Polling and the Public What Every Citizen Should Know Seventh Edition Herbert Asher Ohio State University CQPress A Division of SAGE Washington, D.C. 1 Polling and the Public Americans today are inundated with the results of public opinion polls sponsored by the news media, candidates for public office, incumbent officeholders, and many public, private, and nonprofit organizations. Some of the polling is on matters of immediate national importance, such as the war on terrorism, the health of the American economy, or the war in Iraq. But polls are conducted on almost any conceivable topic, be it a genuine public policy issue such as abortion, education policy, health care, the environment, or gay rights, or something more frivolous such as the guilt or innocence of a celebrity accused of a crime or which alcoholic beverage is the preferred choice of Americans. As a result, Americans are learning more and more about the attitudes of their fellow citizens. Public opinion polling has increasingly become an international phenomenon. The demise of the Soviet Union and totalitarian regimes in other parts of the world gave rise to new opportunities to assess popular opinion in those nations. Throughout the nations of the Middle East, Asia, and Africa, public opinion polling has become almost routine. It enables Americans to learn about the attitudes of people in other nations, including what they think about American popular culture and foreign policy. In 2002 and 2003, during the lead-up to the war in Iraq, many Americans were surprised by the widespread hostility that polls in other countries revealed and by street demonstrations by citizens of traditional allies such as Germany and France against the proposed U.S. action. Polls conducted in Lebanon in 2006 showed that huge majorities of Lebanese citizens did not believe that the United States was playing the role of an honest mediator in the conflict between Hezbollah and Israel. The internationalization of polling is also evidenced by the growing use of polls (and American-style campaign techniques in general) in national elections throughout the world. Not only 1 2 Polling and the Public Polling and the Public 3 have preelection polls become commonplace, but so also has the use of exit polls in areas as disparate as Palestine and Mexico. It is fascinating to learn about the opinions of people in other countries, but it may be difficult at times to evaluate the quality of the polling done in those nations, particularly those without a tradition of public opinion research conducted by reputable survey research organizations. Moreover, American media reports of international polls often focus much more on the substantive results and much less on the methodological and technical aspects of the research. In 2003, the National Council on Public Polls (NCPP) issued a cautionary statement urging that reporting on international polls give special attention to how they were conducted. NCPP observed a number of instances in which claims were made that a poll was representative of an entire country when, in fact, the sample might have been selected from only a few parts of the country or from only a few segments of its citizenry. Although our focus in this book will be mainly on polling in the United States, many of the lessons of the book will apply to evaluating international polls. In the United States, polls are increasingly being used not only to inform, but also to convince and even manipulate Americans in ways advantageous to the polls' sponsors. The aim of this book is to help citizens become more astute judges of polls, so that they will not be misled by assertions made on the basis of polling data. I will accomplish this objective by explaining in nontechnical language the factors that can affect poll results—such as question wording, sampling techniques, and interviewing procedures—and by critiquing various types and uses of polls. The Importance of Polls Why should citizens become more astute consumers of polls? One reason is simply to avoid being manipulated by those who use polls inappropriately to promote their own ends. Other reasons are more positive. Some people make major economic and career decisions on the basis of public opinion polling. For example, the businessperson who commissions a survey on customer preferences or the television station manager who underwrites a survey on audience demographics will use the information obtained to make important business decisions. Potential candidates for public office may commission a poll to assess their electoral prospects before deciding whether to run. In these cases a commercial polling organization is likely to conduct the polling. But the more knowledgeable the businessperson and the would-be candidate are about polls, the better able they are to communicate their objectives and requirements to the survey organization and to apply the results of the survey to their own decision making. Polls also are important for the average citizen. Through the substantial coverage the news media give to polls, citizens can compare their own beliefs with their compatriots'. As citizens use the polls in this manner, they need to be aware of the factors that affect the poll results so that they do not accept or reject them too quickly or uncritically. Polling plays an integral role in political events at the national, state, and local levels. In any major event or decision, poll results are sure to be a part of the news media's coverage and the decision makers' deliberations. How should an international crisis, such as North Korea's or Iran's nuclear threat, be resolved? How should the United States respond to events in Afghanistan or the Middle East or New Orleans? Should a Supreme Court nominee be confirmed by the Senate? What is the best location for a new library in the community? Should state taxes be raised? Because polls may influence how politicians respond to such public issues, citizens need to understand the essentials of public opinion polling. Finally, public opinion polls are playing an ever-larger role in political discourse in the United States because of the improved technology of polling, the introduction of courses in polling methodology in journalism curricula, the widespread assumption (challenged by Benjamin Ginsberg— see chapter 9) that polls are the best way to measure public opinion, and the belief that public opinion polls are instruments of democracy because they allow everyone's views to be represented. More worrisome is the huge growth in the use of pseudopolls—nonscientific and often-biased polls that private and public groups conduct through a variety of mechanisms, including the print and electronic media and the Internet, and that are often confused with legitimate public opinion polling. AH of these factors ensure that even greater reliance on the polls will characterize future political debate. To participate in that debate in an informed and analytical fashion, Americans will have to come to grips with public opinion polls—a useful tool of government and a valuable source of information to citizens and leaders alike. The Pervasiveness of Polls That public opinion polling is a growth industry in the United States is undeniable. The polls most familiar to Americans are those conducted for and reported by the major communications media. For example, major television networks sponsor polls in collaboration with a print news organization: CBS News with the New York Times, ABC News with the Washington Post, NBC News with the Wall Street Journal, and CNN with USA Today and the Gallup Organization. These media-sponsored polls often become an integral part of the print or electronic news story that the media A Polling and the Public Polling and the Public 5 outlet produces. Likewise, the major newsmagazines often incorporate polling information in their news stories. An examination of the stories featured on the covers of the three leading U.S. newsmagazines—Time, Newsweek, and U.S. News and World Report—between 1995 and mid-2003 reveals that about 30 percent of them cited public opinion polls. Other stories also used polling data. Readers of these magazines, then, will be more astute judges of the reporting if they are knowledgeable about the strengths and weaknesses of public opinion polls. The proliferation of polls also is evident in newspaper and television news coverage. Typically the polls the stories cite have surveyed citizens about their views on political issues, candidates, and incumbents (especially the president); their preferences about possible courses of government action; their general attitudes toward politics and the political process; and countless other political and nonpolitical matters. Since late 2001 polls have frequently queried Americans about terrorism and national security matters, and since 2002 Iraq, economic and tax policy, and health care have remained high on the polling agenda. In 2006 other issues such as immigration, wiretapping and domestic surveillance, and stem cell research have emerged to become prominent poll topics. Sometimes survey questions seem to violate standards of good taste. After President Ronald Reagan's surgery for what turned out to be colon cancer, a survey commissioned by Time asked respondents how serious they thought the president's health