How people die vs. media coverage of death | Hacker News Reader

How people die vs. media coverage of death

owenshen24.github.io

221 pointsrkaplan89 comments

They use media coverage and conflate it with "general public sentiment" and how "people think we die". While the wording seems to be carefully crafted that it avoids stating something outright wrong, it does suggest that they are the same.

Are there studies that compare media coverage with surveys on how people think we die?

There are! The most famous one is this one https://scholarsbank.uoregon.edu/xmlui/bitstream/handle/1794... and the results are largely in-line w/ what one might expect.

> Two kinds of bias were identified: (a) a tendency to overestimate small frequencies and underestimate larger ones, and (b) a tendency to exaggerate the frequency of some specific causes and to underestimate the frequency of others, at any given level of objective frequency

Problem with such use of statistics is that we underestimate black swan events.

Humans are intuitively better at it. For example, it would be impossible to predict something like 911 event with such frequentist analysis.

Also impossible to predict some sort of nuclear disaster terrorist act that never happened before and could take a million lives.

Another thing to keep in mind is the "missing life" (dying young) and quality of life after disease. That's why something like Alzheimer's seems a lot worse than heart disease.

Yeah "dying of war" is much more likely than you would think by looking at what Americans died of last 10 years. The same could go for anything that happens rarely but kills a lot of people when it does.

911 was probably too few deaths to make an impact in the statistics though. You'd need something killing tens of millions of people

Hey everyone,

I'm super stoked to see this project getting some more traction!

I was responsible for the visualizations / the scraping, so I'm happy to answer any questions people might have about the whole process.

^_^

I think the concept here is fascinating. One thing I didn't see addressed was that it's not totally unexpected for rare causes of death to be overrepresented in the news (apocryphally, it's been said that "dog bites man" is not news--"man bites dog" is).

That said, it seems like the modern short news cycle has edged out attention being paid to long term trends (kidney disease mortality prevalence was surprising to me as well).

Hello, Owen, from another Owen!

I actually find this pretty funny as overrepresentation of certain causes of death is something I've been talking about to friends and family for years, and I was actually recently thinking to compile some data on it. Hilarious that another Owen beat me to it!

One thing I've been meaning to write about for a while is terrorism specifically. I was thinking of writing something refuting the idea of "so few people die of terrorism because we spend so much, if we spent less it'd be a massive problem." Is that something you've thought of at all?

Have you thought of using this data for anything else? Maybe writing some blog posts about the data or looking at a wider range of publications and comparing them for which are more "accurate" in their articles on death?

Hm. The thing about costs is interesting. (That spending so much on terrorism is why we see so few deaths.) Evaluating counterfactuals is always hard. If I were to tackle it, I'd probably look at what we're spending the terrorism budget on, what the upper bound of deaths might be, and if it'd be reasonable to think that whatever counterterrorism measures we're taking could actually save that many lives.

I think there are more nuanced views that are worth expressing, as some commenters have already pointed out, for example, looking at years of life lost (i.e. controlling for age of death).

While I think the data we got was in broad strokes representative, I'd be curious to see what it looks like in other countries, as I'm wondering if cultural bias plays a large role.

I'm not too sure if I'll write many more blog posts exploring this dataset in the future; the data is all there on GitHub, though, if anyone else wants to play with it. :)

This is great! Can you give any insight into your charts? I have similar data sets that I would love to represent like this. It sounds like you already mentioned using chart.js. Anything else that you use, specifically for the first few charts with the time slider? Anything crazy that you needed to do?

To start with, I didn't know much about charting on the web before this project. (It's much harder than in Python :P)

I briefly looked at all the charting libraries on this list here: https://hackernoon.com/9-best-javascript-charting-libraries-...

If I were doing this project over again, I'd probably use [Chartist](https://gionkunz.github.io/chartist-js/) as out of all the libraries I looked at, it seems to be the easiest to get up and running.

While charts.js is a little more interactive, it was a little painful to get up and running.

The time slider is just an input slider some jQuery I wrote that dynamically loads up a new dataset and calls the redraw function in charts.js to load up a new chart.

Overall, as someone with entry-level experience in both manipulating data and web development, I found the overall process to be a little rough at first, but it got a lot better once I got the hang of the library.

