A SEMINAL TEXTBOOK: COMPARATIVE POLITICS TODAY

Before turning to Powell's substantive research contributions, it is important to recognize the impact he has on the field through his leadership in editing and writing the most widely adopted textbook on comparative politics. While he was a PhD student working with Almond at Stanford University, they began a collaboration that would span nearly four decades. They first collaborated on an introduction to comparative politics that provided the theoretical explanation of Almond's “structural-functional” approach to comparing nations. Structural-functionalism offered a method to compare nations that seemed like apples and oranges, but in essence were performing similar governing functions in different ways.

This first Almond-Powell collaboration was published as Comparative Politics: A Developmental Approach in 1966. The book was written as a graduate-level introduction to the rapidly expanding field of comparative politics, but it became more. The book introduced the structural-functional framework for comparing nations that is still widely used today as a research and pedagogical tool. This framework also introduced a host of concepts that became part of the standard vocabulary of political science. For example, the interests of citizens and social groups were expressed in a process of “interest articulation,” and political parties and other institutions undertook a process of “interest aggregation” to combine these policy demands into governing programs. The book also institutionalized other contemporaneous ideas of how one compares nations. It was the most influential framework for the study of comparative politics, becoming required reading for graduate students, and many faculty, across the country.

After beginning his own professorial career at the University of Rochester, Powell returned to Stanford in the summer of 1975 to collaborate on an expansion of the structural-functional framework to incorporate the public policy dimension of government. Over the next couple of years Powell and Almond revised their first book, reworking the overall framework, exchanging chapters, and incorporating policy making and policy outputs into their presentation. The most impressive feature of this work is the theoretical integration of various elements of the political system to provide a holistic view of comparative politics. Institutional structures were linked to political culture and patterns of interest articulation, and patterns of interest articulation were connected to patterns of public policy. They also tried to integrate many empirical studies, including their own, into the general framework. This collaboration produced a virtually new book: Comparative Politics: System, Process, and Policy published in 1978.

The success of Comparative Politics led to an even larger collaboration between Almond and Powell. Having written a theoretical text for graduate students in comparative politics, they sought to bring this expertise to undergraduates. A slimmed-down version of structural-functional theory was extracted from their 1978 book and provided the introductory theory chapters for Comparative Politics Today: A World View published in 1980. The Almond-Powell textbook was aimed at undergraduate students, and they assembled a group of distinguished country specialists to apply the structural-functional framework to a diverse set of nations: Britain, France, West Germany, USSR, China, Mexico, and Tanzania. This mix of country-study chapters, discussed in a common theoretical framework, became the model for teaching comparative politics. Comparative politics no longer meant the study of nations besides the United States; it became a field that compares nations.

For more than three decades the Almond-Powell textbook has been the leading text in introductory comparative politics courses and, more recently, for AP classes in comparative politics. In subsequent editions the number of country-study chapters grew; from the initial seven to 12 (including the United States) in the current 10th edition. Each edition tried to steer a course that gave students information about other nations, theories about comparative politics, and a framework for understanding this information. Inevitably, the text editions were also shaped by Powell's own research studies on political behavior, institutions, conflict, and representation (discussed more later) and helped communicate them. With Almond's semi-retirement and then death in 2002, the book took on new coeditors (Russell Dalton and Kaare Strøm), as well as the imprint of Powell's direction. With each edition, thousands of undergraduate students expanded their world view and learned to compare nations from the Almond-Powell text.

Textbook authors have to be sensitive about events in the world to share these events with students. When the Soviet Union began to crumble, the world was changing in fundamental ways. Almond and Powell first reworked the language and examples in the core text and encouraged the country specialists to do the same. As a later response Almond, Powell, and Dalton expanded the Comparative Politics Today franchise with a new text, European Politics Today published in 1998. Theoretically, the text continued the structural-functional framework, as well as integrating middle-level theories from rational choice, neo-institutionalism, and other perspectives. It was one of the first European politics texts to compare the new Europe that included nations from both West and East, including four new country studies using the Almond-Powell framework. European Politics Today is now in its fourth edition.

In summary, beginning with his collaboration with Almond on these several projects, Powell has made a major impact on the teaching of comparative politics that reaches far beyond his classrooms at the University of Rochester. His work in this area came at a crucial time when the modern field of comparative politics was developing, and his texts shaped the development of this field. We imagine that, with embarrassing regularity, Powell meets new assistant professors who say that their first comparative politics book as a first-year college student (or high school AP student) was Comparative Politics Today. Along with the substantive contributions of his research, to which we turn in the next section, this is a large part of Powell's legacy as a teacher and scholar.

