Simpson's Paradox
Vložit
- čas přidán 28. 08. 2024
- Thanks to brilliant.org/... for sponsoring this video!
This video is about Simpson's paradox, a statistical paradox and ecological fallacy where seemingly contradictory results are implied by a single set of data depending on how it's grouped. The paradox can arise in medical studies, student test scores, and so on.
Support MinutePhysics on Patreon! / minutephysics
Link to Patreon Supporters: www.minutephysi...
REFERENCES:
Interactive Simpson’s Paradox Explainer:
vudlab.com/simp...
Wisconsin vs Texas schools: educationnext.o...
Detailed analysis of Wisconsin vs Texas schools: iowahawk.typepa...
National US Standardized Test Data: nces.ed.gov/na...
MinutePhysics is on twitter - @minutephysics
And facebook - / minutephysics
And Google+ (does anyone use this any more?) - bit.ly/qzEwc6
Minute Physics provides an energetic and entertaining view of old and new problems in physics -- all in a minute!
Created by Henry Reich
Looks at Texas flag... "Yeah, I can draw that."
Looks at Wisconsin flag... "Umm... I'm just gonna draw their flag as a wedge of cheese on a blue field."
REALITY... FOR US... IS PROBABILISTIC... DETERMINISM IS NOT FOR US.
Wait, that's not the Wisconsin flag?
I think it says a lot about the US when it didn't even occur to me that it wasn't Wisconsin's real flag.
if you google Wisconsin flag, the picture of cheese with blue background actually came up on the fifth row of google images
I am glad I saw this comment because (ignorant of the fact that American states do eveen have their own flags) I thought it was Wisconsin flag:)
“more money makes you a cat”
the true American dream
technically stolen comment
Aristocat!
Wait, is that why americans are CATholics?
the Furries have entered the chat
A fat cat
If a cat loses enough money will it statistically turn into a human?
Yes.
geryon Yes, but I doubt they have enough money to loose to begin with, so I doubt they'll ever turn into humans. But, this makes me wonder why cats exist in the first place... Were they all humans at some point of time and lost all their money and got really poor and eventually turned into cats? After all those extremely poor homeless people on the streets do behave like cats to some extent already, living in trash, eating from trash cans... Omg! It's very likely they were humans before! Jeez!
according to our GOP's understanding of statistics, of course.
Not realy human, more like turning into "The wolf of wall street"
LOLOLOL
This was recommended to me because I've been watching clips from The Simpsons. Thanks CZcams
I don't believe you.
No one looks up clips from the Simpsons on YT.
@@russellfautheree4650 Hello, I am No one. It's nice to meet you :P
Saaaaame
Yeah I thought it would have something to do with the Simpson's!
Russell Fautheree I actually watched episodes
If you put cats in boxes, you don't gotta worry about treating them. Problem solved. Schrodinger was way ahead of his time.
Paradox of paradoxes
just make sure the box's coin flip doesn't make the cats all die anyway
If you put a cat in a box with a poisoned pill, there's a very good chance that the cat won't take the pill.
Put a dog in a box with a poisoned pill, and you've got a dead dog.
@@geraldfrost4710 I didn't need to know this
@@ozmel146 what if you’re in a situation where someone forces you to pick between a cat and a dog of which you have to put in a box with a poisoned pill
I had a math teacher that told one once:
Statistics are very important to learn well in order to know they are useless. (His point was that if you dont understand statistics, you can be fooled more easily).
Thus he assumed you were too stupid to understand statistics
Lies, damned lies, and statistics.
also
Numbers, when tortured, will confess to anything.
@@kraftrichter1599 He was my teacher, not the other way around.
Of course he assumed I was stupid.
Statistics are only useless when no detailed methods and assumptions are supplied and the results aren’t properly reported.
@@KP3droflxp That is a beautiful theory, just like communism: It does not work in practice.
Name me one statistical study that you know where these problems are not present.
"being a cat makes you happier"
so true
cant argue with that
The Commenter have you seen a cat? They have their own towers, they get fed every day, and they get to sleep whenever they like
Eat and sleep
More money makes you a cat
i thought this was a video on how the simpsons predicted the future or something but that’s how little i know about the word paradox
Paradox is when something that cannot be explained by the universe's physics. For example: If you time travel to the past and kill your father, then you wouldnt be born to time travel and kill your father. That's a paradox
ElTech0 XD cheers
Tristan Overholser I thought so too man
@@vrtris A paradox is just something that doesn't make sense
Paradox is also an unexpected answer to something.
"You can misinterpret this graph as saying more money makes you a cat"
I _died_
rip
@@Optimusskyler the ghost of this user appreciates your paying respects
So that’s why furries are rich
well yeah, do you know how much fursuits cost? it's ridiculous!
@@flameindigo8035 i once purchased vegan leather shoes, for 100 bucks, the moment i left the store i almost fell. The ground was wet and i literally fell at least once. But i donot want to make animals suffer for my convinience. Now i am in the Simpsons paradigm? I am so confused.😂
"more saddness makes you richer"
oh boy, where are my Infinite Euros
finally my depression would be useful
Gotta appreciate that Ramiel pfp
you are not sad enough. you need more desparation in your life.
