Republican presidential elect Donald Trump speaks during election night at the New York Hilton Midtown in New York on November 9, 2016. / AFP / JIM WATSON (Photo credit should read JIM WATSON/AFP/Getty Images)
Jim Watson / AFP / Getty Images
A few weeks after the shock of Donald Trump’s victory has worn off and reality sets in, some are alleging that the voting in key Midwestern swing states was somehow “hacked.” These people contend this may be the real reason three states which have voted Democratic over the last seven to eight presidential cycles (Wisconsin, Pennsylvania, and Michigan) narrowly flipped to Donald Trump in 2016.
But Trump really won, and he won in exactly the way you would have expected him to win if he was going to win at all. The results mirror patterns seen across these states, regardless of their type of ballot, and across the country with respect to class and formal education.
The DecisionDeskHQ is a nonpartisan group of elections analysts and volunteers whose job is to tabulate votes on election day and project winners in real time, providing an alternative to the Associated Press. We partnered with BuzzFeed on their election-night coverage and are experts in both the recent and historical voting trends of swing states. We have dealt with this sort of data on a near-daily basis for over four years now.
We will now explain in detail how Trump’s victories in two key states — Wisconsin and Pennsylvania — not only suggest no evidence of fraud, but in fact track with national trends in every conceivable way. Buckle up.
Let’s begin with Wisconsin, as it forms the core of the accusations of vote hacking currently being floated.
According to New York magazine, the Hillary-backing consortium of “prominent computer scientists and election lawyers” making allegations of potential vote fraud offered only one detail: “Clinton received 7% fewer votes in [Wisconsin] counties that relied on electronic-voting machines compared with counties that used optical scanners and paper ballots.” Details beyond this one charge have not been forthcoming, but if this is their “key claim” then it is extremely weak.
Technically speaking, the claim is true: Trump did happen to perform well in many counties that used electronic-only voting. (This handy PDF lists exactly which counties and towns in Wisconsin used optical scanning of ballots versus electronic-only voting.) He also often performed better than the man with whom he shared a ballot: Republican Sen. Ron Johnson, who ironically defeated his Democratic opponent by a larger margin than Trump did Hillary Clinton (3% versus 1%).
So does that suggest that fraud may have been in the mix? No. The data in fact suggests the opposite.
Voting technologies and election processes differ wildly from state to state. Wisconsin in particular has decentralized system run on a county/township level, so any presumptive fraud would have had to be a massively micro-targeted effort involving hundreds of precincts, literally on the machine-level. (Presumably this would require a commensurate level of manpower — it could not be done remotely, as these machines are not connected externally to the internet or networked together.)
Even if all that happened, we would have expected to see inexplicable results in certain regions, as outcomes that had been “hacked” in advance ran bizarrely counter to national trends. And yet the vote shift in Wisconsin follows that of every other state across the entire Rust Belt (specifically Pennsylvania, Ohio, Iowa, Michigan and Minnesota).
That trend is unmistakeable: Donald Trump ignited the white working class of these small towns, exurbs, and rural areas in a way few prior Republican candidates ever had, in many cases converting them from Obama voters to Trump supporters.
If Trump did well in what we think of as blue-collar areas, though, he also tended to run behind his Republican ballotmates in white middle-class suburban regions that in previous cycles supported Republican Presidential candidates, and which — critically, as it provides a direct comparison for us — supported Ron Johnson in Wisconsin in 2016 in line with those same historical trends.
In other words, there is no correlation to whether Trump ran ahead or behind Johnson in Wisconsin on the basis of whether a county uses electronic-only voting. Instead, the correlation is clearly mappable onto geography and demography.
In the educated urban and suburban areas of Wisconsin Trump bled votes to both Hillary Clinton and Gary Johnson that Ron Johnson did not. In Waukesha County — a traditional Wisconsin blood-red GOP stronghold in the suburbs of Milwaukee — Clinton got 8,000 more votes than Russ Feingold (the Democratic Senate candidate), while Gary Johnson pulled a remarkable 4% of the vote on average in the three main Milwaukee collar counties. (Yes, that’s right: many educated white-collar Republicans in the Milwaukee suburbs voted a “Johnson & Johnson” ticket.)
Meanwhile, in points north and west of the Milwaukee collar counties, in rural and working-class areas heavy with white voters lacking college degrees, Donald Trump ran significantly better than Johnson. The fact that many of these voters were likely Democrats is suggested by how Trump outperformed Johnson, a much more traditional Republican candidate.
The “Trump/Feingold” voter was a very real phenomenon outside the Milwaukee suburbs and the northeastern Fox Valley area, especially in the counties surrounding La Crosse and Eau Claire. These are precisely the sorts of places where doctrinaire Republican Senate candidates have historically failed to gain much purchase, yet where Trump’s blue-collar and anti-trade appeals appear to have resonated.