CONFLICT AND CONFLICT PREVENTION

From the very beginning of his career, political conflict and its prevention have been a focal point in Powell's research. One sees this interest at the earliest stages of his career, in Social Fragmentation and Political Hostility, published in 1970, which we discuss more fully as we consider Powell's contributions to the study of mass political behavior. Subsequent research built on that early work to explore micro mechanisms that explain patterns of conflict, both among elites and masses. This research exemplifies Powell's commitment to using careful, theoretically grounded empirical research to make prescriptive arguments about how to manage serious social problems like conflict.

Powell and Stiefbold (Reference Powell and Stiefbold1977) use interviews with local politicians in 50 Austrian communities to present and test three middle-range theories of elite behavior, based on assumptions that anger, bargaining, or political mobilization, respectively, is the driving force of elite conflict. Thus, political violence can be understood as an expression of personal anger or frustration (caused, perhaps, by a sense of deprivation), or as an instrumental outlet for strategic behavior aimed at bettering one's political rewards (as rational choice theorists would generally argue), or it may be a response to mobilizational efforts by political elites such as party leaders. Powell and Steifbold find some empirical support for each of these propositions. Thus, party boycotts of local council meetings in these Austrian communities seemed strategically motivated, whereas personal attacks on one politician by another more commonly reflected personal anger and animosities. Finally, casting dissenting votes in the council was most clearly associated with campaign and mobilizational efforts by the relevant political parties.

Powell's early research on political conflict, based on his Austrian field research, was an important effort to provide micro foundations for a literature that at that point was often structural and mechanistic. Yet, Powell's most ambitious and systematic study of conflict and conflict resolution is without doubt Contemporary Democracies, his widely recognized book first published in 1982, which was awarded the Woodrow Wilson Prize by APSA for the best book in political science.Footnote ¹ In Contemporary Democracies, Powell examines across all contemporary democracies the environmental, institutional, and party system effects on regime performance. His performance indicators include the maintenance of political order, in addition to citizen participation and government stability. All of these are foci of much of Powell's accumulated scholarship, but our focus here will be on the first of these dimensions, the maintenance of political order (through the prevention of political violence).

In Contemporary Democracies, Powell utilizes two different measures of political violence: the mean annual number of riots and the number of deaths to political violence. His analysis carefully examines the connections from socioeconomic conditions, through structures of the political environment, at the elite as well as the mass level, to the eventual conflict outcomes. Thus, Powell demonstrates that political violence, although certainly correlated with economic development, inequality, and ethnic diversity, is perhaps most directly affected by the strategies of political elites, and especially political parties. When even smaller or more extreme parties are incorporated into established political institutions and given legislative and executive representation, the levels of political deaths and rioting will tend to be lower. In contrast, popular support for extremist parties is positively associated with rioting.

In his concluding chapter, Powell discusses themes that would carry over to his subsequent scholarship and especially to the role of majoritarian versus consociational political institutions in containing or preventing political violence. Consistently with his analysis of political violence, Powell acknowledges that “the successes of these examples of consociational practice cannot be ignored” (p. 214), but adds that “the willingness of all the parties to work toward accommodative bargains is critical in the success of consociational democracy.” Representational party systems, which combine nonmajoritarian electoral rules with strong party-group linkages, have some distinctive consequences: they funnel political dissent inside the party system and existing constitutional channels, which leads to lower rates of riots and random violence, though it also means lower government stability and more challenging legislative relations. In many circumstances, Powell would argue, the latter is a price worth paying.

In a subsequent contribution to this literature, Powell (Reference Powell1986) examined two puzzles in the relationship between political elite behavior and political violence. The first puzzle has to do with the interpretation of causality in the well-established association between extremist parties and political instability. Do extremist parties bring about political instability, or does causality run in the opposite direction? In his second puzzle, Powell asks precisely what it is about party extremism that causes problems for democratic stability. Is it distance from the pivotal parties in left-right policy terms, or is it the rejection of the established rules of the game that such parties often represent? Based on analysis of survey data from twelve European democracies, Powell concludes that at least as far as street violence, riots, and so forth are concerned, parties are barometers rather than causes of unrest (in the case of legislative disorder, though, the picture is a bit different). The characteristic of parties that matters most is their alienation from mainstream politics rather than their extremism along a conventional left-right axis.

