Out in the city, into the sunshine
One of the weirder aspects of the extended election post-mortem this year has been the focus on Wisconsin. The state flipping from blue to red was certainly a surprise to all observers (including both campaigns), and it slots into a neat narrative about overconfidence and the “forgotten working class” (or a certain subset of it). Nevertheless, anything that the Clinton campaign did to win Wisconsin would not have mattered a lick if it didn’t also flip at least one other state (Florida), or more likely two more (Michigan and Pennsylvania). So anyone making the argument that Russian hacking doesn’t matter because they didn’t keep Clinton from visiting Wisconsin is, to put it mildly, not engaging in good faith discussion. Anyone not talking about the role of Scott Walker’s vote suppression law is also not engaging with the reality of what went on in Wisconsin this year, as reported last week by the Los Angeles Times:
Starting with John F. Kennedy’s victory over Richard Nixon in 1960, Delia Anderson had voted in 14 straight presidential elections.
She had cast her ballot at the same polling place for years, never with a glitch. This year, however, a volunteer driving her to the polls mentioned that she would be asked to show a state-approved photo ID.
“Don’t these poll people already know who I am?” replied Anderson, who is 77, black and uses a wheelchair, as she frantically sifted through her purse for anything to prove her identity.
It was a lost cause. She had planned to vote for Hillary Clinton. Instead, for the first time in 56 years, she did not cast a ballot.
“Lord, have mercy,” she said. “What happened to voting?”
Such stories abound in Milwaukee, a Democratic stronghold in a state where Donald Trump won by 22,748 votes, a key victory in his path to the presidency.
Now, voting rights advocates, elections officials and political experts have zeroed in on the city as a case study of whether controversial new rules requiring ID for voting — the kind used in several states in November for the first time in a presidential election — blocked vast numbers of largely young and racial minority Democrats from casting ballots and contributed to Clinton’s defeat.
In a state that saw its lowest turnout in 20 years, nearly 248,000 people voted in Milwaukee, roughly 41,000 fewer than in the last presidential election.
“I believe it was voter suppression laws from the state government that crushed turnout,” said Milwaukee County Clerk Joe Czarnezki, one of two officials who oversees local elections. “They tend to hit hardest on people who are poor, who don’t drive and don’t have a license, who are minorities.”
That said, Wisconsin is a very interesting case quite apart from its lack of impact on the overall outcome of the election. The state has long been at the median of many social indicators across the country, and has experienced a precipitous move to the right since 2010. It’s not for nothing that Hillary Clinton did better there than Russ Feingold, who obviously spent quite a bit of time campaigning in the state. There has been a considerable amount of examination of county-by-county results so far, with a particular emphasis on urban/rural splits (and for background on this, Kathy Cramer’s The Politics of Resentment is the place to start), but even in Wisconsin, rural counties are not where most people live. So although the data aren’t superb yet, I decided to explore a bit in the vote records for 2012 and 2016 to see what happened in Wisconsin’s cities.
The ten biggest cities in Wisconsin, as of the 2010 census, are home to 1,449,059 people, or just a little more than a quarter of the state’s population. More than half of that group lives in the state’s two biggest cities, Milwaukee and Madison. That is to say that, like most states in 2016, Wisconsin’s population is concentrated in cities. Unlike most states in 2016, Wisconsin saw its voter turnout hit a 20-year low, as the state cast only 97% of the total number of presidential ballots it cast in 2012 (which doesn’t take population growth into account, so the actual drop in turnout was larger than 3%). If Walker’s attempt to suppress Democratic votes worked like it was supposed to, we would not expect to see that number even across the board, but to generally be lower in cities, and especially to be lower in cities with larger non-white populations.