Best angel
Euros 😂
More money makes you a cat and you can’t change my mind
Steven Crowder in a nutshell
I can if I give you money. Then you'll have a cat brain.
Also: Checkmate, atheists. Evolution's not real. God just creates different species by paying them. God is loaded.
I have no money. I am not a cat. Ipso facto, hic ergo sum, transit sic gloria.
@@jeffallen55 I didn't know Gloria was sick!
Suspiciously wealthy furries 🤔
"if your a cat, you have more money and more happiness"
well, that explains catgirls
Your favorite Haunter :nya intensifies:
Furries
Your favorite Haunter *you're
@@overlordcringe2715 so what
@@Pizzanator-gp2bb q
TLDW;
Having money makes you a cat
Alejo Storni somehow
It's science!
Whatever weirdo makes these videos, he needs to take a basic English Literature 101 class so he can come up with less idiotic analogies. The point of an analogy like the cat-money thing is to explain a complicated abstract concept with a much more simple abstract concept. He did the reverse. He tried to explain a complicated abstract concept with a ludicrous and incomprehensible abstract concept.
Bot having money makes you a black cat
Triumvirate888 Or maybe you just didnt get it. I think it was fairly simple. But then again, so are you.
More doughnuts makes Homer happy, which makes him fatter, which makes him sad, which makes him eat more doughnuts. Poor Homer is only poor and fat, neither happy nor sad.
I thought it was the fact that even though Bart is only ten years old he's been bullied by Nelson for nearly 28 years.
no tv and beer make homer go crazy
Now I want donuts
Doh!
No TV no beer make Homer, something something...
"...and cats are both poorer and happier than people..."
Yup...
Yup.
Not wrong
3:14 what that actually works
my uncle once got like reaaaaally rich and he now is a cat so the implication is true
See what u did there
he is now*
also that's called a "gootraxian"
"More money makes you a cat." This is the best thing. Period.
Ok ... but where is Homer?
Springfield
When it comes to statistics, sample size is everything. If they used the same sample size for all 4 groups, it would clear up the issue.
Small sample sizes are easy to manipulate, larger sample sizes are quite a bit harder.
Nope, this has nothing to do with sample size difference. You can get the paradox with enormous samples of millions as well. It's about categorizing data and he even showed multiple cases with the same sample size yet the paradox applies.
Yeah what the hell, he said "aggregate the data" and then proceeded to bin together 1 treated cat and 4 treated people against 4 control cats and 1 person, like wtf u cant do that
@@sairentov
1. You can, you can aggregate data in various ways depending on what you're studying
2. You rarely get perfectly split statistics, so it happens all the time
It has everything to do with sample size. If the sample sizes were equal, the paradox wouldn’t occur. But in the real world when you collect aggregate data you might not know what the different categories are or if they are relevant.
Forfty percent of people don't believe in this paradox because they wanted to see something about the TV Show Simpsons, Kent.
40% ur statistics ate misleading though
Forfty
Simpson's Paradox isn't about how Homer manages to be alive after all that he has inflicted upon himself?
I tough that was a video about the simpsons
Maaaaarge
Remember the clones of himself from the halloween episode? Oh wait they don't have the belly button nvm
i thought the simpsons paradox was how the show can be on for 29 years and yet bart is still only 10.
I don't really see how that's a true paradox... maybe he's just super tough.
"More money makes you a cat"
Man, I wish I was rich right meow!
[Edit]
Wow over 600 upvotes? It wasn't even that funny or well-thought out. I didn't even get my capitalization correct!?
well, thank you, people. It still feels good to be one of the top commenters.
Someone won the lottery
Lol. I didn’t see that coming. You made my day better. Thank you.
The term fat cats makes much more sense with this in mind
mrrrow? Nyot to rain on nyur parade, but I may be turning into a cat, yet I have nyo wealth to speak of.
Meow...
Err...
Fucking furries.
I choose to believe that more money turns you into a cat.
The first example is actually pretty easy to solve or “correct”
The cats treated should be counted as 4 out of 4 instead of 1 out of 1
The people that are not treated should be counted as 4 out of 4 instead of 1 out of 1
Here we are calculating probability and death rates so we should calculate them as fractions or decimals instead of plain numbers
Yeah, it seemed so stupid to me; I don't see why he didn't simply give the same explanation as you!
or... maybe you just treat the people and not the cats both win and we good
Thats a simple explanation for a problem that can be much more complicated in real-world statistics. He used cats and people to differentiate the groups. Now try it with men and women, asian and white European, people with blue or brown eyes, or even more subtle distinctions in the population that might be FAR from obvious (people who eat nuts versus people who don't eat nuts - unless you're actually looking for it, you might not even know which question to ask of the sample). He used a crude example to make the point - sometimes its almost impossible to know what differentiates two populations in the sample, so its easy to reach misleading conclusions without knowing the correction (interpretation) is necessary.
why would you count 4 cats if you only observed 1 cat. that would totally screw with confidence interval
@@benstallone6784 I interpreted it as having a control group. But I don't know what he truly meant by it.
How did you determine that the socio-economic advantages are the reason for the test score disparity?
School with better funding are always performing better at IQ test or Test score.
Private school have the highest average, Public school with richer residents have better results compared to school in areas where poor people lived.
