Cambridge Catalogue  
  • Help
Home > Catalogue > Degrees of Democracy
Degrees of Democracy


  • 25 b/w illus. 19 tables
  • Page extent: 254 pages
  • Size: 228 x 152 mm
  • Weight: 0.46 kg


 (ISBN-13: 9780521868334)

Degrees of Democracy
Cambridge University Press
9780521868334 - Degrees of Democracy - Politics, Public Opinion, and Policy - By Stuart N. Soroka and Christopher Wlezien

1    Public Opinion and Policy in Representative Democracy

Edmund Burke was a much better political philosopher than he was a politician. Burke was a Member of Parliament for Bristol for six years, from 1774 to 1780. He visited the constituency infrequently during that time, and managed to alienate his electorate through a combination of his rather unpatriotic support for America during the war and a general lack of attention to the principal concerns of Bristol's merchants. It was clear by 1780 – when he withdrew – that Burke stood no chance at reelection.1

Burke's political writings remain fundamental to this day, however, including his Speech to the Electors of Bristol. In his mind, an elected representative should seek and consider the advice of his constituents, but that advice should not be authoritative instruction. It should not serve as a mandate, dictating what a representative does; it should be but one input into his decision making. Burke's philosophizing did not free him from the burden of representing the interests of his electors, however, and his ideas evidently had only limited appeal for Bristolians at the time. That is, Burke paid the electoral price.

The situation is little changed more than two hundred years later. Active representation of the public's policy preferences remains a – if not the – central concern in electoral and inter-electoral politics. Indeed, some argue that the tendency to represent public preferences

has increased as politics has become a career path during the twentieth century, and that politicians have become (out of desire for reelection) more reliably interested in keeping voters happy (e.g., Maestas 2000, 2003).2 Moreover, Geer argues that the onset and development of public opinion polling has made it easier for politicians to know what public preferences are.3 Politicians have both a keen interest in representing the public and the seeming means to do so.

Representation is also at the heart of everyday political debate. Citizens care about what they get from government, whether it's tax cuts or prescription drug benefits or education subsidies. They also care about the degree to which what government does matches their preferences for various other things, whether it's the amount of healthcare, the availability of abortion, environmental protection, or going to war.

Not all politics is explicitly about policy, of course. Citizens also care about outcomes, and many argue that this is true now more than ever before (see, e.g., Clarke et al. 2004). People want a growing economy, safe streets, and high-quality healthcare. Not surprisingly, U.S. presidential candidates routinely ask: “Are you better off today than you were four years ago?” This is not a question about policy per se – it is a question about outcomes. Policy nevertheless is implicated, of course. Politicians do not have magic wands – they influence outcomes using policy. Politicians offer plans to solve important problems, such as a struggling economy. There are policies to reduce crime, improve education, fix healthcare, and so on, and when conditions improve, politicians don't claim to have been lucky. They explicitly credit policy, for instance, when they declare that “Our tax cut policy got the economy moving again,” or that “We put more police on the streets and crime came down.” Policy clearly matters for both outcomes and politics. We want things, and, through policy, politicians try to deliver.4

This opinion-policy relationship is central not just in everyday politics, but in the theoretical literature on democracy and representation as well, from Jean-Jacques Rousseau, to John Stuart Mill, to Robert A.

Dahl. Indeed, for each of these theorists the connection between public preferences and public policy is one of the most critical components of representative democracy. Importantly, for each the opinion-policy link is necessarily reciprocal: on the one hand, policymakers need to react to the public's policy preferences; on the other hand, the public needs to have informed preferences, based in part on what policymakers are doing. To put it differently: If we want politicians to make the policy we want, our preferences must be informed by policy itself. This is the theme that guides our investigation, where “degrees of democracy” are conceived as deriving from the extent to which both “policy representation” and “public responsiveness” are evident.

