Rational Choice and Political Behavior

A theme of Aldrich's earliest work is the testing of rational choice models of electoral behavior. His publications in this area follow a distinct framework. Drawing on multiple rational choice models of elections, Aldrich carefully derives testable implications from theory and seeks out areas where expectations for observed political behavior differ. He then structures empirical tests so as to adjudicate between competing hypotheses—there are no straw men here. The work is unusually circumspect: assumptions are clearly stated, and the weaknesses of both the theories and available data are made plain. In a sense, Aldrich's style makes it easier, not harder, for other researchers to dispute his findings. This practice is designed to advance the field and make knowledge increasingly cumulative.

Aldrich's published record in this area begins with what is, perhaps, the earliest empirical application of the spatial model to voter choice in the literature (Aldrich Reference Aldrich1975). Using data from the Survey Research Center's 1968 survey, Aldrich identifies two issue questions for which respondents report their own positions as well as the perceived positions of presidential candidates (Humphrey, Nixon, and Wallace). With these data, he estimates the concave utility distributions of voters and demonstrates the now-assumed correlation between issue proximity and candidate evaluations predicted by the Downsian model of political competition. Importantly, he explicitly adjudicates between several possible constructions of voter utility functions, each consistent with the McKelvey (Reference McKelvey1975) generalization of the Downsian model. This analysis was repeated and amplified with data from the 1972 election (Aldrich Reference Aldrich1977).

A second focus of Aldrich in this area is strategic (or “sophisticated”) voting in the context of multicandidate elections. Rational choice theory anticipates that voters should be willing to “defect” to their second-most preferred candidate based on the “electability” of their most-preferred option. Of course, observing strategic behavior in American general elections is limited because of the dominance of the two major parties. However, examining data from the 1988 presidential primaries, which included precise measures of the perceived “electability” of various candidates, Aldrich and colleagues show that such strategic behavior is common (Abramson et al. Reference Abramson, Aldrich, Paolino and Rohde1992). Indeed, Aldrich's work in this area shows that American voters are often strategic—consistent with rational choice theory—and that third-party candidates during the last half century have done little to erode support for the Democratic and Republican parties.

Perhaps the most well-known controversy at the nexus of rational choice theory and political behavior is the rationality of turnout. Early theorists identified voter turnout as a challenge to the rational voter premise. Because voting is a collective action, the probability of any individual's vote being decisive is small. Theory suggests it is “irrational” for individuals to vote; of course, paradoxically, many do. Downs (Reference Downs1957) posits that turnout must result from nonrational criteria. Riker and Ordeshook reformulate the calculus of voting “so that it can fit comfortably into a rationalistic theory of political behavior” (Reference Riker and Ordeshook1968, 25). This is done via the inclusion of a “D” term, the expressive benefits of voting often linked to one's civic duty. Yet the D term was theoretically dissatisfying to some: “the decision to vote is rendered rational by recognizing individual errors on the one hand and postulating cathartic or psychic rewards on the other” (Ferejohn and Fiorina Reference Ferejohn and Fiorina1974, 525).

Aldrich joined the fray in an early work that sought to distinguish empirically between the predictions of the Riker and Ordeshook expected utility model and the Ferejohn and Fiorina (Reference Ferejohn and Fiorina1974) minimax-regret decision rule. In an article in the AJPS, Aldrich (Reference Aldrich1976) derives the conditions under which the two models offer distinct predictions, and then seeks to arbitrate between them with data on multiple forms of political participation (voting, campaign participation, etc.) He finds that both models receive some support, but the article's main contribution is to specify the difficulties of attacking the problem given the structure of available survey questions and concomitant degree of measurement error.

Aldrich revisited this controversy in the same journal in 1993. In what was to become his most cited article, “Rational Choice and Turnout,” Aldrich steps back from testing the competing models directly to ask a more fundamental question: under what circumstances should we expect rational choice models to be applicable? Put differently, what are the boundaries to the kinds of behavior we should expect rational choice models to explain? The answer Aldrich provides is that “turnout is, for many people most of the time, a low-cost, low-benefit action” (Reference Aldrich1993, 261). He argues that we should not even expect rational choice models to provide much explanatory power for voter turnout. Instead, the low-cost, low-benefit nature of voting actually implies that theories based on either utility-maximization or minimax-regret will offer limited explanatory power relative to sociological or psychological factors.

