How do I cope with the fact that most people (even ITT) can never understand the correct answer?

File: 64 KB, 772x374, You can solve this, right.jpg [View same] [iqdb] [saucenao] [google]

Anonymous Tue May 5 07:01:09 2020 No.11639020 [Reply] [Original]

How do I cope with the fact that most people (even ITT) can never understand the correct answer?

>>	Anonymous Tue May 5 07:03:32 2020 No.11639028 A because rounding, next question.

>>	Anonymous Tue May 5 07:08:43 2020 No.11639035 >>11639028 you have tall parents and have a 99% chance of being 7' tall, but only 0.1% of the population are 7' tall, how tall will you grow?

>>	Anonymous Tue May 5 07:11:23 2020 No.11639041 Huh, I got about 20%, where did I fuck up? Am I supposed to use bayes rule? I dont have any priors about symptoms to rely on.

>>	Anonymous Tue May 5 07:19:07 2020 No.11639054 can someone tell me why it isn't D

>>	Anonymous Tue May 5 07:22:12 2020 No.11639058 >>11639054 Imagine if the disease affected noone and you tested positive, what is the chance then?

Anonymous Tue May 5 07:23:57 2020 No.11639060

>>11639020
Yah this thread again! Still fucked up!
Thread warning. Thread will be patrolled by an imbecile with a poor grasp on the English language.
Correct answer, according to how the question is worded is: D 99%, Problem is childishly simple, most 6 year olds could solve it
However the imbecile, being a professional Pajeet cocksucker, will demonstrate his mental instability by foaming at the mouth, and start raving on about Bayesian statistics ( which are simple enough when applied to a correctly worded problem )

>>	Anonymous Tue May 5 07:26:07 2020 No.11639063 >>11639020 D.

>>	Anonymous Tue May 5 07:27:38 2020 No.11639067 >>11639058 Then you would be infected with a disease that doesn't affect you. You still would have a 99% chance of being infected.

>>	Anonymous Tue May 5 07:30:35 2020 No.11639072 >>11639058 imagine if the test was 50% accurate and everybody had the disease, what is the chance then?

>>	Anonymous Tue May 5 07:44:06 2020 No.11639094 you people are all pseuds the correct answer is 92%

>>	Anonymous Tue May 5 07:44:06 2020 No.11639095 >>11639054 Because 1% of 99.9% of all people will test positive and not have this disease but only 99% of 0.1% of all people will test positive and actually have the disease. https://brownmath.com/stat/falsepos.htm

>>	Anonymous Tue May 5 07:47:40 2020 No.11639103 >>11639035 99% you are 7ft

>>	Anonymous Tue May 5 07:51:31 2020 No.11639110 File: 12 KB, 1104x248, 1000-dots1.png [View same] [iqdb] [saucenao] [google] >>11639095 the answer is in the question, rephrase the question if you want to talk about your statistic autism

>>	Anonymous Tue May 5 07:54:18 2020 No.11639116 >>11639020 The % of the population it effects literally doesn’t matter in this question. If a test is 99% accurate and you test positive you have a 99% chance of being infected, so it’s D

>>	Anonymous Tue May 5 07:55:56 2020 No.11639119 You guy's dont get that he's just using this to redpill you that covid only affects a fraction of the population under 100 billion years old. Classic 4d chess.

>>	Anonymous Tue May 5 07:57:05 2020 No.11639123 >>11639116 And if 0% of the population was affected? your whole argument falls apart bud

>>	Anonymous Tue May 5 08:02:26 2020 No.11639134 File: 14 KB, 617x369, ss (2020-05-05 at 02.02.01).png [View same] [iqdb] [saucenao] [google] >>11639123

Anonymous Tue May 5 08:04:20 2020 No.11639138

>>11639020
Let TP, FN, TN, FP = the percentage of true positives, false negatives, true negatives and false positives respectively.

Infection rate = TP+FN

Accuracy = TP+TN

TP = x =< 0.001
FN = 0.001-x
TN = 0.99-x
FP = 0.01-(0.001-x) = 0.009+x

Chance you are infected = TP/(TP+FP) = x/(0.009+2x) = 1/(0.009/x+2) =< 1/11 ~ 9.09%

So the answer is indeterminate. It could be A, B, or some other number between 0 and 1/11.

>>	Anonymous Tue May 5 08:06:36 2020 No.11639142 >>11639134 oh so now the population affected does matter

Anonymous Tue May 5 08:07:43 2020 No.11639145

>>11639116
>If a test is 99% accurate and you test positive you have a 99% chance of being infected, so it’s D
Incorrect. If a test is 99% accurate then 99% of test results are either true positive or true negative. This doesn't tell you the percentage of positive tests that are true positive.

Anonymous Tue May 5 08:22:17 2020 No.11639164
File: 18 KB, 328x499, 41xp0Xzej8L._SX326_BO1,204,203,200_.jpg [View same] [iqdb] [saucenao] [google]

1,000,000 people on earth
0.1% affected
So 1,000 sick 999,000 healthy

All people get test
99% correct
For sick people, 990 tested sick, 10 tested healthy
For healthy people, 989,010 tested healthy, 9990 tested sick

You tested sick
Total tested sick 990+9990=10980
You are one of those 10980
Only 990 out of those 10980 are real sick
Chance you are sick 990/10980=9.016%

Anonymous Tue May 5 08:39:36 2020 No.11639211

>>11639164
>For sick people, 990 tested sick, 10 tested healthy
>For healthy people, 989,010 tested healthy, 9990 tested sick
This doesn't follow. If 0 sick people test sick and 99000 healthy people test healthy, then the test is still 99% accurate. There is no reason to assume the test is as sensitive and specific as it is accurate.

Anonymous Tue May 5 08:42:26 2020 No.11639217

>>11639020
You'd better start coping because the vast majority of the population is [math]\textit{never}[/math] going to understand Bayesian statistics. Regardless, your image is misleading as COVID-19 has a higher prevalence than 0.1% based on the serological evidence we have.

>>	Anonymous Tue May 5 08:44:47 2020 No.11639223 File: 26 KB, 460x416, az1zj1m_460s.jpg [View same] [iqdb] [saucenao] [google] >>11639116

>>	Anonymous Tue May 5 08:49:51 2020 No.11639232 >>11639020 What's the false negative rate?

>>	Anonymous Tue May 5 08:49:59 2020 No.11639234 >>11639138 Accuracy is (TP+TN)/(TP+FP+TN+FN) though. But good trolling I guess.

>>	Anonymous Tue May 5 08:53:50 2020 No.11639245 >>11639211 but failing to know about the mechanisms of the test and what makes it fail 1% of total cases, we can only assume failures are randomly distributed because we have no information one way or the other.

>>	Anonymous Tue May 5 08:55:48 2020 No.11639253 File: 51 KB, 600x467, 001.jpg [View same] [iqdb] [saucenao] [google] >>11639234 >Accuracy is (TP+TN)/(TP+FP+TN+FN) though. Yes, and since here TP+FP+TN+FN = 1, accuracy = TP+TN

>>	Anonymous Tue May 5 08:57:13 2020 No.11639258 >>11639217 This has very little to do with Bayesian statistics. See >>11639138

Anonymous Tue May 5 08:58:02 2020 No.11639262

C is statistically the correct answer most often.

The real answer is C. By knowing that each of the given choices isn't accurate to 4 decimal places, their true probability of being the correct answer modulates to 0, and any response needs to be taken as a guess. This is math, this is bayesian statistics.

>>	Anonymous Tue May 5 09:02:29 2020 No.11639272 >>11639060 Conditional probability P(a\|b)=P(ab)/P(b) P(a\|not b)= same formula dipshit Now you find P(ab) and P(anotb) using the formulas above dipshit Then you calculate the odds P(ab)/(P(ab)+P(anotb)) a is test positive b is be sick

Anonymous Tue May 5 09:06:19 2020 No.11639278

>>11639253
TP+FP+TN+FN would only ever 1 if only 1 person was tested.
TP, TN and so on are not ratios but actual case numbers. Or at least that is how they are usually described as.
Accuracy describes how well a test can correctly distinguish between healthy and unhealthy people.
An accuracy of 99% means that of 100 people 99 people will have their status in regards to a certain condition identified correctly.

