He didn't actually explain how Huff was mistaken in his argument about lung cancer, though--which is a crucial part of the lesson, in my opinion. Dismissing Huff's argument on the basis of "well, he supported smoking, so he must be wrong"--which is the sentiment that Harford seemed to be leaning on--is circular reasoning. The reason that Huff was mistaken about smoking is that, unlike the storks scenario, he couldn't provide any third variable which causally explained the correlation between the other two (as far as I'm aware). That's a distinction that should have been explained in the video.
@@mvmlego1212 While I would agree that if they were trying to make the actual argument, you'd be right that they should make that explanation, I think they did fine for a TH-cam video more about telling the history than arguing a side. Because they did already talk about the third variable in the storks/babies case, and hinted at the fact that they tested other possible causes in the smoking case by mentioning the cars theory, so you can kinda make the connection that it was there, even if they didn't say it explicitly.
But how do you differentiate between the two? Remember: you would mistake skepticism for cynicism and vice versa depending on what you ALREADY believe to be true. The only reason, for example, that anyone here calls Huff cynical for saying that smoking causes lung cancer is because they already believe smoking causes lung cancer. If you didn't believe that, you'd call Huff a healthy skeptic.
@@julesmcbride2692 I think that some of what is missed missed in this video, but understandbly so and isn't absent among discussion in the actual field of statistics, is just what maked positive examples positive, which is having a very rigorous methodology which is attentative to all the pitfalls and mistakes one can make and which makes an active effort to both avoid them and make clear exactly how conclusions are being drawn. Going into those more could be very interesting, but I suspect that could be a whole other video (in fullness it is practically a degree tbh, but in summary a video). Minutephysics I think has a great video called something like Correlation Can Equal Causation about how you actually can try and get causal information using statistical analysis by being careful.
As a statistic this video really makes me happy. It's always sad when people dismiss statistics because people have lied with them. But statistics is a really important thing.
I'm a stats myself, I agree. Basic statistics should be taught in grade school with the focus on how to interpret graphs and how to tell when you are being mislead
@@subscribefornoreason542 The "lived long enough..." part? It's from Batman: The Dark Knight. "You either die a hero, or you live long enough to see yourself become the villain." -Harvey Dent
That was a pretty upbeat video until that end bit - glad you included it though. It's just as important to keep in mind that correlation doesn't mean causation as it is to let experts do their job, particularly in the medical industry where statistics really are the best tool for the job.
This video reminded me that maths mean something--not just that they can explain things, like scientific phenomenon, but that they can prove things, or else ruin things if they're worked out poorly. Thank you so much for this.
economics is largely a soft pseudoscience that misuses maths to try to show it's real science, at least mythology studies do not pretend it's scientific (I study economics at uni so I know what I'm speaking about)
I find it lovely that you got an economist to talk about this topic - anyone who has studied economics knows how integral econometrics i.e. regression analysis i.e. statistics is to the practice! Anything empirical in economics is always expressed through statistics.
I think it's important to understand exactly what the flaw was in Huff's argument that smoking was benign so that we're not using circular reasoning. Arguing against the significance of the correlation between storks and babies makes sense because we're able to identify a third variable that correlates and causally explains the other two. On the other hand, Huff never proposed alternative explanations for the correlation between smoking and lung cancer (as far as I'm aware).
I am a fan of "How to Lie with Statistics" (and its sequel "How to Take a Chance: The Laws of Probability"), but I am a bit taken aback to hear of Huff's connection with the smoking lobby.
The problem that the general public has with statistics is that they don't really understand uncertainty or the assumptions that underlie the statistics. When they hear the results of a poll, they either lend it too much credibility or too little. When they hear that some candidate has a 70% chance of winning, they assume that means it's almost guaranteed, not that almost 1 out of 3 times they will lose. It's why it's so important how statisticians and social scientists communicate their findings and why using statistics to lie is so damaging.
I think the stork-baby correlation is flawed. Stork-baby correlation in Eurasia and Africa may be strong, but storks are much rarer in the Americas, Australia and island nations. I propose a different theory: Babies come from cabbage patches. I don't have the actual numbers, but I think the correlation will be stronger than the stork conjecture when all data points are available.
I think Huff was misunderstood because the book was so popular. He wasn't saying you can't take statistics seriously, he was simply making fun of some terrible examples of their use. Edit: my bad that's what happens when you comment before watching the whole video lol.
Can you do a video about what statistical principles explain why the case for smoking causing lung cancer is greater than the case for storks causing babies despite the correlations being similar?
