With another campaign season behind us, it is not hard to make the case that opinion polls pervade politics. Even casual observers of politics are inundated with polling results from media and academic organizations. But just because surveys are ubiquitous does not mean they are useful. In his book, George Bishop makes the provocative argument that just about every bit of evidence that academics, journalists, and policymakers draw from opinion polls is erroneous. This argument will be controversial—indeed this book is clearly intended to provoke controversy—but the force of evidence that Bishop presents demands to be taken seriously by even the staunchest defenders of opinion polls.

Social scientists have long known that the particular wording and order of questions influence the responses individuals give to survey questions. Bishop's review of this evidence is thorough, but he has bigger fish to fry. He argues that the current practices of academic and political pollsters essentially force individuals to “make up” answers to survey questions on the basis of vague understandings of pertinent political controversies. Low levels of political knowledge and ambiguous questions combine to produce survey marginals that reflect illusionary measures of the public will. As a result, Bishop argues, political polls—at least in their current form—are not useful tools for understanding how the public feels about specific political issues.

The author draws upon a large body of data to make his case, ranging from measuring trust in government to approval for stem cell research. This evidence is largely persuasive. Consider, for instance, public opinion concerning the teaching of evolution. A number of recent polls have found that Americans favor by a large margin the teaching of creationism along with evolution in public schools. However, the majority of Americans are completely unfamiliar with the meaning of the term “creationism.” What, then, should we make of the survey findings? Bishop presents question-wording experiments that show that respondents are guided by their general feelings toward religion in public school when they answer the survey question. Thus, he argues, the widely cited polls on creationism do not measure specific attitudes toward the instruction of evolution. Rather, they capture the general desire of the mass public to find a middle way on the question of the separation of church and state.

Given the findings Bishop presents, it would be easy to make the leap that opinion polls are incapable of transmitting meaningful information from the mass public to political elites and should be rejected out of hand. But he argues that much of the problem with polls arises not from the survey method but from the manner in which they are currently conducted by media and academic pollsters. He therefore devotes the final chapter of his book to describing a number of measures that could be taken to improve the information gleaned through polls, including probing the meaning of responses to specific questions, attempting to gauge the strength with which citizens hold particular opinions, and measuring the levels of policy-specific knowledge that undergird the answers people give to pollsters.

While there is much to admire in The Illusion of Public Opinion, there are some problems as well. The book makes an important argument, but Bishop sometimes takes his argument too far. At times, the later chapters border on polemic. For instance, Bishop makes a critical point about exerting caution when interpreting opinion polls on specific issues. These polls, as he rightfully points out, may in fact measure general sentiment on political issues. But scholars and practitioners should not lose sight of the fact that this general sentiment matters in politics. For instance, take the creationism example. The author makes much of the fact that citizens do not have a firm grasp of the creationism controversy (and do not even know what “creationism” means). But this lack of knowledge does not render information gleaned from polls meaningless. The polls on the teaching of creationism tell us that citizens seek a middle ground on the place of religion in public schools. This result is important, both for students of mass behavior and for politicians who make education policy. The bottom line is that while individual polls may be problematic, a consistent message across different polls—here, the desire to find a place for religion in public schools—can be an important and powerful signal for policymakers.

Another difficulty with the book is that Bishop's discussion sometimes divorces polls from their political implications. He catalogs a number of cases where the specific meanings of particular questions have changed over time, including some of the most venerable survey questions, such as presidential approval, trust in government, and ideological attachments. For instance, he notes that the correlation between party identification and liberal/conservative identification has increased notably since the early 1970s; Democrats are more likely to say they are “liberals” and Republicans are more likely to identify as “conservative.” Bishop argues that this increased correlation is not evidence of increased ideological sophistication on the part of the mass public. Rather, he makes the case that the tightened association between these two concepts is an artifact, the result of “a simple increase in the sheer frequency with which the terms “Democrat” and “liberal” and “Republican” and “conservative” have been associated with each other in the mass media” (p. 124). I would argue instead that the increased association of ideology and partisanship represents a meaningful change in how ordinary citizens approach political controversies in this country. Recent work by Nolan McCarty, Howard Rosenthal, and Keith Poole demonstrates that political elites have become more polarized since the 1970s; Democrats and Republicans in Congress have moved toward the ideological extremes. Partisanship and ideology are linked in public discourse because the political environment in America has changed. Thus, what Bishop labels an artifact seems instead to be the public consequence of a change in the political system, and one with serious implications for the practice of politics in this country.

Finally, Bishop's discussion of interpersonal comparability of survey responses illustrates both the strengths and weaknesses of this book. If the same question means different things to different people, it is difficult to make sense of aggregate poll results. Two citizens might give the same survey response but mean different things in doing so. This is an important point, to be sure. But like other arguments in this book, Bishop sometimes takes this point to its extreme. For instance, he demonstrates that measures of trust in government are determined in large part by individuals' assessment of the president. Those who approve of the president are more trusting than those who disapprove of the incumbent. Thus, “trust in government” means different things to supporters and opponents of the administration. The evidence is clear on this point. However, an examination of the data over the last 30 years reveals that the gap in trust between supporters and opponents of the president is fairly steady—on the order of 20 or so percentage points. Among the two groups, the familiar over-time trends—the drop in trust in the early 1970s, the resurgence under Reagan—are evident. Thus, even though mean rates of support for supporters and opponents of the president might be different, the over-time changes in levels of trust are similar. Politicians and political scientists can learn a great deal about political behavior from studying these trends.

These criticisms aside, The Illusion of Public Opinion is an important book that should be read by students of mass political behavior as well as practitioners. Bishop has thrown down the gauntlet, providing a broad and provocative review of the current state of opinion polling. Even those who reject his conclusions must engage his argument.