Monday, September 26, 2016

Demos in Marquette Poll indicate Dem leads may be bigger

I finally had a bit of time to dig into the data from the most recent Marquette Law School Poll of Wisconsin, and get my thoughts together. Bottom line- Charles Franklin and company may be making the state’s races appear tighter than they are.

I wanted to use the registered voter figures to get the picture, and as always, looked to the cross-tabs to see what the real information is (ALWAYS look to the cross-tabs, folks). I’m going with registered voters because that’s a larger data set than “likely voters” and because a sizable amount of those voters were not listed as likely voters even though they said they were “very likely” to vote (over 11% in the latest Marquette Law poll fell into this category, and another 7% said they were “50-50”).

Let’s review the “topline” totals in the poll released last week in the two statewide races for federal office in Wisconsin.

Marquette Law Poll late Sept. 2016
President (2-way Trump vs. Clinton) Clinton 43-39
President (4-way Trump, Clinton, Johnson, Stein) Clinton 39-35
WI Senate- Feingold 46-40

The first item which I wanted to investigate is the Partisan ID question, which was the main reason the race “tightened” in the Marquette Poll that came out earlier in September, and I was curious to see where it fell this time. As a point of reference, I also looked at the 2012 presidential exit polls and the 2014 midterm exit polls, to see what the Party ID was there when comparing self-described Democrats vs self-described Republicans.

Party ID, 2012 + 2014 elections, Wisconsin
2012 exit poll Dem +5
2014 exit poll GOP +1

Now let’s look at the previous Marquette Law Poll, and this current one. I’ll even include the “leaners” among the group who claim to be independent, but usually support one party or the other.

Party ID, Marquette Law Poll
Early Sept. Dem +1
Late Sept. Dem +2

Basically a middle ground between 2012 and 2014, even with it being a presidential year like 2012. Maybe that’s true, but I find myself skeptical of that.

My skepticism of claims of a close race grew when I dug into various demographics of these polls and compared them with the 2012 exit poll results in Wisconsin, when Barack Obama beat Mitt Romney by 7%. In particular, look at two of the constituencies where Dems have traditionally had the largest advantages- young voters and minorities.

Voters 18-29, Wisconsin
2012 election Obama- 60-37
Late Sept. MU Law Poll President 2-way- Clinton 40-35
Late Sept. MU Law Poll President 4-way- Trump 31-26
Late Sept. MU Law Poll WI-Senate- Feingold 38-37

Do we really believe Donald Trump is leading among Wisconsin voters under 30, even with Gary Johnson pulling nearly 30% of that youth vote? And do we really believe that Russ Feingold and Ron Johnson are basically tied with that same group? Call me doubtful.

Same pattern repeats when you look at the white vs non-white breakdown of the Wisconsin electorate in 2012 vs this survey.

White voters, Wisconsin
2012 election- Romney 51-48
Late Sept. MU Law Poll President 2-way- Trump 42-41
Late Sept. MU Law Poll President 4-way- Clinton 38-37
Late Sept. MU Law Poll WI-Senate- Feingold 48-41

So if Clinton and Feingold are doing better with white voters in Wisconsin, why isn’t the Marquette Poll showing a 10-12 point blowout? Here’s why .

Non-white voters, Wisconsin
2012 election- Obama 84-15
Late Sept. MU Law Poll President 2-way- Clinton 50-24
Late Sept. MU Law Poll President 4-way- Clinton 42-25
Late Sept. MU Law Poll WI-Senate- Feingold 36-31

Again, either there’s been a major move by non-white voters toward the party that has a white supremacist running for president and a state GOP that is actively suppressing minority votes, or Dems are likely in a much better position than the Marquette Law Poll says they are.

By the way, I'm not necessarily saying that Franklin and Marquette Law are intentionally keeping this close- it's really hard to get people to pick up when it's a weird phone numbers calling, and that's especially true with the demos I mentioned above. You can only report on the info you get, and when you're dealing with an especially non-responsive group, it can be more likely that the info is misleading.

Might the fact that the race really isn't that close be an explanation why we haven’t seen much in terms of presidential ads and appearances here in recent weeks? Maybe that's changing, as Drumpf and Pence were scheduled (where else?) Waukesha this week, (oops, scratch that Pence visit) but the GOPs also have to come here because their road to 270 electoral votes is so limited that they have to try to get overly-white states like Wisconsin into their column, even if they’re trailing badly. The Clinton campaign only recently has started to bring sent in campaign surrogates, and even those appearances aren’t public events, with the airwaves (thankfully) relatively empty of ads.

Now, maybe Team DNC/Clinton is making a major miscalculation, and the “angry old white guy” strategy of Trump has put this state in play. But the crosstabs of the Marquette Law Poll indicate that if the demographic mix of Wisconsin’s voters is similar to 2012, and youth and minorities vote similar to how they did in 2012, then both Russ Feingold and Hillary Clinton have sizable leads here. And that reality does not change regardless of what the a media reliant on ads and clicks tries to tell you.


  1. Probably fodder for different post, but why are "millenials" attracted to Gary Johnson anyway? Is it just the "not Trump, not Clinton" crowd or is it an attraction to the next flashy thing with no real understanding?I'm not so sure voters in that demographic understand what (an ex-Republican) Libertarian President would do with a Republican Congress...

    1. A lot of it is pot and a little anti-war and some "none of the above" factor. I anticipate a lot of them to come around, especially as it gets more obvious that Hillary will win