This makes perfect sense to me. Statistics can give a big hint as to where to look, but you still need hard evidence to be sure. We took the statistics and based on them looked into HOW smoking might cause cancer. We now know the mechanisms by which it occurs, so we know the statistics are useful and correct in this case.
Indeed, I've heard things about economists (on The Internet of all places, therefore they must be true!) - it's called "The Dismal Science" and not for nothing. It's (allegedly) the only field where two people can get the same top prize for writing papers that come to opposite conclusions.
@@TranscendentBen Fun fact: the phrase "the dismal science" was coined by a 19th-century slave owner who was grumpy at the growing consensus among economists to support abolition. Economists aren't always wrong about everything...
This was a lovely watch full of interesting and important history I didn't know I needed to know but am definitely pleased to know now. The books mentioned seem like amazing reads so I'll definitely be checking them out! Thank you.
At risk of sounding pedantic, I think he also pronounced "causal relationship" as "casual relationship", which I thought was a strange error for a professional to make
@@turtlellamacow do you mean the bit about the senator asking whether the smoking/cancer relationship was as "casual" as the stork/babies relationship? I'm pretty sure that was meant to be casual and not causal. The stork/babies relationship is not causal. It is casual.
@@paulmealor9128 You may be right; I'm not certain it wasn't intentional. But describing a relationship as "casual" is very uncommon (outside of dating!), whereas misreading "causal" as "casual" is a common error among stats students.
@@turtlellamacow Casual is the correct and intended term for describing the smoking-cancer association as being "as casual as" the stork-baby association. Casual can mean incidental or happening by chance or accidental or informal. It seemed to me the economist was quoting a senator asking the witness, "is the relationship between smoking and cancer as incidental/accidental/unrelated as that between babies and storks?" Please don't expect a US senator from 60 years ago to be using "causal relationship" in a sentence. US senators then and now are not noted for their expertise in understanding little things like science and maths.
This and the mega favorite number thing i think its great that theyre starting to collaborate to share as much knowledge with the world as possible (and pay the bills while theyre at it lol)
@@DendrocnideMoroides it's a meme... best not to think about it lest you become one of youtube's hopeless social cripples who can communicate in nothing but memes
Would be interesting to see, what exactly the differences between the statistics for storks/babies and smoking/lung cancer are. How do you make sure that you prove an actual causation between smoking and lung cancer, and not just a correlation?
This was the first video I can remember here on Numberphile that felt unfinished. These videos usually leave me wanting more, but they also usually feel complete.
Well, as is explained, # storks is actually positively correlated to # babys. The correlation is through the land area of the countries. But that does not mean 'storks deliver babys'. But after all, they do correlate with each other, which is what exactly the statistics suggest. Hence statistics do not lie, people may lie by misinterpreting the statistics.
I thought the stork thing came from them being attracted to roofs and chimneys of warm homes. Homes with a new baby would keep fires going all night, so storks would hang out at those homes more.
I first learned of Tim Harford via Malcolm Gladwell promoting Tim's excellent podcast "Cautionary Tales". Now it's come full circle as he's on Numberphile and selling a book through StandupMaths!
@@NatePrawdzik I know, right? _All detergents are synthetic!_ Soap doesn't just grow on trees! Although, there _is_ a strong correlation between the number of trees and the amount of soap...
I remember reading a book that showed that the increase in the sale of ice cream led to an increase in criminal activity. It was then revealed to be an example of using statistics to lie, and that the increase in crime generally happens in warmer weather, when people tend to buy ice cream more often.
"I'm not a student of mythology and traditions. Anyway, let's get back to the *important* stuff, statistics." As someone who loves both math and mythology, I'm hurt
If anyone is curious about how you can tell whether a given correlation is a "storks and babies" situation or a "cigarettes and cancer" situation, you're looking at the field of causal inference. Effectively, it comes down to mapping out all the reasons why two variables might be related (e.g., smoking and cancer might be related because of who chooses to smoke, or the simultaneous rise of cars, or genetic differences, etc. Etc.), and using statistics to isolate only the reason you're interested in (does smoking cause cancer?). Any decent econometrics course will go into this, or for a more computer-science perspective check out Judea Pearl. I also have a series on causality on my channel.
The difference with the correlation between smoking and lung cancer is that the correlation held for numerous different ways of approaching the problem (and, also, eventually there was a mechanism by which smoking could cause lung cancer). It's like the evidence for evolution. Just one correlation does not mean much but, if you keep finding correlations when approaching the problem from many different directions (fossils, geology, biology, anatomy, genetics, antibiotic resistance etc), then eventually correlation can start to look like causation.
