Re: On Rereading Savage's "On Rereading R. A. Fisher" (1970 Fisher Memorial Lecture)

Andy Spragg wrote:
On Mon, 27 Mar 2006 19:27:17 -0500, Jerry Dallal wrote:

Suppose, for the sake of focusing the discussion, you were involved in a multi-million dollar, multi-year, randomized, double-blind, controlled
^^^^^^^^^^^^
trial and there was no rolling enrollment because you have all of the volunteers you need at the start. You get to randomize everyone at once. You do the randomization and notice that everyone who got treatment A is a smoker and everyone who got treatment B is a nonsmoker.

Sorry to interrupt this most entertaining dialogue, but ... I thought the
whole point of double-blinding was to avoid things like that being noticed?

Andy


What is critical is that everyone who has contact with the subjects or has any role in collecting study data or assessing their validity be blinded. Often one of the study personnel who has no contact with subjects or makes any determinations about the measurements (typically one of the statisticians, either attached to the study or the DSBM) knows what's going on. Another common situation has someone knowing who has A and who has B without knowing which treatment is which.

So, as the unblinded member of the team you might see that all the smokers have gotten one while all the nonsmokers got the other without knowing which group got what.
.