I think this serves as a great lesson in statistics being converted into news stories.
The roads are safer.
The roads are deadlier.
Both are true and both are false depending upon if you are looking at the total or the per driver (and perhaps even if you are looking at per driver per minute or per driver per mile).
A better title would have been something like "Traffic fatalities in March drop to ####".
When I was in school, our statistics prof used to put up that daily chart on the bottom of the USA Today front page. He wouldn't start class until we found two or three things wrong with it. By the end of the semester, nobody in the class trusted anything from that newspaper. Great learning experience!
Talking about terrible statistical presentation - I don't know if you saw this graph[1] that Georgia published displaying corona virus impact but if not... there is something a bit off with it.
That one is so bad and so obvious to anyone who takes the time to actually read the labels that it feels like a case of malicious compliance. A boss needed a chart to support a falling case count and someone made him a visually appealing one while also showing the trend is BS.
As someone who has been watching GA's charts closely over the last few weeks, I actually disagree in this case. It was one chart of several on the page, and not even the most prominent[0], the other charts were still in the right order, and this chart still showed an overall downward trend even put in the correct order. As someone who has dealt with coding lots of data into charts and making frequent changes to them and seen how bugs easily sneak in, I find mistake more likely than malice.
That's why everyone need to learn to separate fact and opinions. Ignore the conclusions, instead look at the data. Critical thinking skills are in short supply.
The news is pretty bad in that it's mostly opinions pretending to be facts/information, with occasional data sprinkled in to support their conclusions.
ive been banging on about the subjective nature of "facts" on HN for a while now. perhaps a parable from the movie Charlie Wilson's War will help:
A boy is given a horse on his 14th birthday. Everyone in the village says, “Oh how wonderful.” But a Zen master who lives in the village says, “We'll see.” The boy falls off the horse and breaks his foot. Everyone in the village says, “Oh how awful.” The Zen master says, “We'll see.” The village is thrown into war and all the young men have to go to war. But, because of the broken foot, the boy stays behind. Everyone says, “Oh, how wonderful.” The Zen master says, “We'll see.”
I'm sorry, there is no opinion in either headline. Both are correct, newsworthy and correctly presented. Here are the relevant headlines:
"Stay-at-Home Rules Reduce Fatal Collisions"
Fact. Not opinion. Correct. And important to know, because it's good that fewer people are dying.
"Emptier US roads more lethal"
Likewise a correct fact and not an opinion. And newsworthy because driving right now is more likely to get me killed, and that's something I'd like to know.
And the media got it right, in the headline even. It's posters here that are trying to conflate the two and make them seem like they say opposite things when they clearly did not.
So... which party is the one injecting opinion? It doesn't seem like the media to me. This seems like a bunch of HN commenters trying to spin up a controversy.
> Likewise a correct fact and not an opinion. And newsworthy because driving right now is more likely to get me killed, and that's something I'd like to know.
"lethal" here means more deaths per mile traveled. that might be because roads everywhere are less safe, or it might mean that a greater fraction of total miles are being driven on inherently unsafe stretches of road (with the same local fatality rates as before the crisis). fatalities include pedestrian deaths as well. maybe there is increased risk for pedestrians but not for drivers. the article doesn't say. there's certainly not enough information to say whether you personally are at greater risk driving.
Uh... the detail you're asking for is literally in the text of the article. I still don't understand why you don't think that headline is an accurate summary.
And in any case... I don't understand your criticism at all. If there are more deaths per mile traveled, then every mile driven is on average "more lethal", by definition. So if I drive to the same place I (in the sense of an average "I" across all drivers) am more likely to die. QED.
Basically, you're just criticizing grammar by conflating my use of "I" with me personally and not the abstract reader I clearly intended. Do you make similar quibbles about the abstract use of "you" or "one", or the royal "we"?
I think the argument going on here in the comments is evidence enough that the title is evidence enough that it's misleading. some of that is just HN's special brand of pedantry, but some people are raising valid points.
