> But is there evidence that such a disconnect exists between what we see in the news and what is reality for most us?
is an interesting one, particularly in this case. The leading cause of death for African American males 15 - 35 in 2015 was homicide according to CDC statistics. It seems that’s a very important news worthy focus, especially with respect to policy debates that shape the way US cities are run. Even including caveats in your report would bring an important and nuanced view. I’m curious as to why this was not discussed.
One thing I wished we had done originally was stratify by age. Commenters on the original post mentioned that it wasn't entirely accurate to average over all deaths, esp. as different things are more likely to kill you at different times in life.
The short answer for why we didn't do it is that I didn't realize this until after we finished the project. This was for a university data science project, and we were working on limited time.
I know this is perhaps besides the point: but I'm surprised that traffic incident reporting is so low - but then aren't you scraping more intellectual news media than what I would imagine most people would be reading? Not very familiar with the English/American news situation.
You can do a low-quality check easily. Read some medium that you suspect will be different from NYT/G for a week and count how many traffic accidents were reported in a particular locale, and how many reports there were for each. Then call the police and ask how many there actually were in that week.
I can guess what you'll find: What's chosen for reporting is news or otherwise noteworthy events. Traffic accidents may be a cause of death, but that doesn't make them either news or noteworthy. Traffic jams and their causes are an everyday occurence, even if the cause involves someone being driven off in an ambulance.
The most interesting tidbit for me was how similar the news breakdown was between the two sources: New York Times and The Guardian. The largest deviation was 3.4% for suicide coverage but almost everything was within 1%.
Perhaps these examples are too similar to get a good distribution but if news organizations are all covering basically the same items perhaps there is an opportunity for differentiation. The google search trends shows that what people are interested in knowing doesn't match what media is interested in telling.
I wish they had chosen a different, or a third source though. I rate NYT & The Guardian (though both of decent quality) as fairly similar in editorial outlook.
Even as the person who wrote the scraping code for the original project, I'm a little suspect myself of the news data because of how similar the two distributions turn out to be.
I think the strong similarity is an artifact of our data collection process, rather than reflecting some very deep truth about the similarity of the two sources. Or of sources in general. My priors are that the distribution should have looked more different, but I just didn't do extra verification at the time.
This makes me ponder: The human brain is wired to be interested in novel, relatively unusual things. So, almost by definition, this implies humans are wired to be relatively disinterested in truth. :/
That's why I love that this article also compared against search trends, which is another way of measuring what we are interested in.
Consider "road incidents". Search is disproportionately higher than deaths - we are definitely interested! But news coverage is extremely low, because what is newsworthy about any specific road incident? It's more naunced than "the news reflects what's interesting".
Or maybe NYT and The Guardian are just missing out on a big area of reader interest, and they could increase engagement by running more stories about road incidents!
This data is based on causes of death defined purely physically. It's a shame that in a world as complex as ours, we still don't attach psychological explanations and categorize them as causes too. Why not: he died of loneliness? He died of a broken heart? Or economic ones: He died because he couldn't afford the healthcare he needed? He died because he couldn't pay the heating bill and they turned it off?
Perhaps then we could get a different handle on the issues that face us, especially that face the elderly, and be more responsive.
We don't know, because suicide is the only category on the chart that reflects a psychological factor. If we tracked (more correctly, if we could track) people dying of loneliness but who don't commit suicide, perhaps we'd find the number to be hugely higher. When I look at the sadness of the elderly in large cities, my sense is that number would be a lot higher. But we just don't know.
To some extent we do know. We know people who are lonely tend to be less active, and we know people who are less active tend to have worse health.
On the Greek island of Ikaria for example[1], we know that people regularly live to be 90 years old and that even at that age they remain incredibly social and physically active.
There is a good book I read about this years ago (pre-9/11) called The Culture of Fear by Barry Glassner. It went in depth into the different things that people are afraid of and how they are covered in the media vs what you are actually likely to die from. Definitely made me think a little differently about what I saw on the news for good or bad.
But it's not. Sheeple/masses have been trained into that concept, but Edward Murrow is probably rolling in his grave at the fact. Have you spent time with reporters, specifically TV? The real newsies hate doing the fluff stories, or the must carry by their corporate owners. They want to just do the news. Of course, there are the ones that are just hoping to be the next entertainment star as you have accept news programming to be.
