'There's A Math For That' - The Paradox Of The False Positive

Mathematics is precisely the right language to describe and understand the world around us. In this series, we explore a few instances of how that works. Watch as King's College London Mathematics School Head Designate Dan Abramson explain the real world using mathematics.
Find out more about the King's College London Mathematics School: www.kcl.ac.uk/mathsschool/Home...

[@aztmln]

Great video. Very insightful. Thanks for putting together and sharing/ educating us all .

[@marcusb1704]

What a SUPERB presentation - concise yet crystal clear. Friendly and enthusiastic as well. No wonder Dan Abramson is Head Teacher at the new(ish) King's Maths School. Thank you, Dan!

[@earlgreytea633]

I'm doing my Math IA on this and the video was extremely helpful! Thank you!

[@earlgreytea633]

This video deserves more recognition. I also love the energy put into the video

[@TheGuardian163]

kingscollegelondon I wasn't looking for this but I found it to be amazing great video!

[@rahulsps2009qwerty]

amazing (y) As a med student , I found this very helpful.

[@hamletghukasyan6357]

the best method to explanain the positive predictive values

[@DanielVerberne]

Excellent presentation, very well presented and engaging. As a non-scientist and definitely non-mathematician, my first impulse would be to think that increasing the sample size (number of patients, number of samples, whatever the case may be) would ultimately serve provide ever-greater confirmation of efficacy of a given test, but it seems here that the opposite is true, that the sample size ITSELF leads to the seemingly-odd result.

[@Luiferhoyos]

Hello! It's not about the sample size. If you repeat the exercise using any other sample size, the rate of false positives will be the same. The problem is the low prevalence of the diseases. In this example, if the prevalence were 20% instead of 1%, the rate of true positives would go from 30% to 84%. But, as the prevalence can't be changed, we can only play with specificity and sensitivity. Sensitivity doesn't either impact the correctness of positive tests, as it only affects positive individuals (who have little impact on the final result, as they represent a 1%). Increasing the sensitivity from 97 to 99.9% would practically have zero effect on the rate of true positives. The only way is to increase the specificity. If it could be 99%, the rate of true positives would change from 1/3 to 1/2. I created this Excel which you can play with we.tl/t-7LzyAG8esU

[@DanielVerberne]

@@Luiferhoyos I appreciate the explanation, Luis.

[@tint0rnet]

I liked that, it was pretty cool. Numbers are great

[@Kori-ko]

The music is obnoxiously loud in the beginning, and I can't make out what you're saying.

[@madispokemon193]

I’m just a kid and Trying to learn paradoxes 😂

[@edposs8483]

GURU

[@Skufed]

I commented just to order you to change the title to "there's a maths for that".

[@Grim_Beard]

I'm seeing a lot of people on social media clamouring for 'more testing!' for the Covid-19 coronavirus right now. I'm sharing this video as widely as I can to explain why that's a bad idea.

[@reggaefan2700]

Just don't get tested yourself.

[@Grim_Beard]

@@reggaefan2700 That reduces the _n_ of inappropriate tests by 1. Avoiding this problem requires it to be considered in government strategy, not individual actions.

[@reggaefan2700]

Why don't you go work for the government then and formulate public policy?

[@Grim_Beard]

@@reggaefan2700 Oh sorry, I didn't realise you were new to this planet. Welcome!

[@reggaefan2700]

@@Grim_Beard Diddo.