Weak statistician here

#1
We recently implemented a smoking ban and I am doing a paper on how it has affected bar/restaurant revenue. I surveyed 3 bars or restaurants in a downtown area and have data on whether or not sales for the three meals they serve and the the sales for both the after work bar crowd and nightime bar crowd went up or down or stayed the same. What would be a good test for analysing my findings? The sample size is 3 (approximately 10% of the population and sales are reported as up or down based on a percentage). I know most of the formulas just having a hard time figuring which one to use.

Thanks
ChrisB
 
Last edited:
#2
I would think about paired t-test and test the hypothesis that the change is not equal to zero.

perhaps some one else will offer a better idea though.

jerry