Why are the poorer white students in the bad Wisconsin schools are still doing way better than the black students in the better Texas schools?
Hoerkelis outliers. Everything above applies in general
There is no data showing what proportion of those white students are poor. Standardized tests don't keep track of socioeconomic status. All we see is that scores of white students are higher overall than scores of black students overall. This is attributed to differences in the distribution of socio-economic status by ethnicity.
Here's a study by the NIH (yeah, I know, it's old, but you can find comparable modern ones):
www.ncbi.nlm.nih.gov/books/NBK25526/figure/a2000af96mmm00042/?report=objectonly
Essentially, race can be used as a better consistent marker for test scores because the wealth distributions have a strong correlation with race, which mostly counters the demographic differences between Wisconsin and Texas.
Ned Fred: Actually it seams to be the exact opposite, if you would account for the socio-economic situations, black people would be doing worse then they are now: " After controlling for family income and parents’ education, race uniquely accounts for a substantial share of the variance in SAT scores."
cshe.berkeley.edu/sites/default/files/publications/rops.cshe_.10.15.geiser.racesat.10.26.2015.pdf
"More money makes you a cat" -minutephysics 2017
that may be what goes on this youtube channel's tombstone after this fiasco lol
Thank You for this succinct explanation. This is kind of what has been plaguing us during Pandemic times where people, without breaking down the statistics are reacting to Vaccine data and using arguments against it. This helps us arm ourselves to look beyond the face-value of numbers and diver deeper into all of the confounding factors that make them up. Cheers.
"using arguments against it" - No. They're *defending* themselves with any logical evidence they can think of because they don't want to be personally forced or coerced into a medical procedure.
"this helps us arm ourselves" - Another artifact of an *offensive* mentality.
You've been engaged with an "us and them" mentality and instilled with a vaguely threatened sense of your own mortality. Any fan of history knows what comes next.
@@glitchdigger Gotta love your absolute confidence in psychoanalyzing a complete stranger on the internet based on a couple words. Get real, fool. You're looking for things that aren't there.
bruh 💀
Cov. Vaccines weaken your immunity, thanks, statistics.
Last week I went to a lecture that talked about how statistical data can be misleading in how if 0.1% of vaccinated people die of COVID and 1% of unvaccinated people die of COVID but 95% of people are vaccinated then it'll look like 60% of COVID deaths are from people who are vaccinated and will then be misinterpreted to mean that the vaccine is harmful when it isnt
Why do I find this just now?
I had to spend the entirety of 2018 without knowing that more money makes you a cat.
A really cool video about a really interesting topic 👍
*gets halfway through the video*
RIP comments section.
Most comments I'm seeing are actually about being a cat instead of the political bit in there. Actually pretty cool tbh lol
There's nothing political in there though.
It's just a fact.
+Hakasedess Yeah, but there are some nasty interpretation of those facts that tend to creep up on youtube. Seems this audience is better than that though.
Except that being financially poor and uneducated is directly implied to be caused by a person's race.... as opposed to the individual choices made by individual members of said race.
There are too many poor and stupid members of any race you choose to consider this factual. Just as there are too many intelligent and well off individuals of each race.
If you grow up uneducated and poor in America, then your parents suck and are to blame.
If you stay poor and uneducated as an adult in America then it is you that suck and are to blame for your troubles.
Well, yes, racists do tend to latch onto it and make it political, that much is true.
I think the key is to refuse their racist ideas outright, and with that comes a refusal to acknowledge it as a political issue.
That's how I feel on the issue anyway.
(Robert Seeley is a good example, how convenient for him to drop by.)
Part 2 Simpson's Paradox: "Comments disabled because the discourse failed to remain civil."
I really appreciate for your work and as a long term fan, I'm extremely disappointed with the way you handled that.
#triggered
I was disappointed as well. I expect this from junk channels who are intellectually dishonest. It really surprised me to see this. That the comments were disabled in the first 24 hours leads me to suspect there is disagreement within the subscriber base, from whom my experience is that discussion is very civil, even when it gets heated. I came there after the comments were deleted, but if the part2 comments here are anything to go by, disabling comments is a move I did not expect from MinutePhysics.
I can do nothing more than wholeheartedly agree with aaron4820.
Why not address actual arguments instead of silencing any dissenting voices by dismissing them. If you want the moral high ground, act like it.
I think the part two completely forgot biological average differences between men and women.
For example on average men are more interested in things and women are more interested in people. This might be the reason why women don't choose engineering, even if there was no discrimination.
It wasn't forgotten, it was the elephant in the room everyone was pointing out, resulting in the "uncivil" discussion that ended up getting the comment section closed.
"Comments disabled because the discourse failed to remain civil."
"I censored you because your response didn't fit my narrative."
Thats not a paradox, I prefer to call it bad experimental design
If the data actually came from an interventional study or something then yeah, it would be indicative of bad design. But the paradox can also apply to data from other sources as well.
I think that it is less bad experimental design and more bad statistical analysis. After all, the data can still be good, but if you use it incorrectly or leave out some details, then you get a problem. Of course, bad experimental design can influence what data you get in the first place so it's not completely innocent either.
So, there are true paradoxes which poke holes in the universes and then there are apparent paradoxes which seem like they poke holes in the universe until you take a closer look. People rarely talk about the former.