problems were, and both Time and an ABC News/Washington Post poll asked Americans whether they thought the president was likely to complete his term. A Newsweek poll inquired whether citizens were concerned that the president might not "be able to meet the demands of a second term." Many citizens undoubtedly had questions in their minds about the president's health, and therefore the media thought their readers and viewers would be interested in reading about public opinions on the matter. Ghoulish speculation was the result. When an issue or event becomes visible and especially controversial, the public is usually surveyed to assess its reaction. For example, in 2003 the U.S. Supreme Court, in the case of Lawrence v. Texas, struck down the Texas antisodomy law and affirmed the privacy rights of adult gay citizens engaged in consensual sex. Some reporters wrote exaggerated stories about what the Court decision meant for gay rights and gay marriage, and very quickly the pollsters were asking Americans their opinions about various aspects of gay life. Almost any topic seems amenable to polling. For example, the January 31, 2000, issue of U.S. News and World Report featured a story on "Hell" that described a public opinion poll in which a sample of Americans were asked whether there was a Hell (64 percent said yes) and what they thought it was like. As the final episode of the television show Cheers neared in 1993, the Times Mirror Center for the People and the Press queried Americans about such weighty topics as whether Sam should have stayed single, married Diane, or married Rebecca. That survey also asked Americans who their favorite Cheers character was and which character they would like to see continue in his or her own series (Mills 1993). In 2006, as various Hollywood superstars such as Jennifer Aniston, George Clooney, Tom Cruise, Angelina Jolie, and Brad Pitt were prominent in the news for both professional and personal reasons, a Gallup poll of a representative sample of Americans found that only Tom Cruise had a net unfavorable rating and indeed had lost substantial popularity in the previous year. Clearly, polling is everywhere. The prominent national polls are complemented by visible and reputable state and local polls that focus on state and local matters as well as national affairs. For example, the New York Daily News and the television program Eyewitness News (produced by the ABC affiliate in New York City) have polled people in the city over the years on their views of the New York police, New York mayors, the likelihood of the Yankees and Mets baseball teams' making the World Series, and other matters of local concern. Likewise, the New York Times, in conjunction with WCBS-TV, has conducted extensive studies of race relations in New York City. Many states have first-rate polling organizations, often affiliated with a university or a major news organization. For example, the Eagleton Institute at Rutgers University, working with the Newark Star-Ledger, surveys New Jersey residents about their state government and about New Jersey as a place to live. Publications such as Public Opinion Quarterly have provided summaries of state and local (as well as national) poll results. The polls I have mentioned thus far are the most prominent and probably the most credible to the American public. Their prominence stems from the often-substantial media coverage their results receive; their credibility derives from the public's perception that they are conducted scientifically and that the news organizations and other entities that sponsor them are themselves legitimate and objective. The most critical factor in making these polls scientific (and thus valid) is a carefully selected sample of respondents (most often 1,000 to 1,500 persons)- after all, no polling organization can interview the entire adult American population of more than 200 million. From such a sample, the public and the media can generalize (within certain limits to be discussed later) to the larger population from which the sample was drawn. 6 Polling and the Public Commissioned Polls Although polls by the big news organizations seem the most prominent, they are only a tiny fraction of the public opinion polling done in the United States. Many other organizations commission polls for purposes other than informing citizens. For example, companies may hire a polling firm to gauge the public response to their products, and academic investigators may use surveys in their research. The results of those polls may not attract much public notice, but they still can affect the lives of citizens. An excellent example of a commissioned poll was the one that the Internal Revenue Service JIRS) paid for in 1984 to study the prgblgrnof tax'cheatinc;. Among. "the items in the survey were these statements, with which the respondent was supposed to agree or disagree: Q: it's not so wrong to hold back a little bit on taxes since the government spends too much anyway. Q: The present tax system benefits the rich and is unfair to the ordinary working man or woman. Q: Since a lot of rich people pay no taxes at all, if someone like me underpays a little, it's no big deal. (Sussman 1984) Of the survey respondents 19 percent admitted cheating on their returns; young, upwardly mobile professionals were the most likely to cheat. The IRS study also investigated ways to reduce cheating and found that Americans strongly rejected the use of paid informants to catch cheaters {Sussman 1985b). Although the honesty of tax cheaters' responses to questions about tax cheating is questionable, the IRS probably gained some useful insights into the magnitude of the cheating problem and the feasibility of alternative solutions. The IRS study is typical of thousands that public and private bodies commission to address specific concerns. Indeed, the federal government sponsors many surveys measuring a wide variety of attitudes and behaviors. Often these surveys are longitudinal; that is, they are repeated over time, which allows the government to observe trends and changes. Some of the surveys are based on national samples; others on specialized samples that are more appropriate to the research questions being addressed. Government-sponsored surveys might study the purchasing practices of Americans, or assess the frequency of different types of high-risk sexual behavior, or monitor the use of illegal drugs, or follow the employment history of citizens. Commissioned surveys of this type, whether sponsored by the government or by private or nonprofit entities, are likely to be high-quality enterprises mainly because the sponsors have a genuine need for accurate -o .'a 'ö 3= Q. ° c "d o o £5 O CO 5 E ° Z c c E 8> £ fc .1 3 2 | c O) ■ "to 5 Q. JD CO £ co o w P ** c to n N 3 r§ ° d ü W CO 3 O .Ü O CD P CO CO o E B < jc ~ 5 o c -c co * § .5 ffi E I I M -j e HI o ™ a |P pes » "■o. is o co t as « 8 -J 31 S-5 1 111 Elsa 03 §| p 8 Polling and the Public Polling and the Public 9 information to address some organizational goal or problem. To that end they employ a reputable firm to design and conduct the survey and perhaps to analyze the data and interpret the results. Many other groups, however, conduct surveys for a different reason—not to address a public concern scientifically and objectively but to promote a certain position and to convince the public of the Tightness of that stand. The sponsors design those surveys to yield the desired results, most often through highly loaded questions, although subtler methods also are used. Sometimes in such surveys the people interviewed are chosen to ensure a predetermined outcome. In many cases, the poll itself is secondary to the real aim of the group; to raise money to support its objectives. With the advent of computerized mailings, many organizations have entered the business of raising funds and conducting polls through direct mail. Polling then becomes a device to generate donations—that is, the sponsoring organization encourages recipients of the mailings to make their views known and to contribute to a good cause. Many of these appeals come from political groups, ranging from the Democratic and Republican Parties to various issue-based and ideologically oriented groups and organizations. Until the early 1990s, the Democrats lagged behind the GOP in the use of computerized mailings. But the Democrats soon learned the advantages of attaching a poll to a fund-raising effort. Issue and ideological groups of the left and right have learned the same lesson. Pages 7 and 8 show parts of the mailings from two distinct organizations—Americans for Immigration Control Inc. and Environmental Defense. The kinds of questions included in each survey reveal the agenda of each organization. For example, Americans for Immigration Control asks the following two highly loaded questions: Q: The use of English as America's primary language is being measurably eroded through the growing acceptance of non-English languages in official capacities— such as drivers' license testing, voter ballots and citizenship ceremonies that are now offered in various foreign languages. Should steps be taken to slow this trend toward multilingualism by making English the official language of America? Q: Bilingual programs currently consume over $28 billion education dollars a year because 1 child in 20 now can't speak English adequately—even though bilingual education has been proven to be an ineffective way to teach English. Do you support this use of public education dollars? Environmental Defense likewise asked biased questions: Q: Do you believe that the results of the 2004 election gave special interests and big polluters a mandate to prevent action to curb global warming pollution? 