Opinion Polling

Overview

The stipulation that American democracy requires the consent of the governed was established in the Declaration of Independence, and some scholars believe that opinion polls are one of the best methods for determining that consent. Opinion polling is considered free speech and is protected by the First Amendment to the United States Constitution. When conducting opinion polls, pollsters use samples (1,000 to 1,500 people on the average) of the larger population. If the sample is not balanced the results will be skewed. For example, taking a poll on the website of Mother Joneswould not result in an accurate depiction of American popular opinion because most visitors to the website are politically liberal. Likewise, polls taken by Fox News on their website would skew toward conservatism. To avoid skewed results, pollsters may engage in rebalancing or weighting, that is, considering the significance of variables such as gender, race, age, income, level of education, likelihood of voting, cell phone ownership, and Internet access. Otherwise, the sample might not be reflective of a population. Reliability of polls are determined through a statistical examination of sampling error, which refers to deviations from the general population.

Critics of opinion polling suggest that pollsters may resort to herding. That is, if poll results differ significantly from other polls, a pollster may "follow the herd." Pollsters may also engage in poll aggregation because larger samples tend to produce lower sampling errors. Almost from the beginning, scholars have suggested that a major failing of opinion polling is that popular opinion is not always compatible with good policymaking. Furthermore, minorities may be underrepresented in polls. Major changes in technology have affected poll samples, particularly telephone polling. In 1997, the average response to a telephone poll was 37 percent. Half of all American households subsequently eliminated landlines, and the response rate dropped to only 9 percent in 2016. Caller ID and answering machines enable people to screen their calls, and the prevalence of unwanted telemarketing and "robo" calls have made people less likely than in the past to answer calls from unknown sources.

Pollsters conduct opinion polls to predict the outcome of elections, to inform the public about issues, to provide information to policymakers, and to make their work more interesting. Opinion polling may be either predictive, suggesting what is likely to happen as with election polls, or prescriptive, informing people about how others think or suggesting how individuals should think about an issue. Opinion polling is valid only when opinions are stable, when results can be replicated by others, when large numbers of individuals share opinions, and in the absence of apathy.

Two of the most common criticisms of opinion polling is that polling fails to adequately determine the strength of opinions or differentiate between rational and irrational viewpoints (Ellwanger, 2017). While polling has become increasingly more sophisticated, inaccurate polling continues to undermine trust in its reliability. Some pollsters have been accused of bias because of limits on available answers, composition of samples, statistical miscalculations, slanted discussions of results, and the wording of questions.

Presidential approval polls are consistently of interest, to pundits and politicians alike. They provide an ongoing score card on how a president is perceived by the public. Shortly after Donald Trump took office in January 2017, a poll taken by pollsters at Quinnipiac University found that his approval rating was only 38 percent as compared to a 55 percent disapproval rate. Over the next year, Trump's approval rates vacillated. On March 27, 2018, amid rapid White House changeovers, Robert Mueller's Russia investigation, and sex scandals involving a former porn star, a former Playboy bunny, and a former contestant on The Apprentice, Gallup reported that Trump's approval and disapproval rate were again at 38 and 55 percent, respectively. Some polls suggested that his approval rate had climbed into the low 40s. When approval ratings are divided according to party, pollsters find that a president's base is more likely than others to approve of his job performance. Trump's approval rating among Republicans in September 2017, for example, was 81 percent. Public opinion polls may also indicate how the public feels about a particular presidential action. Despite Trump's claim that only his enemies wanted him to stop tweeting, a Quinnipiac poll reported that 70 percent of Americans believed the president should abandon Twitter (since rebranded X) as a means of making public statements, as his impulsive tweets frequently led to embarrassment or outrage.

Opinion polling in the United States began with straw polls in 1824. In the early years of the twentieth century, advertisers were the first to see the advantages in opinion polling, using them to identify what customers wanted and to learn how to design convincing advertisements. The first book on polling to receive national attention was Lawrence Lowell's Public Opinion and Popular Governance (1913). In 1920, the Literary Digest conducted the first political poll, asking its readers to express their opinion on an election that pitted Republican Warren G. Harding against Democrat James M. Cox and Socialist Eugene Debs. Results were decidedly skewed since only wealthier and more educated Americans were likely to read Literary Digest. The first scholarly works to address opinion polling were Walter Lippman's Public Opinion (1922) and John Dewey's The Public and Its Problems (1927).

By 1936 when President Franklin Roosevelt faced a challenge from Republican Alf Landon, political polls were common. For the first time, polls were conducted using statistical methods. Most pollsters predicted a defeat for Roosevelt, but newcomer George Gallup accurately predicted that Roosevelt would win a second term. By 1940, Gallup and his American Institute of Public Opinion were more trusted than any other pollster or polling organization. Gallup's goal was to use opinion polling to promote democracy by providing information that would result in better policymaking, and he and Saul Forbes Rae published The Pulse of Democracy (1940).

