The Gallup Organization
Address:
47 Hulfish Street
Princeton, New Jersey 08542
U.S.A.
Telephone: (609) 924-9600
Toll Free: 800-888-5493
Fax: (609) 924-0228
http://www.gallup.com
Statistics:
Wholly Owned Subsidiary of Selection Research Institute
Founded: 1958
Employees: 3,000
Sales: $300 million (1999 est.)
NAIC: 54191 Marketing Research and Public Opinion Polling
Company Perspectives:
Gallup's core expertise is in measuring and understanding human attitudes and behavior. Gallup applies this expertise to help companies improve performance by leveraging their employee and customer assets. Gallup also conducts The Gallup Poll, the world's leading source of public opinion since 1935.
Key Dates:
1935: George Gallup begins syndication of 'America Speaks.'
1936: Gallup's reputation made by predicting election.
1948: Gallup picks Dewey to defeat Truman.
1958: The Gallup Organization is created.
1984: George Gallup dies.
1988: The Gallup Organization is sold to Selection Research Institute but maintains New Jersey headquarters.
Company History:
The Gallup Organization was created in 1958 by George Gallup, whose name worldwide is all but synonymous with public opinion polling. Although the Gallup Poll, which monitors political and economic trends, conducted since 1935, remains its most prominent enterprise, the company generates most of its revenues from marketing and management research. Gallup has offices in more than 25 countries.
Polling Background
Although George Gallup did not invent public opinion polling, he virtually created the image of the 'pollster.' He helped to incorporate scientific methodology in the mid-1930s, but almost as important was his gift for promoting the field--and himself. The scores of pollsters that today work in politics, as well as market research, owe a debt of gratitude to his pioneering efforts.
Political polls were conducted in America long before Gallup. The first published presidential poll, based on a straw vote, appeared on July 24, 1824 in the Harrisburg Pennsylvanian. Newspapers at the time were little more than vehicles for the political parties, but as economic pressures forced publishers to become less partisan in order to expand readership, objectivity became a virtue. Straw polls were, by definition, objective; and by the beginning of the 20th century they became a staple of newspapers.
The way straw polls were conducted, however, did not lend itself to accuracy. Some newspapers and magazines printed the ballot within its pages. Readers mailed in or hand delivered their votes, and they were encouraged to stuff the ballot box by purchasing more copies of the publication. Reliable results were willingly sacrificed for a spike in sales. A later technique, the mail ballot, selected names from such sources as telephone directories, registered voter lists, and automobile registrations. Then a 'sample' was created by pulling names at certain intervals, such as every tenth one. It was more difficult to stuff the ballot box, but the sample had an inherent bias against the lower economic strata. The 'personal canvass' proved to be the most reliable method for conducting straw votes. Under this method, 'solicitors' would hand out pencils and ballots to people on the street and collect the votes on the spot. Some newspapers made an attempt to sample a cross section of voters by creating quotas for their solicitors, for example requiring a certain number of white-collar voters from one community and blue-collar voters from another. Although arrived at intuitively, this technique anticipated the scientific polling that Gallup and others would refine.
Most of the early newspaper polls were local or regional. The New York Herald Tribune and collaborating newspapers began to conduct wider pre-election polls in the 1890s. By 1912 they polled in over 35 states. The Hearst newspapers attempted nationwide polling in 1924. Forty-three states were covered, but the average error rate was a high six percent. In 1928, however, Hearst had an error rate of less than three points in 46 states.
By the 1930s the publication with the greatest reputation for accurate polling was the Literary Digest--the Time or Newsweek of its day. The Digest mailed out an incredible 20 million ballots and covered all 48 states. Although some critics questioned the sample, maintaining that the Digest overemphasized the higher income brackets, the results of the 1932 election silenced all doubters. The straw poll predicted a Franklin Roosevelt win with 59.85 percent of the popular vote. The election results gave Roosevelt the win with 59.14 percent of the vote. The straw poll also predicted that Roosevelt would win 41 states, totaling 474 electoral votes. The actual results were 42 states and 472 electoral votes.
The Literary Digest did not hesitate to crow about its accomplishment and was now more than a little confident in its ability to predict election results. Then in the summer of 1936, more than six weeks before the Digest began its massive mailing to poll for the winner of the Roosevelt-Alf Landon presidential race, a little-known pollster from Princeton, New Jersey, predicted that the Digest would be wrong, and he had the further audacity to predict their final numbers. That pollster was George Gallup.
1930s: George Gallup Devises a Poll
Gallup attended the University of Iowa, where he became editor of the campus newspaper. While working one summer for a St. Louis advertising agency that was researching reader satisfaction with the St. Louis Post-Dispatch, Gallup decided that there had to be a more efficient way to measure opinions than to go door to door, neighborhood after neighborhood. He wondered if he could use techniques similar to the ones employed by government inspectors who might test a crop of wheat or a supply of water by taking several small samples then extrapolate the quality of the entire amount.