Glaring omission is deaths from airplane crashes. I would guess that is the most disproportionately covered of any cause of death.
I'm also skeptical of the data on drug overdoses. Seems like the opioid epidemic is pretty well-covered these days but listed as only .4% of "Media Coverage" and 1.3% of searches.
If the news does to proportionally show what we die from, then we all focus on the wrong areas, and vote in politicians who come up with solutions to the wrong areas.
For me, the most likely way I have of dying is a car hitting me. Heart disease is not an issue for me as I exercise every single day and eat, and cancer is not a concern as I do the above and anything else will not affect it significantly.
> Heart disease is not an issue for me as I exercise every single day and eat, and cancer is not a concern as I do the above and anything else will not affect it significantly.
If only it were that simple. Yes, exercise greatly decreases the risk of certain things as a healthy body is just that much more resilient. However, please get yourself checked at your regular 10k, 20k, 30k miles. Life happens in funny ways. Don't be an edge case. Also, if you live in California, you have to be super careful of all the cancer causing things that are only known to the state of California.
Every year in the United States, 225,000 people die iatrogenic deaths.
According to the Journal of the American Medical Association, iatrogenesis is the third largest cause of death in the United States after heart disease and cancer.
Iatrogenic. What is that word? Is it a disease, an accident, or something you get from smoking?
In fact iatrogenic means inadvertent death caused by a doctor or a hospital.
Why don’t they just say that? I guess it isn’t exactly in medicine’s best interests to put it in simple language that anybody can understand.
The oddest one are road deaths to me at 4x below cancer, apparently taken for granted at any stage of life, but mostly avoidable with some policy changes.
The road death category is wrong. What is being reported is "accidents", of which CDC says there were 161,374 out of 2,744,248 deaths in 2016. But even this only gets you 5.88%, under their 7.68% reported in the interactive graph when hovering over 2016.
From [0]
Total Deaths 2016: 2,744,248
Unintended Injuries 2016: 161,374
From [1], a breakdown of those unintended 161,374 injuries above:
Accidental Poisoning and exposure to noxious substances: 58,335
Motor Vehicle: 40,327
Fall: 34,673
Accidental hanging, strangulation, and suffocation: 6,610
Accidental drowning and submersion: 3,786
Accidental exposure to smoke, fire and flames: 2,730
Dividing the 40,327 motor vehicle from the 2,744,248 deaths gets you 1.47%, way less than the 7.68 shown in the interactive chart.
- Start with educating your drivers properly. As a German the US process of acquiring a driver's license looks like thinly-disguised rubber stamping.
- Rethink your roads, signals, and what surrounds them. Use modern techniques that make pedestrians and drivers safer.
The above two should get you from about 7.3 road deaths per billion vehicle kilometers to somewhere in the ballpark of Germany (4.2). Maybe even better than that since some of the planning and signals here still suck...
Finally you're just driving way too much. This becomes obvious when comparing road deaths per 100k inhabitants: just 4.1 in Germany vs 12.4 in the US. Yup, that's a 200% difference. Public transport anyone?
Canada is also doing better on those statistics, but I don't know much about how to acquire Canadian driver's licenses or what their road planning looks like.
I'm skeptical that harsher DUI penalties are the appropriate solution. Most states already have significant DUI penalties, and think that increasing them is largely a PR move. Drunk drivers are involved in about 20% of fatal accidents, leaving 80% to other causes[1]. About 30% of fatal accidents involve speeding [2]. Perhaps we should consider similar penalties for speeding as for DUI.
Drunk drivers do kill more "innocents" than guns, but also keep in mind that very few gun deaths are innocents. 97% of gun deaths are homicide or suicide [3]
The deterrence effect is a function of both the severity of the punishment and its certainty. And while the punishment for driving drunk is already substantial (maybe even too high), the risk of getting caught is not that large. An effective (but maybe not efficient) crack-down would therefore aim to increase the risk of getting caught and punished.
do you have any ideas for cracking down on drunk driving that wouldn't inconvenience and intimidate law-abiding citizens? the only methods I can think of are checkpoints and large-scale surveillance, both of which I would oppose.
I don't. Which is why I called a non-efficient solution. A theoretical solution would be to require ignition interlocks on all new vehicles. Not that I am advocating for that. A more realistic solution should focus more on officers observing and punishing bad driving, including all the effects of driving drunk or with cellphone in hand.
