Election Time

The last few months have been a blur, and I apologize that it’s been so long since I’ve managed to put up a post. I spent the summer and early fall doing voter microtargeting models and analysis. Then, beginning a little over a month ago, things heated up with CNN. As I’ve noted before, I’m part of CNN’s decision team; we’re the analysts you see in the background, studying the way the vote comes in on election night, and who ultimately decide when to project the outcome of particular races.

In the weeks leading up to election night, we would regularly rehearse with simulated data; most of us simply connected via conference call and computer for this. Then, we all spent the weekend before the election in Washington, DC for studio rehearsals. Finally, we were of course up all night on Tuesday the 8th. I got back to my hotel around 4am, went to bed at 4:30am, and managed to get a couple of hours of sleep. Since arriving back home, it’s taken until now to get fully caught up on sleep.

There have been some interesting happenings on the farm in these past months. One story in particular, about a sheep, I’ve been wanting very much to tell. It’s a bit involved, however, so it hasn’t been easy to put a post together. I promise to get this up for you soon.

Back to the election. I’ve been a student of politics, and a fan, my entire life. I live for the campaign experience, and for me election night has long rivaled Super Bowl Sunday. To say that this particular campaign was the most unusual of my lifetime would be a gross understatement.

If the subject interests you, you’ve no doubt read all kinds of analyses about how Donald Trump managed to surprise so many. I think this story from the Los Angeles Times gives a good, highly readable overview of how the candidate coalitions changed this time around, and how the Trump victory was a matter not so much of his mobilizing previously disaffected voters as it was of Democrats failing to turn out for Hillary Clinton the way they had for Barack Obama. For a good story about how Trump flipped the state of Michigan in particular, I’d recommend this one from National Review.

Virtually all of the commentary you will read (including what I’ve linked to above) focuses on vote patterns at the county or state level. Those are fine as far as they go, and certainly help give the big picture, but they share a common problem: counties themselves are often very diverse, both politically and demographically. Looking at how a county as a whole votes can miss a good part of the story. Later this winter, I plan to put together an analysis of Michigan vote patterns at the township level; townships are much smaller, more homogeneous, and closer to the level of the individual voter.

In the meantime, the county where I live (Ingham) is an interesting example. It includes a large urban center with a significant minority population (the state capital, Lansing), Michigan State University (East Lansing) and its affluent liberal neighbor (Meridian Township), semi-suburban bedroom areas at the fringe of Lansing proper, and many highly-rural townships and small towns (one of which is home to our farm). In all, Ingham County has twenty-one separate administrative jurisdictions which tally and report votes independently, and for which Census data are available.

Ingham County is much more Democratic than the state’s median, and Democrats routinely sweep countywide elections. It’s hardly a bellweather for anything, and it’s not a microcosm of the state or the nation. Still, Ingham’s intra-county diversity makes it an interesting place to study, and I think the preliminary township-level results give an instructive look at what happened in the state (and country) at large on Tuesday night. Think of what comes below as a first step in my larger analysis of how changing vote patterns flipped Michigan from the Democratic to the Republican column.

First, a quick overview: Ingham had 208,586 registered voters going into the 2016 general election. A total of 136,160 ballots were cast, for a turnout rate of 65.3%. This was up slightly from 63.5% in 2012. Much to my surprise, given how much we’ve heard about the unfavorability of both parties’ nominees, the percent abstaining in the Presidential race actually declined (from 2.4% to 0.9%). However, the percent of Presidential voters choosing a third-party or write-in candidate did increase sharply: from 1.7% to 6.9%. As a result, although turnout was up, and nearly 5,000 additional people came to the polls, the number of votes for major party candidates actually decreased by over 500. The Republican ticket got 952 fewer votes than last time, and the Democratic ticket improved by 419.

Countywide, the 2012-to-2016 vote totals are very similar. Trump got 35.3% of the two-party vote, down only slightly (-0.6%) from what Romney got. (Because we’re talking about shares of the two-party vote, Clinton is of course the mirror image; she improved by 0.6% over Obama.) But how the candidates arrived at these totals is different from how the 2012 nominees did so — and I think tells an interesting story about what makes this election as a whole quite different.

Let’s start with the county’s one big urban center: the City of Lansing. Lansing accounts for 35% of the ballots cast in the election. Nearly one-third of the city’s adults are black (22%) or Hispanic (10%). Interestingly, Donald Trump got 128 more votes here than Mitt Romney did — but the big story is on the Democratic side. Hillary Clinton won 954 fewer votes in Lansing than Barack Obama. In other words, she was unable to mobilize the urban component of the party’s coalition as effectively as Obama had.

In this year’s election commentary, much has been made of Trump’s support from working class whites (especially those living in more rural areas). Ingham County has eleven jurisdictions which are (1) highly rural, (2) almost entirely white, and (3) in which the percentage of adults with a college degree does not reach 25%. Using this as a rough definition of “rural, working-class white,” in these eleven jurisdictions Trump rolled up 1,115 more votes than Romney — and in some cases posted double digit improvements in his share of the two-party vote. These areas had been open to voting for Barack Obama; he scored 1,239 more votes here than Hillary Clinton was able to manage. Clearly, Trump’s appeal to rural, working-class whites was real — and made a notable impact on the coalitions he and Clinton assembled.

