I think Matt makes some very smart points in his subscription-only review of two new books on the political landscape:
the broader logic of regionalism appears questionable in many respects. If you want to understand the political dynamics in, say, Las Vegas, you should look at a whole class of high-growth Sun Belt cities with large Latino populations, including Houston, Phoenix, and Los Angeles. You don't want to be told that Vegas, like Phoenix, is part of the "Mountains/Plains region" along with rural Utah, Omaha, and Fargo, whereas Los Angeles is in the "Pacific" and Houston is part of the "South."
The impulse to describe American politics in regionalist terms is an understandable reflex for authors who've specialized in the political history of Dixie, and it well suits the recent preoccupation with red states versus blue states. As is well known, however, when you bore down to the county level, there is little evidence of kinds of states, as opposed to kinds of places. College towns, cities, and the inner suburbs of very large cities tend to vote Democratic; less-populated, whiter places tend to vote Republican. The state-by-state aggregates matter, of course, because that's how electoral votes and Senate seats get parceled out. For analytic purposes, however, the point is that Illinois contains a giant city and Indiana does not -- not that Hoosiers, as such, have some characteristics of Republican-ness that demographically similar residents of Illinois lack.
This is a very true, and very undercovered, point. It's long been understood that urban centers go Democratic, rural areas vote Republican, suburban/exurban areas swing/lean Republican, coastal areas trend left, etc, etc. But because presidential campaigns necessarily focus on winnable states -- as you need to win a state to get their electoral votes -- the media has tended to unhelpfully make state-level preferences their sorting category of choice, even as the unit paints a remarkably misleading picture of the actual electoral trends within any one state. California, for instance, is decidedly blue, because folks are clustered in cities and on the coast. Orange County, though, has long been rather red, and inland California is politically indistinguishable from Wyoming. So, in the interest of more accurate political coverage, I think we should move towards popular vote elections.