In total, the entire website took me about a week, with 2 hours of work every day. So something like 15 hours, total?

Wow... nice work. It's not my first foray into data visualization but it will be my first serious attempt at working it from the ground up. Thanks for the info!

I love the graphs, but the way they’re used in this article is very annoying on an iPhone. That scroller is hard to grab and when i zoom i cannot are all the data anymore. No better in landscape mode. Not sure if you can influence that, but thought I’d give the feedback since you mentioned the visualizations.

Ah, I see.

The charting library I used (charts.js) didn't come with an easy way to have a time axis, so I rigged up some stuff with JQuery and a slider.

I guess something that could have helped with accessibility would have been a dropdown box instead to select the years.

Great job! One observation though: I would have liked the last graph to have three parts: under-reported causes on the left, "pretty ok" in the middle, and over-reported on the right. They are mixed up right now, which makes them harder to differentiate.

Oh, yeah, that seems like a good way to help with clarity. Thanks for sharing. I'll see if I can make some minor edits this week to help with that.

The coloring of the chart is awful, alzheimer and terrorism having almost identical color.

I have to concur. I was confused for a moment about the staggering number of deaths caused by terrorism, until I realised what was going on.

That's an easily remediable fix!

Oh you!

Did the database of articles include the obituaries?

Many people are pointing out that it's pretty obvious the news reports more on rarer events. True, but also people don't necessarily realize this when watching the news. I can anecdotally guarantee some of my family members fully believe that homicides are responsible for significantly more deaths than cancer.

100% of people die from being born. If this data is not segregated by age, it is pretty much meaningless. People don't die of "natural causes" or "old age" any more because the doctors are required to list something previously diagnosed on the death certificate (or else the doctor is looking at a lawsuit.) And the family member who happens to be there at the time of death is in no mood to put up an argument. No one wants to do an autopsy and invasive tests for a frail failing elderly patient are even worse.

This would be useful information to have, but I don't agree that this makes the data meaningless. No matter when you are killed, and what by, your life has been cut short. Just because certain diseases (e.g. heart disease) mostly kill older people doesn't mean they aren't most important and worth tackling. If in the future due to medical advancement most people are living to 100+, we will regard dying at 80 an early death and a travesty.

I'm not sure I understand the point you're trying to make.

Are you saying that dying of heart disease and cancer are over-diagnosed? Why does that matter in the context of this analysis? It's a comparison between CDC data and news coverage -- any "over-diagnosis" (or under-diagnosis) should be reflected in all datasets equally, so it doesn't really effect the overall review.

Everyone dies. The only significance of death is if it occurs sooner than otherwise expected. For the vast majority of people that die in old age what to list on the death certificate can be a conundrum, but it is safe to say no one will ever be sued for malpractice if heart disease is listed as the cause of death. 80% of all deaths due to heart disease occur in the 65 and older population.

https://www.cdc.gov/injury/wisqars/pdf/leading_causes_of_dea...

EDIT: Take the data from the above table multiply the deaths by the number of "years of life lost" (65 minus the mid-range age for each column) and ignoring the 65 and older age group and I think you might see a better correlation with news coverage.

Right, but I think the point of the OP is to say "Good news everybody! Despite what you've seen on TV and read in the paper, you're still overwhelmingly more likely to die of old age than at the hands of murderers/terrorists."

And to go along with that, I believe it is trying to ask the question "is it not just as important to follow the trends in so-called 'death by natural causes' as it is to follow trends for 'unnatural causes', at least as far as public awareness/good is concerned?"

If I was over 65 (and I'm close) I would worry more about what put me in the hospital than what appeared on my death certificate. The two are surprisingly unrelated in the elderly.

This is actually a very good point. I was thinking myself something along similar lines that intuitively speaking, death by unnatural causes like terrorism/homicide/suicide are of course more "newsworthy" than the ones due to "heart-disease".

Although the study is trying to do a larger scale comparison between the perceived and actul leading causes of deaths, it would serve it much better if it were to segment the results by age group.

LifeSpan = HealthSpan + GrimSpan

A reasonable expectation for news coverage would be that coverage is proportional to deaths weighted by years of HealthSpan lost.