Anonymous Tue May 5 09:06:22 2020 No.11639279

>>11639258
It's still bayesian
Let P(I) be the probability of infection, and P(T) be the probability of testing positive. Technically you'd have to know the false positive and false negative rate of the test but assuming that 99% accurate means a false positive happens 1% of the time and a false negative happens 1% of the time, then

P(I|T) = P(T|I) * P(I) / P(T)

P(T|I) = 0.99
P(T|¬I) = 0.01
P(I) = 0.001
P(¬I) = 0.99
P(T) = P(T|I) * P(I) +P(T|¬I) * P(¬I) = 0.011

Thus P(I|T) = 9.1%

So yes, it has everything to do with Bayesian statistics.

>>	Anonymous Tue May 5 09:15:49 2020 No.11639303 >>11639245 Why not assume true positives and false negatives are randomly distributed among the sick? Arbitrary assumptions are unnecessary.

>>	Anonymous Tue May 5 09:20:15 2020 No.11639313 >>11639058 Then you are the reason the test is not 100% accurate

>>	Anonymous Tue May 5 09:31:11 2020 No.11639330 >>11639278 >TP+FP+TN+FN would only ever 1 if only 1 person was tested. Incorrect, read my post again: >>11639138

>>	Anonymous Tue May 5 09:33:20 2020 No.11639334 >>11639020 None of the above

>>	Anonymous Tue May 5 09:34:23 2020 No.11639335 >>11639020 50% because there are 2 possible outcomes

Anonymous Tue May 5 09:34:27 2020 No.11639336
File: 355 KB, 812x368, 1560229095770.png [View same] [iqdb] [saucenao] [google]

>>11639060
This problem assumes that the entire population is tested. Then of course the chance you have the disease, even if you test positive, is pretty low. However healthy people usually don't just randomly take tests, they only do it if you're already showing symptoms. Now if you are in the hospital and based on your symptoms the doctors assume you have disease X and then they run a test on you, the chance that you test positive and actually have the disease is pretty much guaranteed.

If you are a completely healthy person and take the same test and it comes back positive, the chance you actually have the disease is almost zero.

Case closed.

Anonymous Tue May 5 09:37:35 2020 No.11639342

>>11639303
it is an arbitrary assumption but the only alternative is to withhold judgment. but then we can't speculate on probabilities.

this whole situation is based on lacking knowledge, only knowing that corona cases are randomly affecting 0.1% of the population and that tests randomly fail 1% of the time.... actually no. we do not know this. we were not told that corona is random by nature and we were not told that the tests rely on rolling dice. it's our own assumption that we are just as likely to be any other random person in the same situation. the randomness comes from the arbitrary assumption.

>>	Anonymous Tue May 5 09:39:29 2020 No.11639348 are you smarter than /pol/, /sci/ ? >>>/pol/256272747

>>	Anonymous Tue May 5 09:40:17 2020 No.11639349 >>11639342 >the randomness comes from the arbitrary assumption. just like in quantum physics

>>	Anonymous Tue May 5 09:46:37 2020 No.11639362 you don't know any of the conditional probabilities. the test's inaccuracy could come from all false positives, false negatives, or any mixture of the two that comprise it's 1% inaccuracy

>>	Anonymous Tue May 5 09:55:39 2020 No.11639382 >>11639362 we also don't know if we caught the virus. not knowing things makes you rely on probabilities

>>	Anonymous Tue May 5 09:56:02 2020 No.11639384 >>11639110 10/10 would english again

>>	Anonymous Tue May 5 10:08:21 2020 No.11639405 >>11639164 What is this jew wizardry? I was just on amazon and this book showed up in my recommendations. I don't even buy nonfiction on amazon.

>>	Anonymous Tue May 5 10:14:43 2020 No.11639416 >>11639330 Well fuck me then. I guess I'm tired and stupid.

>>	Anonymous Tue May 5 10:22:03 2020 No.11639431 How it isn't D? The question literally said "you tested positive" Meaning you already positive according to the test, but because the test is only 99% accurate the answer in 99%.

Anonymous Tue May 5 10:37:28 2020 No.11639468

>>11639431
Because there are 11 times or so as many people tested positive as you.
Point is having a test with 99% accuracy is useless if only 0.1% of tests are expected to be positive.
For a valid result you want to have a decent amount of actually infected people and that is why you only want to test people that show symptoms are had contact with infected people.

Anonymous Tue May 5 10:40:54 2020 No.11639479

>>11639279
>Technically you'd have to know the false positive and false negative rate of the test but assuming that 99% accurate means a false positive happens 1% of the time and a false negative happens 1% of the time
There is no reason to assume that, so Bayes theorem is useless here.

Anonymous Tue May 5 11:06:21 2020 No.11639565

>>11639479
>There is no reason to assume that, so Bayes theorem is useless here.
Lol no. If you don't make that assumption it's still useful. Let [math]x[/math] be the true positive rate, [math]y[/math] be the true negative rate and [math]\hat{y}[/math] be the false negative rate then the probability of being infected given you tested positive is
[eqn]
P(I|T) = \frac{x \cdot 0.01}{x \cdot 0.01 + (1 - y) \cdot 0.99}
[/eqn]
and the probability of not being infected given you tested negative is
[eqn]
P(¬I|¬T) = \frac{y \cdot 0.99}{y \cdot 0.99 + (1-x) \cdot 0.01}
[/eqn]
so yes, it's still extremely relevant.

>>	Anonymous Tue May 5 11:15:08 2020 No.11639586 >>11639123 >And if 0% of the population was affected? how can you make a test for a disease that no one has

Anonymous Tue May 5 11:22:55 2020 No.11639598

>>11639565
Should also say that there's a constraint on [math]x[/math] and [math]y[/math] since 0.99 accuracy means true positives + true negatives / (true positives + true negatives + false positives + false negatives) = 0.99, so
[eqn]
0.99 = \frac{x + y}{x + y + (1-x) + (1-y)}= \frac{1}{2}(x + y)
[/eqn]
So it's possible to eliminate one of [math]x[/math] and [math]y[/math] of you desire.

>>	Anonymous Tue May 5 12:02:12 2020 No.11639706 >>11639278 hes defining TP, TN, etc. as proprotions, nigger. (TP+TN)/ALL = TP/ALL + TN/ALL, then redefine TP to be TP/ALL and TN to be TN/ALL. it all works out. If you don't think so, just multiply everything by 10000 or something.

>>	Anonymous Tue May 5 12:14:41 2020 No.11639732 File: 272 KB, 1280x1348, 4g_tower.jpg [View same] [iqdb] [saucenao] [google] >>11639020 This probability question is uninteresting because there is no possibility of winning a goat.

>>	Anonymous Tue May 5 12:33:52 2020 No.11639784 >>11639565 Your math is wrong. P(I) = 0.001 P(T\|I) = x/0.001 P(T) = 2x+0.009 But why use Bayes theorem when you already understand that these probabilities are ratios of TP, FP, TN, and FN? P(I\|T) = TP/(TP+FP)

Anonymous Tue May 5 13:24:25 2020 No.11639913

Hey, I know a few problems like this:

You have ten bananas in a box, one banana is green. How bananas do you have in the box?
a) 16
b) 3%
c) 10
d) George

Joe is a big dog. What is the name of the dog called "Joe"?
a) Mandy
b) Bert the duck
d) Its a lion
g) Joe

A plane is painted red. It is is flying to New York. What is the color of the plane?
w) soft
h) 9/11
k) red
o) none of the above

...and many more. Hours of wholesome fun for the family during lockdown.

>>	Anonymous Tue May 5 13:26:45 2020 No.11639920 >>11639913 >he thinks accuracy is the same as precision What a brainlet.

>>	Anonymous Tue May 5 13:33:50 2020 No.11639942 There’s not enough information in the question for everyone to come up with the same answer. People are making their own assumptions and getting mad when others don’t have the same.

>>	Anonymous Tue May 5 13:51:01 2020 No.11640007 >/sci/ is actually dumber than /pol/ Embarrassing

>>	Anonymous Tue May 5 13:55:57 2020 No.11640018 >>11639942 wrong. accuracy is well defined.