As general rule in biostatistics, you should not include in your pool of subjects different species because even in the same specie you have a lot of variability. And it is a bad experimental design because part of the experimental design consist in selecting your group subjects.
@@DrKappaDelta sometimes it can be helpful to look at multiple species in an experiment if you want to know how something affects different species, all you need to do is to keep the data separate between the species when analyzing the data.
"More money makes cats sadder"
Citation needed.
Context matters.
There, I just summarised the whole video for you.
"after careful analysis the statistics tell us we're biased and even hint at where those biases are or aren't coming into play know the paradox is that we've remained so reluctant to fight our biases even when they're put in plain sight."
So after explaining how there can seem to be bias but in fact be down to personal choice of the individual (what college course they chose) your conclusion is "we've remained so reluctant to fight our biases even when they're put in plain sight."
Autism Is Unstoppable
You seem to be blaming individual choices while refusing to look for the causes for such choices.
People have free will. There are always going to be more people of a specific group doing a specific thing because people are not the same. For example most nurses are women but most IT engineers are men, this is likely because women have more caring tendencies (because of children) and men have a more logic focused brain.
So that deparments that women prefer are undefunded is someone not a institutional bias?
3:14 you could also make the mistake of assuming a gang of cats is about to fight a gang of humans on a street.
I CANT STOP LAUGHING
I love that you include cats in your example! making it more fun and easier to understand
maybe im just stupid, and i am, I assure you. BUT! it seems like these paradoxs only exists because the data is being incorrectly displayed.
None of the data were "stated incorrectly", it's just that you need context and human input for the data to make sense.
I dont know, the one about being richer would turn you into a cat seems like it was stated incorrectly. The proper way would be to have two lines, one for humans and one for cats. Lumping groups together that have differing variables seems like in it self a fault.
I see it as the three ring magic trick, It only seems like magic if you cover up the break in the rings with your hand and not show it to people. Its no more magic than this is a paradox just because you choose to hide information that would prove otherwise.
The problem is that in most cases subsets are not as rigidly defined and easily identifiable as these examples, and the issue of how to group/ divide them itself injects a subjective aspect into the data. The decision about which variables can or cannot be 'lumped together' is one of the issues (along with identifying potential causal effects from mere correlative data) this paradox is based upon.
Unfortunately the nature of statistics makes it more and more useless smaller the groups under study are; but fortunately breaks in resulting data lines can show where previously unidentified significant variables occur. This is why science isn't just a one-study-and-we're-done project.
@@BoojumFed is exactly correct. The issue is actually even more generic than this because the number of groups overall can be manipulated to tell very different stories.
It gives political pundits the option of finding the story that they want to tell buried in the data.
The gerrymandering of statistics
Wow! That is actually really accurate.
Simpson's paradox: beer is both the cause of, and solution to, all of life's problems.
I got recommended Simpsons videos for watching minutephysics videos before this was recommended
The first one isn't a paradox, it's just that you changed the amount of check marks when switching between the untreated and treated, all the boxes should have 4 check marks
Exactly my thoughts
@@pragmaticduck1772 agree
"More money makes you a cat."
...I have now found my purpose in life.
Minutep- talks about the point of the video
VSauce-first 2 minutes/talks about the point of the video
Rest 18 minutes/talks about assassination of JFK
I think the example is wrong, you joined 100% cats alive as 1 and 25% humans as 1 out of 4, and that's not how maths works, you should represent 100% cats as 4 out of 4, getting as a result that 5/8 of the pacients are alive with treatment and only 3/8 if not treated.
That's the point. The danger lies where you don't know the distinction between "cats" and "humans" exists, so you lump them all together and end up with "Treatment is Bad."
@@heftig0 Then it's not a paradox, right? Maybe I'm just stupid and that's what the whole video was about
@@ide8876 The "paradox" in this problem isn't a classical paradox like the grandfather paradox. This type of paradox, I think so at least, would be a "Falsidical Paradox" while something like the grandfather paradox would be classified as a antinomy paradox.
Basically, there are multiple definitions to paradox and a paradox isn't always something that can't be solved. Just a problem that looks impossible to solve. The paradox here is that the information contradicts itself when viewed from different angles, IE looks impossible to get a proper conclusion.
If you want to know more about paradoxes, I believe one of the Vsauce channels has a video on them.
Monthphysics
Hello fellow planet.
The Earth wrong, this video is not a month long, it's only minutes long
yes but there is only one physics a month
The Earth Fresh air whats next? The rain?
i wish these vids were a month long
"socioeconomically advantaged" - minute buzzfeed, 2017
That's a very commonly used term and it's perfectly valid here. What does it have to do with buzzfeed?
Runstar Homer he's saying it's SJW
Ryan Leimgruber call it what you want, it is literally true that blacks and other minorities have been, and continue to be, sociologically disadvantaged. It's a fact.
Runstar Homer I'm not saying I agree with him I was just answering the other commenters question.
Voltage Drop he disabled comments on the second part of this video. XD
So... who else saw the thumbnail and the title thought that this was going to be about how The Simpsons make so many episodes with unique events that eventually some of theme come to be real?
0:51 ok but the top right and bottom left ✓'s and X's are representing 25% of the population, while the top left and bottom right represent 100%.
