March 25, 2014 | 7
When stats-wiz and political prognosticator Nate Silver’s new venture, FiveThirtyEight, launched last week, it punctuated the rise of “data journalism,” journalism that incorporates actual numerical data into reporting and storytelling! Silver’s star rose through his New York Times blog, which largely focused on political analysis and his ability to predict 50 out of 50 states correctly in the 2012 presidential election. As a standalone venture, FiveThirtyEight, focuses on sports, science, economics, and lifestyle issues in addition to politics, and brings in data and statistical analysis to bear on these topics. That Nate Silver can be heralded as a star, and that a site like FiveThirtyEight even exists is indicative of a culture that has grown increasingly (and thankfully) enamored with data. Alongside data journalism like FiveThirtyEight, the it’s-everywhere trend of big data, the pervasiveness of infographics as a journalistic tool, and the rise of Moneyball-esque advanced analytics techniques in professional sports prove that quantification is IN.
Yet at the same time, the launch of FiveThirtyEight was mostly met with negativity. Although the criticism (much of it summarized here) included some disappointment that Silver’s site didn’t do MORE with data, a lot of it to me smacked of quantiphobia, a fear or disdain of numbers. Much of the backlash also seemed to respond to a proclamation Silver made to Time Magazine earlier in the month as to how he hires:
The x-axis runs from “quantitative” to “qualitative,” the y-axis (top to bottom) from “rigorous and empirical” to “anecdotal and ad hoc.” All FiveThirtyEight employees, he says, need to land in the upper-left quadrant of the coordinate plane, where they are quantitatively inclined, rigorous and empirical. The adjacent quadrant above the x-axis, Silver says, belongs to journalists like some of his former colleagues at the New York Times and Ezra Klein, most recently of the Washington Post. “People call them numbers whizzes, but they’re not that—just very good journalists.” The bottom two quadrants belong to the dregs of American journalism: on the left, sportswriters who cherry-pick statistics without thinking through them, and on the right, op-ed columnists. “That’s the crap quadrant. Two-thirds of the op-ed columnists at America’s major newspapers are worthless,” Silver says. He hates punditry, he hates narratives, he hates bold proclamations — and so too does he hate the media’s most willing vessels for all three.
My sense is that the “crap quadrant” hit too close to home for many, and traditional journalists tend to be wary of data moving in on well-worn territory of (in the Time article’s terms) punditry, narratives, and bold proclamations. Indeed, in Silver’s opening manifesto for FiveThirtyEight, he called out fellow journalists, like Peggy Noonan, who predicted a Romney victory on 2012′s election eve because, in her words, “all the vibrations are right.” A few days after FiveThirtyEight’s launch, a similarly quantiphobic response appeared in the New York Times, with the false dichotomic title, “Creativity vs. Quants.” The article ignores a massive literature on the science of creativity to state that “creativity remains so unquantifiable,” while oddly using Steve Jobs as an example of someone who reached “Eureka!” without relying on numbers. The point of the article is essentially that we can’t reduce a John Lennon song or an Oscar Wilde play, to numbers. Sure. It’s the same point that a lot of anti-advanced-stats folks make in the world of sports about the inability to reduce, say, the beauty of a perfectly executed pick-and-roll in basketball, into a decision tree.
These arguments have existed at least since Mark Twain’s recognition of “lies, damned lies, and statistics,” but they continued to bug me, and I began wonder where such quantiphobia originates. My working hypothesis is that they make objective things that people prefer to be subjective. In other words, numbers make things more fact-like, and facts can evoke discomfort. My thinking on this stems from some research on which I have been lucky enough to collaborate, led by Jordan Theriault at Boston College. The research asks the question of whether people represent morals (e.g., “murder is wrong”) more like facts (e.g., “2+2=4″) or more like preferences (e.g., “chocolate is better than vanilla”), and does so by scanning people’s brains while they evaluate morals, facts, and preferences. Without getting into the details of the currently under review research, both neural and self-report evidence show that people tend to represent morals like preferences more than like facts.
Getting back to the issue of quantiphobia, my sense is that when numbers are appended to issues with moral relevance, this moves them out of the realm of preference and into the realm of fact, and this transition unnerves us. Research on taboo tradeoffs, which I have discussed previously, shows that quantifying sacred values such as religion, national history, or human life leads to moral outrage. Similarly, putting numerical values on issues that have a sacred or moral component to them, like politics to many and creativity, or even sports, to some, can evoke distress, especially when the numbers contradict our existing moral beliefs.
Where this gets interesting is that I think this suggests something for how we might choose to structure our arguments on hot-button issues. For instance, if we intend to persuade others on issues such as abortion, climate change, or income inequality, issues with a moral flavor (and issues in which data journalists like Silver and his crew have begun to delve), we might think that presenting statistical data alongside our argument bolsters our position. In Silver’s terms, presenting data can move us out of the “crap quadrant” and into the domain of rigor and empiricism. However, the more fact-like our argument becomes, the more aversive and polarizing it is likely to be to anyone with an opposing attitude on the given issue. My hypothesis, again a working one, is that making an argument with numbers is more likely to evoke backlash than making an argument without them. I have just begun to collect data to test this hypothesis, and will keep you posted on what happens.
Photo courtesy of Bantosh via Wikimedia Commons