>>	Anonymous Tue May 5 14:00:00 2020 No.11640030 >>11640018 It is but we still don't know specificity or sensitivity of that test which is necessary to know for that problem

>>	Anonymous Tue May 5 14:02:23 2020 No.11640043 >>11640030 that doesn't mean you come up with different answers. if you solve the problem correctly then you can nail down the same range of answers as everybody else. everyone who is getting different answers is misinterpreting accuracy.

>>	Anonymous Tue May 5 14:19:57 2020 No.11640125 >>11640043 If specificity is 100% there would be no false positives by definition meaning you if you get tested positive you can be 100% sure you have it. All inaccuracy would be in identifying false negatives.

>>	Anonymous Tue May 5 14:21:29 2020 No.11640131 Isn't this something to do with Baye's theorem?

>>	Anonymous Tue May 5 14:24:16 2020 No.11640140 >>11639920 Try learning English, Pajeet. Then you dont look as stupid as you already are.

>>	Anonymous Tue May 5 14:25:24 2020 No.11640145 File: 70 KB, 468x284, 1578126362974.png [View same] [iqdb] [saucenao] [google] >>11640131 Yeah I'm 90% sure it is

>>	Anonymous Tue May 5 14:27:04 2020 No.11640150 >>11639942 Its the same twat that posted this same shit last time. Even with the erroneous wording pointed out to him he still doesn't get it. Sort of fun to troll him though because he just cant leave it alone. Like a dog after a bone.

>>	Anonymous Tue May 5 14:31:12 2020 No.11640164 >>11640125 correct. if you solve for the answer, you get a function of one variable and can nail down a range of possible answers. if you solve correctly, you get this range >>11639138

>>	Anonymous Tue May 5 14:31:12 2020 No.11640165 >>11639020 I prayed to the Yin and Yang and they told me that the answer is B. I trust them.

>>	Anonymous Tue May 5 14:39:30 2020 No.11640196 >>11640164 What I get from this is that the 1-accuracy or at least specificity needs to be much lower than the prevalence of what is tested to be useful. So you need to have a reasonably large amount of sick people in your test population.

>>	Anonymous Tue May 5 18:49:37 2020 No.11641022 >>11639058 imagine imagining

>>	Anonymous Tue May 5 18:58:00 2020 No.11641045 >>11640140 >Try learning English, Pajeet. >Then you dont (sic) look as stupid as you already are (sic). Then you won't look as stupid as you already do. What part of my post didn't you understand, illiterate brainlet?

>>	Anonymous Tue May 5 19:56:32 2020 No.11641175 >>11639020 So all of these retards on this thread are trying to get me to believe that a 99% accurate test, has an accuracy rate of 9% if you test positve. Maybe the 5G rumors are true the....

Anonymous Tue May 5 20:41:27 2020 No.11641264

>>11641175
No, the precision of the test, the probability a positive result is true, is between 0 and 1/11. That's the number that answers the question. The accuracy is 99% since the negative version of precision, the probability a negative result is true, is between 990/991 and 1, and there are many more healthy people in the population than sick people.

>>	Anonymous Tue May 5 20:52:51 2020 No.11641292 >>11639020 The answer is0.1%

>>	Anonymous Tue May 5 20:57:03 2020 No.11641305 >>11639041 test 100,000 people 1098 will test positive, of which are 99 sick, 999 are healthy(false positive) 99/1098 = 9.02% correct answer: B

Anonymous Tue May 5 21:00:00 2020 No.11641314

>>11639431
you test 1000 people.
1% will get a false result. so ten people.
0.1% will be infected. so one person.
You now know are in the first group for sure.
What are the odds you are also in the second group?
1 out of 10 people, or 10%.
(If you assume no false negatives)
(if the test is only 95% accurate as in real life, that drops to a mesliey 2% chance)

>>	Anonymous Tue May 5 21:05:57 2020 No.11641325 >>11641305 >1098 will test positive Why not 900?

>>	Anonymous Tue May 5 21:08:15 2020 No.11641335 >>11641325 Not him, but: Out of 100,000 people, 99,900 will be healthy, 100 will be sick. 1% of healthy people will be identified as positive, or 999 99% of sick people will be identified as positive, or 99 999 + 99 = 1098

Anonymous Tue May 5 21:14:27 2020 No.11641349

>>11641325
100000 people tested
0.1% are sick = 100 people
99.9% are healthy = 99900 people
test accuracy 99%
of the 100 sick, 99 will test as positive
of the 100 sick, 1 will test as negative
of the 99900 healthy, 999 will test as positive
of the 99900 healthy, 98901 will test as negative

Anonymous Tue May 5 21:16:08 2020 No.11641350

>>11641335
>1% of healthy people will be identified as positive, or 999
Why? If 900 healthy people are identified as positive and all 100 sick people are identified as negative, then 1% of the tests are inaccurate. Or 1000 healthy people are identified as positive and 0 sick people are identified as negative, then 1% of tests are inaccurate. Or any combination in between those numbers that add up to 1000.

>>	Anonymous Tue May 5 21:17:58 2020 No.11641358 >>11639054 this, what does it matter what percentage of the population has it if the test is 99 percent accurate? Any answer other then D means that the test isn't 99 percent accurate and the information given in the question is not true

>>	Anonymous Tue May 5 21:18:04 2020 No.11641359 >>11641350 https://helpautism.org

>>	Anonymous Tue May 5 21:19:00 2020 No.11641362 >>11641349 >of the 100 sick, 99 will test as positive Why? 0-100 could test positive as long as the false negatives and false positives add up to 1000.

>>	Anonymous Tue May 5 21:19:44 2020 No.11641367 >>11641362 >>11641359

>>	Anonymous Tue May 5 21:20:23 2020 No.11641368 >>11641359 >>11641367 Not an argument, try again.

>>	Anonymous Tue May 5 21:34:23 2020 No.11641407 >>11641350 >Why? Because the test is 99% accurate, anon. That means it will give a true result 99% of the time. In other words, the test has 99% sensitivity and 99% specificity.

>>	Anonymous Tue May 5 21:35:59 2020 No.11641410 >>11641264 So 90 percent of the people in the world who tested positive for the corona virus don't have it, therefore there is even less than 0.1 percent of the population that has it, which further lowers the tests accuracy, ad infinitum

Anonymous Tue May 5 21:36:25 2020 No.11641412

>>11641407
>Because the test is 99% accurate, anon. That means it will give a true result 99% of the time.
Yeah, that's exactly what I said. There are many ways to get a true result 99% of the time.

>In other words, the test has 99% sensitivity and 99% specificity.
Doesn't follow.

>>	Anonymous Tue May 5 21:36:29 2020 No.11641413 >>11639020 how would we know that covid 19 infects 0.1% of the population if by default we’d be detecting 1% of the population as infected?

>>	Anonymous Tue May 5 21:36:52 2020 No.11641416 >>11641410 >therefore there is even less than 0.1 percent of the population that has it The initial 0.1% doesn't include the false positives, dummy.

>>	Anonymous Tue May 5 21:36:57 2020 No.11641417 The answer is D.

Anonymous Tue May 5 21:38:57 2020 No.11641424

>>11641410
>So 90 percent of the people in the world who tested positive for the corona virus don't have it
No, because the test is not given to everyone. Testing is more likely to be given to those who have it.

>therefore there is even less than 0.1 percent of the population that has it
No, 90% of those who tested positive can be healthy while only 0.1% of the population has it. There is no conflict there.

Anonymous Tue May 5 21:41:11 2020 No.11641429

>>11641412
>Yeah, that's exactly what I said. There are many ways to get a true result 99% of the time.
Since the question has no further qualifying qualities for "accurate", it can be assumed that the test is 99% accurate for all potential tests made. It's most likely for 99% healthy people to be true negatives, and 99% of sick people to be true positives, as opposed to something ridiculous like it being 99.1% accurate for healthy people, and 0% accurate for sick people, as you said.

>>11641412
>Doesn't follow.
Yes it does.

>>	Anonymous Tue May 5 21:41:29 2020 No.11641430 >>11641175 imagine being this dense I can't

>>	Anonymous Tue May 5 21:45:16 2020 No.11641441 >>11641358 read the thread. statistics is not intuitive

Anonymous Tue May 5 21:51:58 2020 No.11641463

>>11639020
The interpreted response by people is that the test is 99% accurate on positive results. They ignore true negative results.

