a well-designed experiment should have which of the following characteristics? i. subjects assigned randomly to treatments ii. a control group or at least two treatment groups iii. replication