So really, shouldn't it be 5/8 survival rate with the medicine, and 3/8 without the medicine?
I was going to say exactly this. So the medicine is effective.
2:12 The comment section is going to be fun.
Or scary!
They see me rollin' you got that far?
the comment section is hatin
To be fair, blacks in Texas are smarter cause, texas.
yeah i didnt know the flag was cheese.
Be intellectually honest and activate the comments on the video in which people were rightly pointing out how you simply assumed things regarding gender differences (them being purely social in nature) in order to say that there must be a problem on society overall despite the study showing no sign of bias on the admission process.
Very disappointing, that sort of behavior is part of the cause for why people fail to have honest discussion on sensitive subjects, it's the definition of regression.
Drunken Hobo
You're not supposed to think about leftism, you're supposed to listen and believe.
He simply said that because there was no bias overall in the admission process, then the gender roles in society by definiton is what causes the apparent bias. That is the truth. However, whether or not this societal bias is a "problem" that needs "solved" shouldn't really be a discussion on minute physics. This is in the realm of politics and opinion.
Andrew Whiteman it really doesn't need to be solved, people apply to subjects they like and that shouldn't be changed by quotas etc. Also, as far as I know, women aren't driven out of STEM by men.
Andrew Whiteman If he's making any kind of extrapolation of the data and giving his opinion in the video, I don't think you can fairly argue that the discussion "doesn't belong on a minutephysics video." To me, it looks like unnecessary censorship, and while it's obviously no legal offense, it still greatly degrades the respect I have for the people behind the channel. It's one thing to express your potentially inflammatory opinion, it's another entirely to hide from the flames.
You were so close to realizing something here with the Texas and Wisconsin analysis
I was looking for videos on the Simpsons TV show, and this is not what I was looking for.
But this was a great informational video, nonetheless.
These are not the Simpsons you are looking for.
@@michaelcarter2851 Doh !
Love how the comments are disabled for part 2. That is really a cheap way of avoiding uncomfortable debate of important issues, including sex bias in society. However, I will state this - if you are going to come to the conclusion that bias in certain fields is due to some societal discrimination that happens prior to the selection process at the university level, and not due to individual choices that might have other influences other than society (such as biological or genetic), then you have to show where that bias or discrimination exists and not just assumes it exists.
They literally said that different results mean different opportunities.
Whoever was behind that video clearly has no capacity for reasoning.
Midnight Chemist
Blaming biological or genetic differences requires first disproving every other possibility.
seigeengine
Evidently neither do you.
InfamousArmstrong Not if there is well documented evidence of differences between males and females based on genes and hormones. Most of the social causes others would cite are so nebulous they are difficult define, let alone find to disprove their influence.
It's liberal BS!
"More money makes you a cat."
3:39 *or more sadness makes you richer
And this is why I'm depressed
"More money makes you a cat"
Got it. I am smarter now. I shall spread this news far and wide.
Who else thought that it was related to the The Simpsons show?
Tell Me This me
Nope. But hey what do i know i read about this kind of stuff all day long.
I think Simpson is a common name
me...DO'H
I thought it was something to do with human opinions with simpsons being a prime example.
I've heard this concept several times, and this is the first I actually understand it now. Thank you!
I watched this video to see if it was a video about the statistic likelihood of The Simpsons, which has ran basically about a billion episodes, thus presented the world with about 50 billion unlikely scenarios, being able to "predict the future" simply through huge volume of outlandish scenarios.
This channel should be minute science since there is more than just physics here.
The first paradox isn’t paradoxical, you just drew conclusions incorrectly. When you aggregated the first graph to get the second graph, you added checks and crosses even though the boxes didn’t have the same number of people in them. That would be like finding the mean of fractions without the same denominator. 1/1 combined with 1/4 doesn’t equal 2/5, it equals 5/8.
In fact, this whole video just seems like you are drawing conclusions without enough data or drawing them incorrectly.
That's the whole point of the video! The examples with cats and humans are meant to make it obvious that you can't just add together members of different groups. But the problem is that everyone is unique, so there are no clearly defined groups of people. In a real experiment it's not always obvious which factors are important, and you can also get incorrect results from splitting groups that should be counted together.
What if the groups were something trivial like eye color? If a treatment heals 1/1 brown eyed person but kills 99/99 blue eyed people, do you think we should treat the groups separately and conclude it works 50% of the time?
Reddles37 that example does make more sense. So basically the point of the video is that mistakes are made in statistics and they are supposed to be more complicated than in the videos? That just seems like something i assumed, but that might be because I have taken stats courses before.
The second part makes sense, regarding how context matters and all that, but i have no idea what you were talking about in the first half.
I mean,
With treatment: 4/4 cats live | 1/4 people live
WIthout :3/4 cats live | 0/4 people live
so if we sum those up and make cats equivalent to people you get 5/8 things live with treatment and 3/8 live without. So, its clearly better. Correct me if i'm wrong but what you did makes no sense what so ever, its like, if 1/1 cats live then im adding 1 as having lived and if 3/4 then im adding 3 as having lived! WHAT?!
As i've said please correct me if im wrong or have misunderstood the point of the segment
Youre correct mate, it also felt as though he deliberately moved through it fast to ensure no one saw it.