Powell's scholarly results broadly support those who see power sharing, or more generally political inclusiveness, as the most promising way to prevent or contain civil conflict, especially in its most severe and violent forms. But his arguments are more nuanced, and his analysis more fine-grained, than much of the literature in this field. Political conflicts are not simply reflections of poverty, “ancient hatreds,” or popular alienation, even though such forces matter. Political institutions present opportunities for political leaders and their organizations to enlist popular grievances and demands for constructive or conflictive purposes. Thus, a key feature of this research is the emphasis it places on elite behavior. Institutions, as well as the “demand side” of mass-level politics, influence political conflict. But Powell stresses the constraining effects that party leaders have through their choice of whether to mobilize citizens inside or outside constitutional avenues of political involvement.

MASS POLITICAL BEHAVIOR INDIFFERENT CONTEXTS

From the outset of his career, Powell clearly recognized that one could not study a substantively important topic like violence without considering mass behavior, elite behavior, and their interaction. In this regard, he was outside the mold of typical political behavior studies, which paid little attention to the way in which political context conditioned mass behavior. A central contribution of Powell's work therefore has been his pioneering efforts to bring together cross-national aggregate contexts and individual behaviorthe sort of work that is today embodied in the World Values Surveys and the Comparative Study of Electoral Systems, and the analyses that these databases engender. Powell's central contributions to this end are on the topics of political participation and economic voting.

First consider participation. In his first published article (Nie, Powell, and Prewitt Reference Nie, Powell and Prewitt1969), Powell and his coauthors used multilevel analysis of the Almond and Verba five-nation data to seek out the sources of cross-national variation in political participation. The study was based on the modernization theories of the day, asking “What is the process by which economic and political modernization lead to increased participation?” The study offers what may have been the first explicit multilevel analysis in the field, analyzing the interplay of national social structures, individuals' social positions, and individuals' motivations.

This paper was a thorough piece (43 APSR pages long — a paper that might later have made APSR editor Powell blanch; fortunately, editor Ranney did not). It was path breaking not only in its methodology and its findings, but in the further questions it raised. For instance, the authors presaged the now-classic work of Verba, Nie, and Kim (Reference Verba, Nie and Kim1978), with its conclusion that social inequalities in participation are caused by an absence of cleavage-based political organizations:

Organizational membership, then, could be a political resource for the lower classes. Were this resource equitably distributed, the criteria for pluralistic democracy advanced by Dahl, Key, and others would be more nearly met…. Alterations in the organizational structure, then, can serve to correct the tendency for even the most democratically organized societies to allow a disproportionate amount of political influence to be exercised by the well-to-do. (Nie, Powell, and Prewitt Reference Nie, Powell and Prewitt1969, 826).

A decade later Powell returned to participation, this time to the question of voter turnout, first with a cross-national study (Powell, Reference Powell and Rose1980) that has often provided a structure for later comparative analyses of turnout. His basic structure of socioeconomic environment, constitutional setting, and party system provided the framework, for instance, of the 91-country study by Blais and Dobrzynska (Reference Blais and Dobrzynska1998). It also provided the basis for his other major work at about that time, Powell (Reference Powell1986a), where he concluded from a comparative multilevel analysis that political attitudes in the United States increase participation by several percentage points compared with other countries, but that this effect is swamped by the negative impacts of our registration laws and the party system. “American Voter Turnout in Comparative Perspective” has proved to be one of Powell's most influential works.

Powell's work on economic voting, like his work on participation, was among the first to examine carefully how context affects voting. One should think that when the economy stagnates, the incumbent government would incur electoral loss. But this often does not occur — voters do not hold the incumbent responsible to the same degree in all contexts. Powell and Whitten's (Reference Powell and Whitten1993) work on economic voting set new standard for our analyses of accountability in general and the relationship between economic performance of the incumbent government and voter behavior in particular.

The work's central contribution lies in developing and testing an argument about how “clarity of responsibility” affects the ability of citizens to hold politicians accountable for economic outcomes. On one hand, blame and responsibility depend on political institutions. The more voters perceive the control of policymaking as unified, the more they assign responsibility for policy outcomes to the incumbent. Diffusion of responsibility, on the other hand, helps incumbents to avoid being penalized at the polls.

Powell and Whitten conceptualize responsibility as affected by relatively stable factors (ones whose direct source is institutional design) as well as changing ones. On the former, they include voting cohesion of governing parties — how much representatives or party factions vote together or oppose one another — and inclusion of the committee system in parliament — whether the opposition is granted key roles in the leadership of parliamentary committees. The changing factors include minority versus majority government, divided or unified partisan control under bicameralism, and single- or multiparty coalition governments (although split bicameral control and the number of parties are not found to have an effect).