Early pollsters worked for politicians, or they were entrepreneurs, market researchers, academics, or government hires. The early relationship between entrepreneurs and pollsters grew out of such factors as shared goals and the reaping of mutual benefits. Following World War II, polling organizations broke free and began developing their own polling methods and building up individual reputations. Until 1940, opinion polling was dominated by pollsters such as George Gallup, Elmo Roper, and Hadley Cantril. Newspaper publishers and journalists accepted the notion that polling was newsworthy, but they also saw a danger in promoting pack journalism. As opinion polling became more common, polls began to take on a life of their own, overtaking stories rather than supporting them. Among print media, rules were established by the Associated Press Stylebook and Libel Manual,which mandated clarity and accuracy and allowed for no deviations from rules. Princeton University began publishing The Public Opinion Quarterlyin 1947. The National Council on Public Polls were founded in 1969 to monitor polling organizations.

Further Insights

Noted University of Chicago historian Daniel Boorstein was the first scholar to identify the role of pseudo-events in American politics (The Image: A Guide to Pseudo-Events in America, 1992), contending that the media had begun to create the news rather than reporting on events as they occurred. Since the technical revolution and the widespread availability of Internet access, some scholars have suggested that online polls are constantly orchestrated as pseudo-events, with journalists gathering information from various media sources to reissue them as major news stories. Michael L. Kent, Tyler K. Harrison, and Maureen Taylor (2006) argue that pseudo-events are presented more as entertainment than as scientific opinion polling.

Since 24-hour news sites such as CNN and CNBC are forced to dig for a constant flow of news, critics suggest that polls found on those sites may be unscientific, lacking in real news value, and may be skewed toward opinions of online news junkies or partisan enthusiasts. Online polls and those taken by readers of magazines that appeal to a limited audience are never completely random, and randomness, which requires that the sample be representative of the general population, is a key factor is the reliability of polling. Furthermore, online polls have no influence on the outcome of an election or the handling of an issue. Social media users have become notorious for encouraging their friends and followers, who are likely to share their opinions, to vote in online polls, further reducing randomness and skewing results. During the 2004 presidential election, for example, Free Republic, a conservative online presence, launched an attempt to sabotage Democrat John Kerry and demoralize Democratic voters participating in CNN's Quick Vote (Kent, Harrison & Taylor, 2006).

Robert Northcutt (2015), an expert in the field of predictive science, suggests that predictions may be unsuccessful at times because of the open manner in which social systems operate, the unlikelihood that all possible variables will be identified, and the possibility that reflexibility will occur. Reflexivity is a response to outside factors such as the presence of the pollster or the publication of a poll. Reasons frequently cited for the failure of pollsters to predict elections accurately include the presence of a large number of undecided voters, the popularity of a third party candidate, close elections, and a large number of voters deciding not to vote. In some elections, voters are undecided because they see no difference in the candidates or they do not like either candidate. Highly partisan voters use party as a cue when they do not have a clear preference for either candidate. In any election with a popular third-party candidate, that candidate is likely to drain votes away from the party with which he/she is more ideologically aligned. Some elections are so close that pollsters may disagree on the outcome, and an extraneous factor such as inclement weather may affect voter turnout.

The classic example of pollsters getting a presidential prediction wrong is the 1948 election in which President Harry Truman, who had succeeded to office on the death of Franklin Roosevelt in April 1945, was challenged by Republican Thomas Dewey. The polls and newspapers were so convinced of a Dewey victory that the Chicago Daily Tribuneprepared a front-page banner proclaiming "Dewey Defeats Truman" beforehand. After votes were cast, newspapers featured a smiling Truman holding up the Tribunearticle.

Another example of inaccurate polling occurred during the 2016 presidential election. Hillary Clinton was expected to win, with odds ranging from 70 to 99 percent. However, only 58 percent of eligible voters cast a vote. Democrats proved more likely than Republicans to choose not to vote. In hindsight, it was suggested that, among other things, the media's prediction of an easy win for Clinton caused a large number of voters to believe they did not need to bother to turn out. The most popular explanations for the inaccuracy, according to pollsters, were that narrow voter margins in key states such as Wisconsin, Michigan, and Pennsylvania mislead pollsters, that many undecided voters determined at the last minute to vote for Trump, and that it had been impossible to predict the educational divide exhibited by voters. Pew Research Center noted in 2022 that polling errors for some topics, for example political engagement and personal wealth, are consistently larger. The organization suggested that people are less willing to be surveyed on these matters.

Discourse

Criticism of opinion polling and pollsters has always been common. Political scientist Amy Fried (2013) reports that a popular parody of pollsters in the late 1940s was based on the Yale a cappella singers' Whiffenpoof song: "Gentlemen pollsters up in a tree, / Damned from here to eternity. / God please save us from Tom Dewey… / Baa, baa, baa."

Humphrey Taylor, one of the foremost pollsters of the twentieth century and the head of Louis Harris and Associates, paraphrased the well-known Winston Churchill quote about democracy, contending that "Polls are the worst way of measuring public opinion and public behavior, or predicting elections—except for all the others."