That cannot be exactly right. But coverage of loss of GrimSpan is a separate issue. If hip replacements and cataract operations are reducing the GrimSpan by increasing the HealthSpan that is a happier story than better infection control at Alzheimer's units increasing the GrimSpan. It is hard to state an expectation about news coverage of loss of GrimSpan

Thanks!! It's kind of amazing how skewed our view of reality is. Saying "I don't care about terrorism" gets you dirty looks but is a reasonable approach to life if you live in wealthy country and are are trying to worry about things most likely to affect you.

I care about gun violence and knew it was overreported, but it blew my mind to find out that there are about twice as many deaths per year from just asbestos are there are gun homicides. I wonder how many other misconceptions there are? (Total gun deaths is much higher due to suicide and accidents)

When presented with gun violence statistics ask them to take out gang related and suicide then you can more accurately use the information.

Well, those statistics aren't wrong, they just tell a different story.

Gun suicides matter, but we might address those very differently from homicides.

Gang-related violence... I'm not sure why I'd exclude that? They're human beings. The term also is very unevenly applied and laws meant to target "gang-related violence" have a tendency to burden minorities more heavily. (See also: "terrorism")

https://www.nytimes.com/2015/05/31/magazine/how-do-you-defin...

> Gun suicides matter, but we might address those very differently from homicides.

The main prevention measure for both murder and suicide is to reduce access to methods.

It's fine for a nation to say they make the choice not to restrict access to guns, but they need to be honest about the increased numbers of people dying this will cause.

> reduce access to methods

A sibling questioned the validity of this for suicide, and I agree. Anyone sufficiently motivated to kill themselves will find a way. Knife, rope, drug overdose, threaten a police officer, vehicular "accident", skydiving, move-to-Oregon (for legally assisted suicide, that is -- OR's a nice place :)), sit in the garage while the car is running... every single one of these things is something that a suicidal person could come up with quite easily, and do, even if they don't have a firearm.

The only thing reducing access to firearms will do is change the method of suicide. Humans have been finding ways to do it for thousands of years.

This line of argument "For a sufficiently [blah] ...." rarely adds value, be it "a sufficiently smart compiler ..." or "sufficiently motivated individual".

These add little to no value because in most of the cases where such a line of argument is put forward, there is a sizable set that do not satisfy the "for a sufficiently [blah]" condition. Not only are these sets sizeable, sometimes they are the particularly relevant set.

Many suicidal people have been talked out of it. So its not hard to believe that lack of access to quick measures would have influence.

Digression: I blame movies for popularizing shooting at one's temple. That's how many end up with a botched attempt. If one must, one should take out the brain stem as quickly as possible.

> Anyone sufficiently motivated to kill themselves will find a way.

This is untrue, and it's dangerously untrue.

> The only thing reducing access to firearms will do is change the method of suicide.

Method substitution tends not to happen, or it takes some years to happen.

We know this from some natural experiments. For example, car exhaust fumes used to be a common method, and then we got cleaner engines and catalytic converters and again we saw a long lasting drop in deaths by suicide.

And even if every person did substitute a different method it's likely they'd be using a more survivable method.

Knife, rope, car accident, drug overdose are significantly less likely to be successful than a gunshot. They also take longer, and the longer it takes, the longer the person has to reconsider.

Also, you don't often see a knife, rope, car, drug overdose, skydiving, or assisted suicide used in a murder-suicide scenario. (Murder-suicide by car certainly happens, but not as often as with a gun.)

Hm; I stand corrected then. Thanks, guys, especially for pointing out the car exhaust and other things. All of those things are true (about additional time to be talked out, etc). I was coming from the perspective of someone who followed Sir Terry Pratchett's campaign for legal suicide in response to things like Alzheimer's, where it isn't the kind of thing one is likely to be talked out of. In that case, of course, people are campaigning for legal euthanasia, so it's a bit different.

Reduce access to the methods to end ones life? Is this irony?

We know from interviews with survivors that many people don't have a "diagnosed mental illness", but have something more like "rapid onset dispair". And that more people survive if they don't have easy access to their prefered method.

For example, in the UK this has meant phasing out some ineffective and dangerous medication; reducing the pack sizes of others; and encouraging doctors to safer prescribing of opioids.

