Straw Polls

Straw polls originated as small, informal public opinion “surveys” that later evolved into large-scale, random sample surveys used primarily to determine the viability of potential political candidates. Nowadays, pollsters offer a litany of straw polls before nearly every national election and several state elections.

The phrase straw poll has its origins in the idea of “straws in the wind,” which were used to determine which way the wind is blowing. The earliest use of a straw poll was by a newspaper, The Harrisburg Penn-sylvanian, during the 1824 presidential election. Historically, straw polls were not scientific, and often they were conducted haphazardly. They relied on relatively large and sometimes massive samples to achieve some semblance of accuracy. Early on this included going to public places such as bars, political rallies, train stations and such, where groups of people were gathered and asking them their voting preferences. There was no sampling science at work at all.

However, as straw polling moved into the 20th century, this situation began to change. Yet, even with more than 2 million potential voters surveyed, straw polls were often quite biased, as demonstrated in the often-cited example of the Literary Digest's 1936 declaration that, by a landslide, Alf Landon was going [Page 852]to beat Franklin Delano Roosevelt. (It should be noted that the Literary Digest accurately predicted each presidential election from 1920 through 1932.) Recognizing the folly of the Literary Digest's approach, George Gallup, Elmo Roper, and Archibald Crossley each began using a more scientific approach. Specifically, with the help of demographers, they used quota sampling to build their list of poll respondents. Gall-up's mistake in 1948 caused the major polling organizations to reconsider that approach. Recognizing the downfalls of a quota sample, subsequent polling adopted much more strict probability methods. These probability methods have been modified to fit specifie needs, but they have all held true to one standard: that of random samples.

Straw polls have become commonplace in the news today. They have helped to create the horse race approach to reporting on elections. For example, a Gallup straw poll, conducted December 11–14, 2006, of Republicans and Republican leaders nationwide asked the following: Next, I'm going to read a list of people who may be running in the Republican primary for president in the next election. After I read all the names, please tell me which of those candidates you would be most likely to support for the Republican nomination for the president in the year 2008, or if you would support someone else? With an n of 425 and a margin of error of ± 6, Gallup found that 28% would support both John McCain and Rudy Giuliani. Another 12% would support Condoleezza Rice, and 8% supported Newt Gingrich.

Although the 2008 Republican nomination process would not formally begin for more than a year after this poll was conducted, the results indicated relatively little support for a number of other potential Republican presidential candidates which might well have caused one or more of them to reconsider actually running. A poll like this also may influence potential funders for particular candidates. It also may help the potential candidates themselves refine their campaign to try to capture support among the population that was not supporting them.

...