Definition
Randomisation is the practice of using a chance mechanism — a random number generator, a sealed envelope sequence, a coin — to decide which experimental group each subject enters. Because the assignment is governed by chance alone, it is statistically independent of every characteristic a subject brings to the experiment, whether measured or not.
That independence is the whole point. It converts a collection of people into two groups that differ systematically in only one respect: the treatment they receive. Any other difference becomes, by construction, a matter of chance that probability theory can quantify.
Why it matters
How it works
Before any subject is enrolled, a randomisation schedule is generated — often stratified by key factors such as age or disease severity so the groups stay balanced on those dimensions too. As each subject joins, the next slot in the schedule determines their group; the experimenter cannot override it. Allocation concealment keeps the upcoming assignments hidden so enrolment decisions cannot be gamed.
The result is that any post-trial difference in outcomes has only two possible sources: the treatment, or chance. Statistical inference exists precisely to tell those two apart.