I could certainly support cops focusing more on reckless driving (tailgating, weaving, no-signal lane changes). if a cop sees a driver doing something unsafe with their own two eyes, I have no problem with them making a stop. I bet they would actually end up catching a decent number of drunk drivers this way.
Checkpoints are very common where I live (they're called RIDE checks, which I suppose is acronym for something). They're more common around public holidays, but can happen at any time.
They're not an inconvenience at all, and I'm not opposed to them in any way. It takes 20 seconds to go through them, and all the officer does is ask you if you've had anything to drink or have used any "wacky tobaccy"[1] recently, shine a flashlight into the car to check for open drinks (and presumably, use his nose too), then let you go.
But then again, I'm a white male, and certain minorities might have worse experiences.
[1] That's a new one that I heard just a few days ago.
I'm honestly not sure why this is considered acceptable. officers should not get to stop me and shine a light around in my car unless they have at least some reason to suspect that I am breaking the law. the fact that I'm leaving downtown on a holiday isn't enough.
> officers should not get to stop me and shine a light around in my car
Why not? Which of your rights are they infringing on by doing these routine checks?
It's not like they're free to do anything they want. The Supreme Court of Canada has ruled that during a RIDE check, they can only check for sobriety, and any other searches or questioning will result in the charge being dismissed (unless the evidence of a crime is in plain view, like a gun on the dashboard).
Here's a great summary:
R.I.D.E. stands for “Reduce Impaired Driving Everywhere.” It is an Ontario sobriety program that was established in 1977 with the purpose of reducing the number of tragic accidents and injuries resulting from impaired driving. The R.I.D.E. program has a narrow and specific mandate, which is to detect and deter drunk drivers. Conversely, R.I.D.E cannot be a tool for arbitrarily detaining individuals without reasonable grounds and in violation of their rights under the Charter. In fact, there have been many instances of criminal proceedings where charges were dismissed because officers over-stepped their legal mandate in the course of administering a R.I.D.E. check.
The R.I.D.E. program provides police officers with the legal right to perform planned roadside checks to identify and charge drivers who are under the influence of alcohol. The Supreme Court of Canada has ruled in favour of R.I.D.E. as a justification for stopping and checking drivers. In this regard, R.I.D.E. is unique as it gives officers the right to briefly detain and question driver even if there are no grounds or probable cause for believing that a driver is over the legal blood alcohol limit, impaired or has committed any offence. However, while executing the R.I.D.E. program, police are not authorized to perform other criminal investigations or searches, unconnected with the purpose of R.I.D.E. There are many exceptions to this rule, for example, if illegal drugs or other contraband are in plain view in the vehicle.
Especially considering how often DUI's are repeat offenders, there are plenty of chances to permanently revoke someone's driving privileges with due process. One recent "crackdown" in the law called for mandatory jail time only after the 4th offense. That's a ridiculous number of chances to give someone for something so selfish and reckless.
Not in Ontario, where RIDE checks take place. Canada has quite strict impaired driving laws.
Quoting Wikipedia[1],
The minimum sentences are:
For a first offence, a $1000 fine and a 12-month driving prohibition,
For a second offence, 30 days of jail and a 24-month driving prohibition, and
For a third or subsequent offence, 120 days of jail and a 36-month driving prohibition.
If no one is hurt or killed, and the prosecutor is proceeding by summary conviction, the maximum sentence is 18 months of jail. If no one is hurt or killed, and the prosecutor is proceeding by indictment, the maximum sentence is 5 years of jail.
If another person suffers bodily harm because of the offence, the maximum sentence is 10 years in jail.
If another person is killed because of the offence, the maximum sentence is a life sentence.
The link you cited from nhtsa.dot.gov actually said it was 29% of fatalities, and that was only for BAC of 0.08 or higher. I'm not sure it's tracked anywhere, but I'd be surprised if there weren't a lot more where the BAC didn't rise to that level, but was still a factor. I agree that steeper penalties may not help though.
Thanks for pointing that out. I'm not sure how to reconcile the 29% listed in the intro and table 4, with the 20% listed in table 2 and table 3.
Table 4 shows BAC >0.01 if you are interested.
