Seems like a better approach would be to identify cities with the most startup "clusters"? Identify a cluster, then look at the population(s) of that census tract or zip or whatever.
Using city-level population just makes small cities look great (Boulder) and big cities look terrible (NYC), when I'm pretty sure there are neighborhoods in NYC that blow away Boulder.
Comments
Seems like a better approach would be to identify cities with the most startup "clusters"? Identify a cluster, then look at the population(s) of that census tract or zip or whatever.
Using city-level population just makes small cities look great (Boulder) and big cities look terrible (NYC), when I'm pretty sure there are neighborhoods in NYC that blow away Boulder.
Agreed, a hybrid of absolute numbers of startup people and density is probably the most appropriate.