It’s time once again to dole out FiveThirtyEight’s Data Awards, our annual (OK, we’ve done it once before) chance to honor those who did remarkably good stuff with data, to shame those who did remarkably bad stuff with data, and to acknowledge the key numbers that help describe what went down over the past year. As always, these are based on the considered analysis of an esteemed panel of judges, by which I mean that I pestered people around the FiveThirtyEight offices until they gave me some suggestions.
Prize For Statistical Fortitude
During this year’s presidential campaign, the LA Times poll was the only major national poll that consistently showed Donald Trump leading in the popular vote. That made it a favorite of Trump’s supporters — including the candidate himself, who often touted its findings — and a source of worry for his opponents. The poll was also a frequent target of criticism from polling nerds, who argued that its weird methodology was skewing the results.
After Trump’s surprising win, the LA Times bragged that its poll had seen what others missed. That isn’t quite right. The final Times poll had Donald Trump winning the national popular vote by over 3 points; he ended up losing it by about 2 points.
But whatever the results, the pollsters deserve credit for sticking with their approach. If they had shifted their method to go with the winds of the day, their results would have been essentially useless as data, in part because we wouldn’t have been able to know which movements in the poll represented real changes in the race versus methodological changes. Instead, at a time when many pollsters seem to engage in “herding” — fiddling with their results to avoid standing apart — the LA Times didn’t fear being the outlier. And as a result, we were able to get great data from them about how the race was moving. (The pollsters get extra credit for releasing their underlying data, something few pollsters do.)
Best Use Of Data To Speak Truth To Power
Lots of people know about the water crisis in Flint, Michigan, and many of them know about all the local, state and federal officials who failed at their jobs to allow that to happen. But what a lot of people don’t realize is that much of what we know, we know because Edwards, who had been through this before in Washington, D.C., made information requests about what officials knew and when to the federal government and then made all those documents publicly available.
There are many people who could be recognized for their work in Flint, but Edwards’s contribution is a huge part of why we know everything we do.
Read more: What Went Wrong In Flint
“Word Of The Year” Of The Year
People love the math behind language, and throughout a — shall we say — politically compelling year, the snarky and backed-by-the-numbers Merriam-Webster dictionary has been right there with us. Its website has a feature that highlights the most searched words of the day, the past week and — wouldn’t you know it — the past year. Congrats, “surreal.” For gamely combining data and language, for being easily the sauciest dictionary on a social media site, and for serving as a dispiriting canary-in-the-coal-mine regarding the perceived state of our times, Merriam-Webster truly is the “Word of the Year” of the Year.
Trudeau Prize For Governance
Last year’s Data Awards featured two awards connected to government data. One went to Canadian Prime Minister Justin Trudeau for rescuing his country’s census. The other highlighted a key piece of data that wasn’t being collected: The federal government had no good way to keep track of the number of citizens killed by police every year. (The award honored media outlets and crowdsourced websites that were collecting the data instead.)
Well, this year the government stepped up to the plate, and the Bureau of Justice Statistics published for the first time a realistic estimate of the number of Americans killed by police. The BJS has also committed to enforcing a law requiring law-enforcement agencies to report better data.
The Barest Minimum Of Progress Achieved
2016 began with the keen observation that the Academy of Motion Picture Arts and Sciences had failed to nominate a single person of color for an acting Oscar. Lots of factors are at play here: There’s a fundamental pipeline problem in the movie business that often keeps actors of color from advancing into major roles, for example. And studios could do far more to highlight non-white talent ahead of awards season. But those are hard problems, and solving them would involve addressing serious, systemic issues head-on and may even involve having to spend more money.
Instead, the Academy — last year about 6,400 eligible voters — extended membership offers to 683 people and indicated that they would begin stripping voting rights from inactive members. That’s all well and good, but it happened to only about 60 or 70 people, which won’t do anything to address the problem and instead comes off as cruel. And those 683 new members? They barely changed the overall demographics of the body, with women now 27 percent and people of color only 11 percent of the Academy. This isn’t intended to knock the decision — the move is long overdue and essential on a long-term basis — but when you add almost 700 people, dump 70ish and move the needle 2, maybe 3 points, maybe that’s not the best way to move the needle.
Boldest Sacking Of Experienced Humans In Favor Of Untested Algorithm
Facebook doesn’t want to be a publisher — it wants to be a platform. It also wants to satisfy the news needs of its users. To do this, it had human editors with journalism experience maintain the trending news section. Then, under the belief that these people could be replaced by an algorithm, Facebook sacked them and fired up the bots. Not having journalism experience — and evidently not being coded well enough to emulate someone with journalism experience — the algorithm almost immediately began heavily promoting fake news. All this happened about two months before a major U.S. election. As far as timing on “when to unleash your massive news consumption fire hose without adult supervision” goes, this one isn’t great! Fake news was given a lucrative business model just weeks before people would use information to make an important political decision. The platform has since begun a crackdown, which I suppose is outstanding responsiveness and foresight ahead of the 2020 election.
