A retailer is trying a new concept in 3 of its 200 stores. The trial period is 10 weeks. The goal is to increse sales with at least 10%.
The tricky part is that there is sesonality and campaigns that make it hard to relate current sales to historical sales. A better approach is to compare sales to the old stores.
The question is to test if the sales has increased with at least 10% compared to old stores.
So I know I should probably use a two population hypothesis test for the difference in mean sales. But the sample size will be really small for one of the samples with only n=3 trial stores. Would it be possible to regard the trial weeks as the sample = 3stores×10weeks=30?