Atlas of the 2008 Elections. Stanley D. Brunn, Gerald R. Webster, Richard L. Morrill, Fred M. Shelley, Stephen J. Levin, and J. Clark Archer, eds. Lanham, MD: Rowman & Littlefield, 2011. xiv and 320 pp., maps, diagrams, notes, and index. $84.99 cloth (ISBN 978-0-7425-6795-5).

Not so many decades ago, drawing one choropleth map was a substantial task. To draw a few for a journal article was a major exercise, and for an atlas you needed a dedicated team of specialists working for many months. All that is over—the keyboard has replaced the drawing board—and once you have a shapefile and a data set, creating colored maps, after experimenting with the “best” format building up and printing an atlas, is—in relative terms at least—now a straightforward task, rapidly completed.

This is not to suggest that the Atlas of the 2008 Elections has been assembled at haste: Much effort and care has been put into determining what should be mapped, and how, and providing a substantial textual commentary involving fifty-one different authors. The result is a handsome product. Nevertheless one questions the value of a volume on which work started within weeks of the November 2008 elections, but did not appear until 2011, by which time the Republican contest for the 2012 nomination was well under way. By the time I started assessing it for this review I was much more interested in the next election—and although I could glean hints from the Atlas that might help me understand what was happening where, there were no clear signposts. (Were the 2012 “swing states” the same as those on which the campaigns focused in 2008? Were they the swing states in earlier contests? And so on.)

After a brief introduction the Atlas is in eight main parts—plus a brief conclusion to which I return later. Those eight cover the geographies of the primaries, the campaigns, the “general elections” (i.e., for the presidency—many of the maps are for 2004 and 2008, with a few exploring changing patterns between small groups of elections since 1944—although are standard deviations necessarily meaningful when n = 4?), regional patterns, leading counties and social correlates of the voting patterns, other key elections held on 4 November 2008, nonpartisan referenda also held on that date, and some post-2008 Congressional votes (we are told that these can be compared with how a district voted in 2008—although none of the preceding maps are at that scale—but not why that might be a worthwhile exercise). Each part includes a series of maps (but none with either a scale or a north point), most covering either the entire country or a single state—with counties as the observation units in the latter case—plus associated text.

The choice of topics is sensible but far from complete. Throughout the primaries and the election, for example, there was much discussion of the relative performance of the candidates—both nationally and in particular (especially the swing) states: We are given no information on the time-space variation in support that these uncovered, however, although that strongly influenced short-term decisions on where a candidate should campaign strongly and conversely what areas should be relatively ignored, because victory there was either assured or unlikely. (Contemporary reports suggested that Obama's campaign switched resources in the last weeks to states where victory was unlikely, to draw his opponent's campaign organizers' attention away from the main prizes: Where were those states?) Furthermore, although it is important to know where money was obtained and where it was spent (although data are available at much finer, and more informative, scales than used in the atlas) it would have been interesting to map the places where more was raised than spent and vice versa. Further the campaigns were not just about money: Why no maps of the geography of field offices at different dates, for example? (And why instead have a map showing the “location” of the median voter for each party at the last three elections—all six of which were in southern Indiana: The text gives no indication that this statistic has any political relevance, although we are assured that it is accurate.)

Is a map always the best way of presenting the material, at least without other supporting visuals? (It was good to see graphs used in Gelman's chapter on intergroup voting proclivities.) In several cases we are virtually invited to engage in map correlations (e.g., the maps of campaign endorsements on 41–42), but surely evidence shows we are not very good at that: Why not use scatter graphs to show the extent of any “correlation” between two maps, along with an indication of the residuals, the places where one candidate's endorsements far exceeded his opponent's, for example? Many of the maps in section 6 are accompanied by bar graphs showing, for example, the percentage of the vote cast in a certain type of county (e.g., those with the largest proportion of sixty-five-year-olds) in both 2008 and 2004—but do these show the percentages of the total population there, or (weighted average) percentages in those counties, or something else? And although some of the differences—both within and between years—are substantial, many are not: Why not give some indication of the statistical significance of the differences? With one exception cartograms are not used, so the massive differences in population and number of votes cast across the counties creates a misleading impression: Without an indication of this variation (perhaps by using three-dimensional bar maps, especially across counties within a single state) one would get a much clearer impression of how Obama won, where.