That’s how you cope. You Bayesians can tailor up a specific question that eludes the interpretative abilities of the common man but it has been shown in many studies that if you ask people about Bayesian problems in regular topics with regular numbers, they intuitively know the answer. Their brain works for the numbers and situations they encounter, this question is confusing more than anything. You don’t need to cope with the fact that normies get this wrong. You need to cope with the fact that Bayesian fags will ask intentionally muddled questions to show their intellectual superior to the common man.

Normal people are rational. Bayesians, not so much.

Anonymous Tue May 5 21:52:12 2020 No.11641465

>>11641429
>Since the question has no further qualifying qualities for "accurate"
None are needed. Accurate is a standard statistical term meaning the proportion of true results. This doesn't tell us the sensitivity or specificity as you claim. That's like saying if a*b = 1 then a and b = 1.

>It's most likely for 99% healthy people to be true negatives, and 99% of sick people to be true positives, as opposed to something ridiculous like it being 99.1% accurate for healthy people, and 0% accurate for sick people, as you said.
How is one "likely" and the other "ridiculous"? They are equally arbitrary.

>Yes it does.
How? Since you seem to know what specificity and sensitivity mean, you should also know that they cannot be derived from accuracy and prevalence. Yet you claim they can be. So either you are talking about a topic you lack a basic understanding of, or you're trolling. Can you clarify which it is?

Anonymous Tue May 5 21:53:35 2020 No.11641473
File: 999 KB, 500x229, learnthispower.gif [View same] [iqdb] [saucenao] [google]

>>11641362
>>11641350
because of the Central Limit Theorem, with that big of a sample you know your sample mean is the same as your population mean.
if it is a truly random sample.
Of course if you are being tested because 50% of your office is dead or because you are coofing its different.

>>	Anonymous Tue May 5 21:54:44 2020 No.11641480 >>11641463 The Bayesians are wrong too though, since they interpret accuracy as P(positive test\|positive) because it is expedient for solving with Bayes theorem. But that is not how accuracy is defined in statistics and diagnostic testing.

>>	Anonymous Tue May 5 21:56:08 2020 No.11641482 >>11641480 nobody is looking up a definition of accuracy besides retarded statsfags. the question was designed to be confusing

Anonymous Tue May 5 21:56:56 2020 No.11641488

>>11641473
>because of the Central Limit Theorem, with that big of a sample you know your sample mean is the same as your population mean.
This has nothing to do with what we're talking about.

Whether a test's specifity and sensitivity differ is a property of the test itself, not sample size or sample bias.

>>	Anonymous Tue May 5 21:57:10 2020 No.11641489 >>11641465 https://helpautism.org

>>	Anonymous Tue May 5 21:58:18 2020 No.11641496 >>11641482 >nobody is looking up a definition of accuracy besides retarded statsfags. So not only did you get the wrong answer, you're proud to be wrong. Nice anti-intellectualism, brainlet.

>>	Anonymous Tue May 5 21:59:20 2020 No.11641504 >>11641489 >>11641368

>>	Anonymous Tue May 5 21:59:58 2020 No.11641509 >>11641463 >doubt even /sci/ fags are getting this wrong Please frame the question in a way "normal people" would understand?

Anonymous Tue May 5 22:00:52 2020 No.11641514

>>11641465
>None are needed.
Well, obviously they are as it's otherwise ambiguous. You're given an (effectively) infinite range of values from which to choose the specificity and sensitivity.

The one that makes the most "sense" in the context of the problem, and the one that makes the least assumptions, is that the test's accuracy is simply independent of the health of the person being tested.

Consider this: your assertion would hold true if you randomly tested people from the entire population, but what if you tested the same person repeatedly? With your spread of false results, after repeatedly testing the person, the resulting accuracy would not be 99%.

With the assumption that the specificity and sensitivity are both 99%, however, it doesn't matter if you test one person repeatedly or a random person, or anything inbetween; the accuracy will always trend towards 99%.

>How is one "likely" and the other "ridiculous"? They are equally arbitrary.
See above.

>So either you are talking about a topic you lack a basic understanding of, or you're trolling. Can you clarify which it is?
Neither; it's making due with an imperfect question and choosing, out of all reasonable possibilities, the assumptions that are simplest.

>>	Anonymous Tue May 5 22:00:59 2020 No.11641516 >>11641488 look who I linked to retard, I never linked to the specifity and sensitivity posts

Anonymous Tue May 5 22:04:32 2020 No.11641525

>>11639784
Yeah I misread OPs image as saying P(I) = 0.01, rather than 0.001
>why use Bayes theorem when you already understand that these probabilities are ratios of TP, FP, TN, and FN?
I guess you could intuit that probability of being infected given a false test is TP / (TP + FP), but you're still using Bayes theorem there. You've just intuited the right formula which is Bayes theorem.

>>	Anonymous Tue May 5 22:13:17 2020 No.11641550 >This shit could all be fixed if someone actually fucking defined accuracy unambiguously but this question will just keep getting reposted every couple of days with numbers changed and no real fix

Anonymous Tue May 5 22:15:18 2020 No.11641551

>>11641514
>Well, obviously they are as it's otherwise ambiguous.
How is it ambiguous when it has a specific definition in the context of diagnostic testing? It's the opposite of ambiguous.

>The one that makes the most "sense" in the context of the problem, and the one that makes the least assumptions, is that the test's accuracy is simply independent of the health of the person being tested.
The one that makes the least assumptions is to not assume at all. And I don't see how this makes "the most sense" since specificity and sensitivity are not often equal in real life. There are many assumptions that can be phrased as a symmetry. There is no preferential one.

>Consider this: your assertion would hold true if you randomly tested people from the entire population, but what if you tested the same person repeatedly?
It would only change the accuracy if you actually knew whether the person was sick or not. And your argument is based on accuracy being a specific number too, so what is even the point of this thought experiment?

>With the assumption that the specificity and sensitivity are both 99%, however, it doesn't matter if you test one person repeatedly or a random person, or anything inbetween; the accuracy will always trend towards 99%.
Huh? Sensitivity and specificity don't chasnge regardless of whether they are equal. Accuracy only changes if the prevalance in the tested population changes. So you're not actually making a differentiation between your argument and mine.

>See above.
It's wrong.

>Neither; it's making due with an imperfect question and choosing, out of all reasonable possibilities, the assumptions that are simplest.
You didn't phrase it as an assumption, you said 'accuracy is 99%, in other words specificity and sensitivity are 99%' But that is clearly false.

>>	Anonymous Tue May 5 22:16:51 2020 No.11641557 >>11641504 relax, help is on the way https://youtu.be/zP-u_Rzoa58?t=45s

Anonymous Tue May 5 22:17:23 2020 No.11641558

>>11641516
You linked to my posts, which are showing that sensitivity and specificity cannot be derived from accuracy. Just because I did not use those words doesn't mean that I wasn't talking about them. Do you understand the difference between a concept and its name?

>>	Anonymous Tue May 5 22:19:27 2020 No.11641560 >>11641551 This isn't a question on diagnostic testing, this is a fucking probability question posted on social media designed for people who aren't meant to be familiar with the terminology to be able to solve it

Anonymous Tue May 5 22:20:26 2020 No.11641566

>>11641525
No it's the opposite, you need to first know how to calculate specificity to plug it into Bayes' theorem. But if you know how to calculate specificity from the information given you also know how to calculate precision without Bayes theorem. In reality, most of the people answering this question mistook accuracy for sensitivity and plugged it in to Bayes theorem, giving them the wrong answer.

Anonymous Tue May 5 22:20:57 2020 No.11641568
File: 12 KB, 316x96, negtive.jpg [View same] [iqdb] [saucenao] [google]

>mah specificity and sensitivity fags
classic example of knowing something without understanding it.
top formula is .99/(.99+.01)
this is GIVEN in the question.
the second formula is practically implied because its a multiple choice question. if there was a E. unknowable with given data. you would have to pick that

>>	Anonymous Tue May 5 22:21:28 2020 No.11641572 >>11641557 >>11641504

>>	Anonymous Tue May 5 22:21:59 2020 No.11641574 >>11641558 >which are showing that sensitivity and specificity cannot be derived from accuracy. wrong

>>	Anonymous Tue May 5 22:22:30 2020 No.11641577 File: 50 KB, 645x729, 1515194851321.png [View same] [iqdb] [saucenao] [google] >>11641560 >This isn't a question on diagnostic testing

Anonymous Tue May 5 22:23:05 2020 No.11641578

>>11641551
>How is it ambiguous when it has a specific definition in the context of diagnostic testing? It's the opposite of ambiguous.
It's ambiguous in that both of our interpretations, and many more (as you have pointed out) can result in the same overall accuracy. That's why it's ambiguous.

