>>13608414A lot of the replication crisis is p-hacking in the selection of results and seemingly superfluous parameters.
Basically in any field where there is any room for choice in either the selection of data points, or interpretation, or parameters of the experiment etc all of that is selected in such a way that results are best, and the selection bias is left out.
This is essentially fraud with plausible deniability. Wherever there is vagueness you can assume fraudulent results these days because the incentives are such that if you're the only one not doing it, your results will be honest but they will look worse than the competition, and you won't even get published.
Nobody audits that sort of stuff. The closest there is to an audit is the failure to replicate and there are no consequences to it. There is the equivalent of "it works on my computer" excuse as plausible deniability, and indeed it may well have, and you know why, but you can always pretend you don't.