Following this work, “clarity of responsibility” has become a staple in comparative analyses of voter behavior. But the work makes other important scholarly contributions as well. First, Powell and Whitten show that voters do not simply penalize or reward the incumbent by economic performance. They do so by economic performance compared with that in other countries. Additionally, in analyzing electoral fortunes of a party, they argue, one should take into account swings in support for the party in the recent past, as such swings are often short term and overstate the electoral fortune or lack of support for a party. Lastly, left- and right-leaning governments are punished differently for economic outcomes. While growth helps the incumbent government, the left is penalized for high unemployment rate while there is no clear evidence that the right is penalized for high rate of inflation.

The study shaped the debate for years to come. Hundreds of works that followed Powell and Whitten's study have drawn on their insights, grappled with their findings, and extended their analysis. The work is a natural springboard for almost any subsequent research on the topic.

Although Powell's most influential contributions on political behavior lie in his use of multilevel designs to study political participation and economic voting, it is important to recognize his other early work on behavior, which while lacking the multilevel framework for which Powell is so well known, nonetheless made important contributions to our understanding and interpretation of core hypotheses in the behavior literature. Consider his first book, Social Fragmentation and Political Hostility, published in 1970. This was a close study of social divisions and their management in Hallein, a small city in the state of Salzburg, Austria (and at one time home to Franz Xaver Gruber, composer of the carol “Silent Night”). Powell focused on a classic hypothesis in comparative politics, that cross-cutting cleavages reduce hostility and conflict, while overlapping cleavages (“fragmentation”) increase them. The pillarization of Austrian society into Socialist and Catholic Lager (“camps”), each with its own major party, makes for a clear test of the theory. Unlike the United States, with its profusion of cleavages and identities, Austria has just one primary, dichotomous split in the social structure. Cross-pressured individuals (for example, observant Catholic socialists) are easily identified. Powell conducted his own public opinion surveys to relate the pattern of cleavages and party loyalties to expressions of partisan hostility. Both elites and ordinary citizens were surveyed.

Powell finds, with numerous qualifications, that the theory of overlapping cleavages holds in his city. Political elites in particular express considerable hostility toward each other, they dislike cross-party marriage, and they often find it difficult to communicate and interact. A somewhat muted version of the same pattern exists at the mass level. Powell spells out the implications for the political system, in particular the many deleterious consequences of elite polarization. Writing at a time when many American political scientists still longed for cohesive political parties with sharply differentiated policy positions, Powell sounded a cautionary note that appears all too prescient in retrospect. And he set a standard for historically knowledgeable, politically sophisticated, and inferentially persuasive analysis of observational data.

Powell (Reference Powell1976) followed up the book with a more detailed study of cross-pressure theory in Austria, this time using a national sample. He finds again that the theory is fundamentally correct, but that it often makes less difference than theorists imagine. Individuals turn out to have mixed groups of friends, so that the homogeneous social groups that would give the theory great power simply do not exist for many people. And at the attitudinal level, he finds that about one-half of Austrians do not understand the relationship between the camps and the parties: They do not recognize that the Catholic-oriented conservative party represents Catholics, for example, or that the Socialists represent workers. Thus the cross-pressures imagined by theorists are often weak: many individual citizens have no sense that they are in the “wrong” party. In consequence, the theory holds, but has only modest effects.

Comparative behavior scholars will not be surprised that key issues in politics escape the grasp of many ordinary citizens. But it is rare to see the implications of that well-known fact spelled out for midrange sociological theorizing about politics. Thus Powell's argument is an unusual and very attractive use of micro-level empirical analysis to qualify meso-theoretical claims.

Powell (Reference Niemi, Powell and Stanley1985, with Niemi, Stanley, and Evans) took up a different aspect of partisanship—its development over time in new electorates. Studying newly enfranchised populations in varying circumstances (Swiss women given the vote; blacks in the southern United States), they carry out a careful test of Philip Converse's well-known model of the acquisition of partisanship. Converse posited that parents probabilistically transmit party identification to their children if the parents had it. After that the act of voting reinforces partisan dispositions. In a nuanced discussion carefully distinguishing different cases, Powell et al. show that this framework predicts poorly in several different new electorates. New voters have too much partisan strength. Anticipating subsequent work on adaptive and Bayesian models of partisanship, the article concludes with a discussion of how partisan learning is fundamental, not the act of voting itself. As the authors trenchantly remark, African Americans in the 1960s could learn a great deal about the two political parties even before they were allowed to go to the polls. Thus far from being purely critical, this article pointed the way toward ideas that have shaped several decades of subsequent research.