Political Psychology

The conclusion that voter turnout is a behavior that lies largely outside of a strictly rational-choice interpretation illustrates an important facet of Aldrich's scholarly career. Although a student of the Rochester school, Aldrich's commitment to positivism is not one restricted to rational choice narrowly construed. Rational choice is a theory of choice given preferences, but preferences may have many sources including social identify, racial animus, and inherent cognitive biases. When rational choice theory lacks leverage, other sources for generating theoretical expectations must be brought to bear. As early as 1977, Aldrich argues that “what is needed is a theory about preference formation which can be combined with choice theories to make more realistic deductions” (Aldrich Reference Aldrich1977, 235).

One early example of Aldrich's willingness to step outside the rational-choice box is his joint work with Abramson explaining the decline in voter turnout between 1960 and 1980 (Abramson and Aldrich Reference Abramson and Aldrich1982). They reject the Ferejohn and Fiorina (Reference Ferejohn and Fiorina1974) rational-choice explanation of declining turnout as a function of voters caring less about outcomes. Abramson and Aldrich argue that the primary determinants of declining turnout were lowering levels of party identification and decreasing beliefs in the responsiveness of government—two variables drawn explicitly from the social-psychological model of voting.

Aldrich's research took a more overt turn in the psychological direction during his time at the University of Minnesota, where he had opportunities to collaborate with Borgida, a social psychologist, and Sullivan. His work there advanced several research agendas that sought to draw on the insights of social psychology to provide a more nuanced understanding of the factors affecting voter preferences. So, for instance, Aldrich and his collaborators argue that pocketbook concerns should only outweigh more symbolic (i.e., sociotropic) concerns in the formation of voter preferences when voters feel the issue to be highly salient on their “personal agendas” (Young et al. Reference Young, Borgida, Sullivan and Aldrich1987) and when personal concerns are cognitively accessible (Young et al. Reference Young, Thomsen, Borgida, Sullivan and Aldrich1991).

These works—each published outside of political science—were part of a broader shift in the scholarly consensus about the capacity and competence of voters. In this emerging view, voters' beliefs were neither the nonattitudes outlined in Converse (Reference Converse and Apter1964) nor the ideal concave utility functions demanded by the simplest rational-choice models. Rather, the degree to which any factor will influence voter preferences is highly contextualized (c.f., Zaller Reference Zaller1992).

Aldrich made several contributions to this reconstitution of the social-psychological model. For instance, Aldrich, Sullivan, and Borgida (Reference Aldrich, Sullivan and Borgida1989) directly contest the dual assertions that citizens' positions on foreign policy issues were largely nonattitudes and played little role in presidential voting behavior. The authors argue that foreign policy attitudes were not chronically accessible, but temporarily accessible when primed by the information environment and relevant to voting. Thus, candidates who campaign on foreign policy issues are not simply “waltzing before a blind audience,” they are making their appeals to a potentially receptive electorate. Moreover, while the authors of The American Voter contended that candidate appraisals were primarily projections of partisan bias, Sullivan, Aldrich, Borgida, and Rahn (Reference Sullivan, Aldrich, Borgida and Rahn1990) argue that these affective orientations were constructed from meaningful assessments of candidate competence and personal qualities. Finally, Rahn, Aldrich and Borgida (Reference Rahn, Aldrich and Borgida1994) show that the way that information is presented to voters—whether information was organized by candidate or by policy—can itself determine what factors go into the construction of preferences.

Spatial Models of Elections

The dynamic interaction between strategic elites and semistrategic members of the mass public is a common undercurrent of Aldrich's work. This is most evident in his several contributions to the literature on formal models of elections.