Opinion polling provides a useful tool for politicians, who depend on polls to inform them about possible reactions to running for office, running for reelection, taking particular positions on issues, determining policies, and measuring levels of support for policies and programs. Presidents usually employ their own pollsters to keep them abreast of such issues. In the 1940s, Franklin Roosevelt was the first president to hire his own pollsters, Emil Hurja from the Democratic National Convention and Hadley Cantril of Princeton University. Dwight Eisenhower used Louis Harris, Jimmy Carter employed Patrick Caddell, Ronald Reagan relied on Richard Wirthlin, Bill Clinton depended on Dick Morris, and Barack Obama used Joel Benenson. After the Supreme Court overturned campaign spending restrictions in Citizens United v. Federal Election Commission(2010), partisan political action committees began conducting their own polls.

Professional pollster Kenneth F. Warren suggests that the harshest and loudest critics of opinion polling are conservatives such as columnists William Saffire and journalist Daniel Greenberg. The latter has equated polling with voodoo. In 1964 Barry Goldwater, who was considered an arch-conservative at the time, lost the presidency to President Lyndon Johnson, who had succeeded to office after the death of John F. Kennedy the previous November. Republicans in Congress unsuccessfully attempted to steer an anti-polling bill through Congress. In 1972, after George McGovern lost in a humiliating landslide to President Richard Nixon, Democrats expressed support for anti-polling legislation.

In 1980, three hours after polls on the east coast closed, newscasters, using evidence from exit polls, began declaring Ronald Reagan the victor. Consequently, large numbers of West Coast voters did not bother to vote. That debacle led to changes in exit polling and the calling of elections. Some states passed laws banning exit polling within 300 feet of voting places, but such laws were later overturned by the Supreme Court. In most cases, criticism of opinion polling is most common among those who disagree with the results of particular polls.

During presidential elections, state polls may be more accurate than national polls. Such was the case in 2012 when state polls showed President Obama with a clear lead over Republican Mitt Romney. In national polls, Obama was trailing Romney. When the voters spoke, Obama won the popular vote 4,237,756 to Romney's 4,163,447, and Obama carried the electoral vote 50 to 49 percent. In that election, some online pollsters successfully predicted the votes of most states. Accurate polls were conducted by political scientists at Votomatic and Huffington Post, by neuroscientists at Princeton University, and by non-academics at Five Thirty Eight/New York Times. The neuroscientists and the non-academics correctly predicted the winner in all fifty states.

However, during the 2020 presidential election both national and state polls were inaccurate, leading some commentators to conclude that polling is broken and can no longer be relied upon to accurately predict election outcomes. National polls during this presidential election overstated Joe Biden's lead over Donald Trump. Some state polls showed the race being tight when it was not.

Bibliography

Ellwanger, A. (2017). Reinventing doxa: Public opinion polling as deliberative discourse. Argumentation and Advocacy, 53(3), 181–198. Retrieved May 23, 2018 from EBSCO Online Database Academic Source Ultimate. http://search.ebscohost.com/login.aspx?direct=true&db=asn&AN=123914003&site=ehost-live

Fried, A. (2013). Pathways to polling: Crisis, cooperation, and the making of public opinion professions.New York: Routledge.

Goidel, R.K. (2011). Political polling in the digital age: The challenge of measuring and understanding public opinion.Baton Rouge: Louisiana State University Press.

Keeter, S. et. al. (2021, Mar. 2). What 2020's election poll errors tell us about the accuracy of issue polling. Pew Resarch Center. www.pewresearch.org/methods/2021/03/02/what-2020s-election-poll-errors-tell-us-about-the-accuracy-of-issue-polling/

Kennedy, C., Mercer, A., Hatley, N., & Lau, A. (2022, September 21). Does public opinion polling about issues still work? Pew Research Center. www.pewresearch.org/short-reads/2022/09/21/does-public-opinion-polling-about-issues-still-work/

Kent, M. L., Harrison, T. R., & Taylor, M. (2006). A critique of Internet polls as symbolic representation and pseudo-events. Communication Studies, 57(3), 299–315.

McCutcheon, C. (2015). Political polling: Do polls accurately measure public attitudes? CQ Researcher, 25(6), 121–144. Retrieved May 23, 2018 from EBSCO Online Database Academic Source Ultimate. http://search.ebscohost.com/login.aspx?direct=true&db=asn&AN=119802864&site=ehost-live

Northcott, R. (2015). Opinion polling and election predictions. Philosophy of Science, 82(5), 1260–1271. Retrieved May 23, 2018 from EBSCO Online Database Academic Source Ultimate. http://search.ebscohost.com/login.aspx?direct=true&db=asn&AN=111483425&site=ehost-live

Perrin, A. J. (2014). American democracy from Tocqueville to town halls to Twitter.Hoboken, NJ: John Wiley and Sons.

Shapiro, R. Y., & Jacobs, L. R. (2011). The Oxford Handbook of American public opinion and the media.New York: Oxford University Press.

Warren, Kenneth F. (2018). In defense of public opinion polling.New York: Routledge.