Wednesday, October 12, 2016

MU Law Poll- tight races, if the young and the moderate don't vote

Wanted to give some insight into today's Marquette Law School poll in Wisconsin. The statewide races are closer than I wanted to see, especially on the Senate side.
A new Marquette Law School Poll finds 44 percent of Wisconsin likely voters supporting Democratic candidate Hillary Clinton for president and 37 percent supporting Republican Donald Trump, with Libertarian Gary Johnson at 9 percent and Green Party candidate Jill Stein at 3 percent. Six percent do not express a preference, saying that they will vote for neither candidate, will not vote or don’t know how they will vote...

In the new poll’s head-to-head matchup (as opposed to the four-way race), Clinton receives 46 percent to Trump’s 42 percent of likely voters, with 9 not giving a preference. In the September head-to-head matchup, among likely voters, Clinton had the support of 44 percent and Trump was supported by 42 percent, with 12 percent not expressing a preference.

In Wisconsin’s U.S. Senate race, 46 percent of likely voters support Russ Feingold, 44 percent back Sen. Ron Johnson and 4 percent choose Libertarian candidate Phil Anderson. Five percent do not express a preference. In September, Feingold was supported by 44 percent, Johnson by 39 percent and Anderson by 7 percent, with 10 percent not giving a preference.

In a head-to-head matchup, among likely voters, 48 percent support Feingold and 46 percent support Johnson, with 6 percent lacking a preference. In September, Feingold was the choice of 47 percent to Johnson’s 41 percent, with 11 percent not stating a preference.
As usual, I will eventually dig into the crosstabs of the larger registered voters’ sample to find the real story behind these numbers (the topline numbers really didn’t change much between likely vs. registered anyway). But I want to first note a couple of items on the demographics themselves, because it seems to explain why these races aren’t the blowouts many of us thought they might be.

I’m going to point out a stat that’s consistently predicted the leaning of the topline results, which is Party ID. As I mentioned a few weeks back, the “tightening” in September that we saw in the races was mostly explained by the fact that fewer self-identified Democrats were responding to the Marquette Poll. So let’s dig into this month’s poll, see what we find, and compare it to the November 2012 exit poll in Wisconsin.

Party ID, with Independents included
Early Sept- Dem +2.4
Late Sept- Dem +0.4
October- Dem +0.7
2012 Exit poll- Dem +5

In a time period where many Republicans are embarrassed to admit Trump is their nominee, it seems odd that would there be more Democrats that would fall away into the “Independent” category (unless MU Poll Director Charles Franklin is adding on a curve for WisGOP voter suppression…). Maybe it’s true, but I’m going to be skeptical about that one.

Now, let’s do the same for another key demographic, the youth vote. As you’ll see, the Marquette Poll couldn’t get many of those darned kids on the line.

Voters 18-29 as % of total Wis electorate
Early Sept- 16%
Late Sept- 16%
October MU Poll- 12%

2012 exit age 18-29- 21%
2012 exit age 18-24- 12%

So sure, if we think the share of voters 18-29 in 2016 will be the same as the share of voters 18-24 in 2012, then we can nod, move on, and find it to be accurate. But I don’t (c’mon kids, I have faith in you to step up for democracy!). Again, maybe Franklin is adjusting for voter suppression hitting a disproportionate amount of young voters, or simply can’t get young respondents because who under the age of 30 answers phone calls from weird numbers? But I’d be surprised if 12% was the end result for the 18-29 vote on November 8.

That 18-29 crosstab is especially worthwhile to dig into because the youth vote has been very Dem-leaning in recent years, but young Wisconsinites are showing softer support in this poll for the Dem nominees for president and Senate than we’ve seen in the past.

18-29 voter preference, Wisconsin
2012 exit poll
Obama 60-37
Baldwin 58-39

Oct 2016 Marquette Poll
President 2-way- Clinton 43-32
President 4-way- Johnson (!) 33, Clinton 30, Trump 19

Senate 2-way Feingold 47-42
Senate 3-way Feingold 38, Johnson 32, Anderson 16

If that third party support among young people is soft, and some of those young voters come around to deciding that they want to vote for a candidate that can win (or can’t afford another candidate to win), then that means both Feingold and Clinton have room to grow. However, it also underscores that maybe Russ shouldn’t keep running as a “generic Democrat”, and should emphasize his independence such as his votes against the Patriot Act, Iraq War, and Glass-Steagal repeal, and his stances against oligarchs running elections.

There’s another demographic that I think was under-polled in the October Marquette Poll, and that’s self-described “moderate” voters. One thing that jumped out to me when I looked at demographics in this poll was how many more conservative voters there were compared to what we saw in 2012, but liberals stayed the same, and moderates were much fewer.