> And in any case... I don't understand your criticism at all. If there are more deaths per mile traveled, then every mile driven is on average "more lethal", by definition. So if I drive to the same place I (in the sense of an average "I" across all drivers) am more likely to die. QED.
my criticism is that the average deaths per mile doesn't imply anything about your risk on a typical trip for you. for one thing, the excess deaths could all be pedestrians, and the deaths per mile for drivers could be the same. in particular "So if I drive to the same place I (in the sense of an average "I" across all drivers) am more likely to die." is not implied by the information in the article, because a very possible explanation for the increased rate is that people are not driving to the same places.
let me try to construct a counterexample from my life. in normal times I drive 20 miles each way on the interstate to commute to work. once or twice a week, I drive a mile to the grocery store through a dense urban area. for the sake of the example, let's just assume the interstate has much fewer deaths per mile than city streets do (pretty sure this is true, but don't want to bother looking up a source). I, like many other people, no longer commute to work, but I still buy groceries once or twice a week. so a far greater fraction of my miles are on high-risk stretches of road. the average risk I expose myself to per mile is indeed higher, but this doesn't imply that driving to the grocery store is any more dangerous than it was before. in other words, it is a fact, but not a particularly meaningful one. it shouldn't influence your behavior.
reporting averages without some context or explanation is not very useful and people who aren't familiar with statistics are going to draw all kinds of unsupported conclusions. another random example: I live in a Very Dangerous City if you go by the overall murder rate. but if you dig into where the murders happen, they are tightly clustered in certain areas. I'm no more likely to get shot in my neighborhood than I would be in williamsburg.
> the average risk I expose myself to per mile is indeed higher, but this doesn't imply that driving to the grocery store is any more dangerous than it was before.
It doesn't mean it doesn't either. Someone would need to do the science to show it either way, and that hasn't happened.
I don't see how you get from there (which you could have said in like 5% of the prose as "the variables aren't independent") to your point that the news media shouldn't ever report on scientific results if there is the possibility of a confounded variable.
The standard you want applied would effectively mean that the media can't report on experimental science at all. So I'm going to go back to my original supposition that what you really WANT is for the media not to report on inconvenient facts, and you're hiding behind scientific pedantry to make the case.
There are at least two reasonable readings of that sentence. One is that US roads have more fatal accident risk per mile driven. Another is that the set of US roads is the scene of more total fatalities.
The second one is a more natural reading to me, but presumably not to some others.
Your definition would seem to insist that a nuclear warhead is not a "lethal" weapon, since no one has died to one in three quarters of a century. I guess "reasonable" is an opinion thing, but I think it's pretty clear that the usage in that headline is conventional and correct.
It does not insist on that, but it does mean that “saturated fats more lethal than firearms” would be plausibly correct and “saturated fats more lethal than nuclear weapons” being almost surely correct.
> “saturated fats more lethal than firearms” would be plausibly correct
Then I don't know which dialect you speak, but that's an outrageously weird interpretation. It's certainly not the way the word "lethal" is defined or interpreted in common usage among the folks I've known.
But those are both facts - less people are driving, but if you do drive you're more likely to be in an accident. Are those conclusions? Would a presentation of facts just be the data itself?
Are you more likely to be in an accident? Or are you more likely to be in an accident with a fatality?
It seems like heavy rush-hour traffic is prone to creating a large number of accidents with a low total number of fatalities. When I drive around now, I believe the risk of accident per mile is significantly lower (city dweller used to typically packed city roads), but due to higher speeds, I can’t make the same claim about lower fatality risk.
> Ignore the conclusions, instead look at the data.
Also, do not ignore the opinion, instead look for it specifically, and assume that the data has been curated to support that opinion. Then go find the gaps.
I think the biggest problem with the modern media is that all stories are reported at "equal volume." Any tidbit the media has to fill time is headlined as if it were the most important thing currently happening in the world. There's no moderation of tone or consideration for reporting an accurate overall view of the world.
It's both interesting that traffic deaths are down overall and that deaths are up per mile traveled. But I would argue that they are not anywhere near equally important news. The total lowering of traffic deaths is news. The increase in deaths per mile is an interesting bit of trivia relevant to few people and should be reported as such.
