If you want to reduce the likelihood of any preexisting differences between the groups of participants assigned to different conditions in an experiment you should use
Is to randomly assign subjects to different treatment groups to reduce the chances of preexisting differences between the subjects having an impact (negative) on the experiment.
