Poll

Poll is a term commonly used to refer to a public opinion survey in which the main purpose is to collect information concerning people's opinions, preferences, perceptions, attitudes, and evaluations of public opinion issues. The performance of public officials, confidence in public and private institutions, enacted policies on topics such as poverty, education, taxes, immigration, public safety, the war on terror, same-sex marriage, abortion, or gun control are only some of the topics that polls typically focus upon. Also, the term poll is associated with the measurement of current presidential and congressional vote intention, using either candidates' actual names or generic references such as “Democrat” or “Republican” in order to forecast election results.

A poll is conducted on the basis of sampling principles; that is, a representative group of persons from a specifie population is interviewed in order to generalize results to the whole population. A common practice in a poll is to establish an approximate confidence interval, also known as sampling error or margin of error. For instance, a poll may be reported stating that “with 95% confidence the margin of error is plus/minus 3% around the estimated percentage, given a sample size of 1,000 cases.” This type of statement means that if the poll were taken repeatedly with different samples of the same size, one might expect that 95% of the time (i.e. 19 times out of 20) the poll estimate would be within the confidence interval.

A poll may be conducted using self-administered or interviewer-administered methods or a combination [Page 588]of both, depending on the survey aims as well as budget constraints. Telephone and face-to-face modes are the predominant interviewer-administered methods used for public opinion polling. Mailing, computer-and Web-based methods are self-administered modes that are also used. Such interviewing modes are used either because of financial reasons, time constraints, or at times when public issues are highly controversial or socially (un)desirable. By removing the interviewer's presence, a pollster may elicit more truthful answers; however, having no live interviewers may considerably reduce the willingness of a respondent to participate.

Question wording plays a critical role in a poll; variations in the question stem may achieve different results. For instance, a question asking, Would you vote or not for making marijuana legally available for doctors to prescribe in order to reduce pain and suffering? may yield very different results than a question worded, Do you support legalizing marijuana for medical use? Question order is also an important feature in opinion polling, because respondents tend to answer polling questions within the context in which they are asked; hence, the content of preceding questions may affect subsequent answers.

Mostly, polling questionnaires have closed-ended questions with either ordered response scales or categorical response options. Sometimes polls also include a few open-ended questions, which usually are follow-up questions from previous closed-ended questions. Thus, respondents are provided statements or alternatives among which they can choose, as simple as a “Yes/No” format and as extensive as a numeric 11-point scale ranging from 0 to 10 and anchored with “Extremely Dissatisfied” and “Extremely Satisfied.” Also, differences in the way response options are presented may yield different patterns of response even when the question wording is kept constant. For instance, providing explicit response option categories such as “Don't Know,” “No Opinion,” or “Neutral” may change the overall distribution of the other response options.

...