“If it bleeds, it leads.” – old media aphorism
How alarming would it be if humans used popular horror films to decide what to protect ourselves from? We would spend all our energy avoiding sharks, serial killers, aliens, ghosts, chainsaw-toting maniacs, while being killed by suicides, automobile accidents, and heart disease.
It may sound silly, but we all do something similar when we make decisions based on deaths heavily covered in the news. In this post, we’ll see that the stories that scare us the most — terrorism, mass shootings, plane/train crashes — get the most media coverage and the widest media distribution. But that doesn’t mean they’re common.
Instead, the narrative of a scary story is engaging. The media heavily covers these stories because they know we want to read them and because their business model depends on it. Even in 2017, scary stories are a key way we assess the threats we face.
When Etan Patz was kidnapped in the 1970s, there was so much media coverage that many parents felt like they were living in the middle of a kidnapping epidemic, and parenting was transformed for an entire generation of New York families. Parents protected their children from strangers, without realizing that the vast majority of abductions are abetted by people the family knows.
After 9/11, many Americans chose to drive rather than fly, and since cars are much more dangerous than planes, a thousand more died as a result. After the San Bernardino attack and other mass shootings, a number of Americans made life changes from deciding to homeschool their children to isolating themselves at home — with debatable benefits and at a high cost to their personal lives. After several mass shooting and terrorism incidents, there is a “national anxiety” about mass attacks at odds with the actual risk, leading to numerous airport scares.
Let’s compare media coverage in The New York Times to the actual ways people die. The articles we’ll look at cover 1 in 7 of the articles in the first few pages over 20 months, including international news and many obituaries of the famous. Each article is written as a result of a death.
Just like we noted with terrorism coverage, the point isn’t that social and news media should cover events in proportion to how many they kill. It would be unprofitable to do so and uninteresting for the audience. It also would be problematic, as we would need to understand the future risk (e.g., the risk of a devastating future terrorism attack may be higher, after even a foiled attack).
But by showing the disparity, I hope that readers will find the context valuable when making decisions for themselves and their family.
The Most Covered Death Events
Let’s start by looking at a few of the most covered death incidents and, by contrast, some of the most deadly. The vertical axis is the number of articles in the first pages of The New York Times and the horizontal is number of human deaths:
Source: Wikipedia, The New York Times, custom analysis. This graph excludes the Nepal Earthquake, but it is included in this graph
Western terrorism and mass shootings are covered heavily, compared to the many other ways we can die. In the US version of this graph, you can see, for example, that deadly storms are poorly covered relative to how many they kill.
What inferences would you draw from the world if you focused on just the most covered events? The most deadly?
The most covered (and scary) death is when a human intentionally kills another human.
Some media sources have highlighted that US murders have “surged”, international terrorism deaths have doubled, and a mass shooting happens every day. While these are real risks that Americans face, they are a tiny fraction of the ways we can die. Let’s examine the data.
Today, you have less than half the likelihood of dying in a homicide as you did in the early 1990s — approximately the same likelihood as you would have had in the 1960s. The US homicide rate of ~4 per 100,000 is 40% below the global average of 6.2 per 100,000 people — but still 3-4 times the homicide rate in Western Europe.
Source: US Department of Justice (link)
Each year, ~16,000 Americans are killed in homicides. To visualize this, the number of deaths due to homicides in a single year are five times the cumulative American terrorist deaths since 2000 (~3300 Americans have been killed by terrorism over 17 years).
Let’s examine a few different categories of intentional deaths (actual deaths are on left, coverage on right):
Source: The New York Times custom analysis, CDC data from 2015. Coverage reflects both US and international events. All actual data is sourced in this document.
Like we saw in our last post on western terrorism coverage, terrorism deaths are by far the most heavily covered stories among intentional deaths as they garner the most interest. Similarly, stories of shootings of and by police officers have increased in The New York Times after controversial incidents involving African Americans.
