Good practices dictate that preregistration should always be done nowadays, including for meta-analysis. In practice though, this has very little impact. There

Good practices dictate that preregistration should always be done nowadays, including for meta-analysis. In practice though, this has very little impact. There is absolutely no guarantee that what's written in the paper is a faithful report of what really happened. There's nothing holding back dishonest (or, at that point I guess I should write 'standard') researcher from doing absolutely whatever they want.