4.6.1: The Presidential Election of 1936

A classic example of the dangers of not carefully considering the population of interest comes from the results of the United States presidential election of 1936. Alfred M. Landon was the Republican candidate and Franklin D. Roosevelt was the Democratic candidate. The United States was still amid the Great Depression and Roosevelt was running on a platform that the social programs of the New Deal should be continued to help those who were unemployed and living in poverty. While Landon accepted much of the New Deal, he considered it to be hostile to business, wasteful, and inefficient. Landon also accused Roosevelt of acquiring so much power that he was subverting the Constitution. The candidates also differed on issues of foreign policy, with Landon having a somewhat isolationist attitude.

The Literary Digest magazine conducted a very large survey of approximately 2.4 million individuals in one of the largest political polls ever conducted (Figure \(\PageIndex{1}\)). The magazine had successfully predicted the outcome of every election since 1916. Over the course of several issues, the magazine updated the results of this survey and concluded that Landon would easily win the election, a surprising conclusion to the political experts of the time. Some partial results from this survey are shown from a page in the October 24, 1936, issue of the magazine in Figure \(\PageIndex{2}\). In an astounding defeat for the magazine, Roosevelt easily won the election by securing 62% of the population vote and 523 of the 531 possible electoral votes. The results were devastating to the integrity of the magazine, which went bankrupt shortly afterward.

Figure \(\PageIndex{1}\): Cover of the September 5, 1936, cover of Literary Digest, touting their nationwide presidential poll (Photograph by the author).

Figure \(\PageIndex{2}\): Partial results of the nationwide presidential poll from the October 24, 1936, issue of Literary Digest (photograph by the author).

A large survey would be generally thought to provide reliable results but, in this case, there is a difference between the population that the researchers thought they were observing and the population they were actually observing. To implement the survey, the magazine mailed questionnaires to ten million people using addresses from telephone books and club membership lists (Freedman et al. 2007). Hence, the actual population studied corresponded to individuals who had telephones and belonged to clubs. During the Great Depression this population was overwhelmingly wealthy and supported Landon. This differed significantly from the general population who supported Roosevelt to a greater extent. It is this difference that caused the major error in their poll and the ultimate demise of the magazine.

Interestingly, George Gallup (Figure \(\PageIndex{3}\)), who would become famous for the well-known Gallup poll, correctly predicted the outcome of the election by observing from a population whose makeup was closer to that of the actual voting population for the 1936 presidential election. In fact, Gallup also successfully predicted what the results of the Literary Digest poll would be well in advance of the publication of their results using a much smaller survey.

Figure \(\PageIndex{3}\): George Gallup (1901–1984) was a pioneer of survey techniques and inventor of the Gallup poll, a successful survey for measuring public opinion (public domain image).