Hello, is there a way to do geographical testing for example All of Texas and California would get the same variant vs New York and Colorado?

Hi Alejandro, Yes, you can do this. As Option 2 in this response states, you will need to randomize at the cluster (e.g., Texas and California) id granularity, and then analyze at the user id granularity. As this response states, the uncertainty in your estimates will be too small (and perhaps massively so, given how large your clusters are). So your results will probably provide at best directional signal. It sounds like you have few clusters, so Option 1 is not feasible. I hope this helps. Luke