David Tannenbaum

Got it, thanks Vlad. Looking forward to the morality measurement paper, will keep an eye out for it!

Congrats on the paper.

Would repeated measures within-groups or within-individuals (such as ipsatized scores) get around the issue?

But I agree with your broader point, that we should care what your theory predicts not what you predict.

As a reviewer/editor, don't you think it's makes more sense to ask for some kind of additional corroboration (such as a replication) in (1) than in (2)?

The only difference is (1) is exploratory and (2) is confirmatory. Prereg helps us distinguish exploratory vs confirmatory.

Consider two scenarios: (1) "We ran a study, did a bunch of analyses, stumbled across this interesting 3-way intx seems consistent with theory X" (2) "We were interested in testing a specific 3-way intx predicted by theory X, and designed a study to test it." Assume the same sample size in both.

A big part of preregistration is to make papers more evaluable for editors and reviewers (e.g., your 'predicted' 3-way interaction becomes a lot more credible if you stated your predictions and analysis plan in advance). Author's professed reason for prereg is somewhat besides the point.

"my brilliance is but a vessel for God's truth"

You look like you are being held hostage in that Rady video.

Also pretty cowardly that they won’t put their name on their work. If you don’t think your study was unethical, then fully own up to it.

Memorable one from Herb Clark:

"Half of doing science is writing, so until you write well, you will never be more than half a scientist."

I'm sorry Margaret. Couldn't agree more that dogs are incredible and not sure we deserve them. Rest in peace Henry, may you run forever in the grassy fields of doggy heaven.

Birds aren't real, Quentin

YouTube video by Snazzy Labs Apple's Most Jam-Packed Keynote Ever!

One person who anticipated some of this was @snazzylabs.com. Check out this video (around the 20:30 mark) www.youtube.com/watch?v=C0Yu...

Why I'm quitting the Washington Post Democracy can't function without a free press

This is, or should be, a big deal -- Ann Telnaes, an absolute legend, quitting WaPo due to censorship of a cartoon criticizing billionaires

anntelnaes.substack.com/p/why-im-qui...

Using Conjoint Analysis to Detect Discrimination: Revealing Covert Preferences From Overt Choices | Social Cognition In an effort to continue the development of methods to understand social cognition, we adopt a technique called conjoint analysis that mathematically deduces preferences from the implied tradeoffs peo...

Here’s an interesting application. Not sure if the data is publicly available but you could try reaching out to the authors.

guilfordjournals.com/doi/abs/10.1...

As for terminology, you could call it a form of rent seeking behavior.

To your point, UT Dallas business school created a journal list to rank different b-schools by research productivity. It just so happens that UT Dallas ranks #2 on that list.

ah, I see

In Figure 5, all of the t_-1 & t_0 correlations are identical to the corresponding t_0 & t_+1 correlation. That has to be an error, no?

Ahem

Can someone make a starter pack of all these starter packs?

Correct. Or at the very least, you’d have to do something else, such as use GLS instead of OLS (RE model instead of FE) or exclude the FE terms and recognize that they are confounded with your treatment

I would also add "normally fine so long as your FEs are not perfectly collinear with your predictor of interest". Otherwise the FEs that are dropped (due to collinearity) are going to get absorbed by your predictor and screw everything up.

Not sure I follow the last part… how do open science principles come “at the expense” of creativity and exploration?

A command-line interface for designing/writing a study.

Would love to be able to edit text scripts, rather than a janky graphical interface with a slow back-end that is constantly stalling out.

Sure. But most studies are time stamped, and you’ve now has created a paper trail with the pre-reg.

“A bizarre failure in the review process at PNAS”
statmodeling.stat.columbia.edu/2024/05/16/a...

Love this, thanks for sharing Andrew

www.guidedtrack.com might be worth a look. Doesn't have all the features of Qualtrics, but still fairly capable. Used to be free, looks like they charge now.

If you use gmail, Mimestream is solid