3

I am designing project which requires me to select a Control Group. One the business requirements stipulates that I select as small a sample size as possible for the Control Group so there will be more records for the Test Group.

I would like to select as small as possible a Control Grouo and still be confident that the result would be statistically significant. Does anyone know which test would be appropriate for this?

Kim-Yen

mdewey
  • 16,541
  • 22
  • 30
  • 57
kyg
  • 71
  • 3
  • 3
    You and your client(s) will be unhappy with the answer: assuming you can supply the information needed to solve this problem, then minimizing the control group size is likely to lead to a prohibitively costly test group size. It's a fact that the variability within the control group is just as important for determining statistical significance as variability within the test group. Thus this "business requirement" is counterproductive: you ought to work to replace it with a statistically effective requirement that actually works for the business objectives rather than against them. – whuber Mar 21 '14 at 14:08
  • Hi Whuber. I should have been more specific with the details. The aim is to compare test and control groups to measure uplift in customers' borrowing amount. Unit of measurement is dollars. The business has a limited number of customers for each experiment and requires (for business reasons) as small as possible a control group so that more customers will be in the test group and be treated with a marketing offer. So it is not really a 'business requirement' but rather a 'business constrain'. Kim-Yen – kyg Mar 21 '14 at 16:14

0 Answers0