Sadly this extends past psychology to other scientific disciplines. There are many foods for example that have been shown to both dramatically increase and decrease you risk of cancer in well respected scientific journals. There is a strong incentive to find extreme results and too few peers actually reproduce results.