What is the best geography for longitudinal counts across the widest range of measures?

I’m establishing an open prize competition for the best comprehensive quantitative dynamical model of US society given a standardized table of measures. The openness of the table requires aggregation. The importance of jurisdiction requires geographic aggregation. The dynamics require longitudinal range. Comprehension requires a wide latitude of measures.

It would be best to use the same geographic aggregation across time and measures.

State-level does not provide enough samples.

What would be the best choice of geography?

If a single choice does not suffice to provide a broad range of both longitude and latitude, what would be the best set of geographic aggregations?