https://www.wikiwand.com/en/Randomized_controlled_trial

To get evidence that something causes something else, rather than being just correlated. We break potential extraneous causes of the thing whose causal effect we want to study, by randomizing its application.