Erik van Zwet

Axioms don't have proofs.

When and why (not) to worry about the PO assumption Aim We wrote an article (Long, Wiegers, Jacobs, Steyerberg, & Van Zwet, 2025) about the proportional odds (PO) assumption in the analysis of ordinal outcomes. we use various examples from neurological...

Yongxi Long and colleagues have written an excellent post about when and when not to worry about the proportional odds assumption: discourse.datamethods.org/t/when-and-w... #StatsSky #EpiSky #Statistics #RStats

Evaluating a shrinkage estimator for the treatment effect in clinical trials The main objective of most clinical trials is to estimate the effect of some treatment compared to a control condition. We define the signal-to-noise ratio (SNR) as the ratio of the true treatment ef...

Thanks for the nice blog post about our paper! I'll just add that our shrinkage estimator is a little bit ad hoc, but we did evaluate its performance: onlinelibrary.wiley.com/doi/full/10....

Livin' the Bayesian dream!

Dichotomization I’m writing a paper about “An Empirical Assessment of the Cost of Dichotomization” together with Frank Harrell (@f2harrell) and Stephen Senn (@Stephen). In this paper, we quantify the information loss...

There's a nice discussion at Frank Harrell's datamethods:

discourse.datamethods.org/t/dichotomiz...

Hmm, I get 2x38 for 90% power of a two-sided test at level 10%. But nevermind. I came here to advertize an app which calculates the gain in power if you don't dichotomize. In this case, they would have needed only 2x19. The app is here: vanzwet.shinyapps.io/info_loss/

A New Look at P Values for Randomized Clinical Trials We have examined the primary efficacy results of 23,551 randomized clinical trials from the Cochrane Database of Systematic Reviews. We estimate that the great majority of trials have much lower st...

Speaking of the null hypothesis, here is a much better treatment of the "how many experiments are exaggerating the effect" question.

evidence.nejm.org/doi/full/10....

We still writing up the paper, so all comments are more than welcome!

It has been claimed that there is no such thing as a validated prediction model. We provide empirical evidence for that claim in our new paper here: onlinelibrary.wiley.com/doi/full/10....