Has Polling Killed Democracy?: Polling Through History
An Original Timeline of Notable Episodes in the History of Political Polling.
1820-1829: Public opinion polls begin to play a role in presidential elections.
In 1824, there was no clear favorite to win the presidency and supporters of Andrew Jackson, John Quincy Adams, and the other contenders used polls to certify and increase the popular appeal of their preferred candidate. Non-scientific "straw polls" were taken at political gatherings and other public events and published in the party-dominated newspapers of the day. In one poll that appeared in the The Harrisburg Pennsylvanian on July 24, 1864, Andrew Jackson received 335 votes and John Quincy Adams received 169. In the election, Jackson won the popular and electoral vote but did not capture a majority in the Electoral College. The House of Representatives eventually chose runner-up Adams to be president. Sources: Tom W. Smith, "The First Straw?: A Study of the Origins of Election Polls," Public Opinion Quarterly (Spring, 1990), 21-36.
1850-1860: Straw polling becomes a prominent feature of political reporting and campaigning.
Reporters and political partisans regularly conducted straw polls at political rallies and other public events during mid-century political campaigns. Polls were seen as a good way of gauging popular sentiment, providing intelligence for partisan maneuvering, and as news stories in their own right. The Chicago Tribune, for example, routinely printed straw poll results in a column called "Movement of the Peoples." On September 11, 1860, the column related the results of a straw poll conducted in Unionville, Pennsylvania showing overwhelming support for Republican presidential candidate Abraham Lincoln over challengers Stephen A. Douglas, John Breckinridge, and John Bell. According to the report: "A thorough canvass of the town has been made with the following results: Lincoln 240 [votes]; Breckenridge 4; Douglas 1; Bell 1." Newspapers often published polls that clearly favored their chosen candidate. The Tribune supported Lincoln. Sources: Susan Herbst, Numbered Voices: How Opinion Polling Has Shaped American Politics (Chicago, 1996), 74-79; Chicago Tribune, 9/11/1860.
1916: The Literary Digest, a leading popular magazine of the day, launches the most famous straw polls in American history.
In 1916, the magazine mailed ballots to subscribers in five states asking them their preference for president. By 1932, Literary Digest was mailing over 20 million ballots to citizens in every state. The magazine used telephone directories, automobile registration lists, and magazine circulation lists to determine who received ballots. Although this method of selecting a sample was biased in favor of people from middle and upper socioeconomic classes, the Digest poll correctly predicted the outcome of each presidential election between 1916 and 1932. The 1932 poll was especially accurate. The final Digest poll predicted Franklin Roosevelt would win 59.85 percent of the popular vote, 41 states, and 472 electoral votes. Roosevelt actually won 59.14 percent of the popular vote, 42 states, and 472 electoral votes. Source: David W. Moore, The Superpollsters: How They Measure and Manipulate Public Opinion in America (New York, 1992), 37-42.
1932: Emil Hurja becomes the first political consultant to analyze polls for a presidential campaign.
Hurja, a statistical analyst, used sampling methods and other statistical techniques to analyze published polls and other voter data for the Democratic National Committee (DNC) and Democratic presidential candidate Franklin Roosevelt. Hurja counseled the campaign on which states to devote resources and on what themes to emphasize. He also wrote press releases for the DNC predicting a Roosevelt landslide. In the election, Roosevelt won 57% of the vote and all but six states. After the election, Hurja joined the Roosevelt administration and used his statistical skills to help Democratic National Committee chairman James Farley decide who should receive patronage jobs. Hurja also provided poll analysis for the Democrats in 1934 and 1936 elections. Sources: Melvin Holli, The Wizard of Washington: Emil Hurja, Franklin Roosevelt, and the Birth of Modern Polling (New York, 2002); Robert M. Eisigner, The Evolution of Presidential Polling (Cambridge, 2003), 82-83.
1935: George Gallup founds the American Institute of Public Opinion and starts a weekly newspaper column America Speaks to publicize the results of his national polls.
Gallup offered newspapers a money-back guarantee on their subscriptions if his poll did not do a better job than the Literary Digest of predicting the 1936 presidential election. The Gallup column initially appeared weekly in 60 newspapers but by 1940 it ran several times a week in over one hundred papers across the country. The first Gallup column, which examined the public's views on government spending, appeared in newspapers on October 20, 1935. The poll showed that 6 of 10 Americans believed the government was spending too much to combat the Depression. The second Gallup column showed President Roosevelt's popularity had declined over his previous two years in office. The presidential popularity poll would go on to become a regular feature of the Gallup column and the most influential political poll in America. Sources: Sarah Igo, The Averaged American: Surveys, Citizens, and the Making of the Mass Public (Cambridge, Mass.), 2007, 116-118; Moore, Superpollsters, 31-32.
