Election Polls: Accurate Surveys or Bad Forecasts?

September 10, 2008

In 1948, pre-election polls predicted that Thomas Dewey would defeat Harry S. Truman in the presidential election. Based on these polls, many newspaper editors confidently announced Truman's defeat. As this picture shows, Truman had the last laugh, as he won re-election to the presidency.

Pre-election polls are one of the most visible types of surveys in the news, at least around election time. Are they right or wrong? Good or bad? Have polling methods improved since 1948? Or will a candidate still get to get the last laugh? Interested in better understanding the political polls this year?

Here are some on-line resources to help guide you through sorting through the good and the bad.

The American Association for Public Opinion Research is the main professional organization for public opinion researchers. They have put together a page of helpful information on pre-election polls. http://www.aapor.org/electionpollingresources2
If you really want to learn how to interpret polls in the media, take the on-line course journalists take: “Understanding and Interpereting Polls.” This course by the Poynter Institute is targeted to journalists, but is free and available to anyone. http://www.newsu.org/courses/course_detail.aspx?id=aapor_polling07
Mark Blumenthal of Pollster.com has a great set of FAQ’s.
Rutgers Professor Cliff Zukin has a useful resource talking about pre-election polls: “Sources of Variation in Election Polls: A Primer” (Cliff Zukin). http://www.aapor.org/uploads/zukin_election_primer.pdf

Archive of Articles

Spring Conference Recap

Experts offer Multiple Takes on Technology at Conference

Thomas Guterbock summed things up nicely at the recent “New Technolgies and Survey Research” conference held by the Program on Survey Research at Harvard University.

In his presentation “Strategies and Standards for Reaching Respondents in a New Age of Technology,” Guterbock – the Director of the University of Virginia’s Center for Survey Research – described both sides of the new technology coin: it’s easier than ever for pollsters and survey researchers to get in touch with people – via cellphone, email, instant message, etc.; at the same time, other technologies are making it easier for those same people to avoid surveys—by using caller id, answering machines, etc.

Harvard University Program on Survey Research