This is what the notion of a power of the demos means. The demos is not the population. Nor is it the majority or the lower classes. It is made up of those who have no particular qualification, no aptitude attached to their location or occupation, no aptitude to rule rather than be ruled, no reason to be ruled rather than to rule. Democracy is this astounding principle: those who rule do so on the grounds that there is no reason why some persons should rule over others except for the fact that there is no reason.
Jacques Ranciere (2009, 10)
On July 2, 2014, Quinnipiac University's polling firm released the results of a public opinion poll that found 33% of Americans participating in the survey believed Barack Obama is the worst president since World War II. George W. Bush came in as the second worst, winning 28% of the vote. Ronald Reagan was perceived as the best president since 1945 according to 35% of participants. Bill Clinton came in as second best (18%) and Kennedy as third (15%) ("Obama is First as Worst President," 2014). Quinnipiac is not a fly-by-night polling agency--on the contrary, they are viewed as one of the most reliable in the United States. So why is a respectable firm like Quinnipiac polling the nation with a question like "Who is the worst president since World War II?" (1) What is the point of such a poll?
Scholars in the fields of rhetoric and the social sciences have built up a large body of research on public opinion polling, and most agree that polls have a number of functions. The results of public opinion polls are used as predictors for upcoming elections; they are used to inform the public of what the public thinks (a polite way of saying they are used to modify individual opinions); they are utilized to influence the policy decisions of elected officials; they offer material for additional "news" stories that satisfy the round-the-clock demand for media commentary. In his essay "Public Opinion Does Not Exist," Pierre Bourdieu (1993) notes, "At present, the opinion poll is an instrument of political action: perhaps its most important function is to impose the illusion that there is something called public opinion in the sense of the purely arithmetical total of individual opinions [...] The 'public opinion' that is manifested on the front pages of newspapers [...] is a pure and simple artefact" (150).
Within the sphere of democratic deliberation, there are two major categories of public opinion polls: predictive polls and prescriptive polls. The data produced by both types of polls serve to advance specific claims about public policy and governance. Predictive polls use data analysis to advance explicit claims about trends or potential outcomes in political races. Put differently, predictive polling data purport to tell us what will happen. In contrast, prescriptive polls (sometimes called "issue polls") use data to construct representative models of public opinion at large in order to advance implicit claims about what we should do or how you should think about a particular social problem. In this essay I am primarily interested in the implicit claims advanced by prescriptive public opinion polling. If polling data and the news releases that frame that information suggest a governmental obligation to capitulate to the will of the statistical majority, this study explores that obligation as a rhetorical and political problematic.
The prescriptive question posed in the Quinnipiac poll (2014) is one that is obviously divorced from any political race or policy consideration. Thus, the two interrelated functions of the poll are fairly straightforward: it is intended (a) to influence and (de)legitimize individual opinions about the best and worst presidents of the modern era and (b) to provide material for news stories. But the apparent irrelevance of this poll calls into question the value of all polling. J. Michael Hogan (1985), one of the most prolific rhetoricians writing on polling, describes the core problem with viewing polls as an empirical tool of democratic governance: "American policy-makers routinely assume that 'the will of the people' may be discerned through 'scientific' polling, yet the conceptual and technical limitations of polls call this assumption into question. Polls typically do not distinguish between strong and weak opinions, between rational and irrational opinions, or among shades of opinion on particular issues" (303). Even if the numeric representation of a range of opinion is accurate, a poll cannot adequately represent the range of opinion: the public does not speak univocally. Ultimately, the implicit claims advanced by prescriptive polling undermine fundamental tenets of deliberative democracy: if the job of elected officials is simply to look at the math and implement the will of the majority, then formal public policy debates are merely a distraction from the strict administrative function of government.
New technologies have ensured a vast proliferation and acceleration of polling in the United States over the last 20 years. But these advances have not led to greater accuracy or reliability in the polls. John Zogby (2008) (a well-known pollster) notes that changes in communication technologies have "caused response rates to nose-dive in recent years" (20). Nate Silver (2014), a famed polling analyst and creator of poll-aggregating website Fivethirtyeight.com, explains that the increasing difficulty of running a poll is leading top firms to take both methodological and financial shortcuts. These factors have led to increased public scrutiny and criticism of polling data.
Most criticism of polls attacks the accuracy or bias of polling. People regularly challenge the phrasing of the questions, or the structure of the available answers, or the size of the sample, or the composition of the sample, or the calculation of the results, or the representation of those results in press releases. (2) All of these concerns address the bias inherent in polling and the issue of whether the results reflect the reality of mass opinion. However, questions of accuracy and bias lie outside the focus of this essay: I am concerned with the unintended effects of polling on democracy itself and the way that polling advances not only an image of the public's "opinion," but conflicting conceptions of good governance. Thus, this study takes for granted that there is such a thing as public opinion--a body of pre-existing thought on a given topic with a finite range of perspectives that can be numerically represented in terms of their distribution, prevalence, and intensity. I also assume that public opinion polls do measure this thing called public opinion that they measure it with accuracy, and that pollsters are able to represent and disseminate these results without corrupting the data. In other words, I take as given the outlandish idea that the polls do what they say: that they offer a valid picture of what the public thinks when it comes to the issues.
After a brief history of modern polling, I offer a conventional explanation of how doxa (or public opinion) operates in democratic life before showing how the vision of doxa that is operative in opinion polling represents a dramatic departure from traditional democratic theory. Further, I use raw data from a few issue-oriented polls to demonstrate some practical problems extending from our increasing reliance on prescriptive polling data. Showing that the polls' explicit rendering of doxa is a threat to democracy, I explain the opposing ochlocratic and autocratic implications of governance by polling. Finally, I advocate for a renewal of the classical conception of doxa via a return to more intuitive, indirect methods for assessing public opinion.
The emergence of public opinion polling
In his landmark book called The Image: A Guide to Pseudo-Events in America, Daniel J. Boorstin (1992) explains the development of polling in the United States. He notes that there was straw polling of citizens regarding their preferences for political candidates as early as 1824 (235). This style of polling was unreliable and unscientific. Early in the twentieth century, the scientific approach that defines today's opinion polls was pioneered in the fields of marketing and advertising. The popular magazine The Literary Digest was one of the first outlets to apply these techniques to the political realm beginning in 1920 when it surveyed the public and accurately predicted Warren Harding's presidential victory (Boorstin, 1992, 235). The predictive capacities of such polls were probably the main driver of their rapidly increasing popularity: people seek inside knowledge, and the ability to accurately predict the outcomes of political races by asking citizens who they would vote for seemed to reflect the good health of American democracy. Further, polling firms' recognition of the persuasive capacity of polling data incentivized more frequent polling. Indeed, the still-nascent polling techniques were ripe for abuse: because of the power of print media and the minimal relation between public opinion and truth, false claims about the public's opinion were just as powerful as the scientifically "valid" ones.
In 1936, The Literary Digest and most other polls were showing that Alf Landon would defeat Franklin Delano Roosevelt in the presidential election. George Gallup, a relative newcomer on the scene, not only accurately predicted Franklin Delano Roosevelt's victory, but also famously predicted the degree of inaccuracy in The Literary Digest's poll. By 1940, Gallup had become the most prominent pollster in America. His organization, the American Institute of Public Opinion, was dedicated to developing more scientifically accurate methods of public polling. In The Pulse of Democracy, Gallup (1940) warns of the illusions produced by the "informal" assessment of public opinion and laments their effects on democracy. Emphasizing the...