It’s
easy to make too much of the idea that this election is going to be decided
between two historically and widely disliked candidates. But this description does ring true in
Wisconsin. Voters there genuinely do not like either of the candidates.
Wisconsin provides a very interesting data set
from the primary in a way that other states don’t. All of the states we have
looked at so far have been closed primary states (Florida, Pennsylvania and
North Carolina). What that means is that
while primary data can be somewhat helpful, the doors were also shut on many
potential voters whose intentions we wish to know. In closed primary states, voters not
registered with a major party can’t vote, and if a Democrat or Republican voter
was attracted to a candidate from the other party it was often too late for her
to vote for the candidate she now preferred.
Wisconsin, however, has no such
rule, and as a consequence it provides us with an excellent sample of what the
Electorate will look like. Slightly more
than 3 million people from Wisconsin voted in the 2012 Presidential election,
and more than 2.1 million voted in the 2016 Presidential primary. In many ways
Wisconsin was the largest defeat in the election cycle for both Trump and Clinton.
It was stand alone day with nothing good to report for either one of them. When putting aside caucuses and just looking
at primaries, it was Hillary Clinton’s 5th worst state and 3rd
worst outside of Sanders home New England region. (West Virginia and Oregon
were the other two and both came long after the race was over.) On the Republican side it was Trump’s third
worst state. (Texas (Cruz’s home state) and Idaho were worse, but since they
were on days where Trump won a lot they were easily forgotten.) However, what is interesting is that looking
at these results in total gives us a rough idea of each candidate’s various
advantages and of why Trump confronts a significant mathematical challenge to
do more than make it competitive. Obama
beat Romney here in 2012 by 213,019 votes out of 3,068,434. The 2016 primary saw 2,113544 votes cast. So
the primary is a decent cross-section of the entire electorate, and we can
meaningfully begin the process of talking about the entire electorate by walking
through the primary.
Hillary Clinton received 46,393 more votes
than Donald Trump in the Wisconsin primary. As a reminder no one in Wisconsin
was barred from voting for either them, even by a voter registration deadline
because of Wisconsin’s same day voter registration laws. (Voter ID was an issue
but has been softened for the general, which likely helps Clinton) While this may not seem like a large number
of votes, it was more than four times Kerry’s margin over Bush in 2004, and
more than six times Gore’s margin over Bush in 2000. So it is not nothing. This
brings us relatively quickly to the supporters of Ted Cruz and Bernie Sanders.
Bernie Sanders actually beat Ted Cruz in vote total in Wisconsin by 36,807
votes. Again not a small total but let’s give Donald Trump the benefit of the
doubt and assume that the Sanders and Cruz votes merely cancel each other
out. Supporters of these four candidates
accounted for 90% of all primary votes and about 63% of all votes cast in the 2012
Presidential race. So our analysis so
far has Clinton leading Trump by the same 46,393 by which she beat Trump on
primary day.
It’s the remaining roughly 182,000 primary
votes that are more of an enigma. They are 85 % John Kasich votes, 13% votes for
other Republicans on the ballot and 2% for Democrats who were not Sanders or
Clinton. This is a group in which Trump
desperately needs to make up ground, but one in which he may struggle. The
Kasich voters, based on all of the demographic information we have, are a
serious problem for Donald Trump. This
is likely even more so in Wisconsin where Kasich support was some of his weakest
in the North. Voters there realized that
a vote for Kasich might aid Trump and many people voted for the more conservative
Cruz, even perhaps against their natural inclinations. So the Kasich voters
here might be true moderates. More than 2/3rd of Kasich voters
listed “scared” as their reaction to Donald Trump winning as opposed to excited,
optimistic or concerned. Scared Kasich
voters, who still could not bring themselves to vote for Cruz seem like a tough
demographic for Trump. What is amazing is Trump will almost certainly win this
group of voters overall. The question that
remains is will it be by enough to erase Clinton’s lead? Trump needs to win this group with about 63%
to erase Clinton’s lead. Anything less
and the world looks very cold for Mr. Trump in Wisconsin; however given
partisan loyalty he may just be able to do better. Let’s give him 80% (basically amounts to
everyone who wasn’t scared and 2/3rds of the scared) just for the sake of
argument. This would leap him ahead by about 63,000 votes. For him, it is a start.
