What if the true value of preregistration to social science is that on average it leads to a higher powered study?

How, if at all, would this change your views about the replication crisis?

We all know that the childcare service provided for NeurIPS attendees is really a boot camp for 3yo to learn back prop

'The problem is that one can understand the concepts of preregistration or sufficient sample size while still having a relatively simple mental model of effects as real or fake & questionable research practices as the main source of issues'

Kinda crazy how pervasive this view is

Open science narratives can be highly solutionist, implying that cranking up N or preregistering can guarantee that your study will be informative for your purposes.

I use an analogy to helping someone improve their writing to discuss why this bothers me statmodeling.stat.columbia.edu/2024/04/08/peo…

Why is the rational, selfish, fully informed agent THE benchmark in microecon? Because (save for some philosophical quibbles) one can be rational in only one way but irrational in many, selfish in only one way but non-selfish in many, etc. Uniqueness is key for a good benchmark.

Did a paper about the value of open science practices like preregistration--whose author list includes some high profile open science advocates--really fib about the status of its own preregistration?

It sure looks that way.

statmodeling.stat.columbia.edu/2024/03/27/the…

Nothing quite as magical as reading other people’s preregistrations, at midnight, in Disneyworld, while everyone around you sleeps.

We're looking for a few folks who are comfortable fitting and checking generative models to try out a new R package we created to instantiate a visualization grammar for model checking. Please sign up!

Lots of people in CS are (almost surely) GPT-ing their peer reviews
arxiv.org/abs/2403.07183

If you care about checking model fit and like visualization, please consider signing up for our study!

Ultimately we hope to make it easier to specify and explore the space of visual model checks.