REPRESENTATION: PATHWAYS TO CITIZEN CONTROL OF POLITICIANS

Powell's most recent research program on political representation reinforces the four themes outlined at the beginning of this article that have run through so much of his research. Using a large-n cross-national research design that focuses at once on elite and mass behavior, Powell demonstrates that pathways to citizen control of politicians depend on the subtle interplay of institutions, and that different constellations of institutions create trade-offs in the type of citizen control that can emerge. In addition, a new theme emerges, one where Powell underlines the importance of situating the empirical study of democratic representation within the normative debates in democratic theory.

Democracy is often defended by normative theorists on the grounds that it is a form of government that allows citizens to control the actions of policymakers. But what are the different ways in which this citizen control can occur through elections? And how does the method of control relate to the quality or nature of control? For the past 20 years, Powell has studied these questions and his efforts have resulted in several path breaking contributions.Footnote ²

Center stage in this research is the description of two “ideal types” of democratic control through elections — “majoritarian” and “proportional influence.” The central distinction between the two turns on the “object of control” — that is, on whether a voter uses elections to choose a collective government policymaker (i.e., a single, unified party that can control the levers of government) or a representative agent (such as a smaller cohesive party in a proportional representation system) who then bargains on behalf of the voter. These two distinct objects of control each require their own unique constellation of institutions that provide alternative mechanisms for voters to control policymakers.

To explain how these institutions operate, Powell describes four interrelated stages in the representation process: Electoral competition, electoral outcomes, government formation, and legislative bargaining. For each model, opportunities for citizen control break down if specific conditions are not satisfied. These conditions therefore establish the models of control.

First consider the majoritarian model. Here Powell has Schumpeter's notion of democracy in mind, where political power is concentrated in the hands of identifiable governments that are chosen by the electorate and responsible to it. The electoral competition stage involves a contest between an identifiable incumbent government and an identifiable challenger. Voters evaluate the past performance and future promises of the two contending parties, and choose the one they like most. Effective citizen control therefore requires that party responsibility for past policy outcomes be clear, that two distinct future governments compete, and that citizens use elections used to pick the future government. At the election outcome stage, the most preferred party must receive a majority of seats, which then makes the government formation stage rather obvious — since a single party wins a majority of seats, this party forms a government by itself, sharing no power with other parties. Finally, in the legislative bargaining stage, legislative rules limit the role of parliament, and in particular the opportunities for opposition parties to influence policy. Were this not true, retrospective clarity of responsibility would be lower, making it more difficult for citizens to make choices during electoral competition.

Citizen control works quite differently under the proportional model, where citizens choose agents to bargain on their behalf rather than choosing the collective government itself. The system permits effective citizen control if each citizen is able to select an appropriate agent. The agents represent the varied interests in society, and, to this end, must be able to bargain effectively on behalf of the citizens they represent.

During the electoral competition stage, the proportional influence model requires that each citizen be able to choose an appropriate agent, which requires a wide range of parties spread out across the ideological spectrum. Of course, when this condition is satisfied, the possibility for citizen control by the majoritarian model breaks down, because voters do not have clear retrospective clarity and are not able to elect the next government. But Powell was one of the first to draw attention to the fact that if parties form pre-election coalitions, it is possible to have both a wide range of choice and strong identifiability in multiparty elections. Since each voter needs an agent, during the election outcomes stage, the proportional influence vision requires that parties be represented in proportion to their support in society. The government formation stage then becomes a bargaining process across the parties, with different parties receiving ministerial portfolios. Finally, to ensure that all agents —not just those parties in government— have an opportunity to influence policy, during the legislative bargaining stage, legislative institutions are designed to ensure that all parties have an opportunity to influence legislation.

