The danger of mixing up causality and correlation: Ionica Smeets at TEDxDelft
Вставка
- Опубліковано 4 лис 2012
- Ionica Smeets (@ionicasmeets) is joining TEDxDelft Never Grow Up: A mathematician and science journalist with plenty of media experience. Using her vast knowledge and enthusiasm, she can explain everything about her favorite topics in science and statistics. She does it well on paper and face-to-face: She writes blogs, columns and books and is also asked to appear as a speaker, live, on television and on radio shows.
Since 2006, Ionica has taken on the Internet with interesting and fun mathematics together with PhD Partner in Crime Jeanine Daems on the website wiskundemeisjes.nl. She and Jeanine now write a bi-weekly column in the Volkskrant about mathematics and the website also resulted in a book titled 'I Was Never Good At Math' (Ik was altijd heel slecht in wiskunde) in 2011.
Ionica appears on De Wereld Draait Door to talk about mathematics; trying to explain the most complicated things and developments in the field of mathematics to the host Matthijs van Nieuwkerk and the audience. Even for the most mathematically challenged, it's educative and entertaining to listen to. In 2012 she became a reporter for KRO in a series called 'De Rekenkamer' investigating the financial aspects of giving blood and illegal minor immigrants.
Together with Bas Haring she wrote 'Vallende Kwartjes' which are essays in which scientific processes and concepts are explained with straight-forward, easy-to-get examples and stories. The science journalist has an eye for this sort of thing; defying and destroying the idea that science is boring and/or cannot be explained well.
In collaboration with Govert Schilling (amongst others) she also makes UA-cam videos for the channel Science 101 (Wetenschap 101) where scientific and mathematic concepts are explained clearly in under 101 seconds!
She's charming, adds good natured comments and jokes to her writing and her live appearances (when Matthijs van Nieuwkerk asked her how she celebrated Pi Day, she answered 'by drinking Pina Colada's). What more could be asked for in a TEDxDelft performer?
In the spirit of ideas worth spreading, TEDx is a program of local, self-organized events that bring people together to share a TED-like experience. At a TEDx event, TEDTalks video and live speakers combine to spark deep discussion and connection in a small group. These local, self-organized events are branded TEDx, where x = independently organized TED event. The TED Conference provides general guidance for the TEDx program, but individual TEDx events are self-organized.* (*Subject to certain rules and regulations)
The way she is presenting this stuff is so adorable and amazing.
Simp lol
@@jakevancecastaneda3748 How in any way shape or form is this simping, wow, someone commented positively on the actions of a female, automatic simp, honestly, this generation's common sense is broken, and I know bc I used to think like that, but a classmate of mine made a similar reaction, and the rest of the class was spent talking about what defines a simp bc as it happens, my teacher was female, so we all learned.
@@bryandenis2549 relax warrior
This was great, full of humor but well explained.
Her shirt is so adorable, aw.
Such a lovely presenter and delivery of an otherwise (when stripped down to its bones) boring and dry topic =))
Clearly people need to stop eating ice cream while they're swimming.
Some people still don’t get the message, don’t they.
Clue, it’s the Sun, not the ice cream and not the swimming.
@@gjhardy Well, I think you're the one who misunderstood, bro.
Luckily I’m lactose intolerant 😅
Such a concise presentation of a very subtle concept. "Correlation is not causation" but that is not how our brains, which evolved as pattern recognition machines work.
A 4 year old heard a very loud scary bang at the exact moment a rabbit appeared. He had an irrational fear of rabbits the rest of his life.
Very amazing the way she presents but also highlights the importance of the relationship of causality and correlation. Not every time to conclude on that
Great talk.
Thanks to correlation we have been teaching entrepreneurship all wrong for years.
We seem to think it is a linear planning process were in fact it is an organic process of discovery. Now thanks to Saras Sarasvathy and Nobel price winner Herbert Simon we know better... Effectuation describes how expert entrepreneurs think decide and act.
This is true with an asterisk. While dangerous to short circuit from correlation to causation (as journalists often do), in Six Sigma analyses we use the set of correlations from which to begin our search for causative relationships. So correlations should not be summarily dismissed.
Good point 🙂
Correlation does NOT NECESSARILY imply causation... but it's a good place to start!
We need to ban the weather to stop drowning
it's because of you i understood the difference corelation and casual .thank u very much
very easy to understand this better and grab my attention completely with the way she presented this :)
Very interesting talk!
I'll never eat ice cream again
have mine too, you fatty :-D
Haha, just to be on the safe side !
The correct conclusion is this: warm weather decreases one's ability to swim.
Loved! Thank you!
Excellent!
very insightful talk
excellent explanation
There is actually proof of bright or even dim light interfering with the circadian rhythm or sleep cycle.
not them becoming short sighted tho
It's funny as I see the causality vs. correlation mistake in some TEDTalks - there's one I recall about food sensitivities by Robyn O'Brien which shows a few parallel plots and jumps to conclusions.