Military deaths due to hostile action have less coverage than terrorism, though they might be covered more in previous years when more American troops were stationed overseas in an active role. In the decade from 2000 to 2010, 50% – 70% of American military deaths in a given year are due to illnesses, accidents, and homicide, though these likely get less coverage than military deaths due to hostile action.
There’s another form of intentional killing we should add: suicide. Now, the breakdown looks like this:
Source: The New York Times custom analysis, CDC.
Suicides take ~40,000 American lives each year (2.5X homicides), but are rarely covered even by local news outlets. This may partly be due to the importance of covering suicide appropriately to prevent a contagion, though there are norms that allow coverage without increasing this risk.
We read the news to form opinions and make decisions. As we noted last time, a simplistic model of how humans make decisions might be as follows:
Like we’ve done throughout this series starting with the Space Shuttle Challenger, we can look at the two graphs above and make different inferences:
|Dataset||Inference from Data||Decisions||Result|
|NYTimes Coverage (Selective Data)||Suicides are small compared to homicides and terrorism deaths.||?||?|
|All Data||Suicides kill twice as many Americans in a typical year as homicides, terrorism, and police-related deaths combined. Half of these deaths are people above the age of 50 and half below.||Invest in better mental health or end of life support?||?|
Now, let’s look at automobile crashes. Roughly the same number of Americans are killed by gun-related homicides as in alcohol-related auto accidents — about 10,000 people each year. But as you can imagine, gun murders get way more media coverage than drunk driving.
Source: The New York Times custom analysis, CDC, CDC Impaired Driving.
As an international paper, it makes sense that there’s little coverage of automobile crashes in The New York Times. But even in local papers, these crashes are significantly under-covered.
Like before, let’s compare the inferences you might make between the two graphs:
|Dataset||Inference from Data||Decisions||Result|
|NYTimes Coverage (Selective Data)||Less auto crash deaths happen than those due to homicides or terrorism.||?||?|
|All Data||Auto crash deaths happen much more often than homicides (2.5 times). The number of drunk driving deaths alone are the same as the number of people killed in gun homicides each year.||?||?|
The death rate due to cars is horrific; one in sixty of every human death in the world is from an auto accident. Some 3.6 MM Americans were killed over the last 100 years. That’s 5-6 times the total number of American soldiers killed in battle in every war since the Revolutionary War. (The strong language we use is a technique used in media, but we neglected to mention how valuable automobiles are or how much safer they’ve become)
If we added plane and train crashes to the list above, they would be substantially covered (they can be highly vivid and scary), relative to the coverage for auto crashes. This occurs even though non-automobile transport deaths are less than 5% of the total transport deaths in the US in a typical year.
Common Causes of Deaths
Here’s a simple chart of a few select types of “avoidable” deaths, that we might prevent with the right investment. Above each bar, we’ve noted the ratio of how each type of death compares to homicides:
Combined, falls, drug poisoning, suicide, auto crashes, secondhand smoke, and HIV kill more than 10x as many people as homicides — but receive a small fraction of the news coverage.
While we haven’t included them in our last graph, a bad tsunami or earthquake in a single region can kill a large fraction of worldwide homicides, even though they’ll get much less coverage than war, terrorism or mass shootings.
Here are some examples, all since 2003:
|Event||Number of Deaths||Fraction of World Homicides|
|2004 Indian Ocean Earthquake/Tsunami||280,000||61%|
|2010 Haitian Earthquake||160,000||35%|
|2008 Cyclone Nargis||>138,000||>30%|
|2005 Kashmir Earthquake||100,000||22%|
|2008 Sichuan Earthquake||87,587||19%|
|2003 European Heat Wave||70,000||15%|
|2010 Russian Heat Wave/Wildfires||56,000||12%|
How much coverage do you think a continental heat wave gets in an American paper? A single vivid terrorist attack in Western Europe?
Extreme temperatures have already killed 150,000+ in the last 15 years – but will be poorly covered.
Climate change is predicted to lead to 250,000 additional deaths a year starting in 2030, but will still be less covered than terrorism or homicides. Smoking kills millions, and health and car related drinking incidents kill hundreds of thousands worldwide – and again will have little coverage.
