book reviews 283

According to Gerring, ‘‘each case may provide a Politics in Time: History, Institutions, and Social single observation or multiple (within-case) observa- Analysis. By Paul Pierson. (Princeton University tions (19) ...For those familiar with the rectangular Press, 2004.) form of a dataset [i.e., a data matrix], it may be helpful doi:10.1017/S0022381607080280 to conceptualize observations as rows, variables as columns, and cases as either groups of observations or In Politics in Time, Paul Pierson has written an individual observations’’ (22). Gerring then advises the important book that is engaging, ambitious, and use of experimental or quasi-experimental methods to provocative. Its purpose is essentially threefold: to analyze variation across within-case observations. Yet advocate that political scientists situate arguments in for practitioners of case studies, the advice seems temporal perspective, to illustrate a number of ways problematic; as more and more observations are in which they might do so, and to argue that much of compared, and the inferential leverage afforded by the discipline does not presently take time seriously experimental or quasi-experimental methods is enough. The book is oriented not merely toward brought to bear, it seems less likely that one is really qualitative scholars or historical institutionalists, but conducting a case study—that is, an intensive analysis rather ‘‘those interested in the attempt to develop of one or several instances of a phenomenon (19–20). claims about the social world that can potentially The concluding chapter on process tracing, co- reach across space and time’’ (7)—effectively all authored with Craig Thomas, seems to come closest social scientists who seek to advance generalizable to shedding light on the distinctive contributions of explanations. In his effort to reach such a broad case-study research to causal inference. In process audience, Pierson engages widely with the discipline tracing, not-strictly comparable pieces of information and beyond, drawing upon theoretical insights from are combined in a way that adds up to a convincing Kenneth Arrow to Arthur Stinchcombe, exploring causal account, by rendering alternative explanations causal mechanisms from positive feedback to absorb- less plausible while showing that microevidence is ing Markov chains, and offering substantive examples consistent with theoretical claims. Process tracing from U.S. congressional committees to state building may be akin to detective work; bits of evidence about in early modern Europe. the maid, the butler, and the suspect are combined The publication of Pierson’s book in 2004 oc- to formulate or investigate a central hypothesis curred in the midst of a still-ongoing resurgence of about who committed a crime. In a different but interest in qualitative methods and temporal argu- related account, Collier, Brady, and Seawright (in ments—stimulated in no small part by Pierson’s Rethinking Social Inquiry, 2004) describe how what American Political Science Review article on path they term causal-process observations can provide a dependence published in 2000. Following an initial smoking gun that demonstrates—or rules out—a series of articles on the sources of institutional lock- particular causal hypothesis. Unfortunately, the dis- in, a second wave of scholarship by authors such as cussion of process tracing is a relatively short Kathleen Thelen and shifted the focus addendum to the core concerns of Gerring’s book. to ways in which institutions change rather than Definitional slippages are one distracting feature remain stable over time. Politics in Time, which of this book, despite the inclusion of a glossary and presents revised versions of four previously published careful attention to defining terms. At one point, for articles as well as an entirely new introduction, fifth instance, Gerring notes parenthetically, ‘‘I use the chapter, and conclusion, plays the very useful role of terms proposition, hypothesis, inference, and argu- encapsulating this evolution of the literature and also ment interchangeably’’ (22). Nevertheless, articulat- offering an attempt at synthesis. Ultimately, as ing the distinctive contributions to social science of Pierson argues in Chapter 5, change and continuity case studies has been a core challenge for qualitative must be seen as two sides of the same coin. methodologists, and together with other recent con- Politics in Time begins with a focus on institu- tributions in this area (Brady and Collier, Rethinking tional continuity via path dependence and positive Social Inquiry, 2004; George and Bennett, Case Studies feedback. Examining the mechanisms that sustain and Theory Development, 2005) Gerring’s volume stability over time in economic history, Pierson argues takes on this challenge in a spirited fashion. It will that such processes should be at least as common in provide useful and engaging reading for substantive the political realm. Reasons include the prevalence of researchers of all methodological stripes. collective action in politics; the potentially self- reinforcing accumulation of power asymmetries; the Thad Dunning, absence of a price mechanism to clearly indicate 284 book reviews optimal