The graphic with the two clusters (3:00) perfectly illustrates the maths behind the paradoxon!
3:11 makes me want to start an online journal using statistics to screw with people a la the Onion, and also publicly enlighten people on the dangers of statistics.
I thought that's the whole point of Nutrition Science.
This reminds me of a similar problem in my physics class. I had a 89% in the class(calculated from a weighted average of my test grades and classwork grades)
There would be one more grade before the end of the quarter, and it was a test. I wanted an A, so I figured I'd have to make a 90 or above to bring up the average. However, I made a 87 or so and it brought my overall grade up to 91 or so, because it brought up the average test score.
I thought this would be a good real world example because adding a lower average to a set brought the overall up
how could it bring up the average if the average is 89
@l1mbo69 It's because it's still greater than my test average. For example, imagine if my test grade is weighted to be 50% of my grade and homework the other 50%. Say I made a 40% on my first test and a perfect 100% on my hw. Then my grade is 70, a C. The next test rolls around and I get a 60%, which is lower than my overall grade. Now my test average increased to 50%, increasing my grade to 75, C+.
@@natejack2292 i fr just somehow completely skipped the classwork part
"There are three kinds of lies: lies, damned lies, and statistics."
-B. Disraeli
This seems like bad statistics more than a mathematical paradox. Statistics isn't just math, it also has to do with how you select samples and correlate data.
It's like saying "100% of humans who ate grapes survived and 0% of dogs who ate grapes survived. Grapes carry at 50% risk of death."
"More money makes you a cat."
I finally have my reason to exploit the working class.
That's honestly a really convoluted way to explain a pretty simple idea.
A Box Of Cheese the explanation and premise behind the idea is relatively simple, its the illustration of the problem that makes it sound convoluted, because problems in reality are pretty confusing until you split it down. besides, i think the presentation was fine, but eh, thats just my opinion.
exactly, they've made it way more complicated than it should have been to explain
A Box Of Cheese thats because they explained the statistics behind it. It's totally fine.
I am still stuck at the cat part I will watch the video again. What the hell does cat have to do with anything.
It's pretty easy to illustrate and explain with fully quantitative data like the cats/humans/money/happiness section, but in some ways it's more important to understand in categorical data situations where the paradox isn't as obvious.
I love how you make even complicated concepts and info easy to digest and understand!
Could you do a video on 'a*b*sorption vs a*d*sorption? I'd love to see your take on it, because most other sources give raw data/info without clear examples or fun cat analogies!
Anyway, thanks for making these videos and sharing knowledge in a fun way! I've been a fan and viewer for 5 years or so by now, and hope that you continue for many years to come!
Excellent video on the problem of correlation vs causation, or the “ice cream causes shark attacks” fallacy
Context is everything.
If I find a dead cat on the road, what killed it? Car. Truck. Heat Exhaustion.
If I find a dead cat in the kitchen, what killed it? The chef.
Context. The difference between roadkill and a meal.
I knew it. Bill Gates is a cat.
a very smart cat!
A copycat !
#IllumikittyConfirmed
Why did you turn off the comments on second part of this video? "Are University Admissions Biased? | Simpson's Paradox Part 2".
Turning off the comments means your not open to discussion which is against the scietific methodology .
Caleb Martin I just watched it. Reason why comments are disabled is, because it's liberal BS.
It's a religion.
If you don't believe that all groups are equal; and that evolution may have affected different races and sexes differently; you are an evil evil heathen... or nazi... whatever.
Because the best comments were around "Oh we commenters dared to criticize holy cow of feminism, how could we?"
Lol pretending that youtube "discussion" is scientific in any way.
Because the point of the video wasn’t about university admissions, it was about statistics.
This video is so refreshingly succinct! Gotta love getting all the info you need with no fluff. Keep up the good work!
The only Simpsons paradox is how it's still going when it stopped being funny after Season 9.
Why are the comments disabled on part 2?
Gadiel Nathan Arriesgado - he deviated from the physics and began sharing his political views, of which enough people disagreed with.
I'm here from part 2 (comments disabled) to say that it was a great video. Thanks Mr. Reich.
It reminds me of the statistic that says flying is the safest way to travel when in fact it is not. They use the statistic that more people are killed in traffic accidents than in airplane crashes which is true but it's the wrong measuring stick because your chances of surviving a car crash are pretty good, but your chances of surviving a plane crash are next to zero. So it's not really safer to travel by air.
Again, not a paradox just a counterintuitive result
"more money makes you a cat"
I'll never forget this quote
Since you disabled comments on part 2 of this video...
I do care about equality of opportunity. You don't measure it by the proportion of the outcome.
You also don't measure it by the amount of people in a group who go into a certain field.
When you have two fundamentally different groups (men and women) you cannot in any way expect the proportions to reflect one another. That's because men and women, overall, are different, have different interests and personalities, and will make different choices. Obviously I'm not saying ALL men are this way and ALL women are that way, but overall, there is a difference.
Even if you want to call it a societal problem, then we fundamentally disagree, and you're no longer advocating for equality of opportunity.
It doesn't matter to me if people are stimulated to make one choice or another. At the end of the day the path you take is your own choice and nobody can take that away from you.
