How statistical illiteracy warps our view of reality and perverts public policy

On Sept. 2, 2015, a photograph of a three-year-old Syrian boy lying face down, dead, on a beach in Turkey went viral. His corpse was viewed by more than 20 million people on social media the following day, google searches for the words “Syria” and “Refugees” and Red Cross donations both increased by more than 10,000 percent, and Europe admitted tens of thousands of additional refugees in the following days. Six weeks later donations, media interest and the number of visas processed had fallen to the levels they had been at in the week before the photo was published. Prior to this photograph going viral, a quarter-million people had already been killed in the Syrian Civil war; in the months after it was published, hundreds of thousands more died. The refugee crisis had not ended but the public’s concern for the unphotographed victims had run its course.

Understanding the world through heart-wrenching images or stories made sense when we were hunter-gatherers living in small bands of around 150 individuals, but it is a foolish way to understand reality today, on an interconnected planet with nearly 8 billion people. Numeric competence was unnecessary when the world was small, and all interactions were personal. But we no longer have that luxury, and statistical illiteracy has become our species’ tragic flaw.

Basing your beliefs on emotions is not only a poor guide to the truth, it can actually invert reality. Part of the problem is that the benefits of reacting emotionally are often vivid and satisfying, while the costs can be abstract and statistical — a four-year-old in sneakers who looks like he could be my own kid versus a reasoned financial and political analysis of the costs of intervening in a civil war.

The bigger problem, however, is that the feeling system is not just bad at math, it seems to do it backwards.

The research of Paul Slovik, a psychologist at the University of Oregon, has shown a reliable decrease in empathy as the number of victims increases. This phenomenon, called “psychic numbing,” is the reason that “one man’s death is a tragedy but a million are a statistic” — and why we love stories but are bored by spreadsheets.

You might argue that while evoking empathy is not optimal, at least it draws attention to issues that might otherwise be ignored and is aimed at victims. The problem is that everyone feels this way. Social psychologist Roy Baumeister argues in “Evil: Inside Human Violence and Cruelty” that people who nearly everyone agrees are evil (e.g., Hitler, Pol Pot, and serial killers) saw themselves as victims. It is worth remembering that in Rwanda, it was the historically persecuted Hutus who slaughtered the privileged Tutsis.

Yale psychologist and author of “Against Empathy,” Paul Bloom, shows how empathy can be used to support either side of any issue. Politicians have always understood that a single personalized story relayed in detail evokes far more sympathy than a story about a larger and more anonymous group. Psychologists call this “the identifiable victim effect,” and Reich minister of propaganda Joseph Goebbels used it in the same way that Steven Speilberg does. If you support more open borders, the images of children at the border in cages separated from their parents are excruciating. Oppose immigration, and there’s a riveting story that Trump likes to tell, about “Kate” who was murdered by an undocumented immigrant in San Francisco.

If these examples seem theoretical, we might try something closer to home. Starting with Michael Brown in 2014 there seemed to be an endless stream of white police officers killing black men, which culminated in the murder of George Floyd in 2020. Many of these incidents were captured on video and uploaded to social media. They consumed the world’s attention (nearly a thousand articles in the New York Times mentioning George Floyd in just the first week following his murder) and were offered as proof of a deeply racist police force and society. In a world governed by reason and respect for data, these stories would have been covered by a local paper along with other homicides, or accusations of police misconduct — and the national press would have put them in the context of a nation with 330 million people. Instead, these stories sucked the oxygen out of every other political issue or social cause for months — and were touted as proof of systemic racism, not just in Ferguson or Minneapolis, but in the world.

Anecdotes are data without context.

While a riveting image can show us an intimate view of an event from a certain angle, large datasets are the aggregation of millions of images. Statistics widens our focus. They force us to look at the big picture, and they allow us to see all sides of these stories. But the bigger problem is that the world is big, and these anecdotes are easy to find. Anyone can easily find an example that supports her view, ratchets up the outrage and leaves everyone even more convinced that they are right. But understanding reality is hard. It is unlikely to be revealed by simplistic narratives, or the Facebook “like” button.

The beginning of COVID-19 outbreak seemed opportune for a global statistics lesson. But two years into the pandemic, newspapers like the New York Times are still reporting the raw case counts without a denominator and frequently comparing deaths between countries with vastly different populations.

As thousands of videos from Ukrainian cell phones flood social media, one is sure to excite the algorithms and “go viral,” and we should not let the number of retweets dictate our view of the war, much less our foreign policy. These images are snuff films; political pornography weaponized to arouse self-righteous indignation, and they derange us.

Data, on the other hand, are boring, indifferent to how we feel, and can’t simply be wished away because we don’t like what they tell us. This is their power.

Robert Lynch is a post-doctoral researcher in Penn State University’s Department of Anthropology; he holds a PhD in physical and biological anthropology from Rutgers University. Prior to entering academia, he worked for more than a decade in equities trading.