In the case of Storks and Babies, the "solution" is that there's a hidden correlation--land area--which has a causal relationship to the first two. In the case of Cigarettes and Lung cancer, what ideas did Huff propose as hidden correlations? What hidden factor would cause someone to buy cigarettes and come down with lung cancer?
Easy! Smoking often correlates with unhealthy lifestyle, drinking alcohol, not doing exercises, having stress, doing unhealthy labour, etc. It could be caused by any of these unhealthy activities. D. Huff was right about statistics proving nothing. We know smoking is bad because of medical research, also just by using common sense. This is actually false argument vegans use, to 'prove' meat is unhealthy, by looking at patients with heart diseases (vegan/not vegan). But what they fail to understand, is that vegans follow a healthier lifestyle then obese people with heart problems. If all else is equal, you do all the things vegans do, and you eat healthy types of meat in moderate quantities - then you will be no less healthier then vegans are on average, and probably even healthier.
The problem is that unless the public health authorities start doing a much better job than of late, treating their various ukases with a degree of scepticism is a _perfectly rational_ stance. It's also fair to note that statistics have been perverted by State actors as well as by industry, in exactly the same fashion and for exactly the same reasons. Healthy scepticism doesn't mean a blanket rejection of statistics from whatever source, but it does mean one should be aware of cognitive and other biases, including the all-important question: _cui bono_ ?
I remember watching a football match where the commentators talked for 4 minutes about how the home team never conceded any goal, and they only stopped when the away team scored.
There is, in fact, a correlation between birth rate and storks. That's because there are more storks in rural areas and there's usually more babies (relatively) in rural areas than in cities.
According to a very scholarly and large book I have about the history of the English language, here is the origin on the stork story: Storks were originally called sticks, which makes sense given their long thin legs. The name changed over time to storks. But back then the word stick was also a slang term for, well, a certain part of the male anatomy. So adults were playing a joke on kids with this story of where babies came from.
I remember that there is a correlation between storks and birthrates within countries. Simply because storks live in more rural areas and people in rural areas have statistically more babies than people in urban areas where there are less storks.
So how do you decide if a correlation is significant? How did they know that the correlation they observed between smoking and lung cancer was not spurious? When the judge asked if storks and babies was the same as smoking and lung cancer, I don't like that it's assumed that the view would agree that they are different. Why are they different?
3.13 I would actually not be surprised if there was a correlation with birth rate and stork number. But I would then think that was related to more socioeconomic stuff (Usually richer countries have a lower birth rate, but they might also have a more industrialized countryside, meaning there's less food for storks, thus less storks.)
My favorite quote by my statistics professor (possibly he was quoting someone): it’s easy to lie with statistics. It’s easier to to lie without statistics. Lol
Statistics is a set of mathematical tools. A language. You can use words to tell lies, and you can use words to uncover and communicate the truth. Similarly, statistics have no agency - it's the person who tells you the story, and their method of telling it you should be wary of.
And here I thought this was going to be about how the saying that storks bring babies comes from more babies being born in the spring and summer when storks live here and less babies being born in the fall and winter when the storks are in Africa - a tendency that used to be much clearer back when people still had a work schedule dictated by the change of the seasons. "Storks bring the babies" was never meant to be taken literally - it was just a sign of changing seasons. And only after that did the stork become a symbol for fertility.
Shing-Tung Yau says the conjecture might still be a conjecture, since no one really understands the proof. I don't even understand enough to say who is right about that.
It's not really "nasty cynicism" to question the validity of statistics. It's easy to feel that way with hindsight, but the fact remains that it was important *not just* to establish a correlation between smoking and lung cancer *but also* that the correlation was causal in nature, that the correlation was too strong to be dismissed. This has clear parallels today with Covid-19, the subject of many bogus claims backed up by bogus statistics. It's not a valid argument against scepticism as an approach that in a particular case it suggests that something you believe strongly in may not be true - *that's the point*
Very valuable lesson about the difference between skepticism and mindless denial/cynicism, which can be exploited just as easily as bad statistics.
I don't know. Might just be a coincidence.
He didn't actually explain how Huff was mistaken in his argument about lung cancer, though--which is a crucial part of the lesson, in my opinion. Dismissing Huff's argument on the basis of "well, he supported smoking, so he must be wrong"--which is the sentiment that Harford seemed to be leaning on--is circular reasoning.
The reason that Huff was mistaken about smoking is that, unlike the storks scenario, he couldn't provide any third variable which causally explained the correlation between the other two (as far as I'm aware). That's a distinction that should have been explained in the video.
