Teacher resources and professional development across the curriculum

Teacher professional development and classroom resources across the curriculum

Monthly Update sign up
Mailing List signup
Statistics - Polls: What do 
        the numbers tell us?What Can Go Wrong

Related Web Sites

Polling FAQ
A broad FAQ on polling and its reliability.

Why Do Campaign Polls Zigzag So Much?
The author describes how polling methods caused fluctuations in data reported during the Clinton/Dole presidential race.

Ethical Guidlines for Statistical Practice
This document assists students in learning to perform statistical work responsibly.

Can Anything Go Wrong?

It is rare for any poll to actually ask questions of every person in a particular group. Usually only a few members of the group are questioned as a sample of the population.

Examples of Errors in the Making
Many factors can keep a poll from being a perfect indicator of how all of the people in the population think, feel, or behave. Here are just a few of the ways a poll can end up being a lot less than 100% accurate:

  1. Most polls are conducted by telephone. If we use the phone book to select 500 names at random to call, we will never question people with unlisted phone numbers or people who do not have a phone.
  2. Research indicates that most phones are answered by women or older people. If you interview only the people who answer, will you have a biased sample?
  3. In the United States, about 50% of registered voters do not vote. A poll of 500 registered voters might not include anyone who will actually showup and cast a vote. How might this matter if a poll reported that of 500 registered voters, 70% said that if the election were held today they would vote for Fletcher?
  4. Almost any question will bias the answer to some degree. For example, the pollsters in our story might ask, Do you support Higgins or Fletcher? Or they could ask, Do you support Fletcher or Higgins? Research indicates that the order in which the names are listed in the question will make a difference in the answer a person gives.
  5. Now that most Americans work outside the home, there is a very small window of time when pollsters can reach people by phone. The person who answers may be tired after a day's work, not to mention frustrated by a large number of unsolicited telephone calls. Should we count on what those people say?

In Daily Life

"Statistics" is inspired by programs from Against All Odds: Inside Statistics.


© Annenberg Foundation 2017. All rights reserved. Legal Policy