It ensures the results can be generalized to the population.

It creates comparable groups by balancing out both known and unknown confounding variables.

It eliminates the need for a control group.

It increases the absolute magnitude of the treatment effect.