Formal institutional arrangements play the central role in classifying political systems according to the two visions. At the electoral competition and outcomes stages, electoral laws will be critical in creating opportunities for citizen control under the two visions. The majoritarian vision obviously requires single-member district plurality rule, which is the most effective electoral system for creating two parties and single-party majorities. By contrast, the proportional vision requires PR with large district magnitude to ensure that smaller parties can form and that parties are represented in proportion to their strength. Similarly, during legislative policy making, formal institutions are central to determining whether power is concentrated in the hands of the government, or whether opposition parties have opportunities to shape policy outcomes. In this regard, Powell focuses on elements of the committee structure and of agenda control. Opportunities for opposition influence are strongest when there are a large number of committees corresponding to government departments (so that committees can develop relevant expertise and better control ministers), when committee chairs are distributed to opposition parties,Footnote ³ when the government (as opposed to parliament) controls the legislative agenda, and when limits that exist on the ability of committees to propose amendments. Together, these two dimensions of formal institutional arrangements allow the classification of political systems according to whether they facilitate citizen control through the majoritarian or proportional visions.

A central question is which of the two visions permits better citizen control. Before turning to that issue, however, it is worth underscoring several points about the two visions themselves. First, a fascinating aspect associated with Powell's conceptual and empirical research is the existence of institutional ”equilibria” in the legislative and electoral domain: although there clearly exist countries that are in the “off-diagonals,” it is rather striking the extent to which legislative and electoral rules go hand in hand. Majoritarian electoral systems usually go hand in hand with legislative institutions that limit opposition influence, while proportional electoral systems usually have legislative institutions that enhance opposition influence. To our knowledge, Powell is the first to identify this correlation, and, of course, he provides a clear rationale for why this correlation makes sense.

Second, readers may note a strong relationship between Powell's two visions of citizen control, and Arend Lijphart's research on “majoritarian” and “consensus” democracies (Lijphart Reference Lijphart1984, Reference Lijphart1999.) Indeed, the two frameworks are in many respects related, but it is useful to underline how the two lines of research differ. One is normative. For Lijphart, majority and consensus democracy represent two distinct normative visions. Majoritarian democracy is based on a belief that democracies should represent majorities, and consensus democracy is based on a belief democracies should represent as many people as possible. In Powell's research, there are no competing normative visions, but rather an interest in understanding different pathways to achieving the same objective so central to many normative theorists, which is citizen control of policymakers. A second distinction considers the nature of empirical primitives. In Lijphart's analysis, political outcomes are often used to help classify political systems. These include such things as the existence of single-party majorities or minimal winning coalitions, stable cabinets, two-party systems, and one dimension of political conflict. For Powell, the primitives are institutions themselves — the nature of electoral and legislative bargaining rules. In this respect, Powell's framework is more conducive to institutional engineering, as it is easier to implement institutions than to design outcomes.

To assess whether one institutional constellation has advantages over the other with respect to citizen control, Powell focuses primarily on “congruence” between citizens and policymakers, where congruence is the distance between a country's median voter and the government policymakers in a left-right ideological space. The congruence research has utilized a breathtaking array of data, including dozens of citizen survey data, a range of data that can be used to place political parties in the left-right space, and institutional data spanning long time frames.

From this data, a consistent theme emerges: For almost the entire post-war period, the constellation of institutions associated with the majoritarian vision consistently produce congruence outcomes that are worse than those associated with the proportional influence vision. This is true using a wide range of empirical strategies. Consequently, the mechanism for citizen control in majoritarian systems seems weaker than the mechanism associated with proportional influence.

It is important, however, to bear two things in mind before one gives up on majoritarian institutions for citizen control of policymakers. First, since roughly 1996, congruence outcomes in majoritarian systems have been on par with those in proportional influence systems. This is not true because proportional systems have done worse, but rather because the winning parties in majoritarian systems have moved closer to the median voter, as exemplified by the Labour Party's move to the center under Tony Blair in Britain. Whether this trend will continue is an open question, but recent history at least makes clear that the mechanisms for citizen control in the majoritarian systems are not doomed to fail. Second, while average congruence outcomes are worse under majoritarian systems, there are clearly greater opportunities under such systems to punish governments for bad economic outcomes, as made clear in Powell's research on economic voting described here.

The different outcomes for the two models in the retrospective accountability tests and the congruence tests highlight two general ways in which Powell's research has shaped the study of the institutional foundations of democracy. First, it often makes little sense to try to understand the effects of particular institutions in isolation from others, nor to design them in ignorance of their interactions. For example, adopting majoritarian electoral laws and strong institutions for opposition influence in the legislature would make both congruence and retrospective accountability more difficult. Second, and more centrally, Powell's work always reminds us that the specific institutional arrangements for democracy inevitably create normative trade-offs. There will not be a “best” set of institutions. Rather, the best we can do is identify and understand these trade-offs, making it possible to be explicit about why we might prefer one type of democracy over another.