Polling

views updated

POLLING

POLLING is a form of surveying conducted by the canvassing or questioning of a universe. A universe can consist of a particular group, such as steel workers, or can rely on a more general population, such as a political survey of public opinion in an election year. Polling dates back to 1824 in the United States, when two newspapers, the Harrisburg Pennsylvanian and the Raleigh Star, attempted to predict the presidential election by use of "show votes." By the twentieth century, polls would be taken by magazines, such as Farm Journal (1912) and Literary Digest (1916). However, these polls were mostly local in scope. The first major national poll was conducted during World War I, and asked participants whether or not the United States should become involved in the war.

In 1936, the process of polling would change forever. George H. Gallup, founder of the American Institute of Public Opinion (1935), had issued a challenge to Literary Digest, claiming that he could more accurately predict the outcome of that year's presidential election. At the time, this seemed foolhardy, for the Literary Digest had correctly predicted every presidential election correctly since 1916. Confident in his methods, Gallup had developed a system of interviewing based on quota samples, which employed a relatively small number of people to mathematically determine the views of the public at large. He came up with fifty-four different questions, and considered each question demographically, with such key determinants as age, sex, income, and region. The Literary Digest, meanwhile, had conducted an old-fashioned straw poll, based on telephone and car buyer lists. In the past, such lists had been serviceable, but in 1936, during the depression, they proved to be heavily biased in favor of the Republican candidate, Alfred M. Landon. For this reason, the Literary Digest, having predicted a landslide victory for Landon, was upstaged by the audacious Gallup, who correctly predicted another victory for Franklin D. Roosevelt. Other, lesser-known forecasters, such as Elmo Roper and Archibald Crossley, had also predicted Franklin's victory using similar sampling methods.

However, even Gallup would be proven wrong in the presidential election of 1948. In that year, all major pollsters predicted the Republican candidate, Thomas E. Dewey of New York, would defeat Harry S. Truman, the current president. Gallup had used the same sampling techniques as before, but had made a terrible mistake by concluding his poll weeks before Election Day. Furthermore, Gallup had made the incorrect assumption that the "undecided" votes would split in much the same way as the "decided" ones. This would prove untrue, as most of the "undecideds" either voted for Truman or did not vote at all.

Gallup would learn from his mistakes and recover ever stronger after the election. He improved his sampling techniques, taking into account a greater number of influences, and reanalyzing the effects of the inner city and other regions, which, to his undoing, his interviewers had neglected in 1948. Gallup also made certain that polling was done consistently, a process known as tracking, and that the likelihood of a person actually voting was now also taken into consideration.

With these improvements, Gallup's organization was able to accurately predict future elections. Between 1952 and 2000, the Gallup poll, as it came to be known, achieved an average accuracy of just over 2 percent deviation in presidential elections. In the controversial 2000 election between Al Gore and George W. Bush, Gallup's final preelection prediction had the race as a "statistical dead heat," meaning that the candidates were locked within a range of error of plus or minus 3 percent.

Gallup also introduced polling to social issues. He conducted polls on such complicated topics as the Vietnam War and education. He felt that polls gave the people a voice, and that they were therefore an important aspect of democracy.

Many, however, criticize polling, claiming that it has a "bandwagon effect" on voters, and too much control over politicians. Yet polling continues to play an important role in the political process.

Polling techniques are also extensively used in industry to conduct market research. Companies use sampling in order to determine what products consumers are willing to buy. Such techniques may include random sampling, in which everyone is a potential candidate for an interview; stratified sampling, in which sampling candidates are divided into nonoverlapping groups; quota sampling, random sampling subject to demographic controls; and cluster sampling, in which "clusters," or groups, are selected from various sectors of the population, such as the middle class or working class.