I think with more precise language this would have been easier to understand, the "1 in 1000" metric is referring to the number of people who took the test and actually have the disease. If that was made explicitly clear it would be fairly easy to mentally workout why the odds of having the disease based on a positive test result would actually be lower than 10%.
@gollem148 Says:
I was correct surprisingly.
@Hyndy Says:
So you have a 1 in 100 chance to be flagged as diseased after the test, but after that there is a 1 in 10 chance that you actually have the disease. So 0.01% chance to be flagged positive, and 0.001% to be diseased. So out of the 0.01% flagged 0.009% are false positives, and 0.001% are diseased. So now are total is 10 and our desired event is 1 i.e. 1 in 10 chance, of having the disease, if you are flagged positive. In my attempt to be more rigorous I think I ended up being more confusing but oh well
@ardigo_francesco Says:
anybody got the 50/50 girl's @?
@mike62mcmanus Says:
Great and I watched your video and now I have the disease
@mr.outlokk5493 Says:
You can’t determine the probability without knowing how common the disease is. Also, adding another verified result to the 1,000-person sample changes the observed accuracy. So claiming it’s below 10% isn’t mathematically justified.
@paulhanson6836 Says:
Fact number 1: it effects 1 in 1000 people (not to be mistaken with 1 in 1000 people test positive).
The question was what are the chances that you actually have it, which never changes based on the limited information given.
So if the first statement is indeed fact, then the correct answer is still 1/1000
@KiloWattFPV Says:
my answer is.....the flower lady is dutch....
@ISMsoccerboy Says:
Yeah i didnt even know how to set up this problem lol but its really important for real world application of statistics
@SoSoParanoid Says:
Shout out to the weird math guy.
@Jacky-Poo Says:
99% of the 1/10 or 10/100% so 10%
@anarch1984 Says:
I got this one right o.o
@flarapediati Says:
Hey I actually got this one right, slightly less than 10 because ten precent would occur with 100% test accuracy on the 0.1% of people compared with a 1% false positive with everyone not having it, shuffle the numbers around and it cuts off a slice. I love hunch estimation math.
@YjstY Says:
Why its only 99,900 people not get isnt it suppose to be not 99,901 for people who dont get the disease
@rp-kf8vz Says:
It's a jee question we had the exact same one
Finally someplace I can apply something I hv learnt practically 😅
@fumbled23 Says:
1/100000
@Boluggg Says:
99%
@robsonfujii461 Says:
99% doesn't mean it is correct 99 out of 100. It is an average.
@sagallah Says:
just before i block you, I don't give a damn how you try to twist it to confuse people, if you test positive, yhen you are %99 have it.
@TomGeorgin Says:
I am at the start of the video.
"There is a disease that affects 1 out of 1000 people. [blablabla]. What are the odds that you have the disease?"
Answer: 1 out of 1000 = 0.1%
@HexViccissitude Says:
Horribly explained and fully intentionally.
@LeoH3L1 Says:
No no no, you can't do that, because in reality you in effect didn't test anyone at all who was that 1/100 because there is no functional difference between it missing that person and not testing that person at all.
@joshuaparr3544 Says:
I know this thanks to Hannah Fry and VSauce (Michael Stevens). You gotta really think about the right math.
@Add_Infinitum Says:
So there is both a 1% false positive rate and a 1% false negative rate?
@davidjohanson5911 Says:
This is only true under the assumption that the person being tested is just a random old joe off the street. When a patient takes an examination like this, it should be assumed that they are not just some random joe but that they have already shown other potential symptoms of said disease that gives the doctor a reason to prescribe the test in the first place. Therefore the question in real life actually becomes "If a test is 99% accurate and x amount of people that show enough symptoms of the disease to justify being tested end up having the disease, what is the probability that someone being tested gets a false positive?"
Yes, the example given is an interesting mathematical reality about statistics, but then you get idiots going around thinking that medical examinations are not super accurate, when they are able to correctly diagnose people much more accurately than this example suggests.
