The 2000 presidential election in the United States provides Richard Johnston, Michael Hagen, and Kathleen Hall Jamieson with a “natural experiment” to investigate the dynamics of support for George W. Bush and for Al Gore over the course of the campaign. As we all know, the 2000 election gave somewhat contradictory results: a popular majority for one candidate and an Electoral College majority for the other. Further, the campaign had two distinct elements that these researchers capitalize on for their work—the nationwide news coverage provided by the mass media and the campaign ads running on television chiefly, if not exclusively, in the battleground states. Given the difference between the forecasts provided by previously reliable models of presidential campaigns and the actual outcome of the 2000 contest, Johnston et al. ask whether campaign dynamics may account for the disparity.

Using the 2000 National Annenberg Election Survey (NAES), the authors attempt to “unite two research traditions” (p. 1), conventional political science approaches dealing with economic factors, incumbency, partisanship, and so forth, and political communication, which pays attention to campaign messages and their effects. The authors lay out the phases in candidate support over the course of the campaign that their data allow them to identify. Major demographic factors (party, gender, ideology, race, union family, religion) account for some of the variation in vote choice, but they cannot account for the changes in vote choice over time. Johnston et al. then use interest in the news (and therefore exposure to it) and living in the media market that received broadcast campaign ads, indexed to the volume of ads actually aired, in order to examine campaign effects. They conclude that although fundamental forces affect election outcomes, strategy and campaign dynamics, chiefly involving communication strategies, necessarily influence voters' choices.

If campaign dynamics and strategy make a difference, the data should show differing effects in contested states, compared to states conceded to one candidate or another. If campaign communications make a difference, the data should show short-term responses to campaign strategies, a support for an online model of information processing. The evidence supports both of these suppositions. Had Gore adopted a different strategy in presenting himself and his campaign to voters, he might have made the economy more relevant to voters' thinking. Had the number of advertisements in crucial states in the last weeks been more equal, instead of heavily weighted in Bush's favor, it is likely that Gore's position on Social Security reform would have strengthened his support. It is the campaign, the authors conclude, that makes some factors more relevant to voters than others, and candidates favored by the fundamental forces political scientists normally concern themselves with will not win if they do not make voters aware of those considerations.

The authors pack a great deal into relatively few pages. As a result, The 2000 Presidential Election and the Foundations of Party Politics must be read like a journal article, because much is said in few words. The anecdotal material frequently found in works of this nature is virtually absent: no references to open-ended responses by survey subjects, for instance. That absence is not a shortcoming; it is indicative of the nature of the evidence relied on and the elimination of the marginally relevant. It does, however, put the conclusion about online processing of information by voters in a different light. The conclusions are drawn on the basis of overall patterns in the data, but they are not buttressed by descriptions, in voters' own words, of their thoughts or their deliberations.

Some interesting findings from these data include the conclusion that Bush was able to place himself close to the center on the question of Social Security reform and that in states where news and not ads dominated voter attention, subjects responded less favorably to Bush's proposals. Another finding was that data support the conventional wisdom about Gore's dilemma: identifying with the Clinton administration's accomplishments, especially in the economic realm, while distancing himself from the Lewinsky affair. A third was the authors' dissection of the process whereby questions about Gore's honesty were first raised, then emphasized in ads, and finally left to the news media as a frame for their coverage of Gore. These and other findings form a string of pearls worth treasuring.

The necklace is not without flaws, however. Two spring immediately to mind. One deals with the assumption that the election forms a natural experiment, an assumption derived theoretically from the notion that the competitive states lie in the center of the continuum, that “the hard-fought states were a microcosm of the whole electorate” (p. 13). That may be so. But it would have been nice to see some data on that point from the surveys, specifically comparing respondents in battleground states with those in the noncompetitive states. The “natural experiment” argument would carry a great deal more weight with such supporting evidence. The other flaw is less consequential. I admit that having finished reading the book, the title reference to the “foundations of party politics” still mystifies me. Perhaps there is a sense in which the authors are using the phrase that I have missed. But a reader looking for a discussion of Democratic and Republican Party bases will not find it here. I presume that the authors are actually referring to the traditional political science explanations of partisan differences in voting. It would help to have said so.

On balance, this book is one that serious scholars must read. Its data analysis is detailed, conscientious, and insightful. The authors are especially careful not to interpret beyond the bounds of their data, and they frequently point out the areas where other scholars have taken other approaches. Given these findings, one cannot help wondering whether candidates respond to voter preferences or whether candidates form those preferences in their efforts to get elected. This work, and the NAES itself, will be widely used and cited.