The Democratic Dilemma was published in 1998 by Arthur Lupia and Mathew McCubbins. The book addresses a long-standing question: Does a lack of information short-circuit democratic functioning? This is a question of relevance for multiple pathways of democratic representation: from voters to elected officials, elected officials to bureaucrats, legislators to committees, citizens to jurors, inter alia. The authors in this symposium reflect on the book 20 years after its publication. They do so from a diverse set of perspectives, ranging from the book’s impact on voting research to its role in the development of formal theory, to how it motivates thinking about institutional design, and more. In this introduction, I describe the context in which the book was written, summarize the main arguments, and provide an overview of the symposium’s articles.
CONTEXT OF THE BOOK
Full appreciation of any intellectual contribution requires an understanding of the context in which it was conceived. Accordingly, one must consider both contemporary trends in the 1990s as well as the climate and thinking in the social sciences. Perhaps the most notable feature of the time concerned the radically shifting information environment. Consider that digital cable television became widely available only in the mid to late 1990s, and broadband internet did not become available until 2000. Google launched in 1998—the publication year of The Democratic Dilemma. Lupia and McCubbins thus wrote during a transformative informational era; clearly, politics—at the time of the writing—competed with an increasing number of alternative stimuli (Baum and Kernell Reference Baum and Kernell1999), but the massive information revolution had, at best, just begun.
Institutionally, much can be said about the 1990s, but of particular note was the expansion of the administrative state, which meant increased challenges to keeping bureaucratic actions in check. One scholar goes so far as to claim a “confluence of events and factors, crystallizing primarily in the 1970’s, 1980’s, and 1990’s…culminated in the creation of an unchecked, arbitrary, abusive, and unconstitutional fourth branch of government…referred to as the ‘Administrative State’” (Dobkin Reference Dobkin2008, 362). Bureaucratic oversight became uncertain partially due to the Supreme Court’s 1983 Chadha decision that made ex post legislative vetoes of agency actions unconstitutional and the 1984 Chevron ruling that facilitates agency discretion. These and other developments meant that, when it came to citizens and elected officials, Lupia and McCubbins wrote at a time when “learning what one needs to know” was ostensibly becoming more and more challenging, at least in the United States.
Lupia and McCubbins’ partnership also brought together their individual research agendas. McCubbins coauthored an influential 1984 article with Thomas Schwartz addressing the concern that bureaucrats take advantage of their specialized knowledge to make policy, wresting control from elected representatives (McCubbins and Schwartz Reference McCubbins and Schwartz1984). Delegation to agencies becomes abdication (Lowi Reference Lowi1979; Niskanen Reference Niskanen1971; Weber Reference Weber, Gerth and Wright Mills1946). Evidence along these lines includes that Congress cannot afford to and thus does not often engage in detailed direct investigations of agency activity (Aberbach Reference Aberbach1990). McCubbins and Schwartz (Reference McCubbins and Schwartz1984) argue that oversight need not entail such investigations, which they suggest are akin to “police patrols.” Instead, legislators can learn what they need to know to keep bureaucrats in line by relying on “fire alarms.” These involve third-party signals, from citizens or interest groups, that report cases of bureaucratic slippage or shirking. On receiving such information, legislators can maintain control via sanctions such as slashing budgets and/or changing jurisdictions. This approach is efficient and allows elected officials to claim credit for problems brought to them by their constituents, thereby serving their re-electoral aims. Moreover, agencies anticipate potential fire alarms and, to avoid punishment, act within the bounds of legislators’ desires (Weingast and Moran Reference Weingast and Moran1983). The fire-alarm concept has had enormous influence, but an initial problem with the concept was that it “depends upon the crucial assumption that legislators can learn from fire-alarm activity. Unfortunately, there is neither justification nor evidence for this assumption” (Lupia and McCubbins Reference Lupia and McCubbins1994, 98).