@@mvmlego1212 While I would agree that if they were trying to make the actual argument, you'd be right that they should make that explanation, I think they did fine for a TH-cam video more about telling the history than arguing a side. Because they did already talk about the third variable in the storks/babies case, and hinted at the fact that they tested other possible causes in the smoking case by mentioning the cars theory, so you can kinda make the connection that it was there, even if they didn't say it explicitly.
But how do you differentiate between the two? Remember: you would mistake skepticism for cynicism and vice versa depending on what you ALREADY believe to be true. The only reason, for example, that anyone here calls Huff cynical for saying that smoking causes lung cancer is because they already believe smoking causes lung cancer. If you didn't believe that, you'd call Huff a healthy skeptic.
@@julesmcbride2692 I think that some of what is missed missed in this video, but understandbly so and isn't absent among discussion in the actual field of statistics, is just what maked positive examples positive, which is having a very rigorous methodology which is attentative to all the pitfalls and mistakes one can make and which makes an active effort to both avoid them and make clear exactly how conclusions are being drawn. Going into those more could be very interesting, but I suspect that could be a whole other video (in fullness it is practically a degree tbh, but in summary a video). Minutephysics I think has a great video called something like Correlation Can Equal Causation about how you actually can try and get causal information using statistical analysis by being careful.
Storks deliver normal sized babies. Bigger babies require a crane.
@Neil Gerace and very small babies?
@@1ax oh no!
No wonder my dad didn't give many the details. He didn't want to explain that he was delivered by crane!
Guess that's why it is storks that bring the babies.. because it is a big bird. Simple as that.
Nice pun.
Oh how I yearn for the days of being an infant soaring through the skies with nothing but a blanket on my back and a stork as my pilot.
Given your last name, are you sure it was a Stork? Haha
If you ever write a book please include this sentence as it reads wonderfully well.
I yearn!
@@AxxLAfriku is this a joke that i dont understand?
@@sharven417 ok
As a statistic this video really makes me happy.
It's always sad when people dismiss statistics because people have lied with them. But statistics is a really important thing.
"As a statistic"?
Do you mean "as a statistician"? Or are you referring to yourself as a statistic?
I'm a stats myself, I agree. Basic statistics should be taught in grade school with the focus on how to interpret graphs and how to tell when you are being mislead
it's why you always ask to see the raw data, not the single graph they made from it
It made me rather sad.
@Juan Cespedes I wrote this when I had a seizure
Petition for Tim Harford to write a book "how to tell the truth with statistics"!
Or, even better, "How to Tell Whether Someone Is Telling The Truth With Statistics Or Lying With Them: For Dummies"
@@jaschabull2365 yes please.
Petition for Tim Harford to write a book explaining where babies come from. And narrate the audiobook.
I agree. If not Hartford, somebody should write that book as a spiritual sequel / rebuttal.
@@MGSchmahl spiritual?
I'd pay to see a spiritual sequel, i.e. one where spirits are the subject
This cutesy-sounding subject turned out to be one of the darkest Numberphiles ever 😰
Sounds like economics, baby. The dismal science.
false.
This video was absolute madness. Who would have known that Darrell Huff “lived long enough to see himself become the villain?”
Yeah, I never would've guessed that a guy who wrote a book about lying with statistics would lie with statistics for money.
Is this a reference?
@@subscribefornoreason542 The "lived long enough..." part? It's from Batman: The Dark Knight.
"You either die a hero, or you live long enough to see yourself become the villain." -Harvey Dent
He's not the villain. His work helped curb human population growth and provided value to investors in tobacco companies.
@@medexamtoolscom Is your name Phillip Carvel?
Omg that ad placement right at that cliffhanger lol
Yessss!
I definitely didn't see that coming!
That was a pretty upbeat video until that end bit - glad you included it though. It's just as important to keep in mind that correlation doesn't mean causation as it is to let experts do their job, particularly in the medical industry where statistics really are the best tool for the job.
This video reminded me that maths mean something--not just that they can explain things, like scientific phenomenon, but that they can prove things, or else ruin things if they're worked out poorly. Thank you so much for this.
Stork story exists in Slavic mythology. During winter storks and other birds travel to heaven.They bring souls from heaven to human bodies.
Our version was them just going south to warmer countries. Wait thats not mythology
@The Absolute Madman Hehehe, antique version of Matrix or something :-D
Yes! I remember my grandmother (Polish) telling me the story.
Wait, they only deliver human babies? Dumbo was a lie?
"I''m an economist, not a student of mythology and ritual."
Cheeky nonetheless.
economics is largely a soft pseudoscience that misuses maths to try to show it's real science, at least mythology studies do not pretend it's scientific (I study economics at uni so I know what I'm speaking about)
You could argue that economics is a branch of epistemology.
