Morgan Griffith: Some perspective on public opinion polls
News stories during an election season often focus on the ups and downs of opinion polling. As a decision nears, polls become central to reporting and analysis, and often affect the enthusiasm of the various candidates’ supporters and opponents.
In a republic based on democratic principles, opinion polls occupy a meaningful role in discerning public sentiment. It would be a mistake, however, to let them occupy a preeminent role.
Admittedly, they have come a long way from their predecessors, the “straw polls.” These surveys supposedly got their name from the farming method of telling which way the wind blew by throwing a handful of straw into the air.
The first published presidential straw poll in the United States appeared in the Harrisburg Pennsylvanian ahead of the 1824 election. It found that the residents of Newark and Wilmington, Delaware it had surveyed preferred Andrew Jackson over John Quincy Adams; Jackson did win the popular vote that year, but Adams became president when no candidate earned a majority in the Electoral College and the House of Representatives decided the election.
Over the succeeding decades, polling methods became more sophisticated, but there were still notable missed predictions.
Among the most famous occurred in 1936. The magazine Literary Digest had conducted several presidential election polls by that year and had previously predicted the result correctly. It mailed out postcards asking recipients to respond with their preference for the Democrat incumbent, President Franklin Roosevelt, or his Republican challenger, Kansas Governor Alf Landon.
Based on the survey results, Literary Digest predicted a victory for Landon with 57 percent of the popular vote and 370 electoral votes. That November, Roosevelt won in a landslide with 60.8 percent of the popular vote and 523 electoral votes. Landon carried 36.5 percent of the popular vote and a mere eight electoral votes from two states, Maine and Vermont.
Notably, the polling method of newcomer George Gallup accurately predicted a Roosevelt win. The Gallup poll surveys Americans to this day; Literary Digest did not survive the decade.
Another famous mismatch between poll predictions and election results occurred in 1948. Incumbent Democrat President Harry S. Truman, who had succeeded Roosevelt upon his death in 1945, was seen as a sure underdog to Republican New York Governor Thomas Dewey. Most polls, including Gallup, had Dewey ahead in the race, but Truman won on Election Day.
A famous photograph shows Truman holding up a copy of the Chicago Daily Tribune with the erroneous headline “Dewey Defeats Truman.” The newspaper’s headline was in line with its polling of the race. Afterward, the paper recognized that its poll results were off because it had surveyed a disproportionately Republican sample; the poll was conducted by telephone, and Republicans at the time tended to be wealthier and were more likely to have a telephone.
These examples should not be cited to dismiss opinion polling out of hand. Pollsters have been right, too, as in the victories of Lyndon Johnson in 1964 and Richard Nixon in 1972. They do illustrate the importance of not investing excessive meaning in opinion polls.
Honest polling requires calculation about sampling to properly represent the population, and sometimes parts of the population are difficult to reach. It needs careful wording to not favor outcomes. Even when these criteria are met, the results come with a margin of error that could be meaningful on closely divided questions.
The extraordinary events of this year will also affect the accuracy of polls. The presidential election this year includes an unprecedented amount of early voting, as many states have lengthened the period before Election Day to cast ballots. Polling closer to the election has to account for the fact that more people who answered may have already voted.
My advice would be to read the results of opinion polls but not to let them crowd out other news and information. Placing too much attention on them may lead a person to miss other signals that indicate what is taking place in our country.
Lastly, no matter who you intend to vote for, do not let polling data dissuade you from casting your vote.
If you have questions, concerns, or comments, feel free to contact my office. You can call my Abingdon office at 276-525-1405 or my Christiansburg office at 540-381-5671. To reach my office via email, please visit my website at www.morgangriffith.house.gov. Also on my website is the latest material from my office, including information on votes recently taken on the floor of the House of Representatives.
Morgan Griffith represents the Ninth District in the House of Representatives.