This has been especially bad with something as nuanced and data-driven as COVID. Any new number or stat gets a headline even if it has zero meaning in the grand scheme of things. I've seen news sites with two stories on the same page simultaneously proclaiming that there's no evidence that there is immunity to COVID while also proclaiming that a vaccine trial was successful.
Coincidentally, that's where oldschool print media was better, if only due to space being limited and budgeted. Computers, otoh, are only good at dealing with things that are all alike.
But also because articles were all laid out right there on the pages next to each other and could be compared in physical terms.
"The roads are safer" is not true. It's more dangerous to drive right now. There is no sense in either of these stories which it's safer to be on the roads.
Something like "the roads are killing fewer drivers" would be true.
We're not sure about that conclusion either. One other possibility is that the population currently driving on the roads right now is more prone to accidents than the population currently staying home, and so fatality rates per mile driven have gone up. Perhaps they have older cars without the latest safety features, or their cars have prior accident damage that wasn't fixed, or they skew younger with less driving experience, or they're rushing to get to their jobs, or people are out for joyrides. In that case, fatality rates among drivers staying home have dropped (to zero) and fatality rates among other drivers have probably dropped (less to collide with), but because of Simpson's Paradox overall fatality rates have gone up.
You can't know which is a better explanation unless you have data with more granularity than is typically provided by the press.
or as another poster suggested, people might be driving more accident prone routes right now. my commute to work is a 20 mile point-to-point drive down an interstate. aside from merging, there's not much that can go wrong. a trip to the nearest grocery store is a one mile drive through a dense area of the city with many pedestrians and awkward intersections. the per mile risk of the grocery run is much higher.
These reports only say "motor vehicle fatalities", not whether the drivers are the ones at risk.
In my experience, there are a lot more pedestrians blindly walking or running across the roads in the past couple months. The drop to nearly zero traffic lulled people into a false sense of security. People are treating every road like a giant sidewalk.
The whole jaywalking thing is idiotic. Let pedestrians walk. I live in San Francisco where pedestrians are going to be everywhere and it's really not that much harder. In the UK, you can just cross any non-motorway at any time and it's about whether the situation is safe. Well, fewer fatalities total, fewer fatalities per mile driven, etc. etc.
Genuinely don't think non-highway road crossing needs to be excessively regulated. Human beings will self-organize and the exceptions will not cause significant economic damage. More freedom and less regulation, please.
And that is exactly what the story is saying, which is why organ donation is down. The story does not say the roads are safer. You are drawing that conclusion yourself or being led to it by other commenters here. The words "safe" and "safer" are not present in the article.
Actually, that's the funny thing: drivers don't kill drivers. Not even collisions kill drivers. It's just that some people, post-collision, have their heart stop. So really, if you think about it, their own heart killed them. We should mark them all as suicides, honestly.
In this usage, "road" means more than the asphalt. It includes the cars on it as well as objects that could be crashed into, such as light posts, nearby trees, etc. If it is a road without a guardrail next to a cliff, it includes the ground at the bottom of cliff, since that is probably what's going to kill you, not the road itself.
You could express it in a way that would be correct to all the pedants out there, but it might be a long convoluted sentence.
The drivers aren't necessarily more dangerous. They might just be driving more dangerously with less traffic. It's easier to die going 80 on the freeway than if you're stuck in gridlock.
I think we can at least agree that the statement "the roads are safer" is false.
Plenty of places are increased traffic but not bumper to bumper
This is all rounding numbers that have been rounded already to smooth out uniqueness of a given locale
Water is wet level conclusion that using math correctly by math rules to debate a topic that’s locally unique and out of view for people in far flung regions.
My individual odds of dying in an accident haven’t changed much. Safer or worse in some aggregate measure isn’t that useful to me today or tomorrow. Still gonna do as I do.
The roads are safer.
The roads are deadlier.
Both are true and both are false depending upon if you are looking at the total or the per driver (and perhaps even if you are looking at per driver per minute or per driver per mile).