10 Polling and the Public Polling and the. Public 11 Q: Do you approve of the government's failure to reduce emissions of carbon dioxide, the chief pollutant causing global warming? The American Association for Public Opinion Research (AAPOR), the major professional association of public opinion researchers, part of whose mission is to protect the integrity and reputation of the polling enterprise, has condemned the practices of "FRUGging"—fund raising under the guise of surveying—and "SUGging"—selling under the guise of research. Part of AAPOR's concern about these practices is enlightened self-interest: Sound public opinion research needs respondents who are willing to participate in the polling enterprise. In recent years, respondent participation rates have declined. There are many reasons for it, but an important factor is the behavior of aggressive and sometimes unethical telemarketers, who disturb and disrupt citizens through intrusive telephone calls. Many Americans had gotten so fed up with telemarketers, some of whom pretended they were conducting a public opinion survey when in fact they were trying either to raise money or to sell a product, that they became hostile to legitimate telephone requests to participate in a public opinion poll. In response to growing citizen anger about excessive telemarketing, the federal government established a "do not call" registry. Citizens who registered their telephone numbers would be protected from unwanted telemarketing calls. The "do not call" registry fortunately provided an exemption for legitimate survey research and public opinion polling, and subsequent court decisions have upheld the important distinction between intrusive telemarketing and genuine public opinion research. Today, the "do not call" registry, as well as a federal provision that requires that charitable fund-raisers state at the very beginning of their telephone call that their purpose is to seek a charitable contribution, should dramatically lessen the problems of "frugging" and "sugging." But groups can choose to raise money through direct-mail solicitation, and here the use of a public opinion poll as a ploy to raise money is still widespread. Many groups involved in fund raising, and/or generating survey responses that support their own agendas, mail extremely biased literature and then ask respondents for their opinions. Consider these examples: The Committee Against Government Waste has asked, "Before you received this letter, were you aware of the gross mismanagement and waste of funds in the U.S. Department of Defense's purchase of parts?" The American Farmland Trust has asked, "Were you aware of the gravity of the problem of our vanishing farmland before receiving this mailing?" In case the literature accompanying the poll does not convince respondents of the correctness of the group's position, a carefully constructed question or statement may achieve the same end, as illustrated by the following questionnaire items and their sponsoring organizations: Q: Are you in favor of allowing construction union czars the power to shut down an entire construction site because of a dispute with a single contractor, thus forcing even more workers to knuckle under to union agents? National Right to Work Committee Q: Our nation is still blessed with millions of acres of public lands, including roadless wilderness areas, forests and range lands. Land developers, loggers, and mining and oil companies want to increase their operations on these public lands. Do you think these remaining pristine areas of your public lands should be protected from such exploitation? Sierra Club Q: Do you feel that all of the TV networks are in serious danger of losing the public's confidence and trust because they hire so many liberal Democratic activists as top corporate executives who formerly worked for Ted Kennedy, Walter Mondale, Gary Hart, George McGovern, Mario Cuomo, Jimmy Carter and the National Democratic Party? Fairness in Media Q: Did you know that the United States continues to embrace lousy trade deals with countries that use prison labor, slave labor, and child labor? Fight Back America Q: Are you concerned about the future of the arts and arts education in the United States given that 40% of funding for state arts councils has been eliminated in recent years, and that the share of private donations dedicated to the arts has declined by more than 40% in the last decade? Americans for the Arts Action Fund Q; Although tobacco is estimated to kill at least 418,000 people each year, and to cost you and me and every American taxpayer more than $100 billion, our federal government continues to spend millions to encourage the growing of tobacco and to subsidize the tobacco industry. Therefore, do you think the government should: immediately discontinue all aid to tobacco, phase out all aid for tobacco over a longer period of time, or continue to aid tobacco growers and the tobacco industry at present levels? Action on Smoking and Health All the preceding items were carefully constructed to generate responses s^ffnpjjhptir fr> fho sponsors' nbiectives.Tn fact, for several reasons ffiese enterprises should not be called "polling." First, in most cases the sample is not scientifically selected; instead, the surveys and fund-raising requests are mailed out to lists of citizens who are thought to be likely supporters. Whether the people who actually respond are at all representative of a larger population is of little concern. Second, the questions are often poorly formulated and fundamentally flawed (deliberately so). Third, if the "survey data collected are tabulated at all (and many times they are not), little analysis can be conducted because the original survey was very short and 12 Polling and the Public Polling and the Public 13 omitted key questions about the demographic and political characteristics of the respondents. In other words, collecting opinions is not necessarily valid polling. Orton (1982) has identified similar examples of what he calls pseudo-, polls._For example, the print and electronic media often encourage members of their audiences fb write or phone to express their views. But even with hundreds or thousands of repliesTthese" straw polls" areusually not representative simply because the people who volunteer to participate are likely to differ in important ways from the overall population. They may be more interested, informed, and concerned about the topic at hand and thus hold views different from those of the overall population. In 1992, shortly after President George Bush's State of the Union address, the CBS television program America on the Line featured telephone call-in surveys. At the same time, CBS conducted a scientific poll that included questions identical to those on the call-in survey. The results of the two surveys differed. For example, in response to the question of whether they were better or worse off than they were four years ago, 54 percent of the callers in the phone-in poll said they were worse off, compared with only 32 percent of the respondents in the scientific survey (Morin 1992a). Radio talk shows and call-in polls became more prominent in the 1990s. A 1993 Times Mirror poll revealed that citizens who listened to and called radio talk shows were not representative of the overall citizenry; instead, they tended to be more Republican, more conservative, more male, and slightly more wealthy and educated (Kohut 1993). It is thus not surprising that radio phone-in polls often generate results more conservative and pro-Republican than the outcomes obtained through scientific polling. Other examples of pseudopolls are the questionnaires that members of Congress send to households in their congressional districts. Typically, these are addressed to "Postal Customer," and there is no sure way of knowing just who in the household actually completed the survey. Although thousands of these questionnaires may be returned to a congressional office, it is difficult to ascertain whether the respondents' demographic characteristics and opinions are truly representative of the broader constituency. In some instances the questions are loaded to guarantee responses compatible with a legislator's predispositions and record. This is not to say that completed questionnaires are ignored or discarded; in most cases the results are tabulated and later reported to the constituency in a newsletter. But the major purpose of these questionnaires is to convince voters that members of Congress care about their opinions. The highly publicized surveys on marital relations once conducted by feminist author Shere Hite and by newspaper columnist Abigail Van Jeff Stahler: © The Columbus Dispatch/Dist. by Newspaper Enterprise Association, Inc. Buren, otherwise known as "Dear Abby," are also pseudopolls (Squires and Morin 1987; Smith 1988). Hite distributed 100,000 extensive, open-ended questionnaires