Michael J. Zenz
Public Opinion, Political Representation, and Democratic Choice
For the fulfillment of PhD at The University of Maryland, May 2015
In this dissertation I argue that political representatives have duties to be responsive to public opinion in their policy decisions. The existence of this duty, I claim, is a basic requirement of a truly democratic system of government. In chapter 2, I show that several standard versions of democratic legitimacy require political representatives to ``respect'' public opinion. However, I argue that a particular version of political legitimacy, based upon popular sovereignty and the importance of self-governance, provides an especially useful background for understanding what this ``respect'' must mean. In chapter 3, I argue that respecting public opinion requires political representatives to integrate public opinion information into their policy decisions. According to one of the standard views of political representation, the liberal conception, representatives deciding between policy alternatives should balance what they believe to be in the interests of the public against public opinion. I argue that this is the only adequate theory of political representation. Although this view of political representation is often discussed in the literature, it is less often given a mathematically precise form. Therefore, I present a formal model of such a balancing procedure, and this reveals several important formal requirements that a conception of public opinion must satisfy; most importantly, it must account for instability in the expression of public opinion, individual differences in opinion strength, and it must be representable along a cardinal scale. Standard measures of public opinion do not satisfy these requirements. I argue that if such a model of public opinion cannot be formulated, then the liberal conception of political representation is incoherent. In chapters 4 and 5, I present a model of public opinion based upon Thurstonian scaling techniques that fulfills the necessary formal requirements. Finally, in chapter 6, I discuss several important implications this model has for the measurement of public opinion, the use of public opinion by political representatives in policy deliberation, and other problems in social choice theory.
Political representatives in modern democracies take many considerations into account when making policy decisions. Some are pragmatic, such as how a given legislative vote will be interpreted and evaluated by their constituents, most loyal voters, party leadership, and most likely campaign donors. These people greatly influence a representative's ability to be reelected and progress up the legislative hierarchy, and so it is reasonable to expect a representative to keep all of these individuals in mind when she is deciding upon which public policies to support. Other considerations arise from representatives' beliefs about what is best for their constituents, or what advances the common good. Finally, some considerations arise from their role as representatives who are responsive to the policy judgments of their constituents. Although all of these considerations are likely relevant for explaining the behavior of political representatives, and so are of great interest to political scientists, only a few are relevant to the representatives' proper role in a representative democracy. In other words, only some considerations that influence political representatives are required by democratic theory, while others may actually be excluded by it. In this dissertation I will explore the subset of the considerations that I think democratic theory requires political representatives to include in their deliberations about what policy actions they should take. More specifically, I will argue that democratic theory requires that political representatives weigh public opinion against their estimations of what is in the common good when they make policy decisions. The bulk of this dissertation will be spent showing how such a model of political representation can be made coherent given a number of possibly devastating problems with such a conception of representation.
An examination of any major policy decision quickly reveals the diversity of considerations that representatives must often weigh in their decisions. One decision of particular interest in this regard was the 2010 passage of the health care reform legislation in the United States called the Affordable Care Act (ACA); I will use an example modeled after this piece of legislation throughout this dissertation, because I think it provides particularly salient examples of the weighing procedure between different considerations that I take to be central to the legislative actions of political representatives. A hint of this problem is easily apparent. In the spring of 2010, for instance, Congress was controlled by the Democratic Party after substantial election victories in the 2008 election. The health insurance provisions of the sort provided by the ACA had long been a goal of many members of the Democratic Party, and the legislation was seen by them as being in the common good. Passage also seemed to be supported by what many considered a significant electoral mandate that resulted from the 2008 national elections, which featured health care reforms similar to those in the ACA on the electoral platform of both the Democratic Party and its Presidential nominee Barack Obama. However, these considerations in favor of passage of the ACA were in conflict with some measures of public opinion at the time. The ACA was very controversial, with opinion polls showing that a majority of Americans were against its passage, with the so called "individual mandate" that all individuals buy health insurance being particularly divisive. There was intensive public opinion polling during the passage of the ACA. For instance see the CNN Opinion Research poll released on March 22, 2010. Legislators faced these many conflicting considerations when deciding how to vote. Of particular interest to my project is the set of considerations bearing on their roles as legislators and representatives; situations such as the passage of the ACA make it unclear how political representatives should balance these conflicting considerations so that that they can best perform their role as representatives.