Aldrich's first published model, “A Dynamic Model of Presidential Nomination Campaigns,” appeared in the APSR (Reference Aldrich1980b) and is a particularly fine example of his somewhat unusual approach to theory building. To begin, the model is deeply entwined with Aldrich's understanding of the empirical record (Aldrich Reference Aldrich1980a). The model itself is informed by the successes of the McGovern and Carter campaigns in the Democratic primaries of the 1970s (see also Aldrich Reference Aldrich2009) and by the strategic nature of voters' accounting for the electability of candidates. He develops a “momentum model” of primary campaigns in which candidates whose performance exceeds expectations receive a boost in resources and perceived electability that in turn increases their performance in subsequent elections. Likewise, candidates who underperform expectations are subject to a penalty. The model's predictions are not of a specific equilibrium but rather multiple equilibria and extreme sensitivity to initial conditions. He shows that candidates who beat expectations in early elections start a positive feedback loop, significantly improving their long-term chances. Thus, the necessary conditions for winning a primary are a strong or stronger-than-expected performance in early primaries paired with a substantial organization.

The implications of this model for our understanding of the primary system are enormous. In a system with so much positive feedback, tiny changes in initial conditions can radically alter results in unexpected ways. Thus, the ultimate outcome is determined conditionally by “expectations,” which themselves are not determined in a democratic arena, but rather in the “invisible primary” dominated by activists, donors, and media figures. This theoretical framework has been unusually prescient. No major-party candidate has won their respective party's nomination without winning in either Iowa or New Hampshire (except for Bill Clinton, whose unexpectedly strong performance in New Hampshire was pivotal in establishing him as a legitimate option in 1992).

Aldrich's most important contribution to the field of spatial modeling—to date—was equally inspired by empirical patterns that seemed to require additional microfoundations. Specifically, he was interested in the process by which party activists sort themselves ideologically so that, for instance, conservative activists disproportionately support Republicans, liberal activists tend to support Democrats, and moderates tend to abstain from political activism. He wished to better understand how these party cleavages arise and how they change.

The initial paper on this topic appeared as “A Downsian Spatial Model with Party Activism” in the APSR (Reference Aldrich1983a). Aldrich combines Downs's spatial model with Riker and Ordeshook's calculus of voting. He assumes that the benefits of participation for activists increase as parties become more distinguishable in the policy space and that activists' mean policy positions influence the perceived positions of their respective parties. Although the model predicts a stable equilibrium, it is actually a dynamic system where activists join or abandon parties in response to the decisions of other activists. The conclusion he reaches, with limited assumptions, is that party cleavages will emerge endogenously such that (1) more ideologically extreme activists associate with the parties, (2) moderate activists abstain from the political system, and (3) the parties are separated in the policy space.

A companion article (Aldrich Reference Aldrich1983b) appeared in the same year in Public Choice and was more overtly targeted at the dynamic nature of party coalitions (cf., Sundquist Reference Sundquist1983). Here, Aldrich explored how changes in the saliency of issue dimensions can lead to sudden reconfigurations of both candidate positions and activist coalitions. Subsequently, Aldrich worked with his student, Michael McGinnis, to extend the model in several directions, most notably to more fully accommodate multidimensionality (Aldrich and McGinnis Reference Aldrich and McGinnis1989).

Aldrich's theoretical work in this area is the foundation for several branches of scholarship. First, it is the basis for many important innovations for modeling two-party elections in a multidimensional policy space (e.g., Miller and Schofield Reference Miller and Schofield2003). Second, the need to satisfy a relatively ideological extreme activist base sets the stage for elected officials in each party to work together to maintain a noncentrist party brand, an essential assumption behind the theory of conditional party government.

Why Parties? and Conditional Party Government

Why Parties? (Reference Aldrich1995, Reference Aldrich2011) is John Aldrich's magisterial account of why political parties emerged in the early days of the American republic and why they remain, through varied transformations, a necessary condition for democracy. The book provides a sweeping historical analysis of America's party systems, but it offers much more than history. Aldrich argues that parties are an endogenous solution to problems endemic to republican governance, problems of (1) social choice and voting, (2) public goods and collective action, and (3) political ambition. Aldrich shows that parties provide equilibrium solutions to these problems, thus inducing stability into the political system. Yet, because endogenous institutions are shaped by ambitious office seekers and activists, the form and historical path that parties have taken is highly contextualized.