to women's groups and to individual women who requested one. She received about 4,500 replies, a response rate of only 4.5 percent. In one of her columns, Abby wrote, "Readers, I need your cooperation for an important survey. Questions: Have you ever cheated on your mate? How long have you been together? You need not sign your name, but please state your age and indicate whether you are male or female." She received more than 200,000 responses (Smith 1988). In both the Hite and "Dear Abby" surveys, the sampling method and the questions generated unrepresentative and misleading results, despite the large number of respondents to Abby's poll. (Reputable, scientific national polls typically have a sample of about 1,500 respondents.) Hite found that 70 percent of women married five or more years were having extramarital affairs; 15 percent of Abby's married female respondents claimed to have been unfaithful. As Smith (1988) argues, both surveys 14 Polling and the Public Polling and the Public 15 could not be correct, and indeed both were overwhelmingly likely to be wrong because of the shortcomings in the sample selection and the actual questionnaires. Allowing citizens to select themselves into a survey guarantees biased results because of the motivations that lead people to participate in such surveys in the first place. Since the 1970s magazines have regularly published the results of sex surveys of their readers; the magazines know that sex increases readership. Typically they conducted the surveys by including the questionnaire in the magazine and encouraging readers to complete it and mail it back, although today on-line responses are encouraged. Redbook and Cos-mopolitan in the 1970s, Playboy in the 1980s, and The Advocate in the 1990s are among the magazines that have sponsored such surveys. In some cases the response rate was low, but the number of completed questionnaires very large, simply because of the size of the magazine's readership. For example, the Playboy response rate was about 2 percent, but that translated into 100,000 replies. The Advocate response rate was 18 percent, with almost 13,000 questionnaires returned (Lever 1994, 18). Despite the large number of replies to a typical magazine survey, one must be very careful when generalizing the results to any broader population, whether it be to straight males based on the Playboy survey or to gay males based on the Advocate poll. The reason is that self-selection presents a double problem. First, the readers and subscribers to a particular magazine may not be representative of the broader population of which they are members. Second, the people who actually complete the questionnaires may not be representative of the magazine's total readers and subscribers. Nevertheless, the results of these surveys typically receive a lot of news media coverage (that probably enhances magazine sales). The latest development in pseudopolls is the on-line survey on the Internet. Many businesses, media outlets, and other organizations invite visitors to their Web sites to participate in on-line surveys. Magazines that formerly included hard copy questionnaires in their issues are today more likely to invite their readers to participate in an on-line survey. For example, in its September 2006 issue, Baby Talk magazine published a story on the views of younger versus older moms based on the responses of more than 15,000 readers to an on-line survey. Like other pseudopolls, the on-line survey often generates many thousands of responses, but it is not a valid survey because the respondents selected themselves to participate rather than being part of a scientifically selected sample. They are unlikely to be a representative sample because only those people who are interested in the topic are likely to participate. Moreover, many citizens do not have access to the Internet in the first place and therefore could not have participated in that or three other recent on-line surveys: AOL users on August 1, 2006, could participate in a two-part survey about Mel Gibson's arrest for drunk driving and subsequent anti-Semitic statements: "Will Hollywood turn its back on Mel Gibson? What do you think of Mel Gibson's apology?" Citizens who went to the CNN Web site on August 1 (when Fidel Castro was having major health problems) could respond to the item, "Do you think Cuba ever will be a democracy?" Users of the Columbus Dispatch Web site that day could answer the question, "Should churches tell their followers how to vote on political candidates and issues?" Even though thousands of people responded to these questions, there is no guarantee that they are representative of any larger group of people. Fortunately, both AOL and CNN cautioned their users on this point, although AOL users had to go to the link "Note on Poll Results." If they did, they were told, "Poll results are not scientific and reflect the opinions of only those users who chose to participate." The CNN users were provided a more detailed cautionary note right below the actual results; it read: "This QuickVote is not scientific and reflects the opinions of only those Internet users who have chosen to participate. The results cannot be assumed to represent the opinions of Internet users in general, nor the public as a whole. The QuickVote sponsor is not responsible for content, functionality or the opinions expressed therein." On-line surveys and conditions under which they might provide useful information are discussed in greater detail in chapter 5. The key point, of course, is that pseudopolls are highly flawed and may give misleading portraits of public opinion because of loaded and unfair question wording, sejf-selgction bias in the respondents, outright efforts to stack the results, or other deficiencies. Despite their deficiencies, however, these unscientific enterprises must be discussed and criticized because they are becoming more prevalent in the United States. At times, Americans hear the results of a pseudopoll and pass them along to colleagues as if they were legitimate public opinion research findings, thereby giving the pseudopoll greater credibility and dissemination than it merits. Indeed, there is some evidence that a substantial number of Americans give credence to various kinds of pseudopolls. Brodie et al. (2001, 13) report that 26 percent of Americans believe that a survey in which readers fill out a questionnaire printed in a magazine almost always or most of the time accurately reflects what the public thinks. Twenty-seven percent believe that Internet surveys almost always reflect what the public thinks, and 35 percent give high marks to surveys in which people are mailed a questionnaire and asked to complete it. According to Brodie and her colleagues, fewer than 40 percent of Americans realize that selecting survey respondents at random produces a better, more accurate result than the self-selected samples such as radio call-in or magazine write-in polls rely on. The 16 Polling and the Public Polling and the Public 17 importance of these unscientific Internet polls was brought home to me when I received an e-mail in August 2006 from a friend urging me to support Israel on the CNN Web site, which was asking whether Israel's actions in Lebanon were justified. My friend argued that "this survey is shown all over the world, and will have a strong impact on world public opinion." All of this suggests that bad polling practices and results can mislead many Americans. Citizens are subjected to many different kinds of polls,_aH of which may later attect them in some way through decisions made oh the basis of poll results. That is why it is important that citizens be aware of the gamut of polls and be able to evaluate them. If they are able to recognize unscientific polls and their associated deficiencies (as well as the shortcomings of scientific polls), then people are less likely to be misled by them. This leads to the central concern of this book—the citizen as a potential consumer of public opinion polls. The Citizen as a Consumer of Polls Whatever the quality of the polls, they can affect the attitudes and behavior of citizens. Even media-sponsored polls designed simply to report citizens' attitudes (and perhaps to keep up with the competition and improve ratings) also may help to shape preferences, particularly during a presidential primary season. At such times polling is frequent, and the links among a candidate's poll standing, news media coverage, and primary election fate are pronounced. (The role of polls in elections is considered in chapter 7.) Americans are major consumers of public opinion research on a wide variety of topics. But are they smart consumers? Americans should be aware of the problems and limitations of polls before they "buy" anything from them. Often someone is actively promoting the poll results to generate support for his or her objectives. It might be the president, citing polls to argue that the American people support administration policies. It might be a local builder, waving the results of a neighborhood poll purporting to show local support for a rezoning ordinance to permit his commercial construction project to go through. It might be a regional transportation commission, citing poll results to justify the establishment of bus lanes on freeways. Or it