Wisconsin is also a state that both nominees lost in the primaries, and where neither was especially well-liked. That also suggests a reason for voters to turn away (despite there being a competitive and potentially balance-tipping Senate election on the ballot) or to vote for a non-major candidate. So below I tabulate three things. First, the net change in votes and percentage points for Clinton over Trump compared with Obama over Romney, the net change in third party and write-in votes and percentage points, and the number of ballots cast as a proportion of the total from 2012 (to compare with the percentage of the population that is white).
With the caveat that there’s a lot that raw vote totals can’t tell us, there are several interesting things here. Starting at the right side of the table, turnout relative to 2012 and whiteness are highly correlated (r = .76). This is only the ten biggest cities, of course, and it would be worth extending this analysis to every municipality for which the Census Bureau has race data. It is a compelling finding, though, and particularly so for Milwaukee and Racine, the state’s two blackest cities and the two with the biggest dropoffs in this set. I’ll have more on Milwaukee’s low turnout in a subsequent post.
In the middle columns, we see that every single city gave non-major candidates at least four more percentage points than they did in 2012, up to nearly seven in Waukesha and Eau Claire. Even this maybe undersells it — for Eau Claire that was a 328% increase (from 1.8 percentage points to 7.7). Those big increases are also how Trump was able to win the state with only 47.2% of the vote. But the top and bottom of the list also tell another story. There was a lot of discussion during the campaign among people on the left about whether refusing to vote for Clinton was a move based in privilege (and also whether it was a big free-rider move, but that’s another discussion). The two least white big cities in Wisconsin, Milwaukee and Racine, had the lowest increases in non-major voting; they also had the lowest raw third-party and write-in rates of these cities. That is to say, these electorates appear to be the ones least interested in screwing around with this election. Waukesha, at the other end of the list, is interesting for another reason. It is the center of Wisconsin Republican politics, but it’s not a place that was ever happy with Donald Trump. Its big jump suggests a lot of Republicans there couldn’t bring themselves to vote for Hillary Clinton, but chose to vote for Gary Johnson or write in Paul Ryan instead.
(Aside: We don’t know how many write-in votes Ryan or Bernie Sanders or any other non-registered write-in candidate got, but we do know how many Evan McMullin got. For some reason, he absolutely dominated the write-in vote in Appleton, Kenosha, and Racine, getting 546 of 576 write-in votes in those cities. At the same time, in liberal Madison, write-in votes were up from 251 to 2,099, but McMullin got only 367 of those; I’d bet most of the others went to Sanders.)
Waukesha has a lot to tell us about Clinton’s columns, as well. It and Madison — political polar opposites — are the only two of the ten cities where she had a net gain over Obama’s 2012 margins, 5.1 percentage points in each. In relative terms, Trump did horribly in Waukesha, getting only 51.1% compared with Romney’s 57.1%. Clinton did only a little worse than Obama, and might have done better if a few more of those anti-Trump Republicans had been willing to vote for the only candidate capable of actually beating him. The results in Madison, full of liberals and students, suggest that turning out those two frequently scapegoated groups was not the problem in Wisconsin. Clinton got a higher percentage of the Madison vote than Obama did, Trump did worse than Romney, and Madison’s increase in non-major candidate voting was lower than the state as a whole. Madison was also the only city of these ten that cast more ballots in 2016 than it did in 2012, so the suppression of the student vote does not seem to have been all that successful, especially in light of how much turnout cratered in Milwaukee.
In 2012, Milwaukeeans cast 287,387 presidential ballots; this year it was only 246,617. Both major party candidates were down in percentage point terms from the 2012 nominees, with a relatively small net loss for Clinton. The much bigger loss came from votes that were never cast. If Milwaukee’s turnout drop had been the same as the state average and those extra votes cast in the same percentages as the others in the city, Clinton would have a net gain of a little more than 21,000 votes. But there’s good reason to think the lost votes wouldn’t be cast the same as the votes that happened; they would likely be more favorable to Clinton. In a follow-up post, I’ll dig specifically into Milwaukee turnout to look at how vote suppression manifested in this election.
(Thanks to Locksley for the great song that provides the title of this post.)