Always visualization your data

[shpalman]

Mathematics!

Same Stats, Different Graphs

[Allo V Psycho]

Mathematics!

Same Stats, Different Graphs

Love it!

[sTeamTraen]

To be fair, a correlation of 0.06 is very often uninteresting even before graphing it. And as you increase the correlation it becomes a lot harder to draw pictures.

[jimbob]

To be fair, a correlation of 0.06 is very often uninteresting even before graphing it. And as you increase the correlation it becomes a lot harder to draw pictures.

True, but there are plenty of more straightforward examples that do work

From the NIST/SEMATECH e-Handbook of Statistical Methods, http://www.itl.nist.gov/div898/handbook/ which I point new engineers to when they ask me questions about the background to stats that we use.


https://www.itl.nist.gov/div898/handboo ... /eda16.htm

anscomb4.gif (7.46 KiB) Viewed 769 times

[sTeamTraen]

Results like data set 4 there appear a lot in psychology.

[jimbob]

Results like data set 4 there appear a lot in psychology.

On brand for you, @sTeamTraen

[shpalman]

from the Mocking Religion thread

[jimbob]

from the Mocking Religion thread

Ha,

Don't forget this.

https://twitter.com/mdespriee/status/13 ... 69568?s=20

The person who posted the original graphs deleted his account. It was about phrenology and whether someone seemed trustworthy.

It did inspire me to go fishing and find a stronger correlation than some they were reporting



Yes, that is China is population rank=1 Afghanistan alphabetical rank=1 and yes it is bad.

[Stupidosaurus]

Raaagh! Raaagh! Raaaaaaaaagh!