While survey data is not as reliable as a double-blind clinical trial, to describe such surveys as "completely unscientific" is not justified scientifically. Surveys, passive surveillance of drug safety, retrospective observational studies, etc. are regularly used in medicine and science to collect data in a lower-cost manner.
Observation is a legitimate part of the scientific process. So someone could think to themselves, I seem to feel better when I eat celery, for example. They might chat with 27 of their friends, and some of them might say 'yeah, I too seem to feel better when I eat celery'. So far so good, what you have is an observation that might be interesting. It is, however, not evidence that celery does anything useful. There is a problem if that idea is presented as anything other than an observation. There is a major problem if the treatment is expensive or risky or stops people getting effective help, and it is promoted as useful on the basis of a few anecdotes.
Yes, a survey that isn't part of a blinded controlled trial can be an interesting observation. Being large (thousands of people) and having a sound methodology would make it more interesting. We see many surveys where questions can be written in such a way as to bias answers, for example. There also needs to be care with the selection of the participants - are they all from the celery growers association? If a survey finds that a large percentage of people suddenly were cured, and were able to resume their old lives, and were still well months later, that would be a lot more interesting than several people reporting having a bit more energy.
But we've seen over and over how people can convince themselves that something helps, if they think it will. There is the
Mendus study of CoQ10 for ME/CFS for example. Many people who received the labelled MitoQ CoQ10 supplement reported that it improved their energy, their sleep, their cognition. However, there was also a part of the study where people received unmarked pills - some the CoQ10 and some just an inert pill, a placebo pill. In this part of the study, the CoQ10 result looked just like the placebo result. Therefore, CoQ10, or at least that particular formulation with that particular dosage, did not help ME/CFS. It's a great demonstration of how people can convince themselves that a well-marketed treatment works.
So, you can chat with 27 friends, and conclude that some people think a treatment helps. We do that often here. But it isn't scientifically sound evidence that it does help.
We have threads for quite a lot of the treatments mentioned in the survey you linked. You might like to find them, read them, and add your own observations. If there is a treatment that you think is interesting and that we don't have a thread on, you could start a new thread.