>The one that makes the least assumptions is to not assume at all.
We're both assuming here, and an assumption is required to meaningfully answer the question.

>It would only change the accuracy if you actually knew whether the person was sick or not.
Which is exactly the point. The accuracy would still, in the long run, not tend towards 99%.

>And your argument is based on accuracy being a specific number too
Because it's defined as being such in the question, anon.

>Huh? Sensitivity and specificity don't chasnge regardless of whether they are equal.
I never said they did?

>It's wrong.
No you.

>You didn't phrase it as an assumption, you said 'accuracy is 99%, in other words specificity and sensitivity are 99%' But that is clearly false.
It's not "clearly false", it's an assumption made in order to solve the problem. It's just a far more reasonable assumption than yours.

>>	Anonymous Tue May 5 22:24:38 2020 No.11641586 >>11641568 >top formula is .99/(.99+.01) >this is GIVEN in the question. Wrong. Explain your reasoning so I can have a laugh.

Anonymous Tue May 5 22:26:51 2020 No.11641595

>>11641574
>wrong
Wrong. Those posts show specificity and sensitivity can be in a range of values while accuracy is 99%. I'm sorry you're too dumb to understand such simple concepts that you need everything explicitly labeled for you, but that's not my problem.

>>	Anonymous Tue May 5 22:28:00 2020 No.11641601 >>11641586 accuracy=true positives. true negatives of population=0.99=true negatives of random sample

>>	Anonymous Tue May 5 22:30:43 2020 No.11641614 File: 8 KB, 629x504, roc.png [View same] [iqdb] [saucenao] [google] >>11641586 >>11641595 duggan kruger is strong with you. what do you think Accuracy means, so I can laugh at you

>>	Anonymous Tue May 5 22:34:05 2020 No.11641623 File: 10 KB, 300x258, ROC curves for tests A and B.gif [View same] [iqdb] [saucenao] [google] Just a friendly question for the specificity and sensitivity fag. Which test is better for a screening test for a large portion of the population?

Anonymous Tue May 5 22:39:44 2020 No.11641635

>>11641578
>It's ambiguous in that both of our interpretations, and many more (as you have pointed out) can result in the same overall accuracy.
I didn't say the answer is unambiguous, I said the question is unambiguous. You claimed the meaning of "accurate" needed to be further qualified when it's already as specific as it can be.

>We're both assuming here, and an assumption is required to meaningfully answer the question.
What have I assumed? That the answer lies in a small range of values is meaning enough for me.

>Which is exactly the point. The accuracy would still, in the long run, not tend towards 99%.
No, the accuracy wouldn't change simply by you doing the test repeatedly. The only way the accuracy changes is if the prevalence in the population changes.

>Because it's defined as being such in the question, anon.
Yes, so you agree that your argument is an irrelevancy?

>I never said they did?
Then would accuracy change if they remain the same in both cases?

>It's not "clearly false", it's an assumption made in order to solve the problem.
Saying that 99% accuracy -> 99% sensitivity is a false assumption. Saying sensitivity = specificity is an arbitrary assumption. Neither are necessary to solve the problem, because a range of values or indeterminate are still solutions.

>It's just a far more reasonable assumption than yours.
You still have failed to explain my assumption.

>>	Anonymous Tue May 5 22:40:46 2020 No.11641639 >>11641601 >accuracy=true positives. Wrong. LOL

>>	Anonymous Tue May 5 22:41:48 2020 No.11641642 >>11641614 It means the proportion of true results. It's so simple even a brainlet like you can understand.

>>	Anonymous Tue May 5 22:44:25 2020 No.11641646 >>11639232 1% of the people that tested negative duh so 9,990

Anonymous Tue May 5 22:49:01 2020 No.11641655

>>11641639
>Accuracy = (sensitivity) (prevalence) + (specificity) (1 - prevalence). The **numerical value** of accuracy represents the proportion of true positive results (both true positive and true negative) in the selected population. An accuracy of 99% of times the test result is accurate, regardless positive or negative.
http://www.cpdm.ufpr.br/documentos/ROC.pdf
where is your source fag?

>>	Anonymous Tue May 5 22:49:37 2020 No.11641657 >>11641646 >1% of the people that tested negative duh Nothing in the problem implies that.

>>	Anonymous Tue May 5 22:51:31 2020 No.11641662 >>11641655 me again, let me just add before you rage, given accuracy you can assume sensitivity. HOWEVER given sensitivity you CANNOT assume accuracy. that might be where you are confused

>>	Anonymous Tue May 5 22:53:13 2020 No.11641672 >>11641623 Gee I wonder, which is better, having more sensitivity per false positives, or less? Wow, you really separated the men from the boys with that one.

>>	Anonymous Tue May 5 22:56:55 2020 No.11641686 >>11641655 Your own source proves me right, retard: Accuracy = (TN + TP)/(TN+TP+FN+FP) = (Number of correct assessments)/Number of all assessments) True results, not just true postives.

>>	Anonymous Tue May 5 22:58:21 2020 No.11641693 >>11641642 which in op is 99%/1-.99=99% because .99=is true negatives. true positives/true negatives=proportion of true results

Anonymous Tue May 5 23:00:35 2020 No.11641700

>>11641686
keep reading, you are only 2 paragraphs away
>In addition to the equation show above,
>>>>>>>>>>>accuracy can be determined from sensitivity and specificity, whereprevalence is known. **Prevalence is the probability of disease in the population*** at a given time:Accuracy = (sensitivity)(prevalence) + (specificity)(1 - prevalence

>>	Anonymous Tue May 5 23:02:16 2020 No.11641707 >>11641662 >me again, let me just add before you rage, given accuracy you can assume sensitivity. Totally wrong. Read your own source.

Anonymous Tue May 5 23:04:00 2020 No.11641711

>>11639020
Insufficient information. You failed to explain how accurate it is in false positives and false negatives. You simply say it's 99% accurate. Does it give false positives 1% of the time or 0? Does it give false negatives 1% of the time or 0? A test can be 99% accurate have a 0% false positive rate and a 1% false negative rate therefore giving you a 100% chance of being infected with a positive score. You failed to provide that information with your badly written question.

For further information on the subject that's easy to understand.
https://www.youtube.com/watch?v=M8xlOm2wPAA

Anonymous Tue May 5 23:05:04 2020 No.11641716

>>11641693
>which in op is 99%/1-.99=99%
OP simply says it's 99%. No need to make up numbers to calculate it.

>because .99=is true negatives.
Why?

>true positives/true negatives=proportion of true results
No, (true positives+true negatives)(results) = proportion of true results.

You have no clue what you're talking about.

>>	Anonymous Tue May 5 23:07:13 2020 No.11641723 >>11641657 We know that the test is 99% accurate. We also know that 0.1% are infected and 99.9% test negative. 99% of these 99.9% negative outcomes are true 0.99 x 0.999 = 0.98901

Anonymous Tue May 5 23:07:17 2020 No.11641724

>>11641700
None of this disagrees with anything I said. None of it implies anything you said. You Dunning-Krugered yourself, retard.

Please explain, using math, how any of this shows accuracy = true positives.

Or you know, you could just read your own source, which disproves that.

>>	Anonymous Tue May 5 23:08:58 2020 No.11641728 >>11641723 >and 99.9% test negative Wrong. 99.9% ARE negative. You can test negative without being negative, it's called a false negative. >99% of these 99.9% negative outcomes are true Wrong. You're confusing accuracy with specificity.