Andres Karel That’s untrue
For what it's worth, those hearings did lead to the warning labels being mandated in 1967.
wait. the US government put human lives before profits of large corporations?
@@mikejohnstonbob935 According to Wikipedia, they were the first to do the mandated warning. That's really surprising :o
I like those transitions in your video! Very entertaining to watch for sure!
I'm glad to see Numberphile still making videos. A great story on correlation. I had never heard the one about storks before.
wow, not a single number came up but this is a great math story i gave ever heard
p=0.008 ?
No numbers? Challenge accepted!
0:48 "2"
1:15 "2"
1:20 "number"
2:04 "p=0.008"
3:14 "number"
3:25 "number"
3:41 a graph with numbers
3:57 "1954"
4:35 "numbers"
5:20 "1954"
5:29 "1st"
5:37 "2"
6:02 "20th"
6:19 "numbers"
6:36 "20th"
6:39 "1950s"
7:00 "10s of millions"
7:01 "100s of millions"
7:20 "pi"
7:20 "1960s"
@@magichands135 of course Pi always has to show up
@@magichands135 number is not a number, it's a word
@@mr.cauliflower3536 He said no number.
This went dark very quickly.
His integrity went up in a puff of smoke.
and increasingly relevant to the present day
I find it lovely that you got an economist to talk about this topic - anyone who has studied economics knows how integral econometrics i.e. regression analysis i.e. statistics is to the practice! Anything empirical in economics is always expressed through statistics.
I think it's important to understand exactly what the flaw was in Huff's argument that smoking was benign so that we're not using circular reasoning. Arguing against the significance of the correlation between storks and babies makes sense because we're able to identify a third variable that correlates and causally explains the other two. On the other hand, Huff never proposed alternative explanations for the correlation between smoking and lung cancer (as far as I'm aware).
Also, does Huff consider the correlations between land area and number of storks and between land area and number of babies as equally irrelevant?
It would be nice to see a follow-up video on just this topic
I am a fan of "How to Lie with Statistics" (and its sequel "How to Take a Chance: The Laws of Probability"), but I am a bit taken aback to hear of Huff's connection with the smoking lobby.
The problem that the general public has with statistics is that they don't really understand uncertainty or the assumptions that underlie the statistics. When they hear the results of a poll, they either lend it too much credibility or too little. When they hear that some candidate has a 70% chance of winning, they assume that means it's almost guaranteed, not that almost 1 out of 3 times they will lose. It's why it's so important how statisticians and social scientists communicate their findings and why using statistics to lie is so damaging.
I think the stork-baby correlation is flawed. Stork-baby correlation in Eurasia and Africa may be strong, but storks are much rarer in the Americas, Australia and island nations.
I propose a different theory: Babies come from cabbage patches.
I don't have the actual numbers, but I think the correlation will be stronger than the stork conjecture when all data points are available.
Babies deliver babies because babies have the strongest correlation with babies
Further evidence towards this hypothesis: The existence of cabbage patch kids toys.
Cook a cabbage and eat a baby?
@@fraukevinetabulow4064 I don't like cabbages in my babies.
No matter how rare they are, if there's a nonzero number of them there'd still be a correlation with the size of a geographic area.
OMG, it's Tim Harford, as heard on the BBC WS all the time. Excellent!!
Babies correlate with babies the strongest. Therefore babies deliver babies
Water tight logic.
It's sort of true, it just takes quite a long time.
I think Huff was misunderstood because the book was so popular. He wasn't saying you can't take statistics seriously, he was simply making fun of some terrible examples of their use.
Edit: my bad that's what happens when you comment before watching the whole video lol.
This is probably the most important numberphile video ever considerng the society we live right now.
Can you do a video about what statistical principles explain why the case for smoking causing lung cancer is greater than the case for storks causing babies despite the correlations being similar?
This makes perfect sense to me.
Statistics can give a big hint as to where to look, but you still need hard evidence to be sure.
We took the statistics and based on them looked into HOW smoking might cause cancer. We now know the mechanisms by which it occurs, so we know the statistics are useful and correct in this case.
"I don't know; I'm an economist."
Truer words
Problem is that people who aren't economists cherry-pick whatever data reflects their pre-existing opinions, and then act like they do know.
@@alexpotts6520 why're you excluding economists from that group of people?
Indeed, I've heard things about economists (on The Internet of all places, therefore they must be true!) - it's called "The Dismal Science" and not for nothing. It's (allegedly) the only field where two people can get the same top prize for writing papers that come to opposite conclusions.
