The root problem with many modern sciences is that we look at the result to judge whether the procedure that brought it forth is valid. This heuristic is rarely correct; inproper application is disastrous.
Judging the quality of the method based on the desirability of the results is what fundamentally drives problems like publication bias, questionable researches practices, even fraud. It corrupts the scientific method, as it subjugates it to the outcome of it.
Unless it has been shown beyond reasonable doubt that a certain method must result in a prespecified outcome, using the latter to judge the quality of the former is inherently misleading.
There are certainly cases in which the above holds - when we have an exhaustive (enough) understanding of all the causal mechanisms. In all other cases, a study's outcome has no bearing on how we should judge its methodological qualities.
This is what the scientific method is inherently about - to repeatedly and extensive study phenomena, such that after many trials we uncover regularities. These outcomes are what we are ultimately interested in, but their value is driven by the method, not the other way around.
Contemporary publication practices, hiring policies, award ceremonies, grant funding, and incentives in general should be aligned with the scientific method. That is, to promote good, thorough research methods. Problematically, they are often not.
This is why it is vital to invest in counter-cultural practices such as Registered Reports and open science in general. I wish I could end on a happy note, but I'm not. Shit's fucked. It's going to take a while to fix it.