Besides, you can hardly say women are shunted towards more crowded, less productive careers with the amount of "get women into stem!" projects. Now, personally, I take absolutely no issue with these programs. Why? Because I *don't care* if people want to stimulate other people to make certain choices. There's no issue unless there's force or threats involved (e.g. parents telling their daughter/son they will no longer financially support them if they go into X career)
Rather than trying to socially engineer society to reflect the desired proportion of women somewhere, how about we face the fact that individuals are entitled to their own decisions, and there's nothing wrong with whatever they choose? Point to the actual discrimination, rather than clutching at straws and saying "you can't conclude there is no bias anywhere else in education". The null hypothesis shouldn't be "there is bias against women". Your default position shouldn't be that either, regardless of historical oppression (which, btw, is often way overstated) and whatnot.
Right now, there's actually a measurable bias against men coming out of college. Women are more likely to get hired after they get their degree in STEM.
That's an example of something I *do* worry about, *actual* discrimination. I don't care if that discrimination is in favor of one group or the other. But I'm tired of hearing that women are oppressed with a poor basis to back it up.
Lastly, this statement in your video (again, part 2) somewhat pissed me off.
2:46 "...which is unfortunate if you think women and men should have equal opportunities and/or be paid equally for equal work." (on women generally being "shunted" towards graduate fields that end with lower pay)
Except, it really, really isn't. Women going generally into *different, lower paying fields* means they *ought to earn less*. They had the same opportunity, but made a different choice. Their pay will reflect said choice. If they didn't want to earn less, they would've gone into the higher paying fields, after all...
Please. PLEASE never confuse equality of opportunity with equality of outcome. You're a smart guy, Henry, there's not much else I can say here. What you said there is just objectively wrong.
I guess I'm bumping this so peeps can find this easier. I really wish he didn't turn comments off, it just makes you seem unsure in your own ideas, especially if you only do it on one video.
Thank you for this thoughtful reply. You have said everything that I myself was also thinking. In the second half of this video, the narrator gravely insults women by suggesting that their individual choices (to enter into lower-paying fields) are somehow the result of internalized oppression or the "shunting" of women into lower-paying fields. Choosing to become an elementary school teacher (a lower-paying profession) rather than an engineer (a higher paying profession) shouldn't be framed as a "biased" choice or the result of societal pressures. This takes away the agency from women.
not to mention that there's nothing wrong with making such a choice: favouring family and a social life over money.
I really want him to reply to this comment, as it is clearly a well thought out, completely civil comment that makes good points.
at this point there are so many comments on this video about it, everyone's checking this thread who sees the last video's comments were disabled. the outrage is PRETTY strongly in one direction with very few SJWs apparently watching physics channels. go figure, they dont like STEM, huh
Thanks for this great video which I learned a lot! ...
Simpson's paradox reminds us to be more cautious when we are dealing with these two situations: stratified data for observational study and blocked data for experimental study.
Because the potential "confounding factor" could lead to it if the sample size of each stratum or block is significantly unproportionate.
The real Simpson's Paradox is this: "Beer is the cause of and solution to all of life's problems."
I forget in which episode it was shown (it may have been the vast majority of them), but this is a proven fact.
Since your comments on the sequel to this video are disabled I'll leave my question here: Why should all careers have an equal distribution of different people groups? Is it some sort of unquestionable moral obligation? If anything homogony and equal distribution of careers aren't healthy since demand for different jobs and the backgrounds for specific careers vary widely. Trying to force a child of a specific gender, race or socioeconomic background into a career path towards or against a trend won't help that person at all. They'll probably end up hating their parents or teachers for forcing them into a career they don't enjoy. Just let people go where they feel like and stop dreaming about the perfect distribution of jobs and university majors. Maybe what you call "bias" prior to finding a career or college is just what people find comfortable or find matches their existing interests, is that really a bad thing?
Valuing equal results instead of equal opportunities is always oppressive. Shame they seem to think they're the same thing, as per them saying as much in their second video.
Biases aren't necessarily a BAD thing. There IS a bias for women to pick certain fields and males to pick other fields, and it's undeserved to some point, but it's also very understandable too - more women would pick English to STEM, and that's not a bad thing. The point of the video was that that aiming for a 50%/50% rate actually favours a group because of the stats being misrepresented, and valuing equal results doesn't work when you don't understand the stats and assume the bias is where it's not.
And while it's wrong to force your female child to go for a STEM degree because feminism, if your child actually wants to become an engineer it's equally wrong to force them into a shitty art degree.
I don't understand why all the incels and gaters rage, given if anything the video shows that feminist equity theories don't work because they're predicated on flawed interpretations. Get laid already dumbasses.
Well, we already know employer biases exist. As an example, there were studies where resumes were left unchanged except the name being changed from a white-typical name to a black-typical name, resulting in lower acceptance rates for the same credentials. In the absence of any political goal, profit-motivated companies would try to use some method to avoid this bias since they otherwise would be passing up good employees for bad reasons, and losing low hanging fruit to their competitors.
I believe I've heard of this study before, however, you must call into question what classifies something as a "black/white-typical" name as that in and of itself is a racial stereotype. Also, I'm curious what the metrics would be, what companies they consulted and other factors. Wouldn't the researchers have an obligation to reveal racial biases to companies being tested themselves if not the public? I'm curious if you have a link to it so I can take a look.
