- What people believe;
- How they feel about something; or
- In what way they will act.
The results from public opinion polls are used in a number of ways. They have come to influence what Americans are offered to eat and drink, the kinds of cars they can buy, and the programs that they can watch on TV.
In addition, public opinion polls now play an important role in politics. They are used throughout the course of election campaigns by candidates and by the media to see which candidates are ahead and who is likely to emerge victorious. The results of these polls, in turn, largely determine where future campaign monies are to be spent and where each candidate's efforts will be concentrated until the close of the campaign.
But can the polls really be trusted? In the 1948 Presidential election, for example, the polls predicted certain victory for Republican Thomas E. Dewey. Without waiting for the official count of the votes, newspapers throughout the country proclaimed in their headlines, "Dewey Defeats Truman." The rest is history ... Harry S Truman was elected the 31st President of the United States.
Public opinion analysts and professional polling organizations, however, did profit from this colossal mis-prediction. With their credibility severely damaged, they developed far more sophisticated sampling techniques. Moreover, they made greater use of sociological and psychological research and modern computer technology. In addition, they are careful to point out that their findings apply only at the time the questions were asked and that the results do not predict the outcome of the election.
Nevertheless, in viewing the results of any public opinion poll, it might be useful to ask the following questions:
1. Who Was Interviewed? Generally speaking, the accuracy of a poll depends upon the degree to which the characteristics of the people being interviewed is really similar to those of the group they are supposed to represent. For example, the polling of sixteen-year-olds to predict the outcome of an election would be very questionable since they cannot vote.
Also, as a general rule, the greater the number of people interviewed, the more likely the prediction will be accurate. Everything else being equal, an election poll of 100,000 out of two million voters is more likely to produce accurate results than a poll of 1,000 out of the same number. It is important to point out that large, national polling organizations have small national samples of under 2,000 that predict quite accurately for the entire electorate.
Lastly, those interviewed should have been selected in a random fashion. This is usually done to avoid or lessen the possibility of allowing any "unaccounted for" bias or characteristics ... of those being interviewed ... to influence the results. The accuracy of a poll designed to sample the views of all registered Republicans, for example, would definitely be suspect and have a conservative bias if it interviewed only contributors to Barry Goldwater's unsuccessful presidential campaign of 1964.
2. Under What Conditions Were The Interviews Conducted? Generally speaking, unclear, biased, or emotionally charged questions will produce misleading answers and weaken the accuracy of the results of a poll. Questions such as ... How do you feel about candidate X? or, You are planning to vote for candidate Y, are you not? would be suspect.
Also, if the people being polled are asked to choose from a given set of responses in answering a question, there must be an acceptable number of alternatives from which to choose. For example, suppose those being polled are required to respond to a question ... either "yes" or "no." This practice would eliminate the possibility that some of the people may truly be "undecided" and consequently distort the accuracy of the poll's results.
Finally, polls conducted by telephone or through the mails generally do not tend to be as reliable as personal interviews. This is largely due to the fact that the former measures are not as likely to be able to control for who really participates in the poll, the number who respond, and possible misinterpretation of the questions.
3. When Was the Poll Conducted? It should also be noted that the results of a poll are representative ... however accurate ... of the preferences, views and feelings of a particular group of people at a particular point in time. As a general rule, the more current the poll, the more likely it is to produce meaningful and useful results. A summer poll regarding who should be elected president in 2004, for example, is not likely to be as accurate as a poll taken during election week of the actual election.
4. Who Conducted the Poll? Past reputation and performance can also help an individual determine the validity of the results of a poll. Generally speaking, "novice" pollsters are not likely to be able to compete with professional polling organizations with their large staff's, seemingly unlimited resources, and sophisticated computer equipment. In addition, polls conducted by groups with an obvious interest in the results should be held suspect until proven otherwise. Finally, past performance records of a polling group might be useful in determining the organization's credibility and reliability.
5. What was the Percentage of Error Polling organizations should also indicate what the potential for error of their poll is. Based on the size of their sample it is statistically possible to do sod indicates reliability to the reader.
Based on this analysis, consider the following questions:
- Which of the factors described above in assessing the validity of a poll do you think is most important? Least important? Why?
- Do you think polls are valuable? Why or why not? Would you place any restriction on them in reporting an election? If so, explain. If not, why?
("How True is True? Evaluating and Conducting Public Opinion Polls" was adapted fromBill of Rights in Action, Vol. 10:2 © Constitutional Rights Foundation)
Building Your Own Public Opinion Poll
Although you probably don't have the means to conduct a scientific opinion poll, you can take an informal poll. It can help you learn what people in your school or community think about the election and other issues. There are three steps to conducting a public opinion poll:
1. Create a Public opinion poll
- Make most of your questions multiple choice and yes/no. This will make your public opinion poll easy to tabulate.
- Keep the public opinion poll short and simple.
- Be sure that your questions do not force particular answers. They must be unbiased. Otherwise your public opinion poll results will be open to criticism.
- Test your public opinion poll. Before conducting the public opinion poll, ask someone to check it over. Does that person think it is clear?
2. Select the Population and Sample
- Determine the population. What will your poll results represent? The opinions of everyone in the community? Of a section of the community? Select the population you want the poll to cover.
- Select a sample. You don't have to poll the entire population to get a good idea of how people in the population feel. Try to get a random sample of the population. This means that every person in the population has the same chance of taking the public opinion poll. For example, telephoning the fifth person on each page of the phone book would be a random sample.
3. Conduct the Public opinion poll
- Prepare and practice a brief introduction. When approaching a stranger, introduce yourself, tell them what group you are from, explain the public opinion poll's purpose, and ask whether the person would mind spending a few minutes answering it.
- Be polite. People who answer your public opinion poll are doing you a favor. Don't badger anyone to take the public opinion poll.
- Tell all interviewees that they do not have to put their names on the public opinion poll. Results will be reported anonymously.
- Be as organized as possible. Use a clipboard to hold the public opinion polls and bring extra pens or pencils.
- Wait for each public opinion poll and check it. Make sure the information is complete. If you read the public opinion poll to the respondent and fill it in, write exactly what the person says.