@@TranscendentBen Fun fact: the phrase "the dismal science" was coined by a 19th-century slave owner who was grumpy at the growing consensus among economists to support abolition.
Economists aren't always wrong about everything...
This was a lovely watch full of interesting and important history I didn't know I needed to know but am definitely pleased to know now. The books mentioned seem like amazing reads so I'll definitely be checking them out! Thank you.
Aww man you built it up so nicely and then such a bummer at the end.
Now I don't know what not to believe it is so confusing :)))))))
I heard the word "statistics" about 100 times in this video and "correlation" only once, while this whole video is in fact about correlation.
At risk of sounding pedantic, I think he also pronounced "causal relationship" as "casual relationship", which I thought was a strange error for a professional to make
ok
@@turtlellamacow do you mean the bit about the senator asking whether the smoking/cancer relationship was as "casual" as the stork/babies relationship? I'm pretty sure that was meant to be casual and not causal. The stork/babies relationship is not causal. It is casual.
@@paulmealor9128 You may be right; I'm not certain it wasn't intentional. But describing a relationship as "casual" is very uncommon (outside of dating!), whereas misreading "causal" as "casual" is a common error among stats students.
@@turtlellamacow Casual is the correct and intended term for describing the smoking-cancer association as being "as casual as" the stork-baby association.
Casual can mean incidental or happening by chance or accidental or informal. It seemed to me the economist was quoting a senator asking the witness, "is the relationship between smoking and cancer as incidental/accidental/unrelated as that between babies and storks?"
Please don't expect a US senator from 60 years ago to be using "causal relationship" in a sentence. US senators then and now are not noted for their expertise in understanding little things like science and maths.
Hang on, I just watched Matt Parker interview Tim on his channel... I see he's doing the rounds on the maths yt channels to promote his new book ;)
Maths channels have officially become a Thing now.
This and the mega favorite number thing i think its great that theyre starting to collaborate to share as much knowledge with the world as possible (and pay the bills while theyre at it lol)
Me: Every time I sniff pepper I sneeze.
Tobacco Lobbyists: No relation between those two events because storks deliver babies.
"Do stonks deliver babies?"
The internet has corrupted me
I came here to say, "No, but stonks do." ...what is happening?
what is a "stonks" ??
@@DendrocnideMoroides it's a meme... best not to think about it lest you become one of youtube's hopeless social cripples who can communicate in nothing but memes
knolidj
??
Would be interesting to see, what exactly the differences between the statistics for storks/babies and smoking/lung cancer are. How do you make sure that you prove an actual causation between smoking and lung cancer, and not just a correlation?
I second this question, would love to see a follow-up video
Wow, I was thinking about getting the huff's book but at the end of the video I'm not really sure
Why not? You may learn some tricks and cheats about statistics. I want to get it too.
Huff died in 2001 so it’s not like you’d be giving him money for it.
You could always get it second hand or borrow it from a library. And it's an old, famous book, so you could probably find the pdf online for free!
Using my dad's copy of how to lie with statistics to prop up my small TV screen on my desk. Small world.
Go look up the title and pdf, it shows up
This was the first video I can remember here on Numberphile that felt unfinished.
These videos usually leave me wanting more, but they also usually feel complete.
Well, as is explained, # storks is actually positively correlated to # babys. The correlation is through the land area of the countries. But that does not mean 'storks deliver babys'. But after all, they do correlate with each other, which is what exactly the statistics suggest. Hence statistics do not lie, people may lie by misinterpreting the statistics.
Little did we know that from 0:00 to 2:34 that was all his elaborate pickup routine
I was exited to add a new book to my reading list only to remove it 3 minutes later.
A more or less excellent collaboration between two numberphiles. Sweet!
This video really proves that 'correlation doesn't imply causation'. Loved it!👍🏻
I just finished watching Tim on Stand Up Maths
I mean if storks don't deliver babies then what does? Other birds are either too small or are birds of prey and birds of prey are too scary.
Haast's Eagles. Delivered babies in the past. May have mistakenly eaten some of them
Maybe a swallow could carry it?
Ostriches, Cassowaries, Emus, Rheas, Gulls, Pelicans, Cranes, Swans?
@@imperialguardsman135 It probably depends on if it's unladen or if it's African or European.
@@Vaaaaadim Those either don't fly, are too small or are too few in population. And swans just don't count.
I thought the stork thing came from them being attracted to roofs and chimneys of warm homes. Homes with a new baby would keep fires going all night, so storks would hang out at those homes more.
I first learned of Tim Harford via Malcolm Gladwell promoting Tim's excellent podcast "Cautionary Tales". Now it's come full circle as he's on Numberphile and selling a book through StandupMaths!
