How to Create a Poll

Written Gilbert St. Clair
Professor of Political Science

In this election year, we have all been bombarded with the information gathered in public opinion surveys (a.k.a. polls). The purpose of these polls has varied from those designed to understand human behavior (done by Social Scientists) to the media’s desire to report who is ahead and who is behind in electoral contests to those seeking to manipulate voters’ attitudes or to reveal the mating habits of college students. Often the real intent of these polls was not to inform the public, but either to manipulate or entertain it. Most of what we know about the results of the public opinion surveys comes to us through the filter of the news media.
Journalists who report the findings of public opinion surveys have an obligation to tell us about the technical features of the surveys as well as, present the substantive results and interpretation of the polling data. The description of the survey’s technical features enables the reader/viewer to assess the reliability and validity of the data being reported in a news story. The major polling organizations through their trade associations have adopted the following Principles of Disclosure:

All reports of survey findings of member organizations, prepared specifically for public release, will include reference to the following:

  • sponsorship of the survey
  • dates of interviewing
  • method of obtaining the interview
  • population that was sampled
  • size of the sample
  • complete wording of questions upon which the release is based
  • the percentages upon which conclusions are based

    Any news report on a reputable poll should include all of the above items plus the margin of sampling error and the confidence factor that the respondent sample is representative of the targeted population.

    We need the technical description of the survey for the following reasons:

  • Knowing who sponsored the poll and who conducted it helps us understand the purpose of the poll and the competence of the pollster.
  • When the poll was conducted may include a period in which these were exogenous events that could affect responses.
  • How the data is collected may bias the responses, consequently it is essential to know what method was used to conduct the survey. For example, was it live interviews either face to face or by telephone or self-administered mail questionnaire or Internet polling in which the respondent self-selects to participate.
  • What is the target population – all adults in the United States, registered voters in Texas, unmarried college students, etc.
  • The sample must be large enough to produce a statistically acceptable margin of error and should be a randomly selected set of respondents from the target population not a respondent self-selection sample.
  • Reporting the frequency distribution of the responses is the minimum analysis necessary to understanding any conclusions drawn about the data.

    When given the details about the public opinion survey we will be able to be discerning consumers of the information the survey provides.

  • Bookmark the permalink.

    Leave a Reply