Ideology, Wisconsin voters
Oct 2016 Marquette Poll
Conservative/very conservative 41%
Moderate 31%
Liberal/very liberal 24%

2012 exit poll
Conservative 35%
Moderate 40%
Liberal 24%

Again, with how goofy and amoral Drumpf and Republicans have gotten in the last 4 years, do you think MORE Wisconsinites are calling themselves conservatives these days? Maybe things have changed that much outside of Madison in the Age of Fitzwalkerstan, or maybe that just reflects the fact that more old white people will pick up the phone regardless of who is calling. Regardless of why, that shift to more conservatives vs moderates is a big difference, because Ron Johnson is pulling between 70-75% of the conservative vote, and even Trump still gets 60-70% of them. By comparison, moderates favor the Democrats big time.

Oct 2016 Marquette Law Poll, Moderates
President 2-way- Clinton 56-28
President 4-way- Clinton 50, Trump 23, Johnson 15

Senate 2-way Feingold 56-35
Senate 3-way Feingold 52, Johnson 32, Anderson 5

Bottom line- if the actual voting population on November 8 has an ideological mix similar to 2012, then Dems are clearly in the driver’s seat.

The last stat I wanted to look at involves both of these underpolled groups, and the question of “How comfortable are you with the idea of ______ as president?” In particular, let’s zone in on those who would say they’d be “very uncomfortable” with one of the two main candidates being president. Or as I call it, the “OH FUCK NO!” response.

“very uncomfortable” if ____ was president
Voters 18-29
Trump 58%
Clinton 35% (lowest of any age group)

Trump 63%
Clinton 32%

So if those voters are backed up against the wall, they’re much more likely to be accepting of Clinton winning, and would vote for her to avoid the chance of Drumpf ever being president. So if the race somehow does get close again, I’d anticipate this to be a secret source of Clinton and Dem strength, and that could well translate downticket if Trump spoils the GOP’s reputation.

Now I’m not saying people that want Dems to win should be complacent because the MU Poll may have some dodgy demos. Feingold especially is still in the danger zone of having the election stolen away from him if too many young and moderate voters shrug their shoulders and don’t come out. With that in mind, Dems should use the Marquette poll as a warning, because this thing is not in the bag.

On the flips side, this poll and other recent events also showcases the Dems' opportunity. The MU Poll showed a major shift in the responses for president after the “pussy tape” came out on Friday (Franklin said Trump was WINNING 41-40 on Thursday, which immediately made me question this poll), and now with Trump attacking Paul Ryan in social media, lots of people have various reasons to vote out Republicans. This may perhaps put some House and State Legislature seats in play that previously were not. And if the presidential poll of Dems +7 turns into Dems +12 statewide, that can break the WisGOPs’ gerrymander, and the bottom could fall out. So decent liberal people shouldn't get discouraged by the lame toplines in this poll - just prove it wrong by voting in big numbers.


  1. MLSP properly recontact phone numbers that didn't answer before; since the field period was Thursday through Sunday, the Thursday interviewees will tend towards those who are in a position to pick up their phones on a Thursday. Naturally that tends towards retirees (who are disproportionately Republican), so it shouldn't be too surprising that the state of the race has a dependence on the day within the fielding period.

    If you go back to the early October 2012 poll, the Obama-Biden ticket was winning by 1 point, Baldwin was losing by 1, and the party ID was D+3 (D+1 with leaners). In the end the Obama-Biden ticket won by 7 points, Baldwin by 6, and exit poll party ID was D+5 as you mentioned.

    Either there was a shift in the last month of the campaign, or Marquette's likely voter screen ("I am certain to vote") doesn't accurately capture all the people who will actually vote in the election: perhaps those generally inclined towards Democratic candidates only become sure they will vote as the election gets very near (in the late October 2012 poll, MLSP had Obama+8, Baldwin+4, and a party ID of D+5 with or without leaners).

    Anyway, that's just a working hypothesis. And sometimes you just get statistical flukes - if there weren't ever any, something would be wrong.

    1. Geoff- I didn't know about the "first response/recontact" thing, and that adds up to me, since old white folks will be more likely to be around and pick up.

      I also question the tightness of that likely voter screen, because "yeah, probably" doesn't qualify as a likely voter as it's defined today. But they will be likely to cast to a vote, and I question the demos accordingly.

      Yes, statistical flukes happen. But Charles Franklin and Company also get more hits if the race is interesting. And the media companies that prop them up (and the Bradleys....?) get more cash if certain results come in.

      I also expect an "interesting" adjusting to the last poll that more accurately reflects the electorate.

  2. MU had Burke and Walker tied at this point. lol.

    1. And had Thompson up 1 on Baldwin at this point and also had Obama +1 over Romney.

      Then in MU's last poll in 2012, it shifted to Baldwin +4 and Obama +8- pretty close to the final results. Same thing happened in 2014 with a magic shift towards Walker (although Franklin hedged by it being only Walker +1 with "registered" voters).

      This is why I dig into the crosstabs, and why I will not be surprised if/when Clinton and Feingold lead by relatively comfortable margins in the final MU Pol in 2-3 weeks.

      Polls are media-driven (especially the MU Poll, which has some partnerships with Milwaukee media that wants the click and ads), and are often used as psy-ops to drive up or down turnout and enthusiasm. KNOW THIS.