We develop and empirically test a model of democracy in which opinion-policy relationships are paramount. Following Wlezien (1995, 1996a), we posit a very simple but powerful framework that connects opinion and policy, what we call the “thermostatic” model. The model has as its referent the temperature control systems in our homes, where the public is the thermostat and policymakers the furnace or air conditioning unit. If the model works, we will observe three things. First, the difference between the actual policy temperature and the preferred temperature will cause the public to send a signal to change the policy temperature, for example, to turn up the heat. Second, in response to such a signal, policymakers will increase policy. Third, as the policy temperature approaches the preferred temperature, the signal for change will be reduced. This, we argue, is how a functioning democracy should work. Is it what we actually observe out in the world? That is, do policymakers respond to public preference signals? Does the public adjust its signals in response to what policymakers do?

That these are important questions should be clear. They speak to the capacity of citizens, the quality of governance, and, as we shall see, relationships between the two. They speak not just to the quality but also to the principles and potential merits of representative democracy. They are academic questions, to be sure, but they are academic questions with rather sweeping real-world implications.

The chapters that follow begin to provide some answers. We first set out expectations for public responsiveness and policy representation, varying as a function of issue salience and political institutions, and applying, we believe, across the democratic world. Our theoretical scope thus is rather broad. Our empirical focus is necessarily more

narrow. “Policy” is captured using budgetary data; public preferences are measured using opinion polls. The empirical analysis is limited to the three countries for which extended, reliable time series of both opinion and spending are available: the United States, the United Kingdom, and Canada. We shall argue that these three countries provide a quite strong “most similar systems” research design. Additional leverage is gained by looking at multiple policy domains within the three countries, for a total of nineteen individual domains, across which institutional characteristics differ. The comparative reach nevertheless is limited, as we cannot explore the full range of political institutions with the data at hand. We can take an important step forward, however. We also can consider – as we do in the final chapter – what the broader implications of our findings may be for policy domains and regimes not directly assessed here.

The story that emerges is one in which representative democratic government in some cases works surprisingly well. There is strong evidence that the public actually adjusts its preference for “more” or “less” policy in a thermostatic way, as policy itself changes. There also is evidence that policymakers respond to changes in public preferences over time. The thermostatic model works. The magnitude and nature of both public responsiveness and policy representation do vary across policy domains and countries, and this variation is systematic. As we will show, varying degrees of responsiveness and representation can be explained at least in part by factors such as the salience of various policy domains, as well as by the design of government institutions, that is, by the vertical and horizontal division of powers. That this is true means that we can begin to make some generalizable predictions across different policy domains and political contexts. It means that we can begin to capture, empirically speaking, “degrees” of democracy.

Policy representation

The relationship between public preferences and policy has been the central concern of the literature on representative democracy at least since Jean-Jacques Rousseau's The Social Contract (1762).5

This was among the first considerations of the connection between representatives and their constituents, and many others have examined the opinion-policy connection since. The first explicit consideration of public opinion per se appears roughly fifty years later in the work of Jeremy Bentham. His writings reveal a view of representative democracy very similar to the one we are concerned with below: Bentham conceives of a reactive public that monitors government activities, and to which governments are held accountable (see Cutler 1999). Bentham was an elitist, to be sure, just like Edmund Burke one generation earlier, and John Stuart Mill one generation later. Each believed that certain people's opinions were more worthy than others. But each also focused on essentially the same question: What conditions or institutions are required to produce a representative democracy in which policies reliably reflect the interests of citizens – citizens, that is, whose interests are sufficiently informed to justify representation?

Political scientists are now much closer to answering that question. Or, at least, we are accumulating a body of evidence on when representation seems more or less likely, and under what conditions citizens are more or less informed on political and policy matters. But what exactly do we mean by “policy representation”? There are several different kinds of representation we might look for in a democracy. Drawing in large part on the theoretical literature above, as well as Pitkin's (1967) classic work on representation, our paramount concern is policy representation: policymakers’ active representation of citizens’ (aggregated) preferences. Indeed, while other versions of representation have garnered considerable attention, we suggest that most are valued in large part because they are likely to enhance the connection between what people want regarding policy and what governments provide.