Whereas McCubbins came to the project with this ostensible institutional lens, Lupia’s initial work focused more on voters. In a Reference Lupia1994 paper, Lupia built on a growing literature (Kuklinski, Metlay, and Kay Reference Kuklinski, Metlay and Kay1982; Popkin Reference Popkin1991; Sniderman, Brody, and Tetlock Reference Sniderman, Brody and Tetlock1991) to ask the question: How do voters make decisions given low information? Or, more specifically, can they use cues or endorsements to make the same decisions they would have made if they were extremely well informed—akin to encyclopedias? Building on a formal model that he had developed (Lupia Reference Lupia1992), which shows how uninformed voters in direct-legislation elections could use signals to emulate informed voters, Lupia implemented an exit poll in 1988. The poll studied California voters’ decisions on five complex insurance-reform initiatives. Lupia shows that voters who lacked detailed knowledge about the propositions but knew where the insurance industry stood on each initiative voted in nearly identical ways to those who had the detailed knowledge. In short, shortcuts worked as substitutes for detailed knowledge. An unanswered question coming out of this work concerned exactly when shortcuts would work to compensate for knowledge shortfalls (Lupia Reference Lupia1992, 63–64).
It should be clear that Lupia and McCubbins (Reference Lupia and McCubbins1998) were asking the same general question: When can citizens and elected officials learn what they need to know to make democratic delegation work? Delegation is a defining feature of representative government and, to be clear, Lupia and McCubbins address this question thinking of more than voters and legislators–bureaucrats—they also were thinking of jurors, executives, commissions, judges, and so on (Lupia and McCubbins Reference Lupia and McCubbins1998, 3). Before briefly summarizing their argument, one final relevant note concerns the disciplinary context at the time of the writing. Political science was in the midst of a debate about the worth of rational choice and formal models (Green and Shapiro Reference Green and Shapiro1994), and experiments had not yet fully emerged as an empirical mainstay (Druckman et al. Reference Druckman, Green, Kuklinski and Lupia2006). This makes Lupia and McCubbins’ approach, as I next describe, bold.
It should be clear that Lupia and McCubbins (Reference Lupia and McCubbins1998) were asking the same general question: When can citizens and elected officials learn what they need to know to make democratic delegation work?
THE DEMOCRATIC DILEMMA’S ARGUMENT
The basic questions, then, in The Democratic Dilemma are: What makes for a credible fire alarm or credible cue-provider? Can democratic delegation work? Lupia and McCubbins (1998, 20) offer what can be thought of as a three-part theory. First, they put forth a theory of attention and knowledge. A critical point here is that knowledge—“the ability to predict accurately the consequence of choices”—differs from information which is “the data from which knowledge may be derived…knowledge requires information, but large amounts of information do not ensure knowledge” (20). The distinction underlies the approach that people take when dealing with a world of information saturation. It presaged the now-popular idea of “Google Effects on Memory” (Sparrow, Liu, and Wegner Reference Sparrow, Liu and Wegner2011), which suggests that the nature of memory has changed due to information availability with the internet as a source of transactive memory. It is knowledge and not information that actors need to make “reasonable” decisions.
Second, Lupia and McCubbins (Reference Lupia and McCubbins1998) present a formal theory of persuasion to identify the necessary and sufficient conditions for one person to learn from another. They show that persuasion requires that a receiver (e.g., a voter or a member of Congress) perceive that the speaker (e.g., a politician or an interest group) as having knowledge and being trustworthy. The latter requirement can result from a perception of shared interests or from external forces such as when a speaker will be penalized if he/she lies (e.g., perjury or reputational loss), when the statement will be verified by someone else (e.g., the media or a political opponent), or when there is a cost to making a statement (e.g., taking out television advertisements). The authors expand on this theory to show what people learn, noting that one can learn nothing, be deceived, or become enlightened by obtaining the knowledge needed to make reasonable decisions. For instance, they can learn about agency activity, which candidate for whom to vote, and which policy to support. That is, if enlightenment occurs, people obtain the ability to accurately predict the consequences of their actions and then can make decisions in their interests.
Third, Lupia and McCubbins (Reference Lupia and McCubbins1998) identify the conditions under which democratic delegation works, showing that it succeeds when two conditions are met: (1) the knowledge condition such that the principal can distinguish between better and worse alternative choices (e.g., candidates and agency actions); and (2) the incentive condition such that the agent takes an action that is in the principal’s interest. Stated another way, when a principal (e.g., a voter, legislator, or citizen) delegates authority to an agent (e.g., a representative, bureaucrat, or juror), that delegation works when the agent offers an option that the principal prefers and knows to accept over the status quo (e.g., better administration or better policy).