>>	Anonymous Tue May 5 23:11:40 2020 No.11641739 >>11641362 >/sci/fag can't comprehend sets

>>	Anonymous Tue May 5 23:14:30 2020 No.11641751 >>11641739 >retard can't read a post

Anonymous Tue May 5 23:21:55 2020 No.11641780

>>11641707
yes you can because you can derive it with a somewhat janky system of equations. by plugging in the prevalence=.01
which gives you only a few answers, without dropping values to 0 or 1 the only answers left are Sensitivity = 99/100 Specificity=99/100

>>	Anonymous Tue May 5 23:23:13 2020 No.11641781 >>11641716 >because .99=is true negatives. >>Why? BECAUSE THE PREVALENCE IS .1

>>	Anonymous Tue May 5 23:28:12 2020 No.11641797 >>11641728 You're right, 1% of the total got wrong results and out of that 99.9% are the false negatives. If it was more or less then the accuracy of the test wouldn't be 99%

Anonymous Tue May 5 23:28:58 2020 No.11641800

>>11641780
>yes you can because you can derive it with a somewhat janky system of equations. by plugging in the prevalence=.01
>which gives you only a few answers
Jesus Christ you're dumb.

0.99 = 0.001(sensitivity)+0.999(specificity)

Wow, amazing, a two variable equation. There are infinite solutions. Thanks for destroying your own argument.

>>11641781
Doesn't follow. Go back to elementary school.

>>	Anonymous Tue May 5 23:28:59 2020 No.11641802 >>11641724 >Please explain, using math a=tp/TP+FN), b=tn/(tn+fp), c=(tn+tp)/(tn+tp+fp+fn), c=(a.01)+(b(1-.01)) A=Sensitivity b=Specificity C=Accuracy

>>	Anonymous Tue May 5 23:30:12 2020 No.11641805 >>11641797 >and out of that 99.9% are the false negatives. Nothing in the problem implies that.

>>	Anonymous Tue May 5 23:32:04 2020 No.11641808 >>11641802 Where did you show accuracy = true positives? Are you illiterate?

>>	Anonymous Tue May 5 23:32:41 2020 No.11641811 >>11641805 >A test for it is 99% accurate

>>	Anonymous Tue May 5 23:32:55 2020 No.11641812 >>11641800 what do you think true negative means?

>>	Anonymous Tue May 5 23:33:36 2020 No.11641814 >>11641811 That shows >1% of the total got wrong results It doesn't show >out of that 99.9% are the false negatives

>>	Anonymous Tue May 5 23:35:16 2020 No.11641823 >>11641812 A healthy person getting a negative result. Just read >>11639138 since you're severely confused.

>>	Anonymous Tue May 5 23:38:32 2020 No.11641837 >>11641823 >A healthy person getting a negative result. which trends toward the same as the prevalence rate and sense the question is asking about probability=answer

>>	Anonymous Tue May 5 23:39:34 2020 No.11641841 >>11641808 the answer is left as an exercise for the retard

>>	Anonymous Tue May 5 23:42:09 2020 No.11641849 >>11641814 If 100% of the "wrong" results are positive/negative, then the test isn't 99% accurate

>>	Anonymous Tue May 5 23:43:12 2020 No.11641852 >>11639020 Apply Baye's Rule

>>	Anonymous Tue May 5 23:49:14 2020 No.11641873 >>11641808 just plug it into a equation solver, and set c=.99 there are only a few answers and you can discard ones that set everything to 0 or 1

Anonymous Tue May 5 23:50:35 2020 No.11641878

>>11641837
What the fuck is this gibberish supposed to mean? The rate of true negatives is not even close to the prevalence rate. It's between 0.99 and 0.989.

>>11641841
Retard.

>>11641849
>If 100% of the "wrong" results are positive/negative, then the test isn't 99% accurate
Completely wrong. If 100% of the false results are false positives, then the false positive rate is 0.01, the false negative rate is 0, the true positive rate is 0.001 and the true negative rate is 0.989.

Accuracy = TPR+TNR = 0.99

>>11641852
And get the wrong answer.

>>	Anonymous Tue May 5 23:52:10 2020 No.11641884 also I said accuracy true positives not accuracy =true positives

>>	Anonymous Tue May 5 23:54:04 2020 No.11641893 >>11641884 >>11641601

Anonymous Tue May 5 23:59:01 2020 No.11641899

>>11641878
>What the fuck is this gibberish supposed to mean
take more stats classes.
>The rate of true negatives is not even close to the prevalence rate.
its the same fucking thing. if you test the entire population
1-the prevalence rate=true negative rate
by definition. Otherwise its not a true negative rate
if you sample someone *the probability* of a true negative rate equals the same due to central limit theorem

>>	Anonymous Wed May 6 00:09:04 2020 No.11641916 >>11641802 >>11641873 also you can set tn+tp+fp+fn=1

Anonymous Wed May 6 00:09:05 2020 No.11641917

>>11641899
>take more stats classes.
You've never taken any, and if you have you didn't learn anything.

>its the same fucking thing.
It's not, you mongoloid. Look at >>11639138

The TN rate is 0.99-x which is between 0.989 and 0.99.

>1-the prevalence rate=true negative rate
Still wrong, retard.

TNR = Accuracy-TPR = 1-Prevalence-FPR

>>	Anonymous Wed May 6 00:10:07 2020 No.11641922 >>11641916 I did already, you're wrong: >>11639138

>>	Anonymous Wed May 6 00:11:03 2020 No.11641923 >>11641917 >>1-the prevalence rate=true negative rate >Still wrong, retard. this is the stupidest thing you have said

>>	Anonymous Wed May 6 00:12:38 2020 No.11641928 >>11641922 try again because you are wrong, you left out the fact we know the the prevalence rate which gives a answer to the accuracy rate in terms of specificity and sensitivity

>>	Anonymous Wed May 6 00:12:41 2020 No.11641929 >>11641923 You can't even quote correctly. TNR = 1-Prevalence-FPR You lose. Game over. Stay retarded.

>>	Anonymous Wed May 6 00:13:07 2020 No.11641932 >>11641814 >>11641878 i get it now, i swear this is the second time i fall into that trap

>>	Anonymous Wed May 6 00:13:57 2020 No.11641934 >>11641928 Irrelevant. Specificity and sensitivity aren't known. Fuck off.

>>	Anonymous Wed May 6 00:14:40 2020 No.11641935 >>11641922 where the fuck do you get Accuracy = TP+TN?

Anonymous Wed May 6 00:15:18 2020 No.11641939

is this right?

if it affects 0.1% then it affects 1/1000 people which is 0.001 as a decimal or 0.1%

lets imagine there are 100k people on the earth.

There should be 100 people who have the virus then since 0.1% of 100k is 100

The test is 99/100 accurate and 1/100 borked.

This means if you were tested 100 times for this statistic to be uncovered then 99 times you were positive and 1 times you were negative.

So 100/100k have the virus and we assume all 100k are tested.

1% of 99,900 (does not include real positives) will test erroneously positive due to the error rate of 1%. This is 999 people who are then positive who are actually negative.

99% of 99,900 will test correctly negative, so 98,901.

Aditionally, only 99% of the real positives will be positive and moved to the positive group, and one positive person will move to the negative group.

There are now 999 false positives and 99 true positive people in the positive pool of 1098 where only 100 should be positive according to 0.1% infection rate.

So 99 (real positive)/1098 (total positive cases) will show that if you are tested positive then there is only 9% chance that your positive result is a real positive.

>>	Anonymous Wed May 6 00:15:41 2020 No.11641940 >>11641934 see >>11641802 loser, you failed again kid

>>	Anonymous Wed May 6 00:16:09 2020 No.11641942 >>11641935 Accuracy is the rate of true results over results. TP and TN are the rates of true results over results. Do the math.

>>	Anonymous Wed May 6 00:16:43 2020 No.11641944 >>11641929 haha, I already clsoed the thread, you are dumb, you know answers but dont understand why, its embarssing

>>	Anonymous Wed May 6 00:17:58 2020 No.11641948 >>11641940 I saw it, it says sensitivity and specificity are in a linear equation with two variables. There are infinite solutions, so they're unknown. Fucking retard.

>>	Anonymous Wed May 6 00:19:00 2020 No.11641949 >>11641939 >This means if you were tested 100 times for this statistic to be uncovered then 99 times you were positive and 1 times you were negative. Wrong.

