American Government

In the rare case when public opinion shifts sharply on an issue, research shows that policy often follows suit. However, the public certainly does not always get its way on policies. Sometimes the reason is that certain groups in the populace are particularly committed and intense in their beliefs. A good example is gun control; a large majority of Americans want more of it, but those who oppose it feel much more strongly and are willing to work much harder to achieve their political goals. Another barrier to popular passions is the Supreme Court. For example, although a large majority of Americans prefers laws allowing prayer in schools or banning flag burning, even relatively conservative Supreme Courts have ruled that government would violate the constitutional rights of minorities if it created either policy.

Accurate measurement of public opinion through polls is a relatively recent phenomenon. George Gallup and Elmo Roper first developed statistical techniques for this purpose in the 1930s, but many refinements to their methods have been necessary. For example, Gallup had to change how interviewees were selected after predicting that Thomas Dewey would defeat Harry Truman in 1948. Fortunately, polling organizations learn from experience.

Polling techniques

Television stations often ask viewers to call so that they may express an opinion for or against a particular policy. Newspapers and Internet sites also occasionally indulge in this form of entertainment. These gimmicks may be called "polls," but they are completely unscientific because respondents choose whether to participate, and the group that is motivated enough to do so will not represent everyone else in a community. A key element of scientific polling, by contrast, is the representative sample, which requires that every possible respondent has the same probability of participating. This is accomplished today by using computers to dial telephone numbers randomly and then picking which person in a household to interview using another random method.

Obviously, no poll is perfect. But if the pollster succeeds at generating a random sample, then between 1,200 and 1,500 people will give an accurate picture of national opinion. The level of accuracy is often called the margin of error and indicates how much answers will bounce around the truth from poll to poll. Some people wrongly assume that the margin must include the truth, so if a poll estimates that 54 percent of Americans oppose the licensing of gun owners, with a margin of 3 percentage points, they assume the truth must lie between 51 percent and 57 percent. This is not true, however. One time in 20 a poll will draw an unlucky sample, one that represents national opinion poorly, even if the pollster did everything right.

Avoiding bias

In addition to sampling errors, polls can be biased by the type of questions asked and the way the polls are conducted. Questions must be as neutral as possible to avoid skewed results. "Do you believe serial murderers should be executed?" gets a much different response from "Do you support capital punishment?" Interviewers must be careful not to inject their own views into the process by how they ask a question. A poll is also only as good as the respondents, and its validity clearly depends on their willingness to tell the truth about their positions.

Polls are an integral part of American politics. Besides the polling organizations, the news media routinely conduct and report the results of their own surveys. Pollsters also have high-profile positions on campaign and White House staffs. This concern for measuring public opinion indicates that public opinion is useful in understanding the positions of the American people and what policies they support.