Aggregation causes huge problems of analysis and interpretation. The few and the proud who try to work with spatial data are routinely vexed by the problem. We all know that there are huge differences often just a city block or two away from where we live and work, but we are all the time generalizing about whole metro areas. Urban economists are fond of touting the benefits of "density" when they compare, say, New York to Kansas. But density variations within these places are huge.
Wendell Cox reports recent settlement trends using the Census' latest county and metro data. Notice that he adds his definitions of "historic core counties" because the census bureau's designations of central cities are ideosynchratic.
Look at his Table 4. The suburban counties have been growing steadily at the expense of these historic core areas. Yawn? The conventional wisdom (including from some editors and referees who return my papers) is that long-term suburbanization trends have stalled or reversed. Yes, Manhattan today is not the same as it was in the 1960s-1970s. And a few other downtowns do look better today than then. But let's be careful about jumping to conclusions re historic reversals.