might be a friend or neighbor selectively using poll results to win an argument. Citizens need not become experts at drawing samples, constructing questionnaires, or analyzing data to be better consumers of public opinion research. Instead they can become aware of the steps involved in conducting a survey and the possible consequences of the steps, so that they are better able to reject bad "merchandise" and appreciate good buys. Readers should not construe this book as a condemnation of public opinion research; most of the highly publicized polls, as well as many private polls, reflect high standards of polling. In fact, public opinion polling has improved dramatically over the past sixty years in areas such as sample design, question wording and format, interviewing techniques, and methods of data analysis. Interested readers may consult the fiftieth anniversary issue of Public Opinion Quarterly (lj87JJ-oTdis<^ prances T-iave changed" over-time. A very concisehisfory of polling can be found in Public Opinion: Measuring the American Mind (Bardes and Old-endick 2003, chapter 2); Weisberg (2005, chapter 1) provides an informative overview of the development of survey research as a scientific discipline. Polling practices, of course, continue to change with the advent of the Internet and the increased challenges faced by traditional telephone polling, topics addressed in chapters 4 and 5. An appreciation of the art of conducting and analyzing surveys will leave citizens less susceptible to the intellectual tyranny that can occur when a public opinion poll is deemed by its sponsor to be scientific and its results are presented as therefore beyond question or challenge. Citizens' Views of Polls Historically, ordinary citizens' reactions to public opinion polling have often been positive, although their views of pollsters themselves are much more skeptical. One study rated the credibility of forty-four different professional groups that speak out on public issues; pollsters finished thirty-forth in the ranking (Morin 1999b). But 25 percent of the respondents in a 1996 Gallup survey said they regularly followed the results of a public opinion poll in a newspaper or magazine, and an additional 16 percent said they did so occasionally (Morin 1996b). Fifty-nine percent of the respondents said they did not follow a poll regularly in a print medium; of course, they might sporadically read about polls, or they might be aware of the polls through the electronic media. That Gallup poll and others like it have indicated that Americans held fairly positive views about the accuracy of polls. More than two-thirds of the Gallup sample said the polls were right most of the time. Eighty-seven percent supported polls as "a good thing." In 2001 the Henry J. Kaiser Family Foundation, in conjunction with Public Perspective magazine, conducted an extensive survey on the role of polling in a democracy and on the views of the public, policy leaders, and the media toward polling (Brodie et al. 2001; Witt 2001). The opinions of the public were mixed. Twenty-eight percent of the respondents said that conducting a public opinion poll was a very good way to learn what a majority of people in the country think, and an additional 56 percent said it 75 Polling and the Public Polling and the Public 19 was a somewhat good way. But when they were asked to choose among town hall meetings, polling, talking to people on the street and at shopping malls, or talking to people who call, write, or e-mail public officials as the best way for officials to learn what the public is thinking, 43 percent opted for town hall meetings and only 25 percent for polling. The same study also asked how useful people thought polls are for enabling officials in Washington to understand how the public feels on issues. Twenty-two percent of the respondents said that polls were very useful; 54 percent said somewhat useful; 13 percent, not too useful; and 8 percent, not useful at all. The 75 percent (54 + 13 + 8) who rated polls as somewhat useful or lower were then presented the following question: Q. For each of the following statements, please tell me if you think it is a major reason polls are only somewhat useful or not useful for officials to understand how the public feels about important issues, a minor reason, or not a reason: Percentage rating it Statement a major reason Polls don't accurately reflect what the public wants. 43 Polls don't ask for the public's opinion on the right issues. 39 The results of polls can be twisted to say whatever you want them to say. 58 In the same survey, only 50 percent of the respondents strongly agreed or somewhat agreed that "public opinion polling is based on sound scientific principles," whereas 80 percent strongly or somewhat agreed that "the questions asked in polls often don't give people the opportunity to say what they really think about an issue" (Brodie et al. 2001, 22-23). All of these results suggest that Americans have some genuine reservations about the polling enterprise itself and about politicians' use of polls Other, anecdotal evidence also supports that finding. Most teachers and practitioners of public opinion polling have encountered citizens who have expressed utter distrust of polling. Some complain that they have never been interviewed and don't know anyone who has, and they therefore wonder just how representative samples can be. Koch (1985) found i that people who have never participated in a poll are dubious about the ac- 'j curacy of survey results. |! Others base their doubts on the size of the samples. In talking about | polling with diverse audiences, I repeatedly hear people ask how a sample f. of 1,500 respondents can possibly represent 200 million adult Americans. | DON'T FORGET THAT ELEGTlOW YEAR. 17 AL70 UE-TO-THE- POLL7TEK7 YEAK, Reprinted by permission of the Detroit Free Press And despite my brilliant answer by analogy—a doctor takes only a sample of a person's blood (fortunately), and a chef need taste only a spoonful of soup (assuming the soup is stirred properly) to test its seasoning—much skepticism about polls remains. Indeed, a 1985 Roper survey produced an estimate that only 28 percent of Americans believed that national polls with sample sizes of 1,500-2,000 could be accurate, while 56 percent said they could not be (Sussman 1985c). The 1996 Gallup survey discussed above obtained similar results. Another study found that citizens who questioned the accuracy of polls attributed their skepticism to characteristics of the respondents as well as aspects of the polling enterprise itself (Dran and 20 Polling and the Public Polling and the Public 21 Hildreth 1995). Skeptics worried whether pollsters talked to the right people, as well as whether sample sizes were adequate. (I will discuss sampling and what makes a poll scientific in chapter 4.) Other concerns focused on pollster manipulation of surveys and media misuse of polls. Polls have received much criticism pertaining to their growing role in elections and in election coverage by the news media. The argument is often made that polls have contributed to the packaging of candidates: Aspiring leaders are accused of first consulting the polls and then staking out their positions, thereby abdicating their leadership responsibilities on vital public issues. Polls are seen as encouraging a horse-race mentality among the news media, which concentrate on who's ahead and who's behind, who's gaining and who's falling back, as measured by the polls, instead of focusing on issues and the candidates' qualifications. Candidates themselves have complained about the role of polls. Indeed, the most common reaction from candidates who are shown to be losing in the preelection polls is the assertion that the only poll that counts is the one on election day. In the presidential nominating process, when multiple candidates are seeking a party's nomination, the media cannot and do not cover all candidates equally. Instead they give greater coverage to the viable candidates, and viability is most often determined by how much money candidates have raised and how well they are doing in the polls. Clearly, the polls do more than reflect how well candidates are doing; they may also affect how well candidates will do (see chapter 7 for a more detailed discussion of this phenomenon). Exit polls—interviews with citizens just after they have voted—enable the television networks to project election outcomes even before the voting is finished. This practice has angered many citizens as well as political elites. Newspaper columnist Mike Royko encouraged voters to lie to exit pollsters, and others (such as Munro and Gans 1988) have simply encouraged a public boycott of exit polls. Congress has conducted hearings to try to get the networks to alter voluntarily the ways they report exit polls and election projections. The election night debacle in Florida in 2000 was a major black eye for exit polls, pollsters, and the broadcast media. When television networks prematurely called Florida for Gore early on election night and later, again prematurely, called it for Bush, the excesses and frailties of media election coverage, with its reliance on exit polls, were fully exposed. The premature leaking of exit poll results on election day in 2004 helped generate postelection controversy