There is nothing unique about the decision process that representatives faced when developing the ACA; the question of how political representatives should deal with conflicting considerations is important for understanding how political representatives should act generally. In this dissertation I will examine the way in which political representatives should integrate different considerations into their policy decision process. In particular, I am interested in how representatives should integrate public opinion into their decisions and how this might be balanced against their estimates of what is in the common good. I will argue that political representatives are in a position of leadership, such that they should often use their own judgment about which policies will be for the best of the community at large, but they must also be receptive to the policy judgments of their constituents; I will refer to these judgments as the public opinion. In fact, I will argue that political representatives have a duty to integrate public opinion into their policy decisions. However, as I will show, any precise decision process that abides by those two principles must be capable of using measures of public good that have enough information content such that a weighing between the representative's estimates of the common good and public opinion is possible; it will turn out that this requires that public opinion be representable using cardinal (or numeric) values. Such a model of public opinion will face a number of difficult conceptual and technical challenges that will occupy much of the dissertation. For instance, a suitable balancing is only possible if public opinion is capable of being represented cardinally, something that many claim is impossible. I present a method by which public opinion can be represented cardinally, thus showing how political representatives can fulfill their duties to respect public opinion. By the end of the dissertation, I will be in a position to provide a coherent model of policy decision that is able to account for the integration of the diverse considerations political representatives face when making complex policy decisions.
In chapter 2, I show that a duty of responsiveness to public opinion is implied by a number of popular theories of democratic legitimacy, including the minimalist democratic tradition (Schumpeter, 1950; Dahl, 1991), deliberative democracy (Habermas, 1992; Cohen, 1997; Pettit, 1997), and those theories derived from popular sovereignty. I then argue that a basis for democratic legitimacy in popular sovereignty, inspired by that of Rousseau, should be preferred. In democracies, because the people are sovereign, political representatives must respect public opinion or otherwise they violate the rights of members of the public to govern themselves --- this right is a species of the right to personal autonomy. I show that the existence of a right of self-governance implies the existence of a corresponding duty of some set of political officials to be responsive to public opinion; such officials are typically called political representatives. This theory also provides reasons for political representatives to use their best judgment about what is in the public interest, and not just public opinion, when making policy decisions. Therefore, the duty to respect public opinion requires political representatives to balance these two criteria when making decisions. This is similar to the "liberal" conception of democratic political representation, which originates from Pitkin's (1967) conceptual analysis of political representation in the western context. According to this view, a political representative must balance her "mandate" (public opinion about what should be done) with her "independence" (what she thinks is in the best interests of her constituents). I argue that representation so conceived allows political representatives to use their superior knowledge of policy issues and position within a small deliberative body to make decisions that better track the public good, while at the same time respecting the rights of the public to govern themselves.
I spend the rest of the dissertation responding to the worry that the liberal conception of political representation is incoherent. In chapter 3, I develop a formal model of policy choice that explicates this possible incoherence. This model shows that a mismatch between the levels of measurement of public opinion and measures of the common good is one important source of this supposed incoherence. When one balances one decision criterion against another, that balancing procedure must be sensitive to both differences in weights between those criteria as well as differences in scale magnitudes between the various alternatives. For instance, assuming that two criteria are weighed equally, if one criterion places some alternative x on a scale well above another alternative y and the other criterion places y very slightly above x, then the resultant decision should somehow be sensitive to these differences (and likely rank x above y). Without sensitivity to differences in magnitudes, a system of weights in a decision procedure using two criteria does nothing more than choose which criteria is the dictator. This would force political representatives to choose to follow either their mandate or their independence but would never allow them to balance those decision criteria. The measure of what is in the common good in the model is almost certainly cardinal, given that most cost-benefit analyses (its real-world corollary) provide such cardinal measures. However, public opinion measures are typically only ordinal. I show this by examining three methods by which political representatives typically gain knowledge of public opinion: voting, direct communication, and public opinion surveys. All of these yield ordinal measures of public opinion, which are inadequate for inclusion in a balancing procedure.
