To ensure the researcher collects the same data twice to check for errors

To apply the treatment to many subjects so that results can be generalized and are not due to chance

To make all subjects in the experiment identical

To randomly assign subjects to treatment groups