While teaching journalism at Iowa, Gallup earned his M.A. and Ph.D. His doctoral dissertation, 'A New Technique for Objective Methods for Measuring Reader Interest in Newspapers,' outlined what would simply become known as the 'Gallup Method.' He began to conduct surveys and publish the results in trade magazines, drawing the attention of the newspaper world. In 1932 he was hired by the New York advertising agency of Young and Rubicam as director of research. He would work for the company until 1947.
It was also in 1932 that Gallup's mother-in-law was elected Iowa's Secretary of State, under unusual circumstances. She had been placed on the ballot in honor of her husband who had died during a run for governor in 1926. Despite running as a Democrat in a heavily Republican state, and without even mounting a campaign, she was swept into office with a Roosevelt landslide. Gallup began to wonder if his sampling methods could be used to forecast such drastic changes in public opinion. He used the Congressional elections of 1934 as a test and came within one percentage point of predicting the overall returns. A Chicago agent, Harold R. Anderson, recognized the potential to make money out of the technique and partnered with Gallup to create the American Institute of Public Opinion. They set up shop in Princeton, New Jersey, with the hope that the prestigious Princeton postmark might influence people to mail back their questionnaires to the 'institute' that was in actuality a one-room office in which a handful of workers hand counted ballots.
In 1935 Gallup began writing a syndicated column using his poll results, titled 'America Speaks.' To sweeten his pitch to newspapers, Gallup offered a money-back guarantee that he would prove more accurate than the illustrious Literary Digest poll in predicting the presidential election of 1936. His partner was then able to place 'America Speaks' with 42 newspapers. More than pride was now on the line. Given the cost of a poll sample of 15,000 respondents, ten times the size of what he would one day use, Gallup faced financial ruin. The Digest, suffering so much from the effects of the Depression that it was forced to reduce its mailing to ten million pieces, was more than eager to defend its franchise and to ridicule Gallup's charge that by drawing its sample mostly from the ranks of people who owned telephones and automobiles the Digest would undercount lower-income voters.
The final Digest Poll results were within one percentage point of Gallup's earlier forecast. The Digest gave the election to Landon with 57 percent of the vote. Gallup gave it to Roosevelt with 54 percent. In reality Roosevelt won another landslide victory, taking 61 percent of the popular vote. Gallup was off by seven points, much to his dismay, but at least he picked the winner. The Literary Digest put on a brave front, refusing to acknowledge its errors, but its reputation was shattered, and within a year the magazine folded.
The career of Gallup, as well as those of Elmo Roper and Archibald Crossley, who also predicted the Roosevelt victory, flourished. However, it was Gallup who became the best known of the new scientific pollsters. Eventually 'America Speaks' would be carried by 200 newspapers. He founded the British Institute of Public Opinion, as well as polling operations in dozens of other foreign countries, making Princeton, New Jersey, the public opinion polling capital of the world. The Audience Research Institute, which Gallup founded in 1937, studied public reaction to movie titles, casts, and stories. He began to keep tabs on questions that are taken for granted today, but now provide a historical perspective on the American public and its officials. He kept track of Presidential popularity. He was the first to ask 'Who would you vote for if the election was held today?' He asked Americans if they believed in God and how often they went to church. When Senator Joseph Lieberman was selected to run as the Democratic vice-presidential candidate in 2000, journalists could report how the country had changed in its willingness to vote for a Jewish American, because the Gallup Poll had been asking the question since the 1930s.
Gallup continued to refine his techniques, determined to avoid the large margin of error in his polling of the 1936 election. He no longer relied on mail ballots, because higher-income voters were more likely to return them, which he felt would favor Republican candidates. He sent his people into the field to interview respondents, with quotas based on demographic categories, such as age, sex, geography, and income. The interviewers, however, were given too much latitude. Rather than embarrass respondents by asking their age or income, they often guessed. Interviewers also tended to seek out respondents with whom they felt most comfortable, with the result that working-class interviewers and white-collar interviewers were getting different results. Overall, the Gallup Poll was displaying a systematic bias in favor of Republican positions over Democratic ones, enough to prompt Congress to call in Gallup to explain his election results of 1940 and 1944 that underestimated the Democratic vote in two-thirds of the states. A technical committee criticized him for using a quota system instead of 'probability' sampling, a method that would give everyone a equal chance of being included in a poll. However, probability sampling was both complicated and extremely expensive. Gallup felt that the difference between quota and probability sampling was not large enough to justify the cost. Only in a very close election would it even matter. Unfortunately for Gallup such an election was at hand.
1948: Dewey Defeats Truman!
The presidential election of 1948 would prove almost as devastating to Gallup as the 1936 contest was to the Literary Digest. Because the Democratic Party was splintered, with Henry A. Wallace running for president on the Progressive party ticket and Strom Thurmond representing the Dixiecrats, Harry Truman appeared to have little chance to retain the White House against the bid of the Republican nominee, Thomas Dewey. Gallup and the other major pollsters believed that public opinion only showed dramatic change when responding to important events. Political campaigns were not considered important enough. A poll taken after the political conventions would surely predict the winner. Gallup stopped polling in mid-October, and although he noted a surge in support for Truman, he felt confident that Dewey would win the election. All the experts agreed with him. Truman did not stop campaigning, however; he beat the odds, and won the election.
