The concept of “random” is less simple than it may appear at first.
In practice it will boil down to a sampling process that, at the time of
sampling, is independent of the outcome to be measured.
It is quite likely that the sample will include different subsets
with different responses to the treatment. Some may have a positive
response, some a negative one, and some no response; these may cancel out,
with the aggregate result showing no effect. Consequently, post study
observations should ideally lead to further studies.
The ethical question of conducting randomized studies is a difficult
one indeed. On the other hand, in the 18th century it might have been
considered unethical to randomize patients with septic fever into a
“bleeding” group and an “observation only” group.
Competing interests:
None declared