Benjamin Ginsberg: The Perils of Polling





[NOTE: In April, the Governing America in a Global Era program at the Miller Center of Public Affairs at U.Va. held a conference, "Has Polling Killed Democracy?" to examine the past and future of polling and its role in electoral politics. Among those participating was Benjamin Ginsberg, Johns Hopkins University, the author of this op ed.]

Politicians, advocacy groups, the media and public officials sponsor thousands of opinion surveys every year to assess public sentiment on issues ranging from abortion to Social Security to war in the Middle East. Some pollsters have argued that opinion surveys provide the most scientific and accurate representation of public opinion. George Gallup, one of the founders of the modern polling industry, asserted that opinion polls, more than any other institution, “bridge the gap between the people and those responsible for making decisions in their name.”1

Polling, of course, has no official place in the American governmental schema. The Constitution does not require public officials to follow poll results. During the entire Clinton impeachment process, the president’s standing in the polls remained high as did Richard Nixon’s until the eve of his resignation. Conversely, his declining standing in the polls throughout 2005 and 2006 did not compel George W. Bush to withdraw American forces from Iraq. In fact, Gallup’s view notwithstanding, the virtual representation provided by the polls does little to bridge the gap between citizens and decision makers. If anything, the polls render public opinion less disruptive, more permissive and more amenable to government and elite manipulation.

Polling has become so ubiquitous that commentators make little distinction between poll results and public opinion, but they are not the same thing at all. Public opinion can be articulated in ways that present a picture of the public’s political thinking very different from the results of sample surveys.2 Statements from leaders of interest groups, trade unions and religious groups about their adherents’ feelings are a common mechanism for expressing public opinion. The hundreds of thousands of letters written each year to newspaper editors and to members of Congress are vehicles for the expression of opinion. Protests, riots and demonstrations express citizens’ opinions. Government officials take note of all these manifestations of the public’s mood. As corporate executive and political commentator Chester Barnard once noted, before the invention of polling, legislators “read the local newspapers, toured their districts and talked with voters, received letters from the home state and entertained delegations that claimed to speak for large and important blocks of voters.”3 The alternatives to polling survive today. But, when poll results differ from other expressions of public opinion, the polls almost always carry more credibility. The labor leader whose account of rank-and-file sentiment differs from poll results is not likely to be taken seriously. Nor is the politician who claims that his or her policy positions are more popular than the polls show. In 1999, for example, Republican congressional leaders claimed that the public opinion disclosed by letters and phone calls supported their efforts to impeach and convict President Bill Clinton even though national opinion polls indicated that the public opposed Clinton’s removal from office. Virtually every commentator took the polls to be correct and accused the GOP of disregarding true public sentiment.

This presumption in favor of the accuracy of the polls stems from their apparent scientific neutrality. Survey analysis is modeled on the methods of the natural sciences and conveys an impression of technical sophistication and objectivity. The polls, moreover, can claim to offer a more reliable and representative view of popular opinion than any alternative. People who claim to speak for groups frequently do not. The distribution of opinion reflected in letters to newspapers and government officials is clearly unrepresentative. Scientific samplings of public opinion provide a corrective for false or biased representations of popular sentiment.

Polling, though, is both more and less than a scientific measure of public opinion. The substitution of polling for other methods of gauging the public’s views profoundly affects what is perceived to be public opinion. Polling is what statisticians call an “obtrusive measure.”4 Surveys do not simply record continuities and changes in a naturally-occurring phenomenon. The polls also define how individual opinions are to be aggregated. In opinion surveys, the views of well-informed people usually carry no more weight than those of the clueless.5 Pollsters also choose the topics for which public opinion will be tested. In other words, the data reported by the polls are not “pure” public opinion but the product of an interaction between the opinion holders and the opinion seekers. As surveys measure opinion, they also form opinion.

In the United States, as in other democracies, citizens expect their government to pay close attention to popular preferences. As we saw earlier, most Americans believe that the government does listen to popular opinion most or at least some of the time. This view is bolstered by a number of scholarly studies that have identified a reasonable correlation between national policy and public opinion over time. Alan D. Monroe, for example, found that in a majority of cases, changes in public policy followed shifts in popular preferences. Conversely, in most cases, if opinion did not change, neither did policy.6 In a similar vein, Page and Shapiro found that much of the time significant shifts in public opinion were followed by changes in national policy in a direction that seemed to follow opinion.7 These findings are certainly affirmed by hosts of politicians who not only claim to be guided by the will of the people in all their undertakings, but seem to poll assiduously to find out what that will is.