>>	Anonymous Wed May 6 00:19:43 2020 No.11641952 >>11641939 This is the correct answer. The question is slightly off because it uses the term "accuracy", when it should have honestly just said "the test is wrong 1% of the time". Same general meaning, but less autists would get hung up on it.

>>	Anonymous Wed May 6 00:20:02 2020 No.11641953 >>11641944 No argument left, thanks for admitting you're retarded.

>>	Anonymous Wed May 6 00:20:11 2020 No.11641954 It's D. How do you all not fucking get this? Is this bait?

>>	Anonymous Wed May 6 00:20:38 2020 No.11641958 >>11641942 you simplified wrong. some 3 correct answers 1.)Accuracy = (TN + TP)/(TN+TP+FN+FP) 2.)(Number of correct assessments)/(Number of all assessments) 3.)TP/1 what you did was: TP+TN/1 or TP+TN close but oh so wrong

>>	Anonymous Wed May 6 00:21:05 2020 No.11641961 >>11641954 Smoothbrain

>>	Anonymous Wed May 6 00:21:30 2020 No.11641962 >>11641961 You are braindead. Kill yourself.

>>	Anonymous Wed May 6 00:21:52 2020 No.11641965 >>11641952 >This is the correct answer. It's wrong. >The question is slightly off because it uses the term "accuracy", when it should have honestly just said "the test is wrong 1% of the time". Those are exactly the same thing.

>>	Anonymous Wed May 6 00:23:07 2020 No.11641967 >>11641953 you are the one you admitted were wrong. >>11641948 all that shows me is you didnt try it afried of learning something? and its not two variables

Anonymous Wed May 6 00:24:46 2020 No.11641969

>>11641958
>1.)Accuracy = (TN + TP)/(TN+TP+FN+FP)
Correct.

>2.)(Number of correct assessments)/(Number of all assessments)
Correct.

>3.)TP/1
Wrong.

>TP+TN
>close but oh so wrong
Not wrong, TP and TN were defined as rates in my post instead of number of tests. You once again prove that you're an illiterate mongoloid.

Anonymous Wed May 6 00:25:11 2020 No.11641971

>>11641962
>t. retard who flunked out of highschool stats

>>11641965
>It's wrong.
Nope

>Those are exactly the same thing.
The latter implies the test has a chance of being 1% wrong on every test, independent of anything else. The former is just bait for spergs like you.

>>	Anonymous Wed May 6 00:26:36 2020 No.11641974 >>11641967 >you are the one you admitted were wrong. Learn English, retard. >all that shows me is you didnt try it I did. >and its not two variables a and b are variables. You never solved for them. You're pathetic.

Anonymous Wed May 6 00:26:55 2020 No.11641976

>>11641971
>t. retard who flunked out of highschool stats
If I showed your posts to the CEO of a cancer charity he would shut down all operations.
If you test positive, and the test is 99% accurate, you are 99% infected. The population infected doesn't matter because you're already infected.

>>	Anonymous Wed May 6 00:28:33 2020 No.11641979 >>11641971 >Nope See >>11639138, you're wrong. >The latter implies the test has a chance of being 1% wrong on every test, independent of anything else No, it implies that the test is wrong 1% of the time.

Anonymous Wed May 6 00:34:03 2020 No.11641987

>>11639278
god you are confused.
TP+FP+TN+FN =1
because the range of probabilistics is 0to1, ratios have nothing to do with it.
TP,FP,TN,FN are all subsets of TP+FP+TN+FN (some of which may equal 0)
>>11639278
>TP, TN and so on are not ratios but actual case numbers.
yes, which are expressed as ratios to total case numbers.

Anonymous Wed May 6 00:34:25 2020 No.11641988

>>11641976
>If you test positive, and the test is 99% accurate, you are 99% infected. The population infected doesn't matter because you're already infected.
Holy shit, imagine being this genuinely disabled.

>>11641979
>No, it implies that the test is wrong 1% of the time.
No, it implies that you have a developmental disorder

>>	Anonymous Wed May 6 00:35:21 2020 No.11641989 >>11641976 Accuracy just means the percentage of true test results. We want the percentage of true test results among positive tests.

>>	Anonymous Wed May 6 00:37:09 2020 No.11641991 >>11641988 >No, it implies that you have a developmental disorder Says the guy who can't even read.

>>	Anonymous Wed May 6 00:38:59 2020 No.11641992 >>11641991 Hey man, I'm not the one a few convolutions short of a cortex.

>>	Anonymous Wed May 6 00:40:45 2020 No.11641993 >>11641979 >Accuracy = TP+TN is still wrong >.001+.99-x/.001+(.99-x)+(0.001-x)+(0.001-x) lol, try again

>>	Anonymous Wed May 6 00:43:55 2020 No.11642003 >>11641993 >retard can't even reply to the right post. >is still wrong Because? Use your words like a big boy. >lol, try again How the fuck do you fail at quoting? Just copy and paste you inbred assclown.

>>	Anonymous Wed May 6 00:44:36 2020 No.11642007 >>11641979 infection rate equals .01 stop using TN if you are going to keep impinging it equals zero which is the only way your retarded post you keep referencing works

Anonymous Wed May 6 00:44:37 2020 No.11642008

>>11641965
>>>11641952
>>This is the correct answer.
>It's wrong.
>
>>The question is slightly off because it uses the term "accuracy", when it should have honestly just said "the test is wrong 1% of the time".
>Those are exactly the same thing.

It matches this medical diagnostic test calculation.
https://www.medcalc.org/calc/diagnostic_test.php

>>	Anonymous Wed May 6 00:45:37 2020 No.11642009 >>11642003 I'm not referencing your wrong post near the top any more because it already has too many samefag (yous) from you

Anonymous Wed May 6 00:47:41 2020 No.11642012

>>11642007
>infection rate equals .01
0.001. Can't even write a decimal correctly huh?

>stop using TN if you are going to keep impinging it equals zero which is the only way your retarded post you keep referencing works
Where did I "impinge" it equals 0? Fuck off already, retard. You're late for the short bus.

>>	Anonymous Wed May 6 00:49:22 2020 No.11642015 >>11642009 Good, you have no argument against it so you admit you're a retarded hack.

>>	Anonymous Wed May 6 00:52:21 2020 No.11642022 >>11642012 >Where did I "impinge" it equals 0 when you said >Accuracy = TP+TN

Anonymous Wed May 6 00:57:02 2020 No.11642029
File: 6 KB, 183x275, brainlets.jpg [View same] [iqdb] [saucenao] [google]

>>11642012
Accuracy = (TN + TP)/(TN+TP+FN+FP) = (Number of correct assessments)/Number of all assessments
lets test 10 people slow class
TN,TP,FN,FP=1,2,3,4
1+2/1+2+3+4
= .3
not
3

>>	Anonymous Wed May 6 01:00:43 2020 No.11642033 a a refresher for the slow class (x+y+z)/(y+z)=x/(y+z) + y/(y+z) + z/(y+z)=x/(y+z) +1

>>	Anonymous Wed May 6 01:05:15 2020 No.11642044 > Be retarded OP > Put up problem on /sci/ > People answer according to what the question asks, not what I meant to ask > Be mad Stay mad retard

Anonymous Wed May 6 01:08:01 2020 No.11642046

>>11639211
>This doesn't follow
yes it does. because we are given the Accuracy rate
>0 sick people test sick and 99000 healthy people test healthy, then the test is still 99% 0 sick people test sick and 99000 healthy people test healthy, then the test is still 99% accurate. .
wrong accurate implys--->specificity rate
if accurates is right the other follows or you divide by zero, (but not visveras

>>	Anonymous Wed May 6 01:15:11 2020 No.11642060 File: 118 KB, 1100x792, eidograph.jpg [View same] [iqdb] [saucenao] [google] looks like the mad kid finally left

>>	Anonymous Wed May 6 01:16:29 2020 No.11642062 >>11639028 This

Anonymous Wed May 6 01:36:32 2020 No.11642105

>>11639020
What you guys need to understand is screening tests become useless when a very small proportion of the population has a infection because the false positives are going to be way higher then the false negatives, even with a very high Positive-predictive-value.