Is that why I can get a bottle of 500 acetaminophen tablets in the US or six bucks, but have to pay Boots 4 Euro for 40 of them?? (I don't recall the exact dose but it was the same per table)

I always feel like a drug runner coming back to the US with bottles of generic tylenol and advil.

There's a fascinating documentary called "A Certain Kind of Death",[1] which shows what happens in the US when someone dies with no next of kin.

[1] - https://www.youtube.com/watch?v=ErooOhzE268

So, the media reports on rare, interesting things, not mundane everyday things? I'm shocked!

The news is supposed to keep the public informed... fusing of entertainment and information, along with corporate mergers, has brought us to this unhealthy state of public awareness. Our society is becoming more paranoid because it boosts ratings and profit.

A prime example is the American public's irrational fear of terrorism, which translates to pouring unfathomable amounts of tax dollars into the unending "war on terror". Meanwhile heart disease/cancer silently kills 1.1 million Americans every year.

> A prime example is the American public's irrational fear of terrorism, which translates to pouring unfathomable amounts of tax dollars into the unending "war on terror". Meanwhile heart disease/cancer silently kills 1.1 million Americans every year.

The same thing could be said for the gun control narrative being pushed by the media and politicians. If gun deaths are classified as homicides and using CDC numbers, they account for less than 1% of deaths.

However, looking at the same data on the site for the Guardian and NYT, it looks like the wild west out there and homicides are the leading cause of death next to cancer.

I would elaborate on that as well. The main focus of the media is on "assault style weapons" which translates to semi-automatic rifles. Of that < 1% of deaths, rifles (of any sort) are the weapon type in less than 3%. Literally nearly all gun deaths are caused by pistols, yet the media obsesses on rifles. Even if we filter to mass shootings, the primary weapon is most often, again, a pistol. [1]

How many people are aware of these data? I think that ignorance is going to be primarily the media's fault.

[1] - https://news.vice.com/article/glock-pistol-omar-mateen-orlan...

yeah... focusing on "assault weapons" is just divisive clickbait and legislative bikeshedding. Regulations should keep guns out of crazy and criminal hands, period.

I agree that the media is hyperbolizing gun violence, however that's a false equivalency.

The "war on terror" is costly and only creates more terrorism... stoking public fear feeds back into the problem. Legislation could significantly reduce firearm abuse with negligible cost and no feedback loops.

What legislation? I really wish that whenever people talk about "reasonable gun laws" they would specify what they are referring to.

Fear of terrorism isn’t just fear of dying from terrorism. What was the economic damage or 9/11?

Pretty sure the actual economic damage ($250B) from 9/11 attacks was small compared to the self-inflicted economic damage that followed in the form of market reactions $1T and wars $2T.

In other words, "the only thing we have to fear is fear itself."

Factor in the two wars and you’re talking a further 6+ trillion. Call it self inflicted or not, it seems disingenuous to act like it all wasn’t a direct and predictable result of the attacks. I’m not discussing blame after all, but consequences and what poeple reasonably fear.

This is awesome! You might really like All the News That's Fit to Sell which is a deep dive into the media economics behind why things are covered — it gives lots of color for why coverage is the way it is (there's a famous journalism saying: "If it bleeds, it leads" — which speaks to the readership/profits that come from covering vivid deaths)

If useful, I did a similar analysis, but just for the NY Times vs WHO/CDC (I manually tagged an year of articles by cause of death vs. this article's ability to see across many years):

In 2015-6, the deaths that are most covered are a tiny fraction (<1%) of the way we die: https://www.nemil.com/s/part3-horror-films.html

You can also extend the analysis to see how death coverage varies by region (those who are culturally similar to us get more coverage):

How Media Fuels our Fear of Western Terrorism: https://www.nemil.com/s/part2-terrorism.html

This is super great! Thanks for sharing!

Another good example that "data" is not always meaningful. The graphs help the idea "media are talking too much about terrorism".

All deaths are not equal. Most of old people die naturally from an heart failure or cancer, that the natural process of death. Homicides and accidents are not natural but common. Terrorism is an exception, that's why it so much covered.

100 old people dying from cancer do not worth 1 minute of press coverage, a family killed by a truck in a christmas market a bit more.