But, once we accept the notion that there is some measure of congruence or consistency between public opinion and public policy, we should not take this to mean that the public’s preferences somehow control the government’s conduct. Most citizens do not have strong and autonomous preferences with regard to most public issues. And many lack the basic information that might help them to understand and evaluate policy choices and governmental processes. For example, 40 percent of Americans responding to a recent survey did not know that each state has two senators; 43 percent did not know what an economic recession is; 68 percent did not know that a 2/3 majority in each house is required for a congressional override of a presidential veto; 70 percent did not know that the term of a U.S. House member is two years; 71 percent could not name their own congressional representative; and 81 percent could not name both of their own state’s senators.8 These findings suggest that many Americans can barely describe, much less control, their government.

When it comes to major public issues, many Americans have too much difficulty grasping the substance of the issue and the potential alternative policies to have any serious or coherent preference. For example, during the 2000 election campaign, reform of the Social Security system became a major issue. George W. Bush had proposed partially “privatizing” the system by allowing individuals to invest some of their payroll taxes in personal retirement accounts whose value would be subject to market fluctuations. Bush made a number of speeches on the topic; the idea was highlighted at the GOP convention; and the news media devoted considerable attention to it. After all the attention Bush’s proposal received, however, surveys revealed that most Americans knew little or nothing about it and had no meaningful preferences on the issue. 73 percent of those contacted by Princeton Survey Research Associates after the Republican convention said they knew “little” or “nothing at all” about Bush’s proposal.9 More than half could not say whether Bush’s proposed plan would raise taxes or how it might affect their likely Social Security benefits. More than half, however, favored the proposal–whatever it was.10 The same pattern is apparent in surveys dealing with several other recent political issues.11

Many Americans’ knowledge of contemporary political issues is limited to some half-remembered fact or claim they saw in an ad or heard on a newscast. And once they acquire some piece of information many individuals will retain it long after it ceases to have any relevance. In a 2005 Harris poll, for example, more than a third of the respondents believed that Iraq possessed weapons of mass destruction at the time of the American invasion–this despite the fact that even President Bush had long since acknowledged that no such weapons had existed. Apparently these respondents hadn’t been paying attention.

The unfortunate fact of the matter is that many Americans lack the cognitive tools or basic understanding of political and social realities to understand or to seriously evaluate competing political claims and proposals. Certainly, a minority of affluent, well-educated individuals–perhaps 20 percent of the public according to even the most generous estimates–are knowledgeable about public issues and possess the intellectual tools to evaluate them.12 The remainder are essentially what economists call “noise traders,” that is, individuals whose actions are based upon faulty information and questionable reasoning. This is, after all, a nation in which, according to a 2004 CBS News survey, 55 percent of all respondents reject the theory of evolution in favor of the idea that God created humans in their present form. This is a nation in which, according to an October 2005 Fox News survey, 84 percent believe in miracles and 79 percent in angels. According to the same survey 37 percent believe in astrology, 24 percent in witches and 27 percent in reincarnation. Perhaps we should be relieved that only 4 percent believe in vampires but, alas, 34 percent believe in ghosts. Some scholars have argued that, on the aggregate, the public can possess wisdom even though many, if not most, individuals are foolish. This argument, though, is based upon rather dubious statistical and logical assumptions.13

Americans’ lack of information and basic political knowledge and, frankly, lack of a simple capacity to distinguish fact from fable makes many Americans quite vulnerable to manipulation by politicians and advocates wielding the usual instruments of advertising and publicity. And, though they give lip service to the will of the people, politicians and advocates are quite aware of the fact that most of the time many of the people have no particular will or, for that matter, interest in or understanding of public issues. Their goal, as Jacobs and Shapiro note, is to “simulate responsiveness,” by developing arguments and ideas that will persuade citizens to agree with their own policy goals.14 This effort begins with polling. As Clinton pollster Dick Morris affirmed, “You don’t use a poll to reshape a program, but to reshape your argumentation for the program so the public supports it.”15 The effort continues with advertising, publicity and propaganda, making use of the information gleaned from the polls.

What do these observations mean for the relationship between opinion and policy identified by Monroe, Page and others? They suggest that opinion and policy are related primarily because of their common underlying origins. Rather than providing evidence that public opinion drives national policy, the correlation between the two derives from the fact that the same political forces seeking to shape national policy often find it useful to create a climate of opinion conducive to their goals.

