Is pre-registration a proposal aimed at reducing bias/QRPs? Arguably yes, but I don’t think that’s how we should think of it.
1/n

(Note: this was inspired by one sentence taken slightly out of context of @annemscheel et al.’s preprint: https://psyarxiv.com/p6e9c )

I’ve argued before that we should think of the credibility revolution as having two components: transparency & quality control.
Transparency makes quality control possible, but it is not the same thing. You can do transparent, shoddy research ( http://www.stat.columbia.edu/~gelman/research/published/ChanceEthics14.pdf)
2/n

To my mind, pre-registration is clearly a transparency-enhancing reform. It makes it possible for authors to document what was planned/timing of decisions in a way that is verifiable to readers, and therefore makes it possible for readers to calibrate their confidence.
3/n

Yes, pre-registration makes quality control more possible, and it hopefully has the side effect of reducing bias/QRPs on authors’ part to begin with, but pre-reg doesn’t necessarily increase quality or reduce bias. That’s not the main point. The main point is transparency.
4/n

The danger in saying that “pre-registration enhances quality” or “reduces bias” is that then, a researcher will earnestly pre-register, and then may assume that they couldn’t possibly have let bias creep in and will he shocked & upset if critics suggest possibility of bias.
5/n

So I think we should reframe pre-registration. It might help reduce bias, but the goal is transparency - documenting things that matter (e.g., timing of certain decisions) and making them verifiable. That might actually bring out *more* evidence of bias. That’s the point.
/end