Tribune Print Share Text

America at the polls

Created date

May 22nd, 2012

If you ve ever followed politics, voted in a presidential election, or read the newspaper for that matter, then you have doubtless come across an opinion poll. You ll probably find dozens of articles a day based entirely on some kind of survey. A single Google news search of the term poll turned up ten pages of results that carried headlines such as Gallup Poll: 60 Percent Back Obama s Buffett Rule, Poll Shows Sharp Divide on Income Taxes, Poll: Pelosi Least-Liked Member of Congress. Indeed, opinion polls have been the engine driving American politics for nearly two centuries.

Earliest poll

The earliest example of an opinion survey in U.S. history was a straw poll taken byThe Harrisburg Pennsylvanianin 1824, accurately predicting Andrew Jackson s victory over John Quincy Adams for the office of President of the United States. This survey s success was largely responsible for the rising popularity of polling, which has been part of every presidential election since. By the early 20th century, the once casual practice of surveying the political inclinations of tavern patrons or waiting passengers at stagecoach stops and train stations had evolved into something of a science. During the 1916 presidential race, for instance, theLiterary Digestconducted an ambitious national survey by mailing millions of postcards to prospective voters. The poll correctly predicted a win for Woodrow Wilson. TheDigestdid the same to accurately forecast victories for Warren G. Harding in 1920, Calvin Coolidge in 1924, Herbert Hoover in 1928, and Franklin D. Roosevelt in 1932. I think that polling made the leap from art to science in the first half of the 20th century, says Paul J. Lavrakas, president of the American Association for Public Opinion Research. By the 1920s, you begin to see more sophisticated approaches to sampling and formulating questionnaires for gathering data.

Not always accurate

But Lavrakas points out that polling was far from perfect despite theDigest ssterling track record, and the 1936 presidential election was proof positive of this. One of the two biggest failures in polling history occurred during the race between Republican candidate Alf Landon and Democratic incumbent Franklin Roosevelt, he explains. In this election, theLiterary Digestmade the mistake of sampling the electorate using telephone directories and automobile registration lists. This was in the middle of the Great Depression. Those who had cars and phones were relatively wealthy and overwhelmingly Republican in sympathy. Not surprisingly, the Digest incorrectly predicted that Landon would win. The second and, perhaps, best-known failure in political polling occurred in 1948, when veteran pollsters like Gallup and Roper projected that Republican Thomas Dewey would defeat sitting Democratic President Harry Truman by a landslide. Using a method called quota sampling, Gallup tailored its sample population to fit a specific cross-section of the American demographic and, in doing so, skewed the results to the right of the political spectrum. TheChicago Daily Tribuneso trusted the projection that it prematurely released the next day s edition with the banner DEWEY DEFEATS TRUMAN splashed across the front page.

Sampling methods

The 1936 and 1948 elections really underscore the importance of sampling methods to a poll s accuracy, says Lavrakas. And as demographics and technology change, so do these methods. We re constantly facing new challenges. One of them is the huge increase in cell-phone-only households that has occurred in the last decade. Fewer people using landlines means fewer people listed in phonebooks, and this leaves pollsters looking for new ways to collect data. From the cell phone perspective, nearly the entire nation is unlisted, says Lavrakas. To get around this, pollsters use random dialing programs to generate phone numbers for a representative sample. This is a great example of how opinion polling has changed with society and technology. Aside from the Roosevelt and Truman gaffes, pollsters have refined their trade enough that voters and presidents still take them seriously. For voters, the influence is in exit polls. I ve been studying election polling for years, and I haven t found any evidence suggesting that exit polls sway voters on whom they ll vote for, says Lavrakas. They do, however, affect whether or not a person will vote. If they see that they re candidate is projected to win or lose by a sizeable margin, they re more likely to feel as though their vote won t matter. Presidents, on the other hand, focus a great deal on voter approval polls, which, according to Lavrakas, can be valuable capital when the numbers are high. The bottom line, he says, is that we live in a nation of polls. In a democracy, every person s opinion matters. So long as that s the case, pollsters will keep doing their thing.