Descriptive Representation

Take, for instance, discussions of “descriptive” representation. These typically focus on representatives’ characteristics, specifically, on the extent to which representatives resemble the represented, particularly

with respect to demographic characteristics. Race, ethnicity, gender, language, sexual orientation – any number of characteristics can be salient. “Shared experience” (Mansbridge 1999) can also be critical – experience as a farmer, for instance, or of having grown up in a particular region. According to this view, a representative legislature should be a microcosm of the larger society. Indeed, a number of theorists refer to this type of representation as “microcosmic representation” (e.g., Birch 1971, 2001).

Support for descriptive representation comes in both symbolic and substantive forms. A symbolic argument for descriptive representation focuses on the representation of historically under-represented groups (e.g., women, or racial or ethnic minorities), and stresses the significance of seeing oneself represented in government, and of consequently seeing politics as something that you are or can be involved in. Results include the possibility of increased participation in politics by previously under-represented groups, and – often viewed as more fundamental – the enhanced symbolic legitimacy of a government seen to encompass diverse individuals or groups (e.g., Sapiro 1981, Phillips 1995; Kymlicka 1993). Descriptive representation may also facilitate communication between constituents and representatives, particularly in historically conflictual contexts (Mansbridge 1999). Indeed, there is evidence that some individuals are more likely to contact their representative if they share certain demographic characteristics (Gay 2002).

Proponents of descriptive representation have also argued that it can be a useful proxy for policy attitudes: “When interests are uncrystallized, the best way to have one's most important substantive interests represented is often to choose a representative whose descriptive characteristics match one's own on the issues one expects to emerge” (Mansbridge 1999:644). That demographically similar groups will tend to have similar policy attitudes is thus the heart of the substantive argument for descriptive representation. Bentham's support for microcosmic representation in Britain was based on the concern that a legislature dominated by the upper class would advance only upper-class interests, for instance. Bianco (1994) emphasizes race as a kind of cue for representatives’ values and interests (see also Phillips 1995), a fact that is confirmed by representatives’ own views on how and who they represent (e.g., Fenno 1978; Kingdon 1981; Mansbridge 1986). And the link between demographic characteristics is evidenced in the empirical literature on legislative

behavior: day-care coverage is greater in Norwegian municipalities where there are more female representatives, for instance (Bratton and Ray 2002); black and female representatives in the United States are consistently more likely to introduce or sponsor black or women's interest bills respectively (Bratton and Haynie 1999); the voting records of Hispanic members of the House of Representatives are distinct in ways that may be culturally attributable (Welch and Hibbing 1984; Hero and Tolbert 1995; Kerr and Miller 1997).

That said, the extent to which demographic similarities are a substitute for policy attitudes is unclear, and using a proxy of dubious reliability is less than ideal. Note first that preferences and interests are not the same thing. Many people have a preference for a level of redistributive policy that clearly is not in their best interest, for instance – they are wealthy themselves but support politics that favor the less advantaged.6 Similarly, many white (majority) respondents support policies that benefit visible minorities. In neither case are their preferences in line with what would seem to be their interests.

This gap between preferences and interests is confirmed in work on the actual behaviour of demographically “representative” elected officials. Some work on female legislators argues that women who successfully gain office are not necessarily representative of women's interests – that is, they may provide some kind of symbolic and descriptive representative function, but the extent to which their policy preferences are similar to women's (aggregated) policy preferences can be relatively limited (e.g., Gotell and Brodie 1996).7 A number of authors have expressed further concerns about the way in which descriptive representation encourages the belief that certain groups can be represented only by certain people (in those groups), and alternatively that certain people can only represent certain groups (e.g., Kymlicka

1995; Phillips 1995; Mansbridge 1999). Descriptive representation may actually be destructive to political community; some also charge that “Socially descriptive representation is pernicious because it makes recourse to constituencies unnecessary” (Morone and Marmor 1981:437). We view these problematic features of descriptive representation as in some sense supplementary – supplementary, that is, to the fact that demographic or experiential characteristics will only intermittently and unpredictably be relevant to the representation of policy preferences. Descriptive representation may thus be important in terms of providing role models, connecting to politics, or engendering a feeling of belonging in the political community. Where policy is concerned, descriptive representation falls short.