about the fairness of the election, especially in states such as Ohio and Florida, when early exit poll reports did not correspond with the actual election outcomes. Exit polls are discussed in depth in chapter 7. Other observers have condemned the impact of polls on American politics, none more harshly than Daniel Greenberg (1980), who wrote: Given the devastation that opinion surveys have brought to the American political process, we shouldn't be asking how polls can be sharpened but rather why they are endured and how they can be banished. Polls are the life-support system for the finger-to-the-wind, quick-change politics of our time and, as such, are the indispensable tools for the ideologically hollow men who work politics like a soap-marketing campaign. ... The effect of this—on campaigns, as well as on administrations between campaigns—is an obsession with salesmanship rather than with governance. The nation's political cartoonists, many of whom are syndicated in newspapers that themselves conduct polls, have had a field day attacking the polls, particularly their frequency, duration, and intrusiveness in the presidential selection process. Smith (1987, 209) found that polls were treated negatively in 61 percent of the comics and cartoons he analyzed and worried how this would affect citizens' reactions to polls. More recently, political commentator and writer Arianna Huffington established a Web site called "Partnership for a Poll-Free America." Huffington has been particularly critical of the polling enterprise because of low response rates (a topic discussed in chapter 4}. Various practitioners of polling and survey research have become concerned about what they see as increased skepticism, cynicism, lack of interest, and even hostility toward polls. Prominent pollsters such as Harry O'Neill, of the Roper organization, and Kathy Frankovic, director of polling at CBS News, have called on their industry colleagues to be more reflective about their enterprise and more sensitive to its shortcomings (O'Neill 1997; Morin 1998a). Black (1991) advocates greater sensitivity to the needs of respondents by (1) making the interview itself a more interesting and rewarding experience for respondents; (2) keeping promises made to respondents in such areas as the length of the interview and the provision of final reports if requested; and (3) maintaining high quality throughout the polling enterprise. Lang and Lang (1984) worry that some of the more recent entrants into the polling business may have weaker ties to the profession and a lesser commitment to the high standards that should characterize public opinion polling (Morin 1992c). They urge careful self-policing by the polling industry to protect the profession and its reputation among the public. Others (such as Tanur 1994) recommend better education of citizens as consumers of polls (which is indeed the main purpose of this book). The point is that there is growing unease among many practitioners of polling because the lofty status that public opinion assessment has enjoyed may be in some jeopardy. In May 1997 the American Association for Public Opinion Research (AAPOR 1997a) issued a publication detailing the "best practices" that 13 22 Polling and the Public should characterize public opinion research. The last recommendation called for the disclosure of "all methods of the survey to permit evaluation and replication" and presented a comprehensive list of items that exceeded the standards for minimum disclosure laid out in the AAPOR code (see chapter 6 for a fuller discussion of disclosure standards). AAPOR also condemned a number of survey practices, such as presenting the results of any self-selected poll as if they were the product of genuine scientific survey research. It also sharply criticized "push polling"—an election campaign tactic disguised as legitimate polling. In a push poll, a campaign contacts a large number of voters under the guise of conducting a public opinion poll, presents some negative information about another candidate, and then asks some questions about that candidate. Indeed, one of the major warning signs of a push poll is the large number of respondents contacted, a num- \ ber substantially larger than would be needed for a valid and representative public opinion poll. The aim of push polls is not to acquire legitimate infor- : mation about the election contest but to push potential voters away from a particular candidate. (Push polling is discussed in greater detail in chapter 7.) The Luntz case illustrates how politicians or a political party may use polls to promote an agenda. In April 1997 the executive council of AAPOR ; formally chastised pollster Frank Luntz for violating AAPOR's Code of Pro- j fessional Ethics and Practices. Prior to the 1994 congressional elections, Luntz claimed that his research showed that sizable majorities of Americans supported all parts of the GOP's "Contract with America." But when asked to make public the wording of his poll and other information, Luntz re- I fused. AAPOR rebuked him for failing to meet the standards of disclosure j (AAPOR 1997b). Later it turned out that Luntz's evidence of support for the Contract with America was highly suspect and misleading. Nevertheless, the House Republican leadership used the alleged poll results to build I; legislative support for the items in the program. In general, political lead-ers who use the polls to "prove" that the public supports their positions j; have an advantage in the public discussion of issues. s Although they are sometimes angry or skeptical about poll results, t Americans often think polls are accurate and fair They often resent the in- £ trusiveness and presumed power of the polls, but they eagerly consume the latest public opinion findings about myriad topics. This love-hate relationship || is probably inevitable in the U.S. political system. Americans want their j; voices to be heard, and therefore they attack the polls when they think such § devices are undermining genuine citizen involvement and influence. Yet in a . 'i large and heterogeneous nation such as the United States, the polls may be | the best mechanism for reflecting the diversity of public opinion. The simple | fact that polls generally count all respondents equally bestows on polls a dem- S ocratic character that enhances their appeal in a democratic society. j Polling and the Public 23 Polling and Democracy Different theories and opinions compete concerning the role of polling in a democratic society. Advocates of polls emphasize that polling is an opportunity for citizens to participate in democracy and that it permits quick and repeated assessments of public opinion. Polling is particularly valued by those who prefer a democracy in which the people govern directly rather than through elected representatives. Some advocate polling citizens on their policy preferences and enacting those preferences, thereby circumventing the "middleman"—the elected representative—although such a system would ignore many important features of the governing process, such as dialogue, deliberation, bargaining, and compromise. Many in this group are fascinated with the possibility that technological innovations might facilitate direct governance by the citizenry. The huge advances in communications in recent years have rendered the virtues and feasibility of direct democracy through technology even more tantalizing. Some proponents of the more traditional theory of representative democracy also welcome public opinion polls because they provide systematic information on the preferences of the citizenry. They argue that citizens' opinions should influence the behavior of their elected representatives and that any mechanism that can provide information on citizens' opinions is bound to foster democracy. But the empirical evidence is mixed on the extent to which popular preferences are actually translated into public policy. On the one hand, examples of the government's seeming unresponsiveness to public opinion are numerous. For example, in 2006, polls repeatedly showed majorities of Americans in support of expanded embryonic stem cell research, yet Congress was unable to override the president's veto of legislation that would accomplish that. Likewise, polls in summer 2006 showed Americans increasingly critical of the war in Iraq and increasingly favorable to some kind of timetable for reducing American involvement, yet the mantra from the Bush administration going into the 2006 elections was, "Stay the course." One could, of course, argue that stem cell research and especially Iraq are such complex issues that it would be naive and even misguided to expect that preferences revealed in public opinion polls should automatically be translated into public policy. But the fact remains that the polls revealed a disjunction between popular opinion and government policy. Polls regularly showed an overwhelming majority of Americans favoring some form of gun control, such as handgun registration or waiting periods, for years before Congress finally passed the so-called Brady law in 1993. That measure had first been introduced in 1987. Moreover, the Brady law was the first major federal gun control legislation since 1968. On the other hand, some empirical studies have found substantial congruence 