I am going to conduct a study in a slum area with 94 catchments (clusters). They are randomly assigned one of two treatments, and then within those catchments, 10 women are randomly chosen for surveys at baseline and 1 year.
I am using a software called Optimal Design for the power analysis and must decide if this a 'cluster randomized trial with person-level outcomes', or a 'cluster randomized trial with cluster-level outcomes'. The data we get will be specific to the individual women (e.g. number of healthcare visits they received, number of pregnancy complications), but the overall conclusions will be what effect that the treatment has, which is assigned at the cluster level. My feeling is that I should design this as a cluster randomized trial with person-level outcomes, but am not confident in this.