The first parties, according to Aldrich, originated as a solution to one of the fundamental problems of social choice: majority rule instability. Jefferson, Madison, and Hamilton were unacquainted with Kenneth Arrow, of course, but they had firsthand experience with the Arrovian problem of unstable voting coalitions. Aldrich contends that political preferences during the First Congress were two dimensional. The first dimension concerned how large and active the new central government should be. This “great principle” dimension is evident in the debate about Hamilton's fiscal plan. The second dimension divided legislators by particularized or regional interests and is epitomized by the debate about the location of the new capital. In a multidimensional choice setting, majority rule may lead to “chaos” (McKelvey Reference McKelvey1976; Schofield Reference Schofield1978), where a majority coalition can be overturned by another coalition, potentially creating a cycling of preference outcomes. Aldrich, however, shows that political parties can use agenda control, much like congressional committees (Shepsle and Weingast Reference Shepsle and Weingast1981), to create a structure-induced equilibrium (Shepsle Reference Shepsle1979) to resolve this instability. Indeed, in a 1994 article in Legislative Studies Quarterly, Aldrich provides a formal model of this equilibrium result.

Aldrich's account of the emergence of American parties differs from the standard historical view, which argues that the advent of the first party system marked a radical discontinuity with the politics of the ratification period. Jefferson's Republicans and their antiadministration ideology, it is argued, arose whole cloth in reaction to Hamilton's program, and a new brand of politics was born. Yet, by focusing on the preferences of those involved, Aldrich shows that significant continuity actually existed across the periods. Aldrich and Ruth Grant (Reference Aldrich and Grant1993) argue that the issue dimensions of the First Congress mirror those under the Articles of Confederation. These issues did not come to the fore, however, because the supermajority rules governing the Congress of the Confederation limited the formation of winning coalitions. The switch to majority rule in the First Congress placed those issues on the agenda and incentivized the creation of parties as a means of holding together potentially shifting majorities.

The rise of Jacksonian democracy altered many of the rules governing elections and greatly expanded suffrage. It would no longer suffice for parties to be predominately legislative entities; instead parties became diffuse campaign organizations charged with mobilizing electoral support. In short, the mass political party was born.

As Aldrich notes in Why Parties?, the “problem of creating and sustaining a campaign organization is a … typical example of the theory of collection action, one with large costs and large potential benefits” (Reference Aldrich2011, 107). The Democratic Party, largely through the effort of Martin Van Buren, overcame two types of collection action problems. At the mass level, voter turnout was the fundamental concern. Aldrich—building on his account of turnout and rational choice theory (Reference Aldrich and Grant1993)—argues that the Democrats offset the relatively small costs of voting by reducing voters' information costs and increasing the expressive benefits associated with participation. Both were accomplished through rallies, parades, speechmaking, and the like; campaigns became entertainment. At the elite level, the question was how to convince politicians and activists to affiliate with and maintain loyalty to the party. To be sure, the creation of the spoils system with its provision of selective benefits is vital to telling this story. Yet Aldrich shows that career ambition alone incentivizes office seekers to align with parties. First, “[a]ffiliation with a party provides a candidate with…a ‘brand name’ ” (Aldrich Reference Aldrich2011, 47) and with it economies of scale. Second, because parties offer access to the ballot through nomination, ambitious politicians have a higher probability of election if they affiliate with a party than if they run as an independent, even in cases where parties are weak (Aldrich and Bianco Reference Aldrich and Bianco1992).

