ICC - unclear comment from reviewer I'm new here and not a statistician, so I apologize in advance and I sure hope someone can help!
I'm working on a study where we have 10 locations $(5 \ intervention, 5 \ control)$. We have two measurements of activity for each participant $(n=  700)$ at each location. I've received the following comment from a reviewer: "Typically an intracluster correlation coefficient (ICC) is calculated to adjust for the correlation between participants at a single location" 
I know how to calculate ICC using SPSS, but because there are different numbers of participants at each location, I'm a little confused... can anyone help?
Thanks so much in advance!
 A: *

*Are you analyzing the data as a clustered randomized trial? Well I guess randomization doesn't really matter in this case. But have you addressed clustering in some way? If yes, ignore (2).    

*You need to address clustering in some way. I'm sure there are many ways to do this. The first that comes to mind would be a mixed-effects model with a random effect for location/cluster. (This is a commonly cited reference: Methods for evaluating area-wide and organisation-based interventions in health and health care: a systematic review) But depends on your study and field of research. I've seen location modeled as a fixed effect. This can be useful. More important is what the standard practice is in your field.     

*Usually clustering is addressed in the modeling stage. An actual ICC could be generated, but not sure how it would be useful. This may just be a limitation of my understanding generally.  (ICCs are often used in the sample size estimation stage. But sounds like you're way beyond that now.)