In order to save the balancing conception of political representation, I must show how public opinion might be conceptualized and measured using cardinal measurement scales. This task is far from trivial, and because measurement scales are just special types of utility scales it requires me to show how interpersonal comparisons of cardinal utility might be made significant, thus producing an aggregate (or social) utility scale. This process would therefore produce a social utility scale representing public opinion from individual utility scales representing individual opinions. It is thought by many that the insignificance of such comparisons render the measurement of cardinal social utility impossible. This is the view of standard social choice theory after Arrow (1963). In chapter 4, I show how this problem can be avoided by carefully defining the source of individual utility scales as attitudes about public policies. I formulate an attitude model that explains the source of individual utility differences, and then show how those differences can be measured through series of pair-wise comparisons of alternatives. This serves as the basis for comparisons between individual utility scales. Because all individual attitude sets share similar underlying structures, there exist significant metrics (central tendency and dispersion of utility values) that can be used to normalize all individual utility scales. Therefore, at least within the context of attitude measurement, it is possible to make significant comparisons of cardinal utility. In chapter 5, I show that once a number of normative assumptions are made about which features of attitudes are normatively significant --- and I argue that these assumptions are plausible within the democratic context --- a political representative can formulate a cardinal utility scale representing public opinion. Such a scale could then be balanced against some measure of expected welfare in a social decision process.
This dissertation has implications for a number of research areas. I address these in chapter 6. The most direct implications are in democratic theory and public opinion survey methodology. I argue that a necessary feature of democracies is that some political officials involved in policy decisions respect public opinion. However, a method that is capable of capturing the full information content of the public opinion model that I present would likely require far more demanding measurement methods than are currently employed by survey methodologists, and likely more demanding than is feasible in real-life measurement settings. I present a number of simplification methods that might track the measurement of public opinion that I present, though they may not completely capture the richness of the model.
I next address the implications of the model on deliberative democratic theory. Some might worry that any introduction of public opinion into legislative settings, in which deliberation about policy is central, will detract from the rationality of the policy formation process. After all, if political representatives are to come to a rational policy decision, one that is responsive to the best available arguments and reasons, then public opinion that has been separated from the reasons behind it may in fact make legislative decisions less rational. According to this view, there seems to be no space for public opinion in a deliberative institution. I argue that the use of public opinion can be integrated into decisions within deliberative institutions. I point out that within a society that features informal deliberative settings throughout, public opinion can in fact be seen as the result of a deliberative process, but one that has not been controlled to the degree as that found in a legislature; this should give public opinion some status in the public decision once deliberation has ended and a decision must be made. However, public opinion also gives representatives information about their constituents that might aid them in determining what arguments may ground public opinion. It will often be reasonable to include such arguments within the legislative deliberation process. Finally, communication between representatives and their constituents often resembles a deliberative setting, with representatives both informing the public with their arguments and being informed in return.
I then address an implication of the method of finding social utility on models in social choice theory, where cardinal utility representations are typically avoided as measures of preference. It is generally thought that no significant way exists to make interpersonal comparisons of cardinal utility. I show that if more is known about what the utility is used to represent then it becomes possible to make significant interpersonal comparisons of utility. However, social choice theory is typically neutral to what underlying phenomenon utility represents. I provide a case in which abandonment of that neutrality may allow for a more powerful analysis of social utility using social choice theoretic tools. A similar abandonment of neutrality in social choice analysis has been suggested by Regenwetter et al. (2006), who argue that actual voting patterns take certain stereotypical forms such that they are not susceptible to several voting paradoxes. The method I use in this dissertation adds credence to the view that social choice analyses may often benefit from making certain assumptions about phenomenon being measured.
The central argument of this dissertation can be summarized as follows: Any representative democracy requires that all citizens be represented by some political official(s) who have a role in policy decisions. This fulfills the democratic requirement of self-governance. Because political representatives are also tasked with coming to decisions that are in the best interests of the public, and the public is sometimes misinformed about what policies are in their best interests, the proper representation of the people requires officials to balance what they believe is in the best interests of their constituents with the wishes of those constituents. A failure to adequately balance these constitutes a failure of political representation. However, a number of problems with the basic structure of public opinion, revealed primarily by work in social choice theory and survey methodology, makes such a balancing procedure seemingly impossible. I present a model of public opinion that eliminates the possible problems with the balancing procedure. This includes a model of individual political attitudes, a model that represents individual attitudes in terms of utility scales over policy alternatives, and a social utility model that allows for the interpersonal comparison of those individual utility scales. I show that it is possible for public opinion to have a coherent role in the decision making of political representatives.