This message has been cross posted to the following eGroups: Statistical Consulting Section and Statistics in Marketing Section .
-------------------------------------------
Hello,
I was asked the question below. I can pull a random sample of stores and find the confidence interval for the population mean based on the 2012 scores but I am unsure about the expected increase. Does any one have knowledge on how to obtain the confidence interval based on targets?
Each store in all regions will be given an OSAT growth target of 0.5. This is for improvement of May-November 2013 vs. same months 2012. But we want those targets to have some statistical confidence to them. So instead of telling the stores that the target is 0.5, we want to tell them the target at which we have confidence that increase exists. For example, let's say that a store has 2012 mean OSAT =75 and s.d. = 0.7. Assuming normal distribution, 95% confidence, and the same s.d. in 2013, what would the OSAT have to be in 2013 in order to have 80% confidence that there was at least 0.5 growth? Let's say that at 95% confidence, the 2013 mean OSAT would have to be 77.8. Then the growth target would be 77.8-75=2.8. Rather than calculate this separately for each store, we can calculate it separately for a sample of stores and take the mean.
Thanks,
-------------------------------------------
Kenita Hall
Analytical Consultant
-------------------------------------------