Yes, we need double blind randomized studies. However, I think we need to be careful about dismissing all 'anecdotal' evidence.
Example: medical marijuana. All we need to know is that some people have chronic , painful or terminal conditions and they tell us that it makes them feel better. To me, from a practical and ethical viewpoint, that's all that need be said. If doctors think they have something better, bring it forward and let the seriously ill consumers decide. As it is, I suspect that drug companies are also behind this societal suppresion.
2nd example: If a number of people are documented as having terminal cancer and survive, their survival needs to be counted as more than anecdotal. Are we willing to assert as fact that placebos can cure cancer? If so, a lot more work needs to be done on belief. If the people had faulty diagnoses, then I always wonder, where's the malpractise suits? and how prevalent are they to permit a falsely positive cancer 'cure' to 'appear? If the affected people fail to die, this has to be looked at as more than hearsay.