24 Polling and the Public between the attitudes of the public and the actions of government on certain issues (Erikson 1976; Page and Shapiro 1983, 1992). Although these studies are careful not to hastily attribute government decisions to popular preferences, they suggest conditions under which citizen influence is likely to be significant. Another benefit of polls, according to polling proponents, is the opportunity for citizens to learn about their compatriots and to dispel myths and stereotypes that might otherwise mislead public discourse. For example, Morin reported poll results that challenged the stereotype of evangelical and fundamentalist Christians as monolithic, homogeneous supporters of the religious right (Morin 1993a, 1993c). A USA Today/CNN/Gallup poll conducted in December 1993 showed that stereotypes of the opponents in the gun control debate were misleading. For example, the attitudes of gun own- j ers on various aspects of gun control did not differ substantially from the attitudes of nonowners, particularly concerning less-sweeping forms of gun I regulation. Other public opinion polls have provided insights, sometimes surprising ones, on race and prejudice in the United States. A Louis Harris poll conducted in 1993 showed that the traditional victims of bigotry—blacks, j Asians, and Latinos—often expressed intolerant views of other minority ■ groups. And a study by Sniderman et al. (1993) debunked the simplistic no- I tion that conservatives were prejudiced toward blacks and liberals were not. j Although on many items conservatives were less tolerant than liberals toward blacks, the key point for Sniderman and his colleagues was that the differences were often very small. Polls conducted on specific aspects of the situation in Iraq have yielded surprising results that raise questions about how informed citizens are and ■ how effectively presidential administrations and others can manipulate pub- ! lie opinion. For example, a Gallup poll conducted in October 2004 found a that 42 percent of Americans believed that Saddam Hussein was person- ij ally involved in the September 11, 2001, terrorist attacks, when there was | no evidence of any such involvement. A Harris poll conducted in July 2006 showed that 64 percent of the respondents believed that Saddam Hussein !; had strong links to Al Qaeda, and 50 percent thought that Iraq had weapons of mass destruction when the United States went to war with Iraq. | One has to wonder about the sources of such opinions, a topic we return I to in chapter 9. j] Even more surprising were some of the opinions uncovered in a na- I tional Scripps Howard/Ohio University poll conducted in July 2006. That . | poll found that more than a third of Americans believed that federal offi- | cials either assisted in the September 11 terrorist attacks or took no steps | to prevent them, so that the United States could justify going to war in the f Polling and the Public 25 DOONESBURY © (2002) G. B. Trudeau. Reprinted with permission of UNIVERSAL PRESS SYNDICATE. All rights reserved. 26 Polling and the Public Polling and the Public 27 Middle East (Hargrove and Stempel 2006). Sixteen percent speculated that secretly planted explosives, and not airplanes, were the real causes of the World Trade Center towers' collapse. And 12 percent suspected that the ' Pentagon was struck by a military cruise missile on September 11, rather < than by a plane. One partial explanation for these results was based on where people got their information: Citizens who regularly used the Internet, but did not rely on "mainstream" media, were much more likely to be- i lieve in September 11 conspiracy theories, whereas people who read daily > ■ newspapers or listened to radio newscasts were much less likely to do so j (Hargrove and Stempel 2006). As information sources become more nu- j. merous, diverse, and specialized, and as citizens themselves become much more selective in their sources of information, it will not be surprising if public opinion polls increasingly uncover surprising pockets of opinion on V particular topics. [: In contrast to the favorable arguments of proponents, many critics of p polling worry about the harmful consequences of polls for a democratic po- j; litical system. They agree that citizen influence is a key component of a de- |J mocracy and that public opinion, properly measured, can be useful in | governing. But they argue that polls give a misleading impression of how a democracy actually operates. Public opinion is not synonymous with the j results of public opinion polls, yet today the two are treated as though they i were identical. A focus solely on poll results ignores the dynamics of how j opinions are formed and how they change and often overlooks factors that jj may shape (and manipulate) public opinion, such as the behavior of lead- | ers and interest groups. Polls may present an overall picture of the distri- | bution of opinion, but the reporting and use of polls often ignore important | differences in preferences among subgroups. The result is a misleading pic- \ ture of similar attitudes across different segments of the American popula- \ tion. Margolis (1984) claims that polls may not be the optimal way to j measure public opinion on politically and socially sensitive topics. He ar- 4 gues that in some instances actual behavior provides a more valid expres- | sion of public opinion than verbal responses to survey questions. | A more radical criticism of polls is that they are simply a sop to the cit- I izenry, that they give people a false sense of being influential, when in re- j ality political power is held and exercised by elites who may or may not act \ in the pubtic interest. Social scientist Johan Galtung (1969) makes the point most effectively when he argues that surveys are too democratic: they generally count all respondents equally, whereas people are tremendously disparate in the resources and skills they bring to bear on political decisions. To the extent that a survey is seen as a quasi referendum on an issue, it is misleading because the participants in the referendum have different degrees of opportunity to shape government outcomes. Weissberg (2001) argues that polls should not shape public policies and that policymakers should ignore the polls. He provides an elaborate rationale for this position but essentially argues that poll respondents do not and cannot incorporate all the factors that policymakers have to consider when adopting specific policies. Weissberg claims that although polls measure what citizens want, those wishes do not in most instances take into account the costs or risks of the stated preference. That is, it is easy for respondents in a poll to say that they prefer more spending on education or health care, but such responses do not reflect the costs or the potential problems associated with that increased spending. Only the policymaker has that information, and it is the policymaker who must make the ultimate decision based on much more than citizen preferences. A related criticism of polls concerns their consequences for leadership in the United States. The simplistic version of this argument says that officeholders blindly follow the polls rather than work to educate and persuade the public. That argument is exemplified by an editorial in the Akron Beacon Journal (May 8, 1994) entitled "Foreign Poll-icy." It begins by asserting, "The name of Stanley Greenberg may not be familiar to most Americans. It should be. He is Bill Clinton's pollster, and for all intents and purposes, he conducts the country's foreign policy." Other observers, such as Barnes (1993), have also described the critical role Greenberg played in the Clinton administration, just as he did in the presidential campaign. Polling results in 1998 and 1999 on health care, patients' rights, and prescription drugs showed that Americans cared about those issues, and the findings prompted the GOP congressional leadership to offer its own health care proposals to head off any political backlash from its opposition to the Clinton administration's proposals (Alvarez 1998). Other critics of polling argue that the widespread awareness of public preferences that polls generate has limited the ability of leaders to make unpopular choices. Still others complain that leaders can easily manipulate the polls—for example, by giving a major, televised address that can influence opinions in polls taken right afterward, thereby generating poll results that paint an inaccurate picture of the extent of support for their policies. Some make the fundamental point that polls (and the news media) have altered the style and substance of governance by emphasizing the consequences of politicians' actions for the next election. The result is a shortsighted approach to problem solving. Compare this with poll proponents' defense of the impact of polls on leadership, which argues that officials should have information about citizens' attitudes before they make major decisions and that the polls, whatever their limitations, are the best way to acquire that information. 