Some of the textual material is not especially revealing: It is made up of little more than verbalization of the map(s) it refers to and in many cases assumes knowledge that even the better informed of American geographers might lack—which county is which. A lot of names are dropped, but not always to much benefit; if naming the counties was important, a key showing which one is where would have been valuable. (In a discussion of voting for banning same-sex marriages in Florida, for example, we are told that the issue was defeated in the urban areas, with several of them named [258] but they are not identified separately on the associated map, or any other in the book. Regarding the very next map we are told that the “normal electoral geography of California failed to appear” in the results of a referendum: But what is that “normal geography”? The few lines of text are not very informative.) And should an academic text include reference to a “bigot belt” (129)?

The introduction defends a geographical perspective to the study of elections: The “political system, and therefore, American politics, is inherently geographic” (1)—indeed it “privileges territory” (although in presidential elections that territory is the state, not the county). But are all of the geographies that the Atlas reveals meaningful? The editors suggest two reasons why geography is important, other than the use of states for the counting of votes and the allocation of places in the Electoral College. One is that there are variations in partisan leanings—between urban and rural areas, for example (suburbs get very little mention, but don't they house the majority of Americans?), as well as between states. There is also what they term a further “geographical principle, that of proximity or nearness”: “people tend to make voting decisions similar to those of people living nearby” (1). But are these geographical principles what Agnew (1990) once termed epiphenomenal; if once you know where different types of people live and can “predict” the outcome of the election in each place with considerable accuracy, then you must conclude that geography really contributes nothing to the understanding. (And, as one author at least noted, you are also in danger of committing the ecological fallacy.) Nothing in this volume addresses that important issue: It might be that once you have taken into account race, religion, sex, income, and education you know everything you need to know about why Obama defeated McCain—nothing else was of any significance. The brief conclusion, “Towards a More Perfect Union: Ten Scenarios,” merely increases the suspicion that Agnew was right: With the possible exception of “regional superprimaries,” “reforming the electoral college,” and having “transboundary congressional districts” (an issue hardly raised earlier in the volume), this chapter is very ageographical!

A 1911 statement that “a picture is worth a thousand words” is cited by many authors, and is certainly more than relevant to this book: Without the maps it would be virtually impossible to convey the geographical variations on display here—indeed, at times I wondered if all of the words additional to the pictures really helped. No doubt the book will be used in some teaching—especially if individual maps can be downloaded from the e-edition—but it is very much a historic document and the three years it took to fulfilment means that it will have less impact than it could otherwise have had. Technology is making such productions obsolete: Hopefully instead of a similar successor volume for 2012, electronic formats will make comparable maps available within months of the election.

Of course, 2008 was probably in many ways a much more important and interesting election than 2012, which makes the “stand-alone” nature of this volume disappointing. It would have had much greater value if 2008 had been put in much longer term perspective—as Archer and Taylor's (1981) pioneering book illustrated. On this, the brief text on “Comparison with Previous Elections” is particularly disappointing—it really only goes back to 2004. With such a broader span, this volume would not so rapidly become an archive piece; it could have done much more to illuminate American electoral geographies more generally—context, as geographers increasingly argue, is crucial. Why emphasize statics when your material will have a much longer half-life if it is put into a dynamic context? For that question alone, this handsome book, for all its plus points, is an opportunity lost.

And one final comment: for those not certain, “the” 2008 elections were those held on 4 November in the United States: Just as the British don't need to name their country on their stamps, Americans apparently assume that a book is about their country without telling potential readers!