Thus, for example, a coalition of forces that succeeded in bringing about the elimination of the estate tax in 2001 first made extensive use of polling and publicity over the course of several years to persuade the public that what they labeled the “death tax” was unfair and un-American. This public relations effort was a great success and helped smooth the way for the coalition’s lobbying effort in the Congress. As Graetz and Shapiro show, while the federal estate tax actually affected only the wealthiest 2 percent of the populace, the intensive campaign for its repeal seemed to persuade many naive Americans that the tax actually affected them. One poll taken in the wake of the repeal campaign suggested that 77 percent of the populace believed the tax affected all Americans, and several polls indicated that more than one-third of the public believed they themselves would have to pay the tax.16 When the tax was finally annulled, its elimination was supported by public opinion. But, does this mean that a change in public opinion brought about this change in policy? Hardly. Instead, a particular set of political forces engineered a shift in opinion which helped them to persuade Congress to change national policy. A similar pattern was also observed by Hacker and Pierson when they studied recent changes in tax policy. Citizens, they say, “proved vulnerable to extensive manipulation,” as political elites framed a discussion that generated popular support for policy changes that served the interests of a small minority of wealthy Americans.17

So much for the primacy of public opinion in the American democratic order. Bryce was far off the mark when he called opinion the “chief and ultimate power” in all nations. Indeed, the notoriously cynical Austrian economist, Joseph Schumpeter, was much closer to the truth when he observed that the will of the people was the “product,” not than the “motive power” of the political process.18

1. George Gallup and Saul Rae, The Pulse of Democracy: The Public Opinion Poll and How It Works (New York: Simon and Schuster, 1940), p.14.
2. Scott Althaus, Collective Preferences in Democratic Politics (New York: Cambridge, 2003); Benjamin Ginsberg, The Captive Public (New York: Basic Books, 1986); and Susan Herbst, Numbered Voices (Chicago: University of Chicago Press, 1993)
3. Chester Barnard, Public Opinion in a Democracy, pamphlet (Princeton, N.J.: Herbert Baker Foundation, Princeton University, 1939), p.13.
4. Se Eugene Webb, et. Al., Unobtrusive Measures: Normative Research in the Social Sciences (Chicago: Rand McNally, 1966).
5. For an excellent discussion of information effects on survey outcomes, see Althaus, chs. 4 and 5.
6. Alan Monroe, “Consistency Between Public Preferences and National Policy Decisions,” American Politics Quarterly 7 (1979), pp.3-18. Also, Alan D. Monroe, “Public Opinion and Public Policy, 1980-1993, Public Opinion Quarterly 62(1), (1998), pp.6-18.
7. Benjamin Page and Robert Y. Shapiro, “Effects of Public Opinion on Policy,” American Political Science Review 77 (1983), pp. 175-190. See also, Jeff Manza, Fay Cook and Benjamin Page (ed.), Navigating Public Opinion (New York: Oxford, 2002), Part I.
8. Carol Glynn, et.al., Public Opinion (Boulder, CO: Westview, 2004), p.293.
9. George Bishop, The Illusion of Public Opinion (Lanham, MD: Rowman and Littlefield, 2005), ch.2, p. 35.
10. Bishop, pp.34-35.
11. For a number of examples, see Bishop, chs. 2 and 7.
12. See for example, Jacob S. Hacker and Paul Pierson, Off Center: The Republican Revolution and the Erosion of American Democracy (New Haven: Yale, 2005), p. 67.
13. Some scholars, to be sure, have argued that although many individuals may lack information or coherent policy preferences, public opinion on aggregate may still be reasonable and sensible. Page and Shapiro, The Rational Public (Chicago: University of Chicago Press, 1992) for example, assert that, “public opinion as a collective phenomenon is...meaningful. And indeed rational ...it is organized in coherent patterns; it is reasonable...and it is adaptive to new information. (P.14). As Althaus, however, has demonstrated, this argument rests upon very shaky statistical foundations. Althaus, ch.2. Moreover, the notion that aggregate opinion may be reasonable despite the ignorance of individuals assumes that individuals do not communicate with or influence one another. This condition is usually violated in the case of political opinion where the ignorant influence one another and are influenced by politicians and the media. Anyone who doubts this should listen to talk radio. See James Surowiecki, The Wisdom of Crowds (New York: Doubleday, 2004).
14. Jacob and Shapiro, p.xv.
15. Dick Morris, Behind the Oval Office (Los Angeles: Renaissance, 1999). Quoted in Jacobs and Shapiro, p. xv.
16. Michael Graetz and Ian Shapiro, Death By A Thousand Cuts: The Fight Over Taxing Inherited Wealth (Princeton: Princeton University Press. 2005).
17. Hacker and Pierson, ch. 2.
18. Joseph Schumpeter, Capitalism, Socialism and Democracy, 3rd edition (New York: Harper, 1970), p.263.



comments powered by Disqus