Lupia and McCubbins (Reference Lupia and McCubbins1998) test the details of the theories with a large set of laboratory experiments that asked people to make decisions with some input from a speaker. They show how the external forces described previously lead to correct decisions and successful delegation. Further, they present a survey experiment in which they show that perceived knowledge and common interests—with notable political commentators at the time (i.e., Rush Limbaugh and Phil Donahue)—determine the success of persuasion attempts. Notably, they conducted the experiment in the 1994 Multi-Investigator Study, which was a precursor to Time-Sharing Experiments for the Social Sciences (available at www.tessexperiments.org).
As described, the book may sound abstracted; therefore, I want to accentuate three implications. First, “limited information is not sufficient to strip the reason from our choices or to turn asunder our democratic delegations” (Lupia and McCubbins Reference Lupia and McCubbins1998, 92). Second, citizens and legislators can learn what they need to know to make delegation work by relying on others—when the conditions for enlightenment are met. Third, these conditions involve persuasion and many democratic institutions are designed to facilitate enlightenment. The authors state, “Democratic institutions can…establish the conditions for persuasion, enlightenment, and reasoned choice. Consequently, political institutions can help resolve the democratic dilemma…electoral, legislative, bureaucratic, and judicial institutions can also be the institutions of knowledge” (Lupia and McCubbins Reference Lupia and McCubbins1998, 205). To show this, Lupia and McCubbins describe how electoral systems, political parties, media, campaign-finance rules, legislative procedures, administrative procedures, rules of evidence, and more serve as the external forces that facilitate persuasion, enlightenment, and successful delegation.
The book is a positive endeavor and, as such, one can agree with all that is written and still conclude that citizens do not, in practice, learn what they need to know. Lupia and McCubbins are appropriately cautious overall. Indeed, they point out institutions that are problematic in their view (1998, 225–27), but it is clear that they side with those who believe representative government can and, in many instances, does work well. Low information itself is not a threat and institutions often are set up to facilitate successful delegation. For them, democracy, at the very least, works much better than those who suggest that elected representatives abdicate to the administrative state (in line with McCubbins’ aforementioned work) or that voters are insufficiently informed to make reasonable decisions (in line with Lupia’s aforementioned work)—for them, reason often triumphs.
Low information itself is not a threat and institutions often are set up to facilitate successful delegation. For them, democracy, at the very least, works much better than those who suggest that elected representatives abdicate to the administrative state (in line with McCubbins’ aforementioned work) or that voters are insufficiently informed to make reasonable decisions (in line with Lupia’s aforementioned work)—for them, reason often triumphs.
METHODS AND TRAINING
A notable aspect of the book concerns the range of literature and methods on which Lupia and McCubbins built. Their theory of attention and knowledge draws significantly on cognitive science work, which at that time (Churchland Reference Churchland1995; Holland et al. Reference Holland, Holyoak, Nisbett and Thagard1986) had been largely ignored within the social sciences. They connected that work to formal models of communication, expanding on signaling (Spence Reference Spence1973) and cheap-talk models (Crawford and Sobel Reference Crawford and Sobel1982). Although their contributions to formal theory generated some debate (Austen-Smith Reference Austen-Smith1999; Lupia and McCubbins Reference Lupia and McCubbins2001), the novelty was incorporating various new types of uncertainty (Lupia and McCubbins Reference Lupia and McCubbins1998, 46–47), showing that perceptions of speakers’ interests and knowledge matter and revealing how political institutions can affect those perceptions. Among others, a key link to the cognitive-science literature was revealing how knowledge—as distinct from information—can be relayed using a strategic formal theory. Also novel at the time was the use of laboratory experiments to test predictions from the theory and then to go even further with a survey experiment. As research agendas become increasingly fragmented, the book stands out as a multimethod, multifield approach—and one that lets the questions guide the methodological choices rather than vice versa. Along these lines, chapter 6 of the book provides an exemplary discussion of how to test a theory, emphasizing how a good test serves as a strong analogy to the theory being tested. This, and not mundane realism, is critical. The authors (1998, 99; italics in original) state: “Like all empirical science, our laboratory experiments require an inductive leap. The leap is the assumption that our method of experimental observation is a faithful analogy to our theory. Though many social scientists do not realize it, all scientists make this leap when they use empirical research to evaluate theoretical explanations.”
