Josh Miller
@jdmiller.bsky.social
Psychology professor interested in personality and psychopathology.
Same - haven’t seen major problems with Prolific and we try to test response validity fairly thoroughly.
November 18, 2025 at 10:13 PM
Same - haven’t seen major problems with Prolific and we try to test response validity fairly thoroughly.
Don’t Give Them Any Ideas, Josh!
November 18, 2025 at 3:38 PM
Don’t Give Them Any Ideas, Josh!
Reposted by Josh Miller
Most studies do not have the samples to test interactions.
They are underpowered.
They yield unstable estimates.
Routine IA testing dumps unreplicable noise into the literature.
We already have too much of that.
So just stop.
8/8
They are underpowered.
They yield unstable estimates.
Routine IA testing dumps unreplicable noise into the literature.
We already have too much of that.
So just stop.
8/8
November 14, 2025 at 6:21 PM
Most studies do not have the samples to test interactions.
They are underpowered.
They yield unstable estimates.
Routine IA testing dumps unreplicable noise into the literature.
We already have too much of that.
So just stop.
8/8
They are underpowered.
They yield unstable estimates.
Routine IA testing dumps unreplicable noise into the literature.
We already have too much of that.
So just stop.
8/8
And be prepared for an ugly cry.
November 12, 2025 at 3:33 PM
And be prepared for an ugly cry.