Hari: Dear Christian: I am really very sorry. I should not have rattled that out without thinking. Especially for one who is constantly asking silly questions of this list, & of the PEN list on terminology - the most infamous being in my view the recent pomo thing!. Again - that was stupid of me. Sorry.
i) RCT = Randomized Controlled Clinical Trial. Thus you flip a coin (or a Blue Box Computer nowadays) & this "randomly" allocates a treatment to one (say the experimental arm) or the other arm (say a 'control' arm). If you flip a coin 100 times, & the coin is not weighted etc; it will end up 50% time tails & 50% time heads. i.e. equivalence. Same with the trail. If done (in an ideal world) correctly, & there is not any whit of difference 'tween the 2 therapies - you will get a "non-significant" difference between the 2 arms. If one is in fact superior to whatever your endpoint was to be - then it will (should) turn out significantly different. An arbitrary p value assigned is usually p<0.05.
ii) The Hawthorne effect - When a kindly old General Practitioner (GP) says "I understand everything that you say - have a whiskey tonight" - NO SPECIFIC THERAPY has been performed. What has been done is a compassionate, kindly, (& perhaps not altogether unknowingly) a possibly powerful intervention. This non-specific effect, is supplemented by an even stronger form of the so-called "Hawthorne Effect". So when a new therapy comes along, & someone says take it, the patient naturally "wants to believe". This 'boosts' the propensity to improvement. [Re: propensity to improve: And as an old teacher of mine use to say, What doesn't kill you makes you stronger. So for bouts of illness, by & large, we get over them.]
To sort out all this intensely mind-bending stuff - in figuring out whether a treatment works or not - the RCT should ideally be performed in a "double-blind" fashion. That is to say the pill is labeled X & the control is labeled Y. And - the double bit is that - the patient AND the practitioner should not know to what arm the patient is randomized to. That is 'cos in that little dyad-dance of Dr & patient, BOTH can (& usually do) BELIEVE!
iii) Thus - we get to ethics. For to "randomly allocate" - means that you must "Inform" the patient & obtain her/his consent.
Now for the complexities of the Freudian thing - an RCT would be un-blinded. But there are ways of dealing with the "bias" [Again in this parlance, this is a technical term - it means introducing a systematic deviation form the truth - that might thus be introduced.]
End Clinical Epidemiology 101.
The bottom line was - that I think that an RCT could have been done. There might be some reasons why it never was done. Apart from the fact that the medical community (unlike agricultural scientists who did RCT's much more frequently & earlier) there is the matter of VESTED INTERESTS. Whose? Well, perhaps the billers?
Again I am sorry about my blasé attitude in not spelling out an RCT.
Cheers, Hari