In yesterday's Politico, there's a column by Joel Kotkin and Mark Schill discussing what the authors call the "three geographies" (cities, suburbs and small towns/rural communities) and the political attitudes found in them. They argue that it is these units, and not states, that really affect political behavior. It's a valid point, but it's states that count in the Electoral College, which is why they garner so much attention. At any rate, Kotkin and Schill spend most of the piece making the not so earth-shattering point that Obama is likely to win in urban areas (which they note is where 32% of the population lives) while McCain is likely to do well in rural communities/small towns (17% of the population). The battleground is likely to be in the remaining suburban areas.
Dig just a bit deeper, however, and you find that dividing places of residence into only three geographical units disguises some important differences between subsegments within those units. In fact, the 2004 exit polls revealed that while Kerry beat Bush handily in "big cities" (60% to 39%), the two tied in "smaller cities" (49% each). On the other hand, in rural communities, Bush won by a 59% to 40% margin. But "small towns" were evenly divided (50% to 48% for Bush). True to form, the suburbs were closely divided (Bush won them 52% to 47%). The point is simply that a three category distinction may hide some areas of real competition within what appear to be safely red or blue areas.
Of course, none of this recognizes the existence of the exurbs or the distinctions that can be drawn therein. For an interesting study of this geographical category, written from the strategic perspective of the Democrats, see Ruy Teixeira's 2006 report "The Next Frontier: A New Study of Exurbia."