>>	Anonymous Wed May 6 01:43:13 2020 No.11642120 >>11639020 B

>>	Anonymous Wed May 6 02:00:12 2020 No.11642146 >>11641358 If COVID-19 affects 0% of the population and the test is 99% accurate, what are the odds that you're infected if you test positive?

>>	Anonymous Wed May 6 02:00:56 2020 No.11642148 >>11639913 [math] d\ , b\ , h [/math]

>>	Anonymous Wed May 6 02:41:10 2020 No.11642219 Covid tests have no false positives though.

>>	Anonymous Wed May 6 02:59:47 2020 No.11642238 >>11642219 big if true

Anonymous Wed May 6 04:02:48 2020 No.11642299

>>11642219
of course
no false negatives either
you must accept the covid decision without complaint
it's all for your own good
stay confined in cellar
all opposition is futile
covid tests and authorities are infallible
numerical models have perfect precision
the best decisions ever are taken

>>	Anonymous Wed May 6 04:56:35 2020 No.11642339 File: 221 KB, 285x450, flag.png [View same] [iqdb] [saucenao] [google] >>11642299

>>	Anonymous Wed May 6 04:58:17 2020 No.11642345 File: 32 KB, 1298x539, n.png [View same] [iqdb] [saucenao] [google]

Anonymous Wed May 6 04:58:31 2020 No.11642347

>>11639123
>And if 0% of the population was affected?

If you know that 0% of the population can be infected then you cannot assign a sensitivity to the (or any) test in question of being 99%.

If a test has been validated such that for every person tested it will be correct 99% of the time (if they are infected) then it makes no difference what percentage of the population can potentially be infected - and don't use that 0% nonsense because then the test could not have been assigned a positivity value in the first place.

Anonymous Wed May 6 05:01:07 2020 No.11642350

>>11639095
>Because 1% of 99.9% of all people will test positive

Where does that come from? The question establishes that the number of persons afected is a fixed 0.1%. It seems like you're confusing the comparison of two tests where one has a positivity rate of 0.1% with perfect sensitivity.

>>	Anonymous Wed May 6 05:05:16 2020 No.11642357 >>11639020 50%, you either are infected or you are not

>>	Anonymous Wed May 6 05:08:25 2020 No.11642362 >>11642347 >>11642350 take the poop and relax

Anonymous Wed May 6 05:12:53 2020 No.11642367
File: 104 KB, 1280x720, 1586217025184.jpg [View same] [iqdb] [saucenao] [google]

>>11639020

COVID19 affects 0.1% of the population. A test for it is 100% accurate. What is the chance you are infected?

COVID19 affects 100% o the population. A test for it is 0% accurate. What is the chance you are infected?

>>	Anonymous Wed May 6 05:16:24 2020 No.11642370 >>11642367 Given that I was tested positive in both cases, it's 100% in both. In the second, they are simply independent.

>>	Anonymous Wed May 6 05:34:35 2020 No.11642391 the accuracy of the test has nothing to do with the question, the answer is A

Anonymous Wed May 6 05:36:27 2020 No.11642395

>>11639020
ITT. Trolls trolling trolls trolling trolls...its trolls all the way down.

Nevertheless, OP is a stupid sack of shit who cant even word a simple statistical question correctly, while also trying to push an agenda and failing miserably.

OP if you want to convince people that its a nothingburger then dont baffle them with your 3rd world English. Go and ask your English tutor for some advice. Besides anyone with brains already knows its a nothing burger and there is no convincing the lick-boot dummies who buy into this CV19 crap.. All you are doing is demonstrating your own incompetence.

>>	Anonymous Wed May 6 05:39:51 2020 No.11642396 >>11642395 >>11642339

>>	Anonymous Wed May 6 05:47:55 2020 No.11642403 >>11642391 Bayesed frequentist.

Anonymous Wed May 6 05:49:11 2020 No.11642408

>>11642347
>If you know that 0% of the population can be infected then you cannot assign a sensitivity to the (or any) test in question of being 99%.

Sure you can. Maybe the limited accuracy of the test comes from the dude in the factory putting the right reagents in the box. Whether COVID-19 exists in the population or not is irrelevant to the statistical fact that you're guaranteed to be in the 1% of false positives if there are actually no cases in the population. Think about it a little harder - you can do it anon.

>>	Anonymous Wed May 6 05:59:35 2020 No.11642419 >>11639020 https://en.wikipedia.org/wiki/Modus_ponens

>>	Anonymous Wed May 6 07:35:04 2020 No.11642523 >>11642022 TN = 0.99-TP >= 0.99-0.001 = 0.989 So according to you 0 >= 0.989 What a retarded piece of shit.

>>	Anonymous Wed May 6 07:38:22 2020 No.11642531 >>11642029 I defined those as rates, not number of assessments. Try again illiterate mongoloid.

Anonymous Wed May 6 07:41:24 2020 No.11642538

>>11642046
>wrong accurate implys--->specificity rate
It doesn't, you already disproved this by showing specificity has infinite possible values when accuracy and prevalence are known.

>if accurates is right the other follows or you divide by zero, (but not visveras
Take your meds.

>>	Anonymous Wed May 6 07:41:48 2020 No.11642540 These threads are pure troll and bait for anyone wondering wtf is happening there

>>	Anonymous Wed May 6 07:48:37 2020 No.11642562 >>11642367 0.1% and 100% since you never specified we took the test and got a positive result.

>>	Anonymous Wed May 6 08:34:32 2020 No.11642684 >>11641969 >TP+TN wrong, because i defined + as - and - as + in my personal blog, checkmate

>>	Anonymous Wed May 6 08:38:19 2020 No.11642690 >>11642684 >attempts to respond to post >doesn't read it Not my fault, retard.

>>	Anonymous Wed May 6 08:40:24 2020 No.11642699 >>11642690 are you also defining me as the guy you were talking before? slow down with the assumptions

>>	Anonymous Wed May 6 08:43:05 2020 No.11642708 File: 85 KB, 450x658, 1556429937899.jpg [View same] [iqdb] [saucenao] [google] >>11642699 >Inb4 the 25yo assoomer meme

Anonymous Wed May 6 08:52:56 2020 No.11642723

>>11642699
Ah right. Just by coincidence you write like a retard with no capitalization or periods. So either you're butting in to a conversation without understanding its context or you're the same retard who doesn't understand its context because you didn't read the post you're arguing against. I really don't give a fuck.

>>	Anonymous Wed May 6 09:18:54 2020 No.11642770 >>11642146 99%

>>	Anonymous Wed May 6 11:03:37 2020 No.11643017 >>11639060 Going for this desu. If the test already says you're positively infected, what does it matter how much of the population does or doesn't have it? Why is it not just 99% chance you have and 1% you don't?

>>	Anonymous Wed May 6 11:08:48 2020 No.11643030 >>11639123 mu

>>	Anonymous Wed May 6 11:09:30 2020 No.11643032 >>11642708 kek that pic

>>	Anonymous Wed May 6 11:11:11 2020 No.11643035 >>11643017 the 1% failure rate will show up as a LOT of false positives coming from the healthies group

>>	Anonymous Wed May 6 11:20:46 2020 No.11643061 >>11643035 Ah yes, I see what you mean.

>>	Anonymous Wed May 6 11:47:17 2020 No.11643127 >>11643061 see >>11641349

>>	Anonymous Wed May 6 11:59:50 2020 No.11643148 >>11643127 I actually did get it the first time, but the numbers do put things into perspective

Advanced search
Text to find
Subject [?]Search by post subject. Leave empty for any.
Username [?]Search for user name. Leave empty for any user name.
Tripcode [?]Search for tripcode. Leave empty for any.
Email [?]Search by email. Leave empty for any.
Filename [?]Search by image filename. Leave empty for any.
From Date [?]Enter what date to start searching from. Format is YYYY-MM-DD
To Date [?]Enter what date to start searching until. Format is YYYY-MM-DD
Image hash
Search in	All Posts OPs Only
Deleted posts	Show all posts Show only deleted posts Only show non-deleted posts
Internal posts	Show all posts Show only internal posts Show only archived posts
Order	New posts first Old posts first
Capcode	All Posts Only by Users Only by Mods Only by Admins Only by Developers
Results	Posts Threads
Action	[ Simple ]

/sci/ - Science & Math