Aldrich's most distinct contribution in Why Parties? may be his interpretation of the modern party system. (Indeed, a fuller consideration of modern parties is the main advance of Why Parties? A Second Look (Reference Aldrich2011).) The view of “parties-as-organization,” built to solve the collective needs of its affiliates, is perhaps the lasting image of political parties, according to Aldrich, because it depicts parties during the “golden age” of their influence, but it provides an inadequate account of modern parties. Specifically, Aldrich (Reference Aldrich2011, 275) argues that the 1960s represented a punctuated equilibrium, defined as “long-term equilibrium punctuated by a short, intense period of rapid change, leading to a new long-term equilibrium.” In his view, the decade marks the death of the mass parties of old. “Before the critical era of the 1960s,” Aldrich (Reference Aldrich2011, 281–82) writes, “candidates had no alternative to using the party organization to gain access to office. There was effectively no technology by which an individual, except the very most well known, could create a personal campaign organization.” The advent of television, for example, and emergence of a policy-motivated activist class, allowed ambitious politicians to bypass the party organization and create “candidate-centered” campaigns.

Yet parties—as endogenous institutions—have adapted, and a new party system has emerged. Today's party “is ‘in service’ to its ambitious politicians…not ‘in control’ of them as the mass party sought to be” (Aldrich Reference Aldrich2011, 285). The modern party, particularly at the national level, is highly professionalized and supplies political and technical expertise, as well as financial and in-kind resources to its candidates. This transformation, according to Aldrich, has brought about a nationalization of elections. With local party machines a thing of the past, two-party competition has spread nationally, partisan cleavages have become more consistent across districts and states, and party brand names have greater value. In turn, party cleavages in Congress have become increasingly polarized and party government is in ascendency.

Aldrich argues that parties are equilibrium solutions to the problems inherent to the republican principle. They provide stability, but only in response to the actions and preferences of voters and ambitious politicians. Thus, parties—particularly parties in government—will vacillate in strength. The conditions underlying the periodicity of partisan strength in legislatures are formulated in Aldrich's (Reference Aldrich1994) “A Model of a Legislature with Two Parties and a Committee System.” Periods in which legislative parties seem more consequential vis-à-vis legislative committees are related to instances when there is “relatively extensive shared policy preferences among members of the same congressional party and relatively little commonality with members of the opposing parties” (Aldrich Reference Aldrich1994, 331). In such instances, the majority party is most likely to maintain agenda control. When there is less differentiation in the distribution of interparty preferences, agenda control is most easily achieved by committees. Thus, Aldrich argues that agenda power in a legislature such as Congress is likely to alternate over time between parties and committees as conditions alternate. Note that the conditions underlying partisan strength—the distributions of partisan preferences within the legislature—are not determined in the legislature but instead reflect electoral outcomes.

This dynamic interplay of parties-in-elections and parties-in-government is consistent with Rohde's (Reference Rohde1991) theory of “conditional party government” (CPG). Indeed, by the mid-1990s, the CPG research program had become a collaborative enterprise for Aldrich and Rohde. The theory posits that party governance is most likely under the joint conditions of intraparty homogeneity and interparty heterogeneity of preferences. Under these conditions, election-minded legislators are most likely to delegate strong power to their party leadership to secure shared policy and reputational goals (e.g., Aldrich and Rohde Reference Aldrich, Rohde and Stonecash2010). These leaders serve as agents of the caucus and exercise sway over the agenda to bring proposals beneficial to the majority party to the floor, block those they oppose, and otherwise act to safeguard the party's electoral fortunes.

The CPG research program has progressed in numerous important ways. One early advance was the demonstration that CPG theory was not specific to the reform period of the 1970s and Democratic Party control. For example, Aldrich and Rohde (Reference Aldrich and Rohde1997–98; Reference Aldrich, Rohde, Bond and Fleisher2000) show that partisan governance in the House following the transition to Republican rule in the 1990s was consistent with CPG theory. Likewise, Aldrich and Battista (Reference Aldrich and Battista2002) find empirical support for CPG theory at the state level. Another key advance in the CPG program involved empirical measurement. Aldrich, Berger, and Rohde (Reference Aldrich, Berger, Rohde, Brady and McCubbins2002) use the NOMINATE scaling procedure to create a measure of the historical variability of conditional party government in the House and Senate since the 1870s. Aldrich, Rohde, and Tofias (Reference Aldrich, Rohde, Tofias, Brady and McCubbins2007) expand the measure to account for multi-dimensional preferences. In sum, the CPG theory jointly developed by Aldrich and Rohde has proven to be fundamental to our understanding of Congress and legislative parties broadly.

