search

TengMCing / PhD

https://patrick-li-thesis.netlify.app/
1 stars 0 forks source link

Title of the first paper #2

Open TengMCing opened 2 years ago

TengMCing commented 2 years ago

draft: Conventional tests for diagnosing linear regression models are too sensitive (the use of conventional is ambiguous) Why don't people use statistical tests to diagnose the linear regression model? (we haven't measured the proportion of researchers not using statistical tests) Why shouldn't you use statistical tests to diagnose the linear regression model?

visual inference: a better alternative than numerical tests for linear regression model diagnostics

emitanaka commented 1 year ago

A plot is worth a thousands of tests: a study of regression diagnostics with visual inference

srvanderplas commented 1 year ago

Either A plot is worth a thousand tests: ... or A plot is worth thousands of tests: ...

But I like it!

On Tue, Jan 17, 2023 at 7:03 PM Emi Tanaka @.***> wrote:

A plot is worth a thousands of tests: a study of regression diagnostics with visual inference

— Reply to this email directly, view it on GitHub https://github.com/TengMCing/PhD/issues/2#issuecomment-1386311233, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAQHKFJW2QDH3INLK5OVR2DWS46NRANCNFSM5SYGB6SQ . You are receiving this because you were assigned.Message ID: @.***>

dicook commented 1 year ago

On this thread, I've discovered this paper: https://faculty.washington.edu/gloftus/Research/Publications/Manuscript.pdf/Loftus%20p-values%201993.pdf

emitanaka commented 1 year ago

Nice find! I guess an alternative proposition is: "Show, don't tell: uncovering regression diagnostics with visual inference"