It's not even so much about the rarity of events, but the violation of Informed Consent. The fundamental assumption in modern society is that you know there are dangers and their relative frequency, so that you may make decisions based on that for you to reach a mental equilibrium of risk vs reward. Someone/something acting psuedo-randomly to violate that equilibrium to an extreme (where there are brutally less choices available due to paralysis or fatal event for an individual or group) is what people react to. They want to incorporate these semi-random events into their daily equilibrium in the form of perceived patterns. This eventually leads to more extreme views (some form of theoretical pattern that fits), in lieu of data.

It's actually worse, because you have to factor in media attention, advertisement, bias etc to things that makes the problem worse; e.g. most of those hearts and respitory deaths can be attributed to pollution and lack of exercise, that means the car and all that it brings.

You are inundated with positive coverage of cars all over, they are vehemently supported in most pieces and comments. Pair that with;

    Heart Disease              10.388 Under
    Car Accidents               2.285 Under
    Lower Respiratory Disease   3.520 Under

>"Although all diseases claim almost 1,OOO times as many lives as do homicides, there were about three times as many articles about homicides than about all diseases. Furthermore, homicide articles tended to be more than twice as long as articles reporting deaths from diseases and accidents."

Doesn't that make sense? If diseases kill 1000x more people, they are not news -- they're same old, what has been going on since forever.

Now, if there a novel disease or epidemic, sure, that would get coverage.

This week's episode of Science Goes to the Movies[0] was a repeat from last summer on Emergency Medicine as portrayed in film/TV vs reality:

[0] http://www.cuny.tv/show/sciencegoestothemovies/PR2006291

According to the charts, most people die of cardiovascular disease and cancer.

First, most people think their habits and diet are not leading them to a cardiovascular disease. Then, those conditions that primarily afflict a specific demographic. So it might not be as relevant to someone who is let's say, less than 40 years old.

Which is a good thing, because it means most people are dying of old age.

You can't officially die of "old age" so if you avoid or survive everything else, eventually you are going die from cancer or heart disease. Problem is though, if you improve cancer treatment then more people will die of heart disease, and vice versa.

We really need a category for "died peacefully of advanced age", where we can try hard to increase the death rate.

I'd love to see room for competition in media that took a different path if it wasn't already all monopolies and bribes cough lobbying and the innefectuality of the FCC in it's mandate.

A strong fourth estate is essential to the proper functioning of the constitutional democractic republic.

Huh. Something like 1 in 20 people will likely have their lives end in a car accident.

I think the data is for "unintentional injury" (or death from injury excluding homicide or suicide.)

This category is actually the number one cause of death for ages 1-44. The numbers are fairly constant from ages 15 to 64 but the rate jumps up by a factor of 2.5 after age 65. Assuming that retirees are not just taking up sky diving, this is probably due to increased frailty and hospitalizations caused by falls.

https://www.cdc.gov/injury/wisqars/pdf/leading_causes_of_dea...

EDIT: It seems that in addition to car accidents, poisoning, overdoses, and falls, make up the vast majority of this category.

It's more like 1 in 100 (in America): http://injuryfacts.nsc.org/all-injuries/preventable-death-ov...

Cf. xkcd #1468 How worried you should be when various things happen to you in movies vs. real life.

https://www.xkcd.com/1468/

Seems like there's a joke about movies portraying chest wounds on one's left side being much more serious than on the right side. But I don't get it.

>Seems like there's a joke about movies portraying chest wounds on one's left side being much more serious than on the right side

I'm guessing it's because your heart is on the left (unless you have situs inversus), so he's arguing movies treat a chest wound on the left as hitting the heart

It's a movie cliche. Shot on the left side, you are a gonner. Shot on the right, you'll be just fine. So you don't have to worry about whether your favorite character is going to make it to the end of the movie or not.

The mainstream commercial news in Australia seems to cover every unusual death that has occurred in the past day. I guess those kind of stories rate well.

Would be interesting to see this data for other countries, and comparing for example the overrepresentation factors for violence and terrorism.

Wow, terrorism really is overrepresented compared with how we die these days. I wonder if that's some kind of evolutionary echo of a more violent time when a leading cause of death for most humans was regional strife. We're apparently still pretty interested in who's out there somewhere, scheming to come and kill us.

> Wow, terrorism really is overrepresented compared with how we die these days.

How many people do you know that were killed or injured in a terrorist attack? Second order? Same question for {murdered, illness, natural cause}. Compare magnitudes.

