What does randomisation do?
Randomisation of treatment makes it independent of potential outcomes.
Mean potential outcomes are identical for the treated and the untreated groups.
If you split the population randomly into two groups, what would you expect in terms of attributes?
They would be the same in each population: the average heights, average income, proportion of those with brown eyes etc.
Same goes for potential outcomes.
Implication of random assignment of treatment

Two effects of randomisation