Electoral Representation

More common in the empirical literature is a version of representation which we will refer to as “electoral representation.” Electoral representation is not always distinguished from descriptive representation, since theories of representation focusing on electoral systems are often motivated primarily by an interest in descriptive representation – the work arguing for proportional representation tends to focus on those systems’ capacity to better represent subsets of the population that are demographically different, for instance. (See, e.g., Lijphart 1977; Katz 1997; see also the work on proportional representation and women's representation, e.g., Norris 1985; Matland and Studlar 1996.) Electoral systems themselves have played a particularly prominent role in the comparative literature on representation, however (see esp. Powell 2000). Indeed, the comparative study of representation is mostly based on a concept of representation that prioritizes the vote-seat function over everything else. That is, the electoral representation literature is often agnostic about the representation of groups defined by anything other than their vote. It equates the quality of representation with the efficiency and exactness of the vote-seat function. Lijphart's Electoral Systems and Party Systems (1994) begins with what we regard as a paradigmatic statement of electoral representation theory:

Except in very small communities, democracy necessarily means representative democracy in which elected officials make decisions on behalf of the

people. How are these representatives elected? This indispensable task in representative democracies is performed by the electoral system – the set of methods for translating the citizens’ votes into representatives’ seats. Thus the electoral system is the most fundamental element of representative democracy. (1994:1)

Lijphart's working hypothesis is echoed in much of the comparative institutional literature. For a particularly succinct statement, see Sartori (1997:3): “In the beginning is how a people is made to vote.” But also see much of the literature on vote-seat functions, in which the electoral system is paramount (e.g., Taagepera and Shugart 1989; Cox 1997). Indeed, this work largely reduces democracy to the method by which representatives are selected, not unlike Schumpeter's oft-cited and decidedly narrow view of democracy: “the democratic method is that institutional arrangement for arriving at political decisions in which individuals acquire the power to decide by means of a competitive struggle for the people's vote” (1942:269).8

The particular significance of electoral systems for democracy and representation is relatively clear, of course, and in fact has quite deep theoretical roots. One of the central debates in democratic theory regards the tension between individual equality – leading to majority rule – and the protection of minorities.9 The tension is central in the Federalist Papers, where the American system of checks and balances is laid out with the primary objective of avoiding a tyranny of the majority in the face of individual political equality, a difficult and pretty much unattainable task. It is also prominent in J. S. Mill's argument for proportional representation, for instance, and Toqueville's concerns about democracy in America. The preoccupation with electoral systems follows almost directly from the fundamental concern with the equality of individuals in liberal democratic systems (see, e.g., McGann 2006).

Like descriptive representation, however, electoral representation only captures a part of what we might like in representative government. If descriptive representation is about the representation of (mainly)

demographic characteristics, electoral representation is principally about the representation of votes. In both cases, representation is – in Pitkin's terminology – “standing for.” Electoral representational theory is interested in the extent to which a legislature mirrors the electorate, where the most salient characteristic of voters is for whom they voted. It thus focuses on how the distribution of seats reflects the distribution of votes. This is not to say that representation as “acting for” is entirely ignored; as with descriptive representation, electoral representational theory is premised on a connection between the representation of votes and the resulting representatives’ legislative activity.10 Some of the preceding work has demonstrated such a link, and our own models will provide further evidence of the connection between electoral representation and policy representation.

© Cambridge University Press
printer iconPrinter friendly version AddThis