Just as “democracy is unthinkable save in terms of parties” (Schattschneider Reference Schattschneider1942, 1), our scientific knowledge of political parties is unthinkable save in terms of Aldrich's research. Over his career, he has contributed greatly to the study of parties as conduits between the governed and their government. In recognition of its contribution to the field, Why Parties? The Origin and Transformation of Political Parties in America was awarded the APSA's Gladys Kammerer Award in 1996, which is awarded annually to the best book published in the field of US national policy. In 2008, Aldrich received the Samuel Eldersveld Award for career achievement from APSA's section on political organizations and parties.

Political Methodology

Although Aldrich's primary interests have always been the building and testing of theories of political competition, he has also made several important contributions to political methodology. For a generation of political scientists, John Aldrich provided their introduction to statistical models with dichotomous dependent variables (Aldrich and Cnudde Reference Aldrich and Cnudde1975; Aldrich and Nelson Reference Aldrich and Nelson1984). Aldrich's first published article with Charles Cnudde (Reference Aldrich and Cnudde1975) admonished the discipline against the use of OLS regression when the dependent variable is binary, advocating instead for a probit model. Although originally developed by the statistician D. J. Finney in 1947, probit was then rarely used in political science. Indeed, a search of the JSTOR archive shows that the probit model (along with logit) was used in political science only 18 times before 1975. In the decade that followed Aldrich and Cnudde's article, the use of binary choice models increased sharply, with more than 320 citations. Aldrich's “little green” Sage monograph with Forrest Nelson (Reference Aldrich and Nelson1984) expanded the treatment of the subject, offering students a critical introduction to both the probit and logit models, and became—for many years—the primary text by which statistical models with dichotomous dependent variables was taught to political scientists. This text alone has been cited more than 3,000 times.

A second lasting (and arguably underappreciated) methodological contribution of Aldrich's is his APSR article with Richard McKelvey (Reference Aldrich and McKelvey1977), “A Method of Scaling with Application to the 1968 and 1972 Presidential Elections.” The Aldrich and McKelvey (A-M) scaling procedure uses information from voters' own policy positions and their perceptions of candidates on those same issues to generate estimates of candidate locations and voter ideal points. The method is exceptionally flexible, not even requiring that respondents interpret questions identically (King et al. Reference King, Murray, Salomon and Tandon2004).

By explicitly applying the spatial model to observable statements of preferences to make inferences about the ideological position of respondents and candidates, the A-M technique foreshadowed the roll call scaling that became so dominant in the congressional literature (cf., Poole and Rosenthal Reference Poole and Rosenthal1997; Clinton, Jackman, and Rivers Reference Clinton, Jackman and Rivers2004). In a recent retrospective on the life and work of McKelvey, Keith Poole states that he “has repeatedly prodded his methodological friends to read the A-M paper…because of its mathematical elegance.” He continues, “it is a beautiful tour de force” (Aldrich and Poole Reference Aldrich, Poole, Aldrich, Alt and Lupia2007, 100).

Another methodological achievement comes in the area of survey methods. In Aldrich's early work testing implications of the spatial model for public behavior, he was forced to rely on survey questions that were poorly suited to the task. Aldrich and colleagues, therefore, set out to amend the standard questions in the ubiquitous American National Elections Study to better facilitate testing in a spatial framework (Aldrich et al. Reference Aldrich, Niemi, Rabinowitz and Rohde1982). The modified questions set out specific policy alternatives, clearly identified the status quo policy, and implemented a branching technique to measure reliably the positions of both respondents and the perceived position of candidates. This now-familiar question format still appears on prominent national surveys.

Aldrich's continuing interest in the structure of surveys is reflected in the prominent role he has held in two of the most important political survey projects. He has served on the board of the ANES since 2002, and beginning in 2009 he has served as its chair. He also serves on the planning committee for the Comparative Study of Electoral Systems and edited a recent book on improvements to the ANES (Aldrich and McGraw Reference Aldrich and McGraw2012).