I love eating tomatoes, but I have read that people who were born in 1860 and ate tomatoes during their lifetimes have a mortality rate of 100 per cent. :-O Are you sure there's no causality? ;-)
Thanks, I never knew that! Glad i wasn't born in that period of time :)
Tomatoes were clearly far more toxic then.
Daniel Roy If this is the claim, I believe the main point is missed here. She is saying that even if you see there is direct correlation, you still have to prove "why it is happening?" and "how it is happening?" to show causality (basically cause and effect relationship) which is the essence of understanding the underlying principles of a mechanism (can be a sociological, biological, dynamic, socioeconomic, .....etc.). In other words you have to prove it.
Yeah, I know. I just meant to underline what she said.
Of course we can always research causality but that is not the purpose of correlational research. It is two variables that rise together, fall together, or have no relationship at all. There is a positive correlation between baby boomers and hepatitis but that does not mean that being a baby boomer will cause you to have hepatitis or the having hepatitis means you are a baby boomer. It is only a relationship. Sure there is a cause of each, but that is a different type of research. Mortality rate of those that were born in 1860's and ate tomatoes, is only a relationship. You can not say, confidently, that because any one born in 1860 and ate tomatoes, will have a high mortality rate. Again relationship, not cause.
wetenschap is dat voor te weten x
BAM!!! Thank you!
Good stuff!
I read a study where they tested self-esteem in third and sixth grade. High self-esteem in third grade was not a predictor for good grades in sixth grade. But good grades in third grade predicted higher self-esteem in sixth grade. So from this it seemed fair to conclude that good grades cause self-esteem and not the other way around.
Thank you
Elaboration 💯
Amazing
Thank you very much, I didn't know I needed this for my thesis. Thank you once again.
What thesis?
she is great
I was hoping she would go into what actually constitutes causality. I'm sure we're all aware of badly-written newspaper articles claiming causality for research that only shows correlation, but what would actually prove causality?
lambd01d I think, nothing can prove causality of apparent causes. No basis other than mere intuition. On the contrary, we can think of "evidences" which disprove causality. A centuries long discussion though!!!
Yusuf Osmanlioglu Actually, you demonstrate causality through well-crafted experiments based on theory. You can learn more about demonstrating causality by taking some courses on research methods.
lambd01d To illustrate causality you have to isolate & test the variables independently., In other words you keep everything the same except for what you are testing.
Examples;
If you want to see if a plant food works you need a group of plants that get the plant food & a group that doesn't and you need to keep all other variables the same. They need to be in the same soil, they need to get the same amount of light & water. Both groups need to be treated exactly the same in every way except for the plant food. Then, if the group that gets the plant food grows better you have shown causation.
With things like the ice cream sales & drownings it gets a bit more complex.
If you wanted to illustrate ice cream causing drownings in city A you would have to look at the base rate of drownings for a year & then ban ice cream sales & see if there are fewer drownings.
What you are essentially doing is comparing city A (with ice cream)
to city A (without ice cream)
We are trying to keep all other variables the same as much as possible.
Obviously, with an entire city or a country there are other factors that you just can't control. What if there was a cold snap in the weather? What if they hired more lifeguards? Those things could reduce drownings & possibly prompt an incorrect conclusion.
That's why we may study the same thing several times.
"Prove" Is a very strong word & we could debate if anything was ever proven but, if you control for variables properly, you can offer strong evidence for things.
Bradford Hill criteria for causation:
1. Strength of Association - the magnitude of the association.
2. Consistency - relationship can be demonstrated on many occasions and in different
situations (e.g. countries, times).
3. Specificity - a 1-to-1 relationship between cause and effect; a factor leads to one
disease, or disease is due to one factor. Must be acknowledged, however, that
most exposures have multiple effects and some diseases have multiple causes.
4. Appropriate time relationships (TEMPORALITY) - a proposed cause MUST precede an effect (exposure
before outcome), and the latent period between the earliest exposure and manifestation
should be appropriate.
5. Biological gradient (DOSE RESPONSE) - causality is further enhanced if the risk of disease
increases as the dose of exposure increases. The absence of a dose-response
relationship is often taken as evidence against causality; however, the
presence of a dose response relationship may be caused by confounding.
6. Biological plausibility - existence of a physiological mechanism by which a
proposed cause exerts its effect.
7. Experiment - natural experiments occur or semi-experimental evidence is available e.g.
Nuclear plant exposes workers to radiation.
8. Analogy - if similar
situations or findings are present regarding related chemicals, then the case
for causation is much stronger.
9. Coherence of the evidence - summary of all other criteria for causation and its
integration into a view as to whether a true association exists and, if so, its
magnitude and what should be done about it. Final decision is always an
informed judgement.
IF !A THEN !B. This is Causality.
So if happen A without B, or if happen B without A, then, there is no causality, just correlation. But in the Universe nothing seems to be so pure. And then we need to try to isolate the variables. We start to look less to 0% and 100% and start to look more in between. The earlier comment looks like a pretty good start.
Now you know the definition. But the way is eternal! Good luck!
This topic is very important 👍🏻
awesome!
Prof. Saeb's class sent me here.
Can anyone tell me what this song is at the beginning of this video?
I would like to know how Ionica Smeets defines "causality" if not by correlation.