1936: Gallup correctly predicts Franklin Roosevelt's victory over Alf Landon while The Literary Digest wrongly forecasts a Landon victory.
Gallup predicted Roosevelt would win the election with 54% of the vote. The Literary Digest forecasted Republican Alf Landon would win 57% of the vote, with Roosevelt tallying only 42%. Roosevelt won 61% of the vote and the presidency. Before the 1936 election, most people equated survey accuracy with large sample size. But unlike the Literary Digest, Gallup employed scientific sampling methods and based his prediction on a much smaller number of respondents. In addition, Gallup supplemented mail-in ballots with in-person interviews. Opinion researchers Elmo Roper of Fortune magazine and Archibald Crossley of the Hearst newspapers also predicted an FDR victory. But Gallup's bold prediction made him a household name. Until his death in 1984, Gallup tirelessly promoted the methodological and democratic virtues of polling. Gallup believed polling would establish a new age of "scientific democracy" in which public opinion would continually guide the policy choices of elected officials. Sources: Moore, Superpollsters, 47-55; J. Michael Hogan, "George Gallup and the Rhetoric of Scientific Democracy," Communication Monographs (June, 1997), 161-179.
1948: Gallup, Roper, and Crossley and other pollsters wrongly predict presidential challenger Thomas Dewey will defeat incumbent President Harry S. Truman.
After the election, pollsters and academic researchers undertook an intensive effort to determine what went wrong. Three main reasons were identified. Pollsters stopped surveying too early and therefore missed last minute shifts in opinion. Pollsters did not adequately account for uncommitted voters and non-voters. There were errors in sampling and interviewing. In the 1950s, opinion researchers focused on improving their techniques and making polling more reliable. To make pre-election polls more accurate, pollsters improved their sample designs, began to screen out unlikely and uncommitted voters, and conducted surveys later in campaigns. The average margin of error for Gallup pre-election polls dropped from 3.7 percentage points from 1936 and 1950 to 1.4 percentage points from 1954 through 1968. Sources: Moore, Superpollsters, 68-72; Harold Mendelsohn and Irving Crespi, Polls, Television, and the New Politics (Scranton, PA., 1970), 73.
1960: John F. Kennedy becomes the first presidential candidate to hire a pollster to conduct private surveys for his campaign.
Lou Harris, a veteran political pollster, conducted and interpreted dozens of national and state polls for Kennedy from the pre-primary period through the general election. During the Democratic primaries, JFK relied on Harris' polls of voter's religious attitudes to develop a state by state strategy for dealing with the issue of his Catholicism. Harris' polls helped to convince Kennedy to confront the issue in West Virginia in a series of speeches and TV ads. Although Catholics only accounted for 5% of the population West Virginia, JFK won the primary and went on to win the nomination and the presidency. Harris did some private polling for President Kennedy, but, in 1963, Harris quit the campaign polling business to concentrate on public policy polling. He started a weekly newspaper column, The Harris Poll, and became a leading public pollster. Sources: Eisinger, The Evolution of Presidential Polling, 86-88; Theodore White, The Making of the President 1960 (New York, 1961), 100, 106; Moore, Superpollsters, 89.
1967: CBS becomes the first news organization to establish its own polling unit.
CBS hired Warren Mitofsky, a statistician at the U.S. Census Bureau, to design a procedure for accurately forecasting voting results on election night. Over the next few years, Mitofsky developed a new method, known as the exit poll, for surveying voters who had just cast their ballots. In 1972, Mitofsky conducted the first national exit poll for CBS. Exit polls are widely used to gather information about voter demographics and voting behavior. They have proven to be an especially valuable tool for political journalists and analysts. Veteran Washington Post reporter and columnist David Broder has called exit polls "the most useful analytic tool developed in my working life." Sources: Moore, Superpollsters, 251; David S. Broder, Behind the Front Page: A Candid Look at How the News Is Made (New York, 1987), 253.
1968: Polls and controversy over poll results occupy a central place in the race for the Republican presidential nomination.