Another aspect you have to consider is job-culture fit. Some companies, though they might like the employee and they have all the right credentials, might not like or mesh well with other workers regardless of their background. And even beyond that sometimes people don't work well with coworkers they can't relate with. Anyway, my point being that while diversity of a workforce is important, it's more important that a team works well together and that its members are competent enough for their roles. At least that's what investors look for in smaller, startup companies. They don't care as much about the estimates as they do about how solid the team is and what they've already achieved.
Yeah I don't understand. The first half of that video shows that women are actually given more oportunity than men per individual department (revealing that this is indeed a case of simpson paradox), only to later assert that women have less opportunity than men (!!!)
Nice video, i learned something. Apparently my room is painted in a shade of "socioeconomically advantaged", wasn't aware that was the name of the colour.
My main take away from this video is that non-white people are poor and stupid. I don't agree with that, but the video implies it.
Thought this was about the Simpsons... The TV show :(
@Simon Dewitt The more it gets renewed the worst it gets. So many paradoxes...
glad I wasn't the only one
"More money makes you a cat" I'm dying 🤣
not sure if i can stay subscribed to a channel that doesn't allow discussion, a real shame tbh
why would you stay sub'd to a science channel that doesn't believe in science?
i feel almost the same but i would be ok with an apology and just
*STICKING TO PHYSICS, MINUTEPHYSICS* ffs
If the best comment is aroudn "Oh we commenters dared to criticize holy cow of feminism, how could we?", who would not turn off comments?
RIP Snowball 1
This comment!!
3:37 "More Money = More sadness , less money = more cats"
I'm so happy you added "more money makes you a cat."
Tip of the day :
Just be a cat, cat is always happy.
"There are lies, damned lies, and statistics."
-Mark Twain
“More money makes you a cat”
This is the strongest endorsement of capitalism I’ve ever heard
Today I learned that bill gates is a cat.
Minute physics evolved to four minute physics 👍 awesome.
This episode was more like four minute math:D
NetAndyCz .Yeah but math is used to describe physical phenomena, so I think it doesn't bug me that much. Or should I say that I am familiar with 'mathematics' of physics. It's not easy though. But once you get familiar with it, it is so beautiful.✌😋
But but if you look at it like this, was not it more about math being used to describe biological phenomena?
NetAndyCz .LOL yes😂 i guess it is a way of looking at it. But you see even some theorems like that of Bayes' involves a lot of examples of biological phenomena. To be honest, there are no two branches of science that are completely separable(having no connection at all between them).
I don't remember where it was but I recall Henry saying that he deliberately made the choice to rename from 'One Minute Physics' to 'Minute Physics' for that exact reason. I you go back and watch the first 2 videos on the channel, the title card does say 'One Minute Physics'.
I see that you've deleted comments for part 2, stating that the discussion was too uncivil. If you decide to do that permanently it would be a huge knock on the reputation of MinutePhysics, please don't do that!
The discussion format is an excellent way to learn more, and I always scan through the discussion for thoughtful remarks. I concede that there are a lot of remarks that aren't carefully thought out. However, the benefit of healthy and free discussion is the foundation of CZcams. As an author you can easily delete select remarks if they cross the boundary of racist or insulting behavior, and you can even screen remarks before they appear. Most authors will have the backbone to let remarks from those who disagree with some aspect of their video remain. So, it also represents a forum for healthy discussion. Now that CZcams have deleted the author's ability to post annotations within the video, it is an important way to update thoughts or mistakes that are present in the video, short of deleting the whole thing. Advertisers have to know that the author is not responsible for remarks in the discussion forum, we can't let advertisers force us into accepting CZcams as a lesser format just because some person left a bad word.
True
Thank you for your thoughts. I couldn't agree more.
"comments are disabled for this video"
immediate red flag about everything said in the video shoot up into the air. This is the internet, if your gonna complain about the lack of civility of peoples comments don't put up videos with controversial statements in them. I'm certain if you did a study, you would find that videos that say controversial things are much more likely to have a verbal brawl in the comments section than not, though you shouldn't need a study to figure that one out.
Yeah I noticed that as well... I think he reserves the right to disable comments if he wants to, but that doesn't necessarily mean it;s the best choice.
spelunkerd I agree with nearly everything you said here, except for your comment of allowing a "healthy discussion". There is nothing healthy about the trolling, bullying, insults, and racist comments that pop up in every single CZcams video comments section, discounting the small channels that only get 20 or less comments.
The CZcams comments section is a nasty place and is no place for a "Healthy discussion". But I agree that the comments should be left there unless they start getting unmanageable using the tools CZcams provides creators. If it gets so bad you can't control it then sometimes the nuclear option is the only option.
I cringed when he started mentioning the test scores and different races thing, even though he was being perfectly factual and professional, you know there is someone somewhere who got triggered.
Thank you for highlighting the old adage "there's lies, damned lies, and statistics." The phrase was never meant to mean you couldn't trust statistics, only that some groups will twist their meaning in ways like you described as being errors in logic, just to support an agenda. Sad part is a lot of them think they are doing it for the right reasons and believe their own distortion. Always look at the data yourself, make a good judgement, rely on people you trust to help you interpret it.
Still one of the best and clearest explaining video about this topic.