You and your colleague are designing a study to try to determine the ideal mean cost for discretionary service. You will market your service to a very large population. Your colleague thinks you will get the best data if you take lots of small samples. You think the data will be neat reliable if you take one or two very large samples.