But this is where things become
even more difficult for Mr. Trump. There still remains a pool of roughly 950,000
voters who limit voting to presidential years.
This group voted for Barack Obama by about a 2-1 margin. Let’s be kind to Mr. Trump with our
math. We can assume that instead of
going to Obama 627,000 to 323,000, these 950,000 voters are not the same group
from 2012 to 2016. Some of the 950,000
from 2012 may have voted for Obama last time but voted in the R primary this
time. So let’s drop the margin to 570,000
to 380,000. Let’s also assume Democratic demoralization, so instead of a pool
of 950,000 there are only 700,00 and with the same margin. That would net
Hillary Clinton 140,000 votes and a relatively easy victory, even factoring in
Trump’s prior 63,000 lead. Even if you
cut the available pool down to 450,000 that would still net Clinton a 90,000
margin from this group. Netting this
against the Trump lead produces a Clinton 27,000 vote margin. It’s not much for
comfort, but it is also not nothing.
Reviewing these figures, it would seem the major potential source of
improvement for Trump might be in the Sanders/Cruz exchange, particularly with
more 3rd party bleeding from among the Sanders voters than among the
Cruz voters (even though 55% of Cruz voters were also scared of Trump). But hoping for such an advantage seems very
risky for Trump. This may be why the
Clinton campaign, despite close polls, has not gone up on television in
Wisconsin.
We would be remiss if we did not also provide a little bit
of county level flavor, which these posts have been keen on. When looking at a
county level, we quickly see that Trump’s problems stem in particular from the
three most Republican counties in the state. These three counties (Ozaukee, Washington and
Waukesha) have been the three most Republican counties in each of the last four
Presidential elections. They are the heart of Red Wisconsin and contain all the
Milwaukee suburbs. (Unlike other swing state suburbs, they have not budged in
their redness one bit). They were also far and away Donald Trump’s worst
counties in Wisconsin. He averaged about 23% in them when he received 35%
statewide. He got about 38% in the rest
of Wisconsin. It was these counties which dragged him down.
What is particularly interesting is
that, outside of Trump’s two worst primary states (Texas because of Cruz and
Idaho because of Latter Day Saints), he only got less than 25% of the vote in
18 counties total across all states in all primaries. Three of them were the
Ruby Red Three in Wisconsin. Known collectively as the WOW counties. These
counties make up the very base of the Republican Party in Wisconsin. The
provided Mitt Romney with a 132,536 vote margin. Since Obama’s 213,109 margin is what Trump
must make up, bleeding anywhere further is dangerous for him, yet he does look
likely to bleed at least a little in the WOW counties. (Seeing him winning them
by only 100,000 would not be a shock.) Otherwise Obama saw gains over Kerry in almost every
county in the state, and seeing bleeding back toward the Kerry-Bush number is
possible. Turnout might also be down, but Kerry still won so you need to see
deeper bleeding than Kerry faced and likely a lot of it to make up for WOW
problems.
Northwest Wisconsin is showing an albeit minor
trend back toward Republicans, but the Democratic gains in the Counties around Dane
(Madison), such as Columbia, Sauk, Richland, Green, and Iowa seem to make up
for any weakness in the Northwest. It
should be noted that Western Wisconsin, once you get west of the Dane County
area, was Hillary Clinton’s worst region in the primary. Democratic strongholds
like La Crosse (Obama 58% in 12) and Eau Claire (56% in 12) are places of
particular concern for HRC. However
Trump’s weakness in the WOW counties likely negates any strength he can pull
here. The WOW counties also have far more votes: 374,744 against 118,268 for LaCrosse
and Eau Claire combined. The gap would
remain even if you were willing to add in more Western Wisconsin Counties. Basically so long as the Dane Region and
Milwaukee proper stay true to form and there is bleeding in the WOW’s, there is
not much a Republican can do to win Wisconsin. Trump is not out of it here, and
it may even be that this is a state in which his odds are slightly better even
than in Pennsylvania. But getting 2/3rd
up the hill is not the same as getting all the way up the hill. We don’t quite
see how Trump gets all the way up the hill here and obviously neither does the
Clinton campaign.