The Focus on Specific Human Deaths in the News
Let’s recap: 1 in 7 articles in the first few pages of The New York Times between January 2015 and August 2016 is due to a death event. Of these articles, ~65% reflects terrorism, homicide, war, and plane/train crashes. In reality, those types of deaths represent 0.7 percent of the 2.6 MM American deaths each year.
The breakdown looks like this (as you can see, homicide is the only one that registers in the “US/World deaths” data):
Source: WHO, CDC, The New York Times analysis. Actual data is sourced from here.
Many causes of deaths included in this dataset are considered “natural deaths” (e.g., Alzheimer’s, cancer) and therefore may be unavoidable. Even if we counted only avoidable deaths, the most covered deaths still only represent 2.8% of the total but ~65% of coverage.
As you can see, intentional deaths like homicides and terrorism are covered heavily compared to how many they kill. Many other types of death that kill millions each year are poorly covered.
Like before, let’s compare what inferences we might make with just the selective data versus all the data:
|Dataset||Inference from Data||Decisions||Result|
|NYTimes Coverage (Selective Data)||Terrorism, homicide, plane/train crashes, and war kill a substantial portion of the people who die each year||?||?|
|All Data||The most heavily covered deaths in media are a tiny portion of the way you (or any human) are likely to die – even if we exclude deaths like old age||?||?|
Source: WHO, CDC, The New York Times analysis.
Think about how you might feel looking at the actual graph above compared to reading a story that highlights that more than one mass shooting occurs each day. They’re both accurate, yet lead to very different perceptions and threat levels for the reader.
“Scariness” and Coverage
The scariest ways to die are the most covered. We can quickly see this by examining the characteristics that make an event scary:
(Scarier vs Less scary)
|Dread||Anything that kills us in a scary way worries us||Shark attack vs heart disease|
|Novelty||New risks are more frightening than old ones||Zika vs influenza|
|Catastrophic or chronic||Things that happen less often but kill many are scarier than those that happen regularly||Plane crash vs car crash|
|Uncertainty||Concern will be higher if all the answers are not available||Genetically modified crops|
|Control/Choice||Lack of control or choice worries us||Plane crash vs car crash, self-driving car vs. human driven car|
|Man-made (not natural)||Man-made dangers are scarier than natural ones||Genetic modification vs hybridization of species|
|Children||We get more scared by risks faced by our children||Child abduction vs. our own abduction; childhood vaccines|
|Awareness||The more aware we are of a risk, the more concerned we will be|
|Risk benefit trade-off||If something feels like it provides value, it will be less scary||Accepting a Syrian refugee vs driving a car|
Adopted from David Ropeik, The Consequences of Fear (link).
Compare the most covered types of deaths in our previous chart to the characteristics in this table. How scary are the most covered deaths? The most prevalent deaths?
Even in 2017, so much of our decision making around danger is driven by a story we want to read rather than a thoughtful consideration of the risks we face. In the context of war, one New York Times journalist highlights a critical characteristic of a good story that influences coverage:
“Conflicts gain sustained American attention only when they provide a compelling story line … most of all … good guys and bad guys.”
Using the Media to Determine Personal Safety
Throughout this series, we’ve shown the large difference between the coverage of deaths and actual deaths. Just like I pointed out with the Space Shuttle Challenger, we need to look at all the data if we seek to make good decisions, not just selective data.
Would you make different voting or personal safety decisions by looking at reality rather than coverage? How interesting would reality be compared to the coverage? How profitable would reality be compared to coverage?
This highlights a societal paradox. If you figure out how to make people exercise consistently, you’ll likely have saved more lives (or increased lifespans) than nearly anything anyone else can do today. And yet, there will be very little media coverage of this (especially the front pages), as the effects are too diffuse and the narrative too weak.
More dangerously, media coverage may influence what our society focuses on — meaning that public policy or research funding or political decisions may be suboptimal.
Next time, we’ll see why selective media coverage occurs by examining the invisible hand of the reader — and the business model of content creation/distribution.