I have a relative who has been repeatedly arrested for drunk driving. He had already severely injured someone and will kill someone sooner or later. The courts do nothing, he walks free with his driver's license every time. Even walked after assaulting an officer during one of the arrests. They will not put him in prison or take his license away.
This is in a midwestern U.S. state.
The laws may or may not be strong in theory, but even if they are strong they are not applied consistently. The courts are failing to protect the public from people who present a clear danger to the public.
This person, for example, is white. I am 100% sure that if he was not white he would be dealt with more harshly (and more appropriately, in this case).
It probably helps that mass transit is much more prevalent. In New York state for example, where the population is heavily weighted towards New York City where mass transit is more accessible, the % of traffic fatalities related to alcohol drops to about 5%, not far out of line for all of UK. (Though urban areas like London in the UK might be even lower
Where did you get the idea that the news is only supposed to talk about the government?
Even if that's true, how can the masses be expected to make informed political decisions on things like healthcare funding vs. war on terror funding if they're constantly being terrorized by the news about a negligible threat while ignoring much greater immediate threats?
I heard an ICU physician say that here in Australia a cause of death must be put on the death certificate and that the usual choice is "myocardial infarction" (heart attack). Whereas a better choice might be simply "old age", but this is not an acceptable cause. So I guess that this is one of the reasons why "heart attack" is listed as the most common cause of death.
Why is "old age" a better choice? Usually there's some kind of a disease that can't be handled by the aging body. Is it because it's not worth it to identify what the disease is?
A whole chain of things need to happen for you to die, but they typically culminate in starving the brain of nutrients or oxygen. That usually happens due to organ failure, and that can certainly be the eventual consequence of old age. So tying 'cause of death' to a single factor is almost always a gross oversimplification. There are many causes for every death, and it's only important to look at specific ones should you care to prevent them or assign blame for them.
They run from 1999-2016 for both the news organizations, and from 2004-2016 for the Google searches. 9/11 happened in 2001, so of course there were going to be lots of articles about that, and less searches if you exclude that time period.
The deaths were selected from 2016 alone, which is maybe reasonable, but then why select articles from a range of years elsewhere? Certainly there were more deaths in 2001 from terrorism than any other year in this range, so excluding it from some things while including it in others seems like an act of Hanlon if nothing else.
A pattern I noticed is that media coverage is biased towards the things that kill us that we don't have as much control over.
I imagine people know heart disease is a huge killer but they also know they can make lifestyle changes to prevent it and they know what those changes need to be.
> biased towards the things that kill us that we don't have as much control over.
It's not about what we have control over, but what we have information about. Informed consent is the bedrock of western society, from the enlightenment period. I suspect fear mongering (and terrorism) is effective precisely because westerners are trained to define unknowns and mitigate danger.
Some go so far as to be outraged that someone hasn't preternaturally predicted how an unknown could have been dealt with.
This entire analysis is predicated on ignoring tail risk. Most causes of death are pretty normal events; if we took no measures to prevent any of them then chances are most of them would increase by very small amounts; maybe doubling for the most dramatic cases.
Terrorism is especially terrifying in this regard because if left unrestrained, I think it is pretty intuitively obvious that there is no realistic bound on the amount of deaths that it would cause. The lack of deaths is not a function of desire on the part of terrorists to cause death, but just their ability to do so.
In some ways this is similar to the reaction to nuclear power -- even though practically speaking it is very safe, if it were to go wrong, it could got wrong on a scale that is not really possible for almost any other source of energy. Regulations and safety practices can go a long way towards mitigating this, but the tail risk remains a huge unknown, not in the probability sense, but in the negative impact sense.
>Terrorism is especially terrifying in this regard because if left unrestrained, I think it is pretty intuitively obvious that there is no realistic bound on the amount of deaths that it would cause.
I think you rebut this one yourself - terrorism has a natural limit in the ability of terrorists to launch attacks. Not a hard limit, a statistical limit, but important none the less.
It is important to recognize the order of magnitude of the probability of the risk. At CERN they don't publicize that there is a long-tail risk of causing the destruction of the world because the probability, though non-zero, is exceedingly tiny (and people freak out, unable to think about tiny probabilities.)
>terrorism has a natural limit in the ability of terrorists to launch attacks. Not a hard limit, a statistical limit, but important none the less.
