When research is intended to reveal very precisely what a population is like, careful sampling procedures must be used. This requires defining the population and sampling people from the population in a random fashion so that no biases will be introduced. In order to learn what elderly people think about the medical services available to them, for example, a careful sample of the elderly population is needed. Obtaining the sample only from retirement communities in Arizona would bias the results because these individuals are not representative of all elderly people in the population.
Thus, when evaluating survey data, a researcher must examine how the responses were obtained and what population was investigated. Major polling organizations such as the Gallup organization typically are careful to obtain representative samples of people in the United States. Gallup polls are frequently conducted to survey the voting public's opinions about the popularity of a presidential candidate or a given policy. Many other surveys, however, such as surveys that are published in popular magazines, have limited generalizability because the results are based on people who read the particular magazine and are sufficiently motivated to complete and mail in the questionnaire. When Redbook, for example, asks readers to write in to say whether they have ever had an affair, the results may be interesting but would not give a very accurate estimate of the true extent of extramarital sexual activity in the United States. An example of an inaccurate sampling technique was a survey by Literary Digest (a now defunct magazine) sampling almost ten million people in 1936. The results showed that American presidential candidate Alfred Landon would beat Franklin D. Roosevelt by a landslide in that year's election. Although it was large, the sample was completely inaccurate.
Was this article helpful?