[lucozade]

Presumably pre-registration could also help with positive results bias.

Assuming the registration was made public, one could mandate that something had to be produced from the study, if only a statement that nothing was found.

Wouldn't solve the problem but would presumably aid meta-studies?