I believe terrorism generally creates more terrorism, so the limit you suggest would be amplified by the ability of terrorists to create more terrorists. However, I would interested in seeing some hard stats on this.
I guess the relevant end-state here is civil war. For which the historical upper bounds are pretty high indeed, like 50%. (Unless you've defined terrorism so narrowly that it only counts if it's directly ordered from a different continent, or something.)
> I think you rebut this one yourself - terrorism has a natural limit in the ability of terrorists to launch attacks. Not a hard limit, a statistical limit, but important none the less.
The limit is not “natural”. The limit is at least partially because we are excessively vigilant about terrorism, and put measures, effective and ineffective, in place to anticipate and prevent terrorism. The far reaching and extensive nature of those measures is a result of the same risk calculus that I describe here, and is a manifestation of the same phenomenon that the research linked here found.
If a terrorist acquires a nuclear weapon and detonates it in NYC, what happens to the death rate? The point about tail risk is that probabilities that seem rare, aren't as rare and defined as you'd hope.
A gun style nuclear bomb isn't that complicated. If terrorists were able to get enriched uranium, then they could probably make one. If there were no counter-terrorism efforts at all, then there is a reasonable chance that they would be able to get enriched uranium.
The problem with terrorism is that if left unrestrained, they may accomplish their political goals. Those goals are generally to let them have free reign in the region they're in.
For example, ISIS was busy setting up their caliphate, and if we had not defeated them militarily, they'd have their own little hellhole nation to run.
The amount of death such regimes can inflict on their own people is pretty significant.
It doesn't even require death squads or anything like that; I was at one of Sadam's palaces and remarked to an Iraqi that the pools were empty. Well, it turns out, once Sadam fell, the water was redirected back to the local tribes.
Because to punish recalcitrant tribes, Sadam would just cut off their water.
(None of this means major powers necessarily ought to be getting involved in all these areas; the point is that terrorists aren't just killing people for kicks, they have a reason for the things they do and there are consequences when they're successful.)
>The lack of deaths is not a function of desire on the part of terrorists to cause death, but just their ability to do so.
I don't necessarily agree. Terrorists are people with a point to make who view violence as the preferable way to make it (usually out of desperation).
Numbers are pretty much never the point (any more than companies set out to kill as many people as possible with pollutants). The symbol is what's important. The world trade center. Mosques. Gay bars. All representative of what the attacker views as a corrupting influence on society in their own worldview.
I also suspect that if it was less widely disseminated a different strategy would be used, i.e. just actually going for military or government infrastructure rather than trying to sway public opinion. Terrorism is a political act,it is all about sending a message.
To add another point, not only does this Steven-Pinkerism or whatever name this ideology deserves ignore tail risk, it also fails to analyse things from any point of view that isn't ten mile birds-eye utilitarianism.
The primary motivation behind terrorism, is as the name suggests, not material violence but psychological damage and attack on the nerve centers and political systems and ideologies of targets. The physical violence is only secondary and a tool. So to evaluate terrorism in terms of body count, and not generated fear, is quite myopic.
It also of course ignores that systems aren't uniform and can take blows at any point with equal consequence. Terrorism at a political focal point of 10k people can collapse a nation state a thousand times as large, ask Franz Ferdinand and Europe. The effects aren't linear.
It's quite frightening to see how shallow and superficial the analysis of people has become who are apparently in charge of our institutions, because this thinking is extremely prevalent.
This is a really interesting point. It reminds me of the story of the American soldier meeting a former Vietcong many years after the Vietnam war. The American said, "Every time we met you guys in the field we beat you. You never won a battle." The Vietcong replied, "That is true. It is also irrelevant." Vietnam was a great example of how misleading measures and statistics could be.
This article was about causes of death. They perfectly reproduced the causes of death versus the reports.
The interpretation that we should not discuss certain causes of death as much was only added in this thread. The only implication discussed in the article seemed to be "Perhaps we should provide more context".
I would say that people dying from outlier events is newsworthy because they are outliers, not that we should cover events in the proportion of their occurance in real life. That'd be awfully dull.
> if we took no measures to prevent any of them then chances are most of them would increase by very small amounts; maybe doubling for the most dramatic cases.
