What's a reasonable evidence standard (alpha) using NHST to require for claims of interactions for ordinary individual-level research? (I.e. more data relatively easy to collect). Assume that we switch to the new standard of .005 for main effects. #statistics #replicationcrisis
-
My doubt here is: Assume the interaction is fake news. Then, lowering p will mean (per Gelman) that when you do detect an effect, it will be misleadingly large. If it is true, but small, then lowering p might lead to not detecting it. Its trade offs all the way down
Most variation is due to main effects, so we won't be losing much detail. Main effects are the way to bet because the others are too small to find anyway. Similar to the bet on sparsity. https://web.stanford.edu/~hastie/StatLearnSparsity/ …pic.twitter.com/jf9M9YU0Cw
-
-
I mean, I don't disagree!
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.