I knew a person whose brother in law was killed in 9/11 and another guy whose best friend was shot down in Afghanistan during the Soviet occupation. Trouble is I really can't compare it to the scads of relatives of friends who have undoubtably died of cancer or heart disease, because those just aren't the kind of stories people swap on a fishing trip.

> when a leading cause of death for most humans was regional strife

What leads you to believe that was ever a major cause of death for more than a tiny handful of people? It would seem the hardships of living before modern medicine and (any) agriculture would have killed far more people. Given how expensive/risky it was to collect food, why would any group of folks risk most of their healthy, strong brood to wage regional conflict? If they lost, all of them would lose by virtue of losing the most capable members of the society.

I think the fears of terrorism is less about how many people it kills, as how unsafe it makes people feel.

i am one of those people who think that we should allocate resources against causes of death in close proportion to the number of lives claimed. my guess is that many other engineering types also take this position.

recently however, i have come to worry that this position lacks empathy. especially in light of recent mass shootings in the US, i am forced to consider that some types of deaths have far worse second order effects than others. if ten people die on different days out in the woods, ten sets of family and friends are devastated. tragic, but if ten students are killed at school in the same day, it seems that the lasting damage is far more severe and widespread.

to some extent you can say that's just because people are emotional/irrational and they are blocking optimal allocation, but maybe the optimal solution does need to take into account how people feel about things. maybe things like terrorism and mass shootings are actually a lot more harmful than the numbers suggest.

The issue is more that due to highly asymmetric reporting, people wrongly estimate what the risks to themselves and loved ones are. This means that we don't act rationally in our own interests, to best avoid harm. I think most parents would care more about road safety, carbon monoxide poisoning, mental health, etc. than about terrorism or school shootings, if they were aware of the statistics.

And how useful that feeling is to manipulate for media companies.

It’s right there in the name.

Yeah, and people likewise discount the deaths caused by cancer and heart disease because these are diseases of old age. People generally think humans aren't supposed to live forever, so they see old people dying of these diseases as "natural".

"Natural" perhaps. "Not really possible to prevent" is something like a steelmanned version of it; maybe something like an artificial heart could keep people alive, but maybe such people's bodies are usually worn down enough that the best medicine in the world couldn't keep 90% of them alive for 5 more years.

This brings me to the question: What are the data on deaths in certain age brackets? If, say, media coverage was proportional to the number of people under 18 or under 25 who died, or perhaps some sum of "f(age(person))" where e.g. f(under 18) = 1, f(18-25) = 0.9, f(65+) = 0.01... then that would be a very striking result. (I guess if the smallest weight is 0.01, then no disparities exceeding 100x, such as the terrorism ~4000x disparity, can be explained.)

Edit: Well, here we go. https://www.cdc.gov/injury/wisqars/pdf/leading_causes_of_dea...

Unintentional Injury is cause #1 for ages 1-44. Suicide is #2 for 15-34, #3 for 10-14, and less for higher ages. Homicide is #3 for 15-34 and 1-4 (!), and #4 for 5-14. Cancer is #2 for 5-14 and 35-44, #4 for 1-4 and 15-34. Congenital Anomalies is #1 for <1, #2 for 1-4, and #3 for 5-9. Terrorism isn't listed, although if it were counted it might go under homicide. Heart Disease is #5 or less for 0-34, and #3 or higher starting at 35+.

If you figure that the news media only cares about young people, it goes some distance towards explaining matters...

Ok but that's just the definition of fear. There's no reasoning behind those feelings.

It wouldn't make much sense for the news to report on things which were not newsworthy. That people die from very typical things people always die from just isn't news.

But people have always died from violence. It's hardly news.

Fewer people die to violence now than they ever did. That’s what makes it news.

So by the argument, the more terrorism kills, the less it should be reported?

I think it's clear it's not due to its infrequency that it's reported on so much.

Are you sure about that? I haven't actually lived through the Troubles in England, but my understanding is that while terrorist acts were obviously reported by the news, there was much less hysteria than for today's much rarer terrorist events.

Our brains look for/create drama + drama gets views. To the first point, I’m reading the book Factfulness, and it covers this stuff pretty eye openingly we’ll.

Reply on news.ycombinator.com