Example 4: Polling The 1936 presidential election proved a decisive battle, not only in shaping the nation's political future, but for the future of opinion polling as well. The Literary Digest, the venerable magazine founded in 1890 , had correctly predicted the outcomes of the 1916, 1920, 1924, 1928, and 1932 elections by conducting polls. The 1936 postal card poll claimed to have asked one fourth of the nation's voters which candidate they intended to vote for. Cards were sent to a sample of Americans drawn from telephone directories and car registration lists. In Literary Digest's October 31 issue, based on more than 2,000,000 returned post cards, it issued its prediction: Republican presidential candidate Alfred Landon would win 57% of the popular vote. In actuality, only 37% of voters supported Landon. How would you categorize the sampling design? Do you think such a large sampling error was primarily due to sampling variability or do you suspect there was some type of sampling bias at play?
The sampling design used in this example can be categorized as convenience sampling. The sample of Americans was drawn from telephone directories and car registration lists, which may not accurately represent the entire population of voters.
The large sampling error in this case is likely due to sampling bias. The sample was drawn from telephone directories and car registration lists, which may have introduced a bias towards certain demographics or socioeconomic groups. Additionally, the mailing of post cards may have reached only a specific subset of the population who were more likely to respond, leading to a potential bias in the results.
The sampling design used in the 1936 Literary Digest poll can be categorized as a non-probability sampling method. The magazine sent out postal cards to a sample of Americans drawn from telephone directories and car registration lists. These lists were not representative of the entire population as they excluded people without telephones and cars, which likely introduced a sampling bias.
The large sampling error observed in the poll's prediction, where they estimated 57% support for Alfred Landon but only 37% actually voted for him, is primarily due to sampling bias rather than sampling variability. Since the sampling method used was not representative of the population, the results of the poll were skewed towards certain demographics or socio-economic groups that were more likely to own cars or have telephones. This bias led to an inaccurate prediction of the election outcome.