If by behavior we can reduce heart disease and cancer by half, that extends far more lives than eliminating terrorism, homicide and suicide entirely. Even the tail risk of terrorism is on the order of 10^4/year, while heart disease plus cancer is 10^6. So adjusting for tail risk, the focus of coverage on violence remains disproportionate.
There's a more substantial tail risk in global warming, nuclear war, infectious disease or SMOD. If the coverage of those was proportional to their tail risk they would crowd out most other news.
We can delay heart disease, cancer, and strokes, but not prevent the deaths. Those 1e6 numbers aren’t moving very far.
I assume you’re calculating the tail risk of terrorism based on backtesting, and ignoring the fact that preventing terrorism is a disproportionately high priority of the governments of the world. The extent and scope of measures taken cannot be ignored when talking about the impact.
Yes global warming has tail risk; it doesn’t show up in this analysis because it isn’t causing deaths, but I think the level of coverage is pretty high considering the low death toll.
Similarly nuclear war - especially during the Cold War, thermonuclear weapons had killed fewer people than falling pianos, yet for some reason was very high on people’s mind.
I posted this sentiment in another thread and got a huge amount of down votes. The counter point, at least to nuclear, is that we did have a severe nuclear incident @ Chernobyl and Fukushima and neither amounted to millions of deaths.
That is, if you are a Black Male in the US between 15 and 35, yes homicide is how you are more likely to go. And that news becomes very important to Black mothers, fathers, grandparents. The real — as in US government stats — on infant and maternal mortality are also worse than many Caribbean countries.
My takeaway is that there are different “world”s within countries — especially for marginalized groups where the country doesn’t have the will or resources for a safety net.
> "What's interesting is that Americans search on Google is a much closer reflection of what kills us than what is presented in the media..."
This is because the key driver of what the mainstream media reports is how well a given topic will increase the bottomline.
Pardon the editorial, but this is what passes for journalism currently. Sadly the media drinks their own Kool Aid, believes their own press releases, and continues to trumpet its own importance, even when the data/evidence consistently tells otherwise.
> This is because the key driver of what the mainstream media reports is how well a given topic will increase the bottomline.
Counterpoint: there are news sources that don't have a profit motive, e.g. BBC News. They still do this same kind of reporting. For them, it's not about their bottom line. So... what is it about?
Yes, but the question is, why are these metrics chosen in a way that incentivizes BBC News reporters to follow exactly the same kind of sensationalist stories as for-profit news sources?
Career, ego, "conventional wisdom", riot mentality, etc.
All those aside, it's naive to think the BBC doesn't track ratings, do market research, etc. They might not have a traditional purely financial bottomline but they still have a bottomline, as well as someone to report to about results, etc.
The BBC is a bit of a laughing stock at the moment because of their wall-to-wall coverage of right wing Brexit types. Unsurprisingly because many of the "journalists" have connections to the Conservative party.
They also suffer from the "two-sides" problem. They'll get a scientist who's spent their life studying climate change, then on the other side someone who says "It's a hoax". That's not investigative journalism.
Should media coverage accurately reflect what we die from? I don't think so. Cause of death is not the only reason to be interested in terrorism, homicide, and suicide. These charts are eye-candy, and while I respect the work that went into producing them, I believe they are simply trying to spin a story instead of addressing the title's topic.
It probably wouldn't differ that much from the NY Times. I would also expect MSNBC to be about the same. Its the angle that is the difference and not so much the subject. The big networks pretty much fall in line on topics. I think there was a study posted a long while back here on HN, but it had something to do with the NYT.
Yep. Capitalism-induced feedback loop. That said, more people are going to pay attention to a newscast about terrorism than one about, say, how the yellow dye used in legal pads is poisonous.
If you think about it, I'm actually more likely to accidentally ingest a legal pad.
There are two parts to terrorism: doing something terrible, and spreading the word to upset and frighten lots of people.
It's one of the oddities of modern Western societies that you can end up in jail for donating to what you think is a liberation movement in another country, but when the New York Times and The Guardian are making money by doing the second part of terrorism in their home countries, nothing happens.
meh...
- "Are car reviews watched proportional to car sales?"
Seems Ferrari has really bad sales reps...
- "Do food-related articles reflect what we eat?"
My pasta-with-a-canned-sauce is still waiting to make it to the cover of Saveur.
Happy to answer more questions people might have about methodology or the analysis in the original study.