So, where did Hillary Clinton make up the difference? In the highly-educated liberal enclaves of East Lansing and neighboring Meridian Township. In East Lansing alone, Clinton garnered 1,310 more votes than Barack Obama, more than cancelling out her decline (from Obama’s total) in the rural working-class jurisdictions. Trump’s total was down 412 votes (or 3.8 percent of the two-party total) from Romney’s in East Lansing. But by far the biggest swing was in bordering Meridian Township, where 60% of adults hold a college degree. Clinton improved by 1,440 votes here over Obama, while Trump got nearly 1,300 votes fewer than Romney (for a 6.3% smaller share of the two-party vote).

The pattern was similar, but to a lesser degree, in the more Republican Williamstown Township, the next jurisdiction out from Meridian. Here, fully half of adults hold a college degree, and the median income is the highest in the county. Romney won Williamstown Township with 53% of the two-party vote, but Trump narrowly lost it (winning only 49.8% of the two-party vote).

There are some additional jurisdictions in Ingham County which are rural but with more adults holding college degrees, or less rural but with a good number of working class whites. In these jurisdictions, both candidates tended to be down a bit from their 2012 counterpart, or they did basically the same as the 2012 nominees did. While not exactly “suburban,” these tend to be bedroom communities (or within fairly easy commuting distance) to Lansing, such as Delhi Township and the City of Mason.

The full table of results:

Ingham Results.jpg

Obviously, given that Clinton improved slightly over Obama’s performance in Ingham as a whole, this county’s usefulness in explaining Trump’s statewide improvement over Romney is limited. As much as Trump improved in rural, white working class areas, there just weren’t enough such voters in Ingham County to keep pace with Clinton’s improvement in places like East Lansing and Meridian Township. Fortunately for Trump, statewide, there were more working class whites and fewer college town voters than in Ingham County. That’s why this analysis is meant as a simple preview, and is more suggestive than definitive. Still, these shifts in voting patterns (even in a Democratic stronghold such as Ingham) give a good hint as to how the party coalitions in Michigan changed this year.

On a less scientific, and more purely on-the-ground observational note: Trump’s victory didn’t surprise me very much. Especially down the stretch, both campaigns were treating the state as if it were in play. A lot of money was being spent here, and the candidates were making more visits than we usually see in a Presidential year.

The biggest “early warning” indicator to me, however, was the observable difference in enthusiasm between supporters of the two candidates. It wasn’t just a question of yard signs, though that was part of it. My daughter and I spent a lot of time this summer and fall on our bikes out on rural roads. We were struck by more than the sheer number of Trump signs (especially compared to the paucity of Clinton/Kane signs). People were also making their own homemade Trump signs. Some of these were HUGE. Also, out in the country, it’s not uncommon to see flag poles in people’s yards. What is unusual is to see a Presidential candidate flag flying on a flag pole — and we saw a lot of Trump flags. I’ve been a student of politics my whole life, and I’ve lived in a lot of different places. But I confess that I’ve never before seen someone (let alone many someones) literally flying a flag for a candidate.


But the most colorful anecdote is something that happened just around the corner from where we live. In late summer, one of our neighbors put up a sign in her yard which read, in large all-caps, TRUMP THAT [FEMALE CANINE]. Then, in smaller print at the bottom: “Before It’s Too Late!” I’ve obviously cleaned up the language; I don’t use words on the blog that I wouldn’t let my kids use. Suffice it to say that it shocked our family that someone would put something up in their yard which was so provocative. And, given the provocative language, I expected that sign to get torn down within a day or two. After all, every election year, political signs get torn down all the time.

This one didn’t. Day after day, we drove past that sign. Then week after week. Still there, still taunting every Clinton supporter in the township (and this was a fairly well-traveled road). When the sign had remained in place for two weeks, I suspected there had been a serious shift in the electorate — because in any other year, the passionate supporters of any other candidate would have destroyed a sign like that within hours. Or possibly within a day or two at the most. Instead, this sign remained up for two full months. And when it did finally disappear, shortly before the election, the owner took to the local Facebook page to express her anger that someone had taken it. Turns out, she’d bought that sign herself. In the comments on her post, a large number of people commented as to how much they’d liked it.

Let that sink in for a moment: she bought the sign herself. Don’t campaigns usually supply yard signs? When was the last time you heard about someone passionate enough to buy her own campaign sign? As I thought about it, I realized those Trump flags I’d seen flying must have also been purchased. (Turns out, you can still buy them on Amazon. And they’re not that expensive.) The more I reflected on this, the more I realized just how much of a “voter passion gap” was emerging in my state’s rural electorate.

All of this is a long-winded way of saying: I wasn’t terribly surprised when I saw the Michigan votes roll in for Trump last Tuesday. Once I get the more complete township-level analysis put together this winter, I will share it here on the blog.