@beecraft5975 Says:
Bro the same question is there for my 12th maths in Bayes theorem
@anthonyanglim7147 Says:
I can't believe the girl said: "Well then it's 50%, you either get it or not." 😂
@kylesebring Says:
Shouldnt it correctly identify all 100 people with the disease if it only produces false positives? And if so it would be exactly 10%.
@supreetsahu1964 Says:
"I promise I studied" said the 50% lady 😂😂
@oscargama8727 Says:
That not 99% accurate if 99 test were correct and 999 wer false
@gracecalis5421 Says:
I have a clarification/correction for the test. You said that it was 99% accurate with a 1% false positive rating. Yet in your answer, only 99 of the 100 people that actually have the disease have positive results.
That doesn't add up to your initial question because you implied that the test only has a 99% accuracy rate BECAUSE of the false positives. Yet that 1 undetected person with the disease implies it has a 1% false NEGATIVE rate. False positives and false negatives are not interchangeable and represent very different domains in a test kit's applicability. False negatives are a result of a failing of sensitivty. False positives are a failing of specificity.
That you didn't bring up the rate of False negatives occurring implies that false negatives do not occur. Which means 100 out of those 100 people that ACTUALLY have the disease should be detected, and the failure of accuracy occurs EXCLUSIVELY in the population WITHOUT the disease; AKA the 999 people without the disease but with a positive result.
@101arg101 Says:
This is also the exact reason why police can’t use facial recognition for nation-wide surveillance. The rate of doppelgängers is about 1 in 80,000, and number of people with active arrest warrants is over 2 million. This means there’s on average 25 people with arrest warrants that look just like you.
@mcthunderstick2374 Says:
1/1000 is not 99%
@motion1picture809 Says:
Strictly mathamaricly sure, but this is just a question to catch people out. In the real world you only get tested for something there is a chance of you having as you show symptoms. There are so many factors to take into account in this. I hate these types of questions as they are just made to make people feel dumb. If you want to talk about the maths talk about the maths.
@haeschensk8er Says:
Yet no one wanted to hear that mass testing is absurd a couple of years ago... make it make sense
@ariabehforouz7584 Says:
"Accuracy" is not a correct terminology though! It's either specificity or sensitivity. From your explanation in the second half of the video, I guess you meant that both are 99%.
@burntwallaby Says:
This assumes that everyone gets the test, so it isn't very applicable to the real world
@xyttra Says:
Is it not 1/11?
Edit I assumed there was a trick and it wasn't going to be that but turns out it was 😁 I'm sure I wouldn't solve this on the spot if I wasn't sitting on the toilet in the comfort of my home tho 😁
@YO_BOY-t1i Says:
Classic bayes
@gnkk6002 Says:
Bayes theorem
@AD-cf8ru Says:
Sensitivity/1-specificity
@siddharthatwal6633 Says:
It depends on pre test probability
@umanathjha5762 Says:
Why complicate
99% accurate ---> 1 in every 100 which implies 10 in 1000
but as said 1 in 1000 actually has the disease
so the test says 10 persons has disease while actually only 1 has it and therefore 10% is the chance that the person might actually have the disease
NOT A GOOD TEST BTW
@Tapanfrom2025 Says:
Math is not mathing
@adriansauer7178 Says:
I would use 1000 people. 1 has the disease, 10 are false positive. That's 1 out of 11 = 9.1%, in the Video 9.01%. Where does the small change come from?
@MelfiortheOne Says:
I think the broad statistics doesn’t affect the chance, when you have another factor in play: the positive test. It shrinks the probability to only test accuracy.
@codeconcepts3194 Says:
class 12th math question on bayes theorem.
@danieltownley7436 Says:
The trouble with videos like this is there's only a 50% that what he's saying is correct - it either is or it isn't. 🤷🏻
@tijnjansen7609 Says:
This explanation is bad. I am most worried about negative tests that are actually positive where do those results come in. They die of a lack of hartbeat i guess.
LATEST COMMENTS