The half-life of knowledge in psychology is thought to be one year, so by now most of the famous experiments (and their findings) in the field are considered flawed.
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.
http://en.wikipedia.org/wiki/Half-life_of_knowledge