It doesn't help that many of those experiments are deeply unethical to run and so simply can't be reproduced anymore anyway for institutional and moral reasons.

It doesn't help that many of those experiments are deeply unethical to run and so simply can't be reproduced anymore anyway for institutional and moral reasons. The Stanford prison experiment was particularly egregious in that the experimenter actually played the role of superintendent in the prison, contaminating the data. I'm always cautious of conclusions drawn from studies with these sorts of ethical issues. If the researcher was comfortable causing harm to other human beings (whether undergrads or little albert or anyone else) in order to get a result, then why would we expect them to be honest about what the data showed? I'm sure this is some sort of cognitive bias that's been explained by a hopefully replicable study and that the willingness to cross a moral/ethical line in running a study doesn't have to imply willingness to cross a moral/ethical line in fudging data and drawing conclusions, but combined with the lack of reproducibility, I think those studies should be largely discounted as sources of psychology knowledge