Bridging the Divide: The Creation of EITM

In July 2001, the National Science Foundation's (NSF) Political Science Program held a workshop to foster dialog between the formal theory and empirical modeling communities under the leadership of Jim Granato, an Aldrich student who was then the program's director. During the 1990s in particular, theoreticians and empiricists were each growing in sophistication and were doing so largely independently of one another. Advanced graduate training in one area typically occurred at the expense of advanced training in the other. The NSF workshop, “Empirical Implications of Theoretical Models” (EITM), hoped to bridge the divide and foster research linking formal theory with empirical modeling. The workshop gave rise to the EITM Summer Institutes, intensive month-long programs aimed at providing graduate-level training that links formal and empirical analysis and creating a network of scholars who recognize the value of a closer relationship between rigorous theory development and empirical causal inference.

It is no surprise that Aldrich was an integral player in the emergence of the EITM movement. While some may think of EITM as a new approach to the study of politics, Aldrich undoubtedly recognized it as consistent with the positive tradition in which he was trained. Aldrich was not only a participant in the formative NSF EITM workshop; he was the principal or co-principal investigator on each of the three NSF grants that funded the Summer Institutes between 2002 and 2012. He also served as an instructor and mentor at each Summer Institute held during that time.

The close relationship between formal theory and empirical analysis has always been at the heart of Aldrich's work. To be sure, the Aldrich-McKelvey scaling method remains an exemplar of EITM-type work. Yet the best statement of Aldrich's approach to political science research is taken from an excerpt found early in his book, Before the Convention. The excerpt is restated in a recent essay on the origins of “the EITM approach” (Aldrich, Alt, and Lupia Reference Aldrich, Alt, Lupia, Box-Steffensmeier, Brady and Collier2008) in The Oxford Handbook on Political Methodology:

A Life in Political Science

In May 2013, Aldrich was awarded the University of Rochester's Distinguished Scholar Award, given in recognition of alumni whose careers in academia, industry, government, or the arts have exemplified the values of the university. Aldrich, undoubtedly, has done great honor to that university and to the political science graduate program envisioned a half a century ago by William Riker.

Aldrich's contributions to the political science discipline have been duly recognized. He is a fellow of the American Academy of Arts and Sciences (2001) and received the John Simon Guggenheim Fellowship in 2012. In addition to the Eldersveld Award cited earlier, Aldrich also received the 2012 Frank Johnson Goodnow Award from the American Political Science Association for his career of service to the discipline. His scholarly work has earned APSA's Gladys Kammerer Award, the Heinz Eulau Award, the CQ Press Award, and the Pi Sigma Alpha Award. He has been co-editor of the American Journal of Political Science (1985–88), and he has served as vice president (1995–6) and president of the Southern Political Science Association (1998–99) as well as president of the Midwest Political Science Association (2004–2005). He has served as the secretary of APSA (1993–94) and chair of the APSA Task Force on Interdisciplinarity (2006–07). He has also been a Fellow at the Center for Advanced Study in the Social and Behavioral Sciences (1989–90) and at the Rockefeller Center in Bellagio, Italy (2002).

These recognitions reflect a career that has left a significant imprint on the discipline, contributing to some of the most important intellectual trends of the past four decades. His work has advanced formal modeling, data collection and analysis, and the marriage of the two. His research helped revive political parties—and political institutions more generally—as a focus of scientific inquiry. He has contributed to our understanding of the dynamic interactions between political elites and voters, and helped revive the thinking—if not always rational—of the voter as central player in American democracy. Even more, he has built bridges across disciplines, helped develop and maintain important intellectual institutions, and served as a mentor and a role model to young scholars from around the world. In 2013–2014, Aldrich will serve the discipline as president of the American Political Science Association. In this role, we expect that he will continue to have a positive influence on our discipline.

Acknowledgments

The authors would like to thank Brendan Nyhan and David Rohde for providing commentary on an earlier draft of this paper. As students of John Aldrich, we thank him for being an excellent mentor and friend.