Another unique aspect of the book is its concluding chapter, which is one page long—Lupia and McCubbins (Reference Lupia and McCubbins1998, 229) say they prefer not to “wax nostalgic about [their] intellectual travails….” This is fair enough but, as someone who worked on the book as a graduate research assistant, I am nostalgic for two reasons. First, production of the book involved scores of student research assistants; whereas some of this hopefully helped Lupia and McCubbins, much of it—I can see retrospectively—was to train those of us involved. We learned how to connect vague ideas to extant literatures, how to identify big questions while also funneling down to specific research applications, how to develop theory, how to test theory, how to analyze data, and how to write. We also learned how to work together; this type of group or laboratory approach is not common in political science but it certainly can be powerful. I learned a tremendous amount just from witnessing conversations between Lupia and McCubbins, whether at the department at the University of California, San Diego (where the book was written), at one of their homes, at lunch (which McCubbins treated us to on a near-daily basis), or at the Dairy Queen in Carmel Valley, California. In the current era in which spending time in person with collaborators seems unnecessary—given all the electronic communication media available—it is easy to forget the value of simply spending time together. Second, the general atmosphere at UC, San Diego, at the time is worth mentioning. The intellectual excitement was contagious; there were regular speakers from across the social sciences (e.g., some of this is reflected in the Lupia et al. Reference Lupia, McCubbins and Popkin2000 edited volume), and the faculty produced some of the most influential books and articles of the previous two decades (e.g., Beck and Katz Reference Beck and Katz1995; Cox Reference Cox1997; Gerber Reference Gerber1999; Kernell and Jacobson Reference Kernell and Jacobson1999; Lijphart Reference Lijphart1999; Lupia and Strøm Reference Lupia and Strøm1995; Shugart and Carey Reference Shugart and Carey1992).
WHY REFLECT?
One can certainly ask what can be gained from reflecting on the book now. I hope, at this point, that the answer is clear. Lupia and McCubbins wrote when the world and the social sciences were in a very different place than they are today. Yet, the book provides a general theory of knowledge, democratic delegation, and institutional design. Thus, one can ask not only how the book influenced subsequent work but also how its lessons apply in today’s context. The authors of this symposium address these questions.
Articles by Gailmard and Krupnikov, in distinct ways, discuss the book’s aforementioned normative statement. Gailmard bemoans the lack of subsequent explicit normative debate between rationalist and behavioral approaches and how this lack of back and forth has hindered an understanding of democratic institutions. Krupnikov delves into what the book implies for partisan divides today, suggesting that its core argument provides an often-missed perspective in recent work on preference formation and behavior. In their articles, Boudreau and Landa explore how subsequent literatures have been influenced by the book—the former focusing on sophistication, endorsements, and neuroscience and the latter discussing the formal literature on social learning. Rosenbluth considers the book’s argument in light of the rise of populism around the globe, and Esterling explores the approach to research and implications for institutional design.
It is remarkable to recognize the range of the contributing scholars; indeed, Gailmard and Krupnikov come from distinct intellectual perspectives (respectively, formal theory and political psychology) to accentuate the need for scholars to consider Lupia and McCubbins’ optimistic take on democracy. Analogously, Boudreau and Landa are engaging in a cross-field conversation about the importance of sophistication in preference formation. Moreover, whereas Esterling reflects on the philosophy of institutional design, Rosenbluth builds on the book’s argument to highlight dysfunctional institutions. In short, Lupia and McCubbins stimulate cross-field and cross-method conversations that range from approaches in the philosophy of science to the discussion of contemporary political institutions. It is an agenda-setting book that ignored disciplinary boundaries. We are fortunate that the symposium concludes with reflections from Lupia and McCubbins on the book’s argument, considered in what is a very different world from the one in which they wrote.
ACKNOWLEDGMENTS
I thank Kevin Esterling, Paul Gronke, Adam Howat, and Amanda Sahar D’Urso for helpful comments.