally, and behind it when he was pushing for the lead. Carson almost never falls short of his polling numbers on election day either.
John Kasich 9.9%
This is the biggest guess of them all. Kasich was a total non-factor each of the times Minnesota was polled. Jeb and Kasich combined for 9% in January (7 points from Jeb.) The lazy thing to do is just add those together, assuming that Rubio and Bush voters aren’t that similar in Minnesota.
That’s what I did. It’s also roughly in line with where it looks like Kasich will wind up in a few northern states over the next week or two, with a mild deduction for being a caucus.