28 Polling and the Public The cartoon on page 25 is a jab at the Bush administration and the president himself for his reliance on public opinion polling. In the 2000 presidential election, candidate George W. Bush criticized the Clinton/ Gore administration for slavishly following the polls and claimed that in a Bush administration decisions would be made on the bases of principles and values and not because of polls. In hindsight, there are striking similarities between the Clinton and Bush administrations, and one marked difference, when it comes to polling and political research in general (Harris 2001a, 2001b). The similarities are simply that both administrations used polls, focus groups, and the like extensively to ascertain what Americans were thinking, how citizens would respond to various policy alternatives, and what were the best ways to package those alternatives. The major difference between the two was that Clinton himself was actively involved in reviewing the poll results in weekly meetings with key members of his administration (Harris 2001a), whereas Bush relied more on recommendations from his chief political operative, Karl Rove. Harris discussed how polling results affected Clinton administration policies. For example, a needle exchange program for drug addicts, to combat the spread of AIDS, was dropped when polls indicated substantial public opposition. In another example, when the Clinton administration enjoyed a federal budget surplus, it did not want the Republican Congress to spend that surplus on a tax cut. Public opinion polling helped the Clinton administration learn that the public overwhelmingly preferred devoting the surplus to Social Security rather than to tax cuts. Thus the president challenged Congress to "save Social Security first," a theme that put the Republicans on the defensive on tax cuts until George W. Bush became president. These two examples raise important points. Surely the Clinton administration was using polls for political purposes, to build support (on how to use the surplus) and avoid voter anger and opposition (on the needle exchange proposal). Yet it was also the case that the polls were telling the administration what popular preferences were, and the administration then followed those preferences. Obviously the question for leadership is: To what extent does one follow the polls and public opinion, and to what extent does one lead and try to change citizens' opinions? Any presidential administration does both. Harris points out that the Clinton administration took major actions on trade policy, military intervention in the Balkans, and other matters that went against the grain of American public opinion at the time. Murray and Howard (2002) found that the Reagan and Clinton administrations polled extensively from their first days in office, whereas the Carter and first Bush administrations polled only sporadically in their first three years but became very active in polling in their fourth year. They conclude that the public opinion polling operation has become institutionalized Polling and the Public 29 i. in the White House but that there are differences in how each administra- tion uses polls and political research in general. The evidence from the George W. Bush presidency shows that polling, focus group, and political research activities are as central as they were during the Clinton administration despite claims to the contrary (Harris | ■ 2001b; Tenpas 2003). For example, Washington Post reporters Peter I Baker and Dan Balz (2005) analyzed how the tone of President Bush's ; prime-time address on Iraq was very much influenced by public opinion re- search. The Bush administration brought Christopher Gelpi, an academic expert on public opinion during wartime, onto the National Security Council staff as a special adviser for strategic planning. Gelpi and his colleague, Peter Feaver, had conducted research on the determinants of popular support for military missions. They found that one of the key factors in maintaining public support for military action was a perception that the war could and would be won. The costs of the war and the circumstances of how it began were less relevant in influencing public support than was be-| lief that the war would have a positive conclusion. That finding influenced | the president's speech and his subsequent messages on the war in Iraq: £ progress was being made and the war would be won. | What, then, is the verdict on opinion polls in the United States? They I are now an integral part of the political and social landscape, and they are ! likely to become more prominent in the future, even as the technology of \ polling changes. Polls can provide citizens and leaders with useful infor- f mation; they also can be highly misleading and inaccurate. Polls may en-l hance the opportunities for citizen influence; they also can -serve to i manipulate the public. In 1965 George Gallup wrote optimistically about j the future of the polls: \ As students, scholars, and the general public gain a better understanding f of polls, they will have a greater appreciation of the service polls can per- I form in a democracy. In my opinion, modern polls are the chief hope of >' lifting government to a higher level, by showing that the public supports j the reforms that will make this possible, by providing a modus operandi \ for testing new ideas. . . . Polls can help make government more efficient \ and responsive; they can improve the quality of candidates for public of- I fice; they can make this a truer democracy. (Gallup 1965-1966, 549) i \ More than four decades have passed since Gallup made those claims, j and discourse about the polls has become much more critical. Neverthe-| less, as citizens become wiser consumers of polls, Gallup's lofty aspirations are more likely to be realized. The chapters that follow consider polls in detail and raise some method-| ological points, often in the context of important substantive examples. 30 Polling and the Public Polling and the Public 31 Chapter 2 addresses the problem of "nonattitudes"—they are what polltak-ers measure when citizens do not have genuine opinions on a topic and yet answer the poll questions anyway. Despite pollsters' best efforts, citizens often respond to questions on which they have no real opinions, so that the poll yields misleading results. Chapter 3 discusses the wording of questions and their order and con- j text. Examples of poorly worded questions likely to produce skewed results 1 have already been cited, but the wording of a question is not the only important consideration. A survey is, after all, a series of questions, and their order and context can greatly affect the results. Chapter 4 focuses on various sampling techniques and their advan- {■ tages and disadvantages. It also deals with sample size and error. Chapter ' 5 explains in detail how different interviewing procedures can affect results. ij Chapter 6 examines how the news media report the polls, and chap- ] ter 7 analyzes the role of polls in elections. Because Americans learn about polls primarily through the mass media, how they cover polls greatly influ- \ ences public opinion. This influence is particularly interesting in the case of j high-visibility national polls because the organization that reports the polls \ also is responsible for conducting them. Chapter 7 argues that the polls t have come to play an intrusive role in elections and that their use by can- | didates and their reporting by the media often ill serve citizens and the elec- | toral process. Elections are the most visible opportunity for citizens to i influence their government, and to the extent that polls affect elections, cit- \ izens should be sensitive and wary. | Chapter 8 explains that the analysis of poll results is more an art than \ a science—and one that affords many opportunities for manipulative in- \ terpretation and dissemination of poll results to sway public opinion. Chap- j. ter 9 ties together the various themes, offers suggestions about better ways } to use polls, and discusses the effects of polls on the American polity. 'j Exercises 1. Keep a one-week log of the public opinion poll results that appear in your local daily newspaper. Note the topic of the poll and the source. Also note how much information was provided about how the poll was conducted. What conclusions can you make based on these findings? Do you think adequate information was provided about how the poll was conducted? 2. Pick one of the national newsmagazines (Time, Newsweek, or U.S. | Netus & World Report) and examine how it used and reported polls 1 during one calendar month. Keep a record of the topics of the polls and how much information was provided about how they were conducted. Also note how the poll results were used. Were they incorporated into and discussed in the story, or did they appear-as a sidebar? 3. For one week, go to AOL and to the CNN Web site each day and record the daily poll questions that they ask their users. Also record the results of the poll questions. Then try to find genuine public opinion polls that have asked similar questions around the same time. Compare the results of the Internet and the scientific polls on similar topics. How similar or dissimilar are the results of the two? How would you explain this? 4. A number of survey items that the author claimed were biased are shown on page 11. Do you agree that these are biased questions? If so, what do you think makes them biased? Be specific in your critique of the items.