Oops… I really thought that this terminology would be well-known in a community that is so closely related to medical research…
I seem to have been wrong in my assumption…
The background of this terminology is that there are a lot of pitfalls if you want to find out if something works, and you “just try”. These pitfalls work both ways: there is a high likelihood that you will think that something does not work even if it does, and there is an even higher likelihood that you will think that something does work even if it does not.
Decennia of trying, evaluating, and learning have resulted in what is now considered a “golden standard” of medical research.
If you want to have meaningful results, any test setup must be:
You do research on a lot of test persons – the more the better. Half of those gets the treatment you want to test, the other half either gets a placebo treatment or a treatment with a known cure. Randomised means that patients are assigned to one of these two groups in a totally random way.
Neither the doctor nor the patient may have any knowledge about the treatment. That is, neither of the two may have any way of knowing whether that particular patient receives the treatment that is tested, or the other treatment.
Everything that is done is recorded, from the random assigning of each patient to one of the two groups, to the actions that are taken, to the results, and any possible side effects. In addition, if at all possible, anything else that the patient does in parallel to the investigation is recorded.
This is the only way to do research. If you forget one of these three parts of the golden standard, your results will be unreliable.