On the other hand, serious kudos to Facebook for mounting what appears to be a remarkably well-done and effective voter registration drive.
The “Are We Still Doing This?” Award For Willful Misinterpretation of Government Statistics
The president-elect gets this one for claiming that the unemployment rate is actually higher than the government is telling you. And not just a little higher — Trump has on multiple occasions said the unemployment rate could be as high as 42 percent.
That claim — and others like it, including from Trump’s choice to run his National Economic Council — has been debunked so many times that I won’t waste the pixels on a detailed rebuttal. But the short version: “Unemployment” is 42 percent ONLY if you count as unemployed people who have retired, or are staying home to raise kids, or are getting a degree, or don’t want or need a job. The U.S. government, like governments around the world, instead counts people who are actually trying to find jobs. And, no, the definition hasn’t changed recently, and, yes, the unemployment rate is falling even if you prefer an alternative definition. For dragging out the notion that economic data is rigged, Mr. Trump wins this prize. If he’s looking for other kooky economic ideas, though, have I got the one for him with loads of support.
Prescient Data Of The Year
One big story of the election was Hillary Clinton’s losses in Wisconsin, Michigan and Pennsylvania. If you were watching the polls, you saw that the race was close in those three states; if you were watching where the candidates were campaigning, the Clinton team appeared to think she had those in the bag. But if you were watching where the candidates were spending their advertising dollars (and who was watching those ads online), you got a pretty good idea that Trump was in a strong position in the Midwest. Data available from two different sources, Kantar Media for television and YouTube’s real-time election widget, both revealed Trump’s media dominance in the region. This isn’t to say that there’s a direct link between this and the outcomes — we’re going to be mulling over this election for some time — but these numbers seem worth paying attention to in the future.
The Volkswagen Prize For Insidious Data Manipulation
Last year, the prize for putting the thumb on the scale went to Volkswagen, which cheated on emissions tests on millions of its diesel vehicles. It was difficult to fathom what could be worse, until Theranos, a bloodtesting company, was found to have been doing something similar to the human body. Basically, there was substantial evidence to doubt the accuracy of the company’s Edison blood testing machine, as reported by John Carreyrou of The Wall Street Journal. It’s been a trip since:
- In March, a government report came out that found the company’s work lacking.
- In April, we learned that federal prosecutors had “launched a criminal investigation into whether Theranos Inc. misled investors about the state of its technology and operations.”
- In May, Theranos had told “federal health regulators that the company voided two years of results from its Edison blood-testing devices.”
- In July, “regulators decided to revoke its license to operate a lab in California because of unsafe practices and to ban founder Elizabeth Holmes from the blood-testing business for at least two years.”
- In October, Theranos said it would “shut down its blood-testing facilities and shrink its workforce by more than 40%.”
- In November, partner and investor Walgreens sued the pants off of Theranos.
The company has truly set a new bar for this illustrious prize.
Ashley Madison Memorial User Data Leak of the Year
Imagine a guy who has decades of computer experience, the latest in information security, uncrackable passwords, impenetrable cyber-defenses, is impervious to phishing attempts and lives his life through a dozen VPNs for online privacy.
Next to Yahoo, even Hillary Clinton looks like that guy.
The company announced in September that there had been a data breach of around 500 million accounts in 2014, a colossal screwup in information security from what was once considered a leading technology company.
Three months later, we learn from Yahoo that more than a billion accounts may have been stolen in a hack from 2013. That’s the kind of incompetence that lands you an Ashley Madison Memorial User Data Leak of the Year prize.
Significant Digit Of The Year
This prize goes to “1 percent” for two major reasons. First, the national conversation about the yawning gap between the top 1 percent of American earners and the bottom 99 percent was never louder than in 2016. For math geeks like us, this is almost as sweet as the time America re-elected Richard Nixon even after he referred to the third derivative in a speech. Second, multiple forecasters gave Trump around a 1 percent chance of winning the presidency. That estimate appears to have been low.
FiveThirtyEight Person Of The Year
Here at FiveThirtyEight, we pay close attention to gambling lines. After all, what better way to gut-check our forecasts than to compare them to the predictions made by people with real money on the line? For the sports desk, we have Las Vegas. For everything else, we have overseas gambling sites, which for tax reasons are predominantly located in Ireland and outlying British possessions. And we’ve never loved Irish gamblers more than this year, FiveThirtyEight’s first U.S. presidential election without the late, lamented political betting market Intrade. So whether I’m checking Oscar odds on Paddy Power or Nate is looking to Betfair for a check on the presidential race, European gamblers were there for us.
So why are the Irish the winners in particular? Because they were the winners no matter what: A few weeks before the election, Paddy Power decided to pay out all bets on Clinton. Then when Trump won, they had to pay out on those bets too. In other words, everyone who gambled on the election at Paddy Power got paid, unless they were some schmuck who went for Gary Johnson.
Anna Maria Barry-Jester, Carl Bialik and Ben Casselman contributed to this story.