behavior; and the centrality of formal in- argument Pierson develops in the fourth chapter. stitutions, which have large set-up costs and generate Here, Politics in Time takes aim at ‘‘actor-centered learning effects, coordination effects, and adaptive functionalism’’—the rational-choice claim that insti- expectations. When political actors begin to travel tutions exist because farsighted, purposive, and in- down a self-reinforcing path—even one that is strumental actors benefit from them. Drawing upon eventually considered inferior to the road not tak- numerous empirical examples, Pierson lays out six en—reversal is unlikely because of the short time limitations to this perspective. Institutions may have horizons of many political actors and the obstacles to multiple or unanticipated effects, such that their change built into many political institutions. existence cannot be explained by a simple reading Having established the importance of positive of their creators’ preferences. Designers also may feedback mechanisms, the second chapter of Pier- adopt institutional forms for noninstrumental reasons, son’s book examines the closely related issue of such as diffusion or cultural specificity. Actors may timing and sequence. Here, he advocates an approach have short time horizons, creating institutions with that combines the insights and precision of rational undesirable long-term effects, or their preferences may choice models of legislative cycling with the attention change over time as institutions remain stable. Finally, to large-scale social changes that have been the political actors themselves may change; a new gener- mainstay of historical institutionalism and American ation may inherit institutions that reflect previous political development. Sequencing arguments, he actors’ preferences rather than their own. maintains, need not be ‘‘Dr. Seuss-style,’’ ad hoc In recognition of these limitations to the rational explanations in which it ‘‘just so happened’’ that one choice perspective, many historical and sociological event followed another. Rather, the best of these institutionalists have begun to articulate mechanisms arguments illustrate, often via comparison, how the of how institutions change over time. Yet in Chapter order of particular events matters crucially for the 5, Pierson argues that shifting from an assumption of eventual outcome. Examples include the self-rein- static institutions to an exclusive focus on fluidity forcing incumbency advantages enjoyed by a party involves too much of a swing of the pendulum. that initially ‘‘fills up’’ a political space, such as by Research on institutional change, which has identi- organizing the working class. Future competitors for fied mechanisms such as critical junctures, layering, this constituency will have greater difficulty than if conversion, and diffusion, places undue emphasis on they had been the first ones to occupy the same the malleability of institutions at the hands of political niche. entrepreneurs or the losers from previous rounds of The third chapter of Pierson’s book shifts the political competition, and it does not predict when focus from arguments emphasizing continuity to we should expect one type of institutional change those that focus on long-term change. Many topics versus another. Most importantly, arguments about in political science, Pierson argues, are those in which institutional change often lose sight of the sources of cause immediately precedes effect and both occur institutional stability, such as coordination problems, relatively quickly. The tendency to limit time hori- veto points, asset specificity, and positive feedback. zons of research topics in this fashion, however, Thus, Pierson concludes by proposing five distinct ignores a number of important processes, such as research agendas on processes of institutional devel- cumulative causes, threshold effects, and causal opment that seek a synthesis between arguments chains, in which the cause or the outcome unfolds about change and continuity. over a significant period of time or there is sub- As mentioned at the outset, the objective of stantial temporal separation between the two. Pierson Pierson’s book is not only to illustrate ways in which argues that the trend toward research involving only political scientists can incorporate the temporal short-term independent and dependent variables dimension into their research, but also to argue that risks excluding many of the classic subjects in much of the discipline does not presently take time political science, such as party-system change and seriously enough. While acknowledging that rational state building. An exclusively short-term focus may choice scholars and quantitative methodologists often even contribute to errors of causal inference if can model long-term historical processes, Pierson scholars mistakenly treat slow-moving causal varia- insists that theoretical possibility does not translate bles as mere background conditions. into typical practice. The main evidence to support An excessively static approach to political science this claim is presented in Chapter 3, where he shows can occur not only in the selection of research topics, that in several top political science journals, articles