LESSWRONG
LW

Comic endorsing Bayesian statistics

by grouchymusicologist

1 min read6th Apr 201113 comments

9

Well, that might be editorializing on my part, but you'll see what I mean ...

Comic endorsing Bayesian statistics

11Vladimir_Nesov

New Comment

13 comments, sorted by

Click to highlight new comments since: Today at 4:15 PM

[-]Sniffnoy13y150

Definitely editorializing on your part.

[-]David_Gerard13y20

Yep.

[-]Alicorn13y10

Definitely

95% confidence? :P

[-]Sniffnoy13y00

Well, OK, if I had to actually assess a probability I'd go with something more like 7/8 or 8/9...

[-]DanielLC13y110

And the moral of the story is to never stop playing Minecraft.

[-]bentarm13y40

You don't have to be a Bayesian to do a correction for multiple comparisons. The scientists in the comic just did their job badly.

Comic endorsing Bonferroni corrections!

[-]jsalvatier13y40

More of a critique of hypothesis testing than an endorsement of Bayes stats. It is a pretty tough criticism though.

[-]Vladimir_Nesov13y110

Come on, it's about journalists' sloppiness in understanding technical statements. What the minecraft people said was certainty correct.

It's the file-drawer problem in comic form.

[-]Pr0methean10y00

Makes me wish initiatives like U of Ottawa's JI-R would turn into real journals that could reliably publish at least quarterly.

[-]HonoreDB13y00

It's an excellent comic in that you can see room for improvement in every step. With better hypothesis generation, we control the number of false positives by controlling the number of false hypotheses--maybe they should have just kept playing Minecraft?

[-]Psy-Kosh13y00

I just came here to post the same comic. :)