"And his name? You may have seen it coming-"
"YOU'RE NOT A DISH, YOU'RE A MAN!"
One of the most annoying commercials on YT! Can't click off that fast enough.
@@NatePrawdzik I know, right? _All detergents are synthetic!_ Soap doesn't just grow on trees!
Although, there _is_ a strong correlation between the number of trees and the amount of soap...
I've been known to yell "oh f*ck off, I am TOO a dish!" at my phone.
I got an ad for Disney+ at that point.
I remember reading a book that showed that the increase in the sale of ice cream led to an increase in criminal activity. It was then revealed to be an example of using statistics to lie, and that the increase in crime generally happens in warmer weather, when people tend to buy ice cream more often.
I just love this channel
Correlation doesnt imply causation is something the every human being should know
"I'm not a student of mythology and traditions. Anyway, let's get back to the *important* stuff, statistics." As someone who loves both math and mythology, I'm hurt
Amazing! Never thought I'd be finish watching this video!
I remember one from Denmark in the early 70ies that showed a correlation between the number of TV antennas and the number of car accidents.
If anyone is curious about how you can tell whether a given correlation is a "storks and babies" situation or a "cigarettes and cancer" situation, you're looking at the field of causal inference. Effectively, it comes down to mapping out all the reasons why two variables might be related (e.g., smoking and cancer might be related because of who chooses to smoke, or the simultaneous rise of cars, or genetic differences, etc. Etc.), and using statistics to isolate only the reason you're interested in (does smoking cause cancer?). Any decent econometrics course will go into this, or for a more computer-science perspective check out Judea Pearl. I also have a series on causality on my channel.
Super amazing and hugely dark ending.....well done
The difference with the correlation between smoking and lung cancer is that the correlation held for numerous different ways of approaching the problem (and, also, eventually there was a mechanism by which smoking could cause lung cancer). It's like the evidence for evolution. Just one correlation does not mean much but, if you keep finding correlations when approaching the problem from many different directions (fossils, geology, biology, anatomy, genetics, antibiotic resistance etc), then eventually correlation can start to look like causation.
In the case of Storks and Babies, the "solution" is that there's a hidden correlation--land area--which has a causal relationship to the first two.
In the case of Cigarettes and Lung cancer, what ideas did Huff propose as hidden correlations? What hidden factor would cause someone to buy cigarettes and come down with lung cancer?
People who buy cigarettes drive cars more often than people who don't, which in turn gives them cancer
Easy!
Smoking often correlates with unhealthy lifestyle, drinking alcohol, not doing exercises, having stress, doing unhealthy labour, etc.
It could be caused by any of these unhealthy activities.
D. Huff was right about statistics proving nothing. We know smoking is bad because of medical research, also just by using common sense.
This is actually false argument vegans use, to 'prove' meat is unhealthy, by looking at patients with heart diseases (vegan/not vegan).
But what they fail to understand, is that vegans follow a healthier lifestyle then obese people with heart problems.
If all else is equal, you do all the things vegans do, and you eat healthy types of meat in moderate quantities - then you will be no less healthier then vegans are on average, and probably even healthier.
Using statistical wizardry to "prove" absurd theories
*storks*
stonks
Great to see the More or Less guy on this channel
The problem is that unless the public health authorities start doing a much better job than of late, treating their various ukases with a degree of scepticism is a _perfectly rational_ stance. It's also fair to note that statistics have been perverted by State actors as well as by industry, in exactly the same fashion and for exactly the same reasons. Healthy scepticism doesn't mean a blanket rejection of statistics from whatever source, but it does mean one should be aware of cognitive and other biases, including the all-important question: _cui bono_ ?
"Will they get a reply?" --- "Maybe!" Instant
I remember watching a football match where the commentators talked for 4 minutes about how the home team never conceded any goal, and they only stopped when the away team scored.
I really love the fact that the book 'naked statistics' has written 'Best selling author of naked statistics' on the cover!
There is, in fact, a correlation between birth rate and storks. That's because there are more storks in rural areas and there's usually more babies (relatively) in rural areas than in cities.
What is the airspeed velocity of an unladen stork?
According to a very scholarly and large book I have about the history of the English language, here is the origin on the stork story: Storks were originally called sticks, which makes sense given their long thin legs. The name changed over time to storks. But back then the word stick was also a slang term for, well, a certain part of the male anatomy. So adults were playing a joke on kids with this story of where babies came from.
I remember that there is a correlation between storks and birthrates within countries. Simply because storks live in more rural areas and people in rural areas have statistically more babies than people in urban areas where there are less storks.
