What are the inherent advantages of comparing the means of two populations using sample data ? Can we be sure that the comparison is accurate ? Why or why not ?