All the major Republican presidential contenders developed their overall electoral strategies based upon polls. George Romney started campaigning early because polls showed him ahead and quit the race before the first primary when polls showed him way behind. Nelson Rockefeller did not enter Republican primaries because polls showed he could not win them. Rockefeller gambled he could take the nomination on the strength of his Gallup Poll numbers alone. Nixon ran in the primaries because polls showed him so far ahead of the other Republican challengers among the party's rank-n-file. The Republican "Battle of the Polls" came to a head on the eve of the Republican National Convention when the Gallup and Harris reported conflicting results, with Harris showing Rockefeller doing better against likely Democratic presidential candidates and Gallup showing Nixon ahead. Political partisans on both sides worked overtime distributing the poll favoring their man and explaining away the unfavorable one. Gallup and Harris issued a joint press statement trying to clarify the apparent discrepancy. In the end, Nixon won the nomination and in the words of one pollster: "the polling profession in this country has got a real black eye." Sources: Mendelsohn and Crespi, Polls, Television, and the New Politics, 59; David Broder, "Poll Jolts Stop-Nixon Drive," The Washington Post, 7/30/1968
1969: The National Council on Public Polls (NCPP), an association of professional polling organizations, is created to establish standards for conducting and reporting political polls.
The increasing political manipulation of polls prompted George Gallup, Archibald Crossley, and other leading pollsters of the day to found the NCPP to protect the public and the reputation of the polling profession. For example, in 1967, President Lyndon Johnson leaked biased popularity polls to unsuspecting reporters to counter his declining Gallup numbers. This episode helped to convince pollsters and the media of the need to improve the media's understanding of polls. In addition to adopting guidelines for the public disclosure of polls, the NCPP, which is still in existence, has sought to educate the media and the public about how polls are conducted and how to interpret polls results. Over the years, media organizations have adopted poll-reporting standards and reporters have developed a better understanding of polling. But critics still maintain that the news media can do a much better job reporting polls and assisting citizens to become more informed poll consumers. Sources: Herbert Asher, Polling and the Public: What Every Citizen Should Know (Wash., D.C., 1988), 78-83; NCPP website: http://www.ncpp.org/
1975: CBS and The New York Times join forces to become the first media outlets to establish a permanent polling operation.
The first CBS/ Times poll asked Americans whether they would support the use of federal funds to help New York City out of a fiscal crisis. The CBS/Times poll found 55% of Americans around the country favored federal government assistance for New York. This finding challenged an earlier Gallup Poll that showed that 49% of Americans opposed federal assistance to New York. President Gerald Ford initially refused to support a "federal bail-out of New York City," but he eventually extended the city billions of dollars in short term loans after it adopted budget-cutting measures. The success of the CBS/Times poll led to an explosion of media polling in the 1970s and 1980s. By 1990, almost half of all daily newspapers, almost all large circulation papers, and more than half of all local TV stations reported on their own polls. Not everyone has welcomed the growth of media polling. Some critics claim media polls are merely a "form of corporate advertising" for news organizations, not legitimate news stories. Source: Moore, Superpollsters, 275; Gerald R. Ford, A Time to Heal (New York City, 1979), 331; Thomas R. Mann and Gary R. Orren, eds., Media Polls in American Politics (Wash., D.C., 1992), 4, 9.
1980: For the first time, a major television network uses exit polls to predict the winner of the presidential election.
Three hours before the polls closed in California, NBC projected Ronald Reagan to be the landslide winner over Jimmy Carter based on exit polls. The use of exit polls to forecast winners on election night has generated controversy over the years as some have claimed that releasing projected results can influence the voting behavior of people who have not yet voted. Before the 1982 and 1984 elections, Congress passed resolutions requesting the networks not call elections before the polls closed in all the states except Alaska and Hawaii. Source: Moore, Superpollsters, 262-264.
1990: ABC, NBC, CBS, and CNN join together to create the Voter Research and Surveys (VRS) to conduct all their exit polling.
Exit polling is very costly and complex undertaking. On election day in 1990, VRS, headed by Warren Mitofsky, conducted and tabulated almost 70,000 interviews from 1,200 precincts across the country. To this day, media outlets continue to pool their resources to conduct exit polls. In 2003, six major media outlets (ABC, Associated Press, CNN, CBS, Fox, NBC) formed the National Election Pool (NEP) and contracted with outside polling firms to conduct exit polls and provide them with poll analysis on election night. Some view the consolidation of exit polling as a dangerous development because it removes checks on the accuracy of the polls and makes the results more vulnerable to political manipulation. Sources: Moore, Superpollsters, 265-269; Edison Media Research and Mitofsky International website: http://www.exit-poll.net