Great examples of how correlation does not mean causation.
Well many times it does, but not ALL the time. It's interesting because Tedx will always use a disproven correlation/causation argument when it comes to legalizing drugs, or pushing the evolution theory.
so in other words, you have to be knowledgeable in the physical domain in order to interpret the data/stats
Correlation and cause and relation are two different things...
Her tone inflection was a little difficult but because I could rewind I got all the important points. Good presentation, but she must have been nervous :)
I had exceptional grades in high school but it does not reflect my self confidence.
tho causation is correlated to the principle of sufficient reason transcendentally, (speculations and unproved philosophems on a side), namely the correlate of freedom (empiric aka practical reason) is the Will (phenomenological)only as an abstraction or a subject-object representation aka relation etc.
I want that shirt she's wearing
can someone cite this for me?
Luvely
4:02
GRIFFITHHHHH!!!!
Too bad more people don’t use this simple logic when it comes to the premise of descent with modification.
So obviously we need to ban good weather.
"Look mario, its a question blotch"
WHAY ARE YO SAYD MARIO?
Correlation does not imply Causation
The phrase should be "Correlation not ALWAYS imply causation", because there are times when it does imply causation.
what if i eat icecream in the desert?
Then it would be dessert.
crisp
is that so wow some people would nt and couldnt eat foods or stop its to hard my taste buds and still hungry and ice cream thats been my favorite since i was 9 years old. i barely gave up or stoped eating candy or choclate like edible candy everything ing the world being edible to me any way
Do you English?
GRIFFFFFITHHH!"
Probleem is dat Ionica geen echte wetenschapper is maar een fantast.
Journalists should wach this before wrinting articles🤣🤣🤣 as they use to turn correlations to facts by habit.
Yes---one needs to think about our snap conclusion about things...she is very charming ......but the marriage example is flawed.
We would need to analyze the data based on one income level. Ex rich men who do not marry and those who don't. Poor men that marry/vs those that don't.
lol you shouldn't ban ANYTHING that is dangerous cars are the most dangerous thing in the world!
i just. got a pack of medium nails i am working on in the next few and today. and
wat
Lethal icecream
This video made me feel depressed
Why? Because of the married men segment?
Maybe it is the reason of superstitions. People derive moronic causation from a correlation.
الدحيح
This lady is talking about people drowning to death with a smile.
5:24 "female bankers and financial crisis" 🤣
The financial crisis I thought were caused by men - oh wait correlation but no causation? Thank you doctor!
Stream ice cream
Why she sound like Stephen Hawking ?
Man made global warming could be another example!
Wrong on "life expectancy"
A Married man takes better care of his health (because his wife insists on it) and therefore he lives longer.
Her accent is hard I cannot understand what she is saying
Agreed, she has an accent, but I doubt this is an issue here, and it certainly does not detract from the quality of the talk.
I presume you're being too hard on her either because of an accent based bias or because you're not proficient in English. This is not my native language and I was able to understand her thoroughly.
I was able to understand everything she said.
Awesome!! Now do Climate Change…
So there's no correlation between smoking and lungs that leads to the causation of lung cancer?
Correlation simply means the co-relation of two or more subjects. Conditions simply don't 'happen'. That's like saying the Principle of Cause and Effect isn't true.
If somebody stabs my leg with a knife and I yell in pain and bleed - then the CAUSE stems from the relationship between the knife and my torn flesh. I'm not going to sit down and say "Well gee, I want to say that this excruciating pain is the result of a knife tearing into my leg. But correlation is not causation. So I can't prove that this is the case."
Why the hell do people not understand the obvious?
Kiko Joseph There is a correlation between smoking and lung cancer. The causation comes from the study of the various mechanisms by which this is possible. The example you provide is a silly one, because in that case the cause is blatantly obvious because you have reliable evidence that the knife caused it (e.g. you saw or felt the knife stab you.) In other cases, like the ones mentioned in this talk, the situations are much more complicated and as a result the real cause may be less intuitive for the interpreter. This is why mistakes are made... so the answer to your last question is that it is not always obvious.
smoke is like tiny knives poking at your lungs
In the knife scenario You would know why you're yelling, it would be quite clear to You but to an outsider it should technically be difficult to tell. It could happen that something else is causing you the pain and it coincidentally occurred at the same time you were being stabbed and also happens to be of greater pain magnitude, hence you're yelling. Without your acknowledgement, I can only go on correlation, I'll know the cause once you tell me.
The phrase should be *"Correlation does not ALWAYS equal causation but there are many times it does"* It's interesting that Ted talks will draw causation ONLY when it suits their libertarian left-wing agenda such as evolution, legalizing drugs, and legalizing prostitution.
From your many, give me your top 3 examples
Infering causation from correlation is a logical fallacy.
It never even implies causation
Let's talk, what would be the consequences of legalizing drugs?
What would legalizing look like?
Except for the snarky comment about vaccines it was a pretty interesting video...not pro or anti, just neutral on the topic, but there is also no proof that it DOESN'T or CAN'T cause autism...
This was 10 years ago
Ok Mojica.