2

I’m not sure if this question has been asked before as I don’t know how to phrase it in a search.

But essentially I’m trying to test if there are differences between store locations that have trained their employees differently to see if there is an increase in profitability after training. So for store A in Utah that trained their employees using method 1 vs Utah Store B that trained their employees using method 2.

For this example you would use a T-test if there were only two groups or an ANOVA if there were more than 2 groups. I get it. But what if I would like to compare 3 stores in Utah using method 1 vs 3 Stores in Utah using Method 2. I would imagine this better to do than comparing store vs store. What is the best method to use? Layman terms would be much appreciated.

What if I wanted to take into consideration stores using either of the two methods but from other states? Should I just compare state to state?

DataTx
  • 477
  • 1
  • 7
  • 19

1 Answers1

0

It sounds as though you interested in effects potentially induced by the differences in the variable $\small \text{Method}\in\{1,2,\dots\}$ (sloppy notation), but you want to account for the potential variability between $\small \text{Store}\in\{\text{A,B,C},\dots\},$ and also potentially the variable $\small \text{State}\in\{ \text{NY,NJ,AZ},\dots\}$.

This can be accomplished with a nested (hierarchical) ANOVA or a mixed-effect regression model.

The idea is that the variable $\small\text{Method}$ would account for fixed effects (i.e. effects you care about), while the other two variables would constitute random effects.

Community
  • 1
Antoni Parellada
  • 23,430
  • 15
  • 100
  • 197