*YES*
Next question please
So how do you decide if a correlation is significant? How did they know that the correlation they observed between smoking and lung cancer was not spurious? When the judge asked if storks and babies was the same as smoking and lung cancer, I don't like that it's assumed that the view would agree that they are different. Why are they different?
3.13 I would actually not be surprised if there was a correlation with birth rate and stork number.
But I would then think that was related to more socioeconomic stuff (Usually richer countries have a lower birth rate, but they might also have a more industrialized countryside, meaning there's less food for storks, thus less storks.)
Almost didn’t watch this one, really glad I did!
8:20 Albert Einstein!
No, wait...
When I was an undergrad, I took a course based on How to Lie with satistics, which was designed to teach statistical literacy.
"There are three kinds of lies: lies, damned lies, and statistics." - Popularized by Mark Twain who questionably attributed it to Benjamin Disraeli.
''Quotes in youtube comments are usually exactly by whom the commenter sais they are'' - Abraham Einstein
My favorite quote by my statistics professor (possibly he was quoting someone): it’s easy to lie with statistics. It’s easier to to lie without statistics. Lol
Statistics is a set of mathematical tools. A language. You can use words to tell lies, and you can use words to uncover and communicate the truth. Similarly, statistics have no agency - it's the person who tells you the story, and their method of telling it you should be wary of.
@@JackFlashTech Lying is easy regardless, but it's easier to convince people that a lie is true if you use statistics.
There are four kinds of lies: lies, damned lies, and statistics, false attributions and being unable to count to five. ;)
Is there also a significant correlation between number of trees and GDP? Because if there is, does that mean money grows on trees?
Midwives used to use a small pair of scissors during delivery called stork scissors to cut to umbilical cord.
I have subscribed to you long ago my dude, I 💖 Numbers
And here I thought this was going to be about how the saying that storks bring babies comes from more babies being born in the spring and summer when storks live here and less babies being born in the fall and winter when the storks are in Africa - a tendency that used to be much clearer back when people still had a work schedule dictated by the change of the seasons. "Storks bring the babies" was never meant to be taken literally - it was just a sign of changing seasons. And only after that did the stork become a symbol for fertility.
I like to think that there's an asterisk after "Best Selling Book About Statistics". Something like "According to random person we interviewed".
For a split second I read the title as "Do Storks Devour Babies?"
Yes STONKS
Best comment xD
So, do storks actually deliver babies or not?
You heard him! He asked you to email him!
Yesn't
I just took ap stats last year and got a 5, and this makes it look easy.
There is a correlation between ice cream sales and people drowning
Ice cream sales per what? People drowning… per what?
Fantastic Video, but missing an important detail: How to tell if the statistical data is worth using OR what to investigate to prove correlation.
Now do one on why statistics are important. And how most of the tech world runs on statistics.
I enjoyed Tim’s last book!
It's so weird seeing Tim, I've only heard him in podcasts until now...
Me too, didn't realize his voice until the very end
Last time I was this early, The Poincare Conjecture was still a conjecture.
Haha
I need your Phone number because of lotto
Shing-Tung Yau says the conjecture might still be a conjecture, since no one really understands the proof. I don't even understand enough to say who is right about that.
This answer only raises more questions. Where do storks get the babies? Do they bring babies for other species? Where do newborn storks come from?
The newborn storks come from eggs, of course))) and eggs are found in the cabbage! 😆
10/10 what a fabulous video.
"pi span" lolol
2:39 - Poland is the one on top. Turkey is the one on far right a little below.
Sounds great!
Now I know why the _stork exchange_ was a complete riot today.
_Storks went up!_
Wow! this can be made into a thriller movie!
We need more statistics!
We need fewer staticians working for liars.
It's not really "nasty cynicism" to question the validity of statistics. It's easy to feel that way with hindsight, but the fact remains that it was important *not just* to establish a correlation between smoking and lung cancer *but also* that the correlation was causal in nature, that the correlation was too strong to be dismissed.
This has clear parallels today with Covid-19, the subject of many bogus claims backed up by bogus statistics. It's not a valid argument against scepticism as an approach that in a particular case it suggests that something you believe strongly in may not be true - *that's the point*
Covid-19 statistics: the more you test, the more cases you get :)))
@@nowonmetube As the rate of testing goes up, so does the number of confirmed cases. So to stop covid, we must stop all testing, right? :)))
@@Johnnyb3g00d how did you know?
@@nowonmetube Well, in order to achieve this conclusion, the first step I needed to do was turn off my brain. It works wonders
Can someone explain how/if we can now tell the difference between the two correlations being linked in reality or just in statistics?