'The pollsters became national laughingstocks,' according to Michael Wheeler, author of Lies, Damn Lies and Statistics, 'and Gallup, the most famous pollster of them all, took the hardest fall. Others were more graceful in their embarrassment, but Gallup was indignant. How, he sputtered, could scientific surveys be expected to take into account `bribery of voters' and `tampering with ballot boxes.' Gallup was also accused of favoritism, a charge that could prove devastating to a man whose business depended on impartiality. Although Gallup vehemently denied that he rigged polls to favor Dewey, he admitted that he considered Dewey to be a close friend and had been in contact with him throughout the 1948 campaign.
There was no doubt, however, that Gallup's reputation had been tarnished. Many newspapers, unwilling to accept his explanation that this was an election that would happen only once in a generation, threatened to cancel their contracts. Gallup managed to survive, became more scrupulous about maintaining distance from political candidates, and improved his polling methods. Still, the influence of political polling in America was diminished until the 1960s.
Market Research in the 1950s
While other polling operations began to focus on the more profitable area of market research, Gallup maintained his academic approach. Finally in 1958 he created the Gallup Organization and moved his company into market research, but he never achieved a significant share of the business. Although in 1966 his son, George Gallup, Jr., became president of the Gallup Organization, the elder Gallup continued to serve as chairman of the board and was actively engaged in its day-to-day operations. He vowed that he would never retire, and he never did. He died in 1984 of a heart attack at his summer home in Switzerland.
Gallup's final years were not, however, without controversy. In 1968 two of Gallup's interviewers were discovered to have falsified data in a poll of Harlem residents conducted for the New York Times. More troubling were charges that Gallup's people maintained improper ties to the Nixon administration. Poll numbers were provided before publication, allowing Nixon to prepare the public and the put the best possible spin on the results. Nixon's aides also suggested questions for the Gallup Poll, thus influencing public opinion from the outset. The Nixon administration used both Gallup and rival Louis Harris, misleading the pollsters' associates into thinking that Nixon would not make improper use of early poll results. At the very least the pollsters were naive. The fact that Gallup officials only met with Nixon aides in a hotel rather than the White House was a tacit admission that if such contact were known to the public the company's reputation for objectivity would be compromised.
New Leadership in the 1980s and Beyond
With the loss of its founder in 1984, the Gallup Organization struggled. According to George M. Taber writing for Business News New Jersey, 'One of the most difficult transitions any corporation faces is to move away from the structure and style established by the founding father and become a company that has a life bigger than he. Corporations as large as the Ford Motor Company and as small as a country grocery store have faced that challenge. It was a transition, however, that the Gallup Organization did not successfully make.'
The Gallup name did retain brand value, and in 1988, after discussing a merger for three years, the Gallup family sold its private company to the Selection Research Institute, a marketing research firm based in Lincoln, Nebraska. Gallup's data processing and interviewing operations were moved to Lincoln, but the company's headquarters remained in Princeton, considered the 'epicenter of the polling world,' according to a 1997 article in Business News New Jersey. The Gallup Organization continued to produce the Gallup Poll in conjunction with print and television news organizations, and became more aggressive in performing surveys for corporations and marketers.
Under its new parent, as well as the leadership of co-chairmen Alec Gallup and George Gallup, Jr., the founder's sons, The Gallup Organization saw increasing revenues in the late 1990s at an estimated annual rate of 30 percent. In a 1997 interview for Marketing News, the two men discussed the future of the research industry in an age of rapid technological advancement, agreeing that although the polling process has become quicker and more sophisticated, the 'fine art of crafting a good question' remained integral to successful polling. Regarding Gallup's prominence in its field, one in which competition was heating up in the late 1990s, Alec Gallup noted, 'We do have an advantage--and its an important one and we exploit the hell out of it--and that is the fact that the Gallup name is well-known and has credibility.'
Principal Competitors: McKinsey & Company; Harris Interactive Inc.; Opinion Research Corp.; Towers Perrin.
Further Reading:
Gallup, George Horace, The Pulse of Democracy, New York: Greenwood Press, 1968, 335 p.
Jacobs, Lawrence and Robert Y. Shapiro, 'Presidential Manipulation of Polls and Public Opinion,' Political Science Quarterly, Winter 1995/1996, p. 519.
McCullough, David, Truman, New York: Touchstone, 1992, 1,117 p.
Miller, Cyndee, 'Gallup Brothers Analyze the Research Industry,' Marketing News, January 6, 1997, p. 2.
Moore, David W., The Super Pollsters, New York: Four Walls Eight Windows, 1992, 388 p.
Pace, Eric, obituary, New York Times, July 28, 1984, pp. 1, 9.
Sussman, Barry, obituary, Washington Post, July 28, 1984, pp. A1, A7.
Smith, Richard D., 'In Polling, It All Started with Gallup,' Business News New Jersey, April 21, 1997, p. 5.
Taber, George M., 'Gallup Polls Sells Out to a Nebraska Company,' Business News New Jersey, October 11, 1988.
Wheeler, Michael, Lies, Damn Lies and Statistics, New York: Liveright, 1976, 300 p.
Source: International Directory of Company Histories, Vol. 37. St. James Press, 2001.