but also in the explanation of institutional origins—an investigating short-term causes and outcomes vastly book reviews 285 outnumber those in which change in an independent Colin Elman and Miriam Fendius Elman have produced or dependent variable occurs over a long period of a refreshingly coherent, thought-provoking, and engag- time. Yet in attributing this outcome to the rise of ing volume dedicated to the use of Imre Lakatos’s rational choice analysis and quantitative methods, methodology of scientific research programs (MSRP) Pierson moves into the realm of assertions that— for appraising progress in the field of International while plausible—he does not have data to support. Relations (IR). After the editors’ detailed overview of Similarly, Pierson argues that statistical tools for MSRP and the controversies it has generated in the estimating complex temporal relationships ‘‘fre- philosophy of science community and other disciplines, quently go unexploited’’ (168), and that while it is Part I of the book moves into a series of applications by possible to incorporate long-term sociological pro- leading figures in IR. Robert Keohane and Lisa Martin cesses into rational choice explanations of compara- reconstruct institutional theory according to MSRP, tive politics, ‘‘they are not the kinds of hypotheses followed by Jonathan DiCicco and Jack Levy on power that these analysts typically go looking for’’ (100). Yet transition theory, Andrew Moravcsik on liberal IR it is questionable whether typical practice is so theory, James Ray on the democratic peace, Stephen myopic at a time when Beck and Katz’s 1995 article Walker on operational code analysis, Robert Jervis on on time-series cross-sectional methods is the ninth the debate between realist and neoliberal IR theories, most cited article in the history of the American Randall Schweller on neoclassical realism, and Jack Political Science Review, and Acemoglu and Robin- Snyder on the empirical aspects of normative research son’s (Economic Origins of Dictatorship and Democ- in areas like ethnic conflict. Part II contains a mixture racy, 2005) rational choice analysis of long-term of commentary on the aforementioned applications, processes of democratization has generated so much critiques of MSRP, and arguments for alternative excitement among comparativist formal modelers. rationalist models of scientific development provided Quantitative and rational choice scholars certainly by David Dessler, Roslyn Simowitz, John Vasquez, approach the temporal dimension in a different and Andrew Bennett. The coherence of the book manner than does Pierson, but his point is that derives in part from strong editorial guidance in the it is rare for these scholars to take time seriously overview of MSRP, but also a conference held in at all. His critiques of tendencies in the discipline Arizona in 1999, where the ideas that shaped this may ultimately be accurate, yet they would be book were debated in spirited fashion. much stronger if they were accompanied by a This book is a fine example of methodological more convincing empirical demonstration of actual work that bridges the quantitative-qualitative divide. practice. The model derived from Lakatos’s dense prose is Despite overstepping the bounds of its own clearly operationalized by Elman and Elman in evidence when critiquing current practice on the Chapter 2 to guide subsequent applications (see also quantitative and formal side of the discipline, Pier- the handy brief guide on pages 19–20). The following son’s book nonetheless offers an excellent demon- chapters describe key elements of theories or stration of best practice on the qualitative side as well successions of theories contained in well-known IR as a convincing claim that all political scientists research programs. The contributors charged with should take time seriously. Some will certainly dis- applying MSRP reconstructed their research pro- pute the book’s arguments, but one suspects that that grams adductively by comparing observations with is precisely the author’s intention. Indeed, the great- the model. est value of Politics in Time lies in the fact that is In keeping with the adductive method, many of certain to stimulate (and indeed, has already begun to the contributors questioned the operational defini- stimulate) an important process of debate about tions created by the editors as they were piecing temporal processes in politics and the ways in which together evidence and theory in their respective different research traditions can better address them research programs. There were disagreements over in the future. the editors’ preferences for heuristic novelty as the operational definition of novel facts, with Moravscik Taylor C. Boas, University of California, Berkeley and Bennett making strong cases for the use of background theory novelty. Most contributors noted Progress in International Relations Theory: Appraising the difficulty of identifying all of the elements of a the Field. Edited by Colin Elman and Miriam scientific research program in a nonarbitrary manner. Fendius Elman. (MIT Press, 2003.) Determining what counts as the hard core of a doi:10.1017/S002238160708022X research program is complicated by Lakatos’s