And here's the updated preprint: doi.org/10.31235/osf...

With @stijnruiter.bsky.social and Wim Bernasco

No, I don't think I do. :)

Not sure what a grown-up discipline is, nor how does a discipline become grown-up, but perhaps severely testing theories with tools like preregistrations and registered reports helps us get there?

I haven’t read the full piece, but the argument in the quote seems to be of the form “some good were produced without this” (leaving unsaid “and quite a lot of bad results, too”).

But the argument for pre-registration isn’t that it leads to good results, it’s that it makes results more trustworthy.

I don't think many people think that either... Preregistration is most useful in confirmatory research, when it's important to know if a hypothesis is being severely tested. Not necessary for exploratory research, if you ask me. It still adds a layer of transparency though.

But from an error statistical philosophy, it "has the goal to allow others to transparently **evaluate the capacity of a test to falsify a prediction, or the severity of a test**" (Lakens, 2019, p. 221).
www.jstage.jst.go.jp/article/sjpr...

And It don't see the author challenging that point.

2/2

Agree. It could be argued that the author is a bit off regarding the main value of pre-registration.

Author: "... preregistration was designed to increase replicability and transparency, and evidence so far suggests it has done neither of those things."

1/2

Try being a Real Zaragoza supporter...

Haven't tried it. Sounds useful!
To me, the most important thing with these AI tools is to understand how they handle the information that you provide (the content of your paper). If data processing is fair and transparent, then good! If it's not or unclear, then burn them with fire.

Should we have a call? :)

We’ve preprinted for a reason: we want feedback. So feel free to reach out, comment, or suggest ways to improve!

Cc: @stijnruiter.bsky.social