In Sauk County, a Democratic-leaning small-town and rural county, both Clinton and Feingold won. But, tellingly, Clinton did far worse than Feingold, nearly losing the county 47.4-46.9%. Meanwhile Feingold won handily, 52.5-44.5%. The clear difference-maker? Trump/Feingold voters.
The key thing about Sauk County: They use optical scanners and paper ballots.
Trump’s overperformance can’t be attributed to electronic-only balloting and a mysterious “hacked vote.” It is but one example of Trump’s overperformance among white midwestern voters outside of urban/suburban enclaves, one that was replicated all across the rest of Wisconsin. In nearby Iowa County (whose larger towns use optical scanners instead of electronic voting) Clinton and Feingold both won but Trump again ran ahead of Johnson by a point. The same pattern shows up in Rock County and Green County (two southern Madison exurbs); they are optical-scanner Democratic counties where Trump nevertheless performed notably better than Ron Johnson on the same ballot.
This pattern — of Trump overperforming more traditional Republican candidates in Democratic-leaning working-class regions while underperforming them in the GOP’s historical educated suburban heartlands — is repeated county-by-county, precinct-by-precinct, all across the entire Rust Belt region from New York through to Minnesota. It happened regardless of whether that county in question used paper ballots, optical-scanners, or electronic-only voting technology. It was an undeniable national trend. It’s also a logical, demonstrated explanation for the election’s outcome in these areas.
Trump won Pennsylvania thanks to a massive surge in white working class voters in every conceivable pocket of the state. While Clinton indeed suffered a drop in votes in Philadelphia city proper, it’s striking to note that even if she had hit the same numbers as President Obama did during his 2012 reelection, she still would have fallen short statewide.
There have been a variety of allegations of nefarious activity, but many of them concern the manipulation of votes in areas with high concentrations of minorities.
But Pennsylvania is an overwhelmingly white state. As in, mayonnaise-on-Wonderbread white. Outside of Reading, Bethlehem, Allentown, a few scattered communities around Pittsburgh, and about half of Philadelphia-proper, the variance in Pennsylvania voters runs along class, not race. The state is whiter than North Carolina, Florida, and Michigan. There simply aren’t enough minority voters in the state to determine its fate — Democrats had performed well enough with white voters to hold the state for six elections.
In every place you look, areas with high concentrations of white voters with with no college degree, or some college experience, broke very heavily for Trump, even in pockets of counties that broke toward Clinton (like Delaware, Montgomery, and Allegheny), and these changes directly affected the fate of the vote.
County-level comparisons are insufficient for us at the DecisionDeskHQ, where we have spent years arguing that Republican prospects in Pennsylvania were greater than the conventional wisdom held. If you want to get a real feel for the voters and the various demographics, you need to look at things on a municipal level. Even in areas where Clinton improved on President Obama’s numbers, like the Collar Counties (Bucks, Delaware, Chester, and Montgomery), Trump’s white working class voters surged:
See the interactive chart here.
When you look at the data within the areas of the state, you find things you might expect based on demographics: that Clinton netted more votes out of both Reading (a township in Berks where the Latino population is now larger than the white one) and Pittsburgh (erroneously thought of as a white working-class town, when in fact over a third of its residents have college degrees) than Obama during his re-election.
In municipalities where college degrees abound, Clinton soared; in areas where they did not, Trump did. Clinton’s Achilles heel in Pennsylvania was the simple fact that whites without college degrees vastly outnumber whites with them in the state.
The decline in Democratic numbers in minority-heavy areas coinciding with mini-Trump surges wasn’t consistent across the board, either. For example, Penn Hills township is about a third black and saw a dip in Democratic votes — but saw a larger dip in Republican ones. Pittsburgh saw an increase in Democratic votes even when Philadelphia saw a decline.
Trump won because he solved a problem that bedeviled Republicans for almost three decades: He solidified white working-class voters in the North. He carried every conceivable one of them he could find, not just in the counties that blinked red on the map, but in the blue ones, too. Clinton’s campaign failed from weaknesses not just in one section of the state, but everywhere. When you look precinct after precinct, municipality after municipality, you see the same pattern — white working-class voters moving rightward, college-educated voters moving a bit leftward.
This is a case where the simplest explanation is the correct one: Donald Trump won because he did exceptionally (indeed, historically) well with the white working class, a bloc that until 2016 was resistant north of the Mason-Dixon line to voting Republican en masse. These voters are concentrated in the Rust Belt and Pennsylvania, which is why Trump swept every state therein except Minnesota. But the hallmarks of this shift were evident in states well beyond their borders (Coos County, New Hampshire, for example). It is all too human to be uncomfortable with outcomes that we do not predict. It is dangerous, however, to stare at reality in the face and insist that what is real cannot be true, simply because it was unforeseen.
Brandon Finnigan is the founder of the DecisionDeskHQ. Jeffrey Blehar is an elections analyst at the DecisionDeskHQ.