It would possibly benefit a concept called triple blind, which I've learned about in the latest Slate Star Codex linkfest: it's about studies where the metodology is submitted for peer-review before beginning the experiment. This at least gives a better focus on the approach and the design of the project.