Presidential Polling Inaccuracies
Are the Problems of Accuracy in Political Polls Only in the Past?
History has shown that political polls are not consistently accurate. A dramatic example is the presidential election of 1948, when the top pollsters in the U.S. mistakenly estimated that Thomas Dewey would defeat Harry Truman. They predicted victory for Dewey with a 5 to 15 percentage point margin, but Truman eeked out a 4.4 percentage point margin of victory. But, are those problems relegated to the past, or have polls been fixed within the past sixty years?
The Haunting Problem of Inaccuracy
Today, Zogby International is recognized as one of the hottest polling enterprises. In 1996, John Zogby became nationally known by precisely predicting Bill Clinton’s margin of victory in the presidential election. However, he made a serious error. In the presidential election of 2004, he predicted that George W. Bush could win a slim popular vote, but John Kerry would significantly win the electoral college vote.
This is from the man who came within one-tenth of one percentage point of the actual outcome of the 1996 election. It was heralded as the best performance by any contemporary pollster and earned Zogby the respected title of: the "prince of pollsters" and "America's Pollmaster General." The Christian Science Monitor even proclaimed him as "Champion Pollster."
How could he have been wrong? A few days after the election, Zogby International posted a statement on their Web site: "We feel strongly that our pre-election polls were accurate on virtually every state… I thought we captured a trend, but apparently that result didn’t materialize.” Like Gallup’s gaffe in the 1948 election, this prediction was an embarrassment for Zogby and his organization. However, polling organizations are alive and well and pumping out public opinion percentages for the populace.
Refinements in Polling
Understandably, most advances in polling and sampling have been shaped by academics. After the Truman victory in 1948, Dr. Gallup was stunned but not defeated. He is quoted as stating: "We are continually experimenting and continually learning."
From 1948, the technique of “quota sampling” was trashed as an accurate method of measuring public opinion. In the very next presidential election, the pollsters used the newer technique of “modified probability sampling” or “random sampling” which incorporates a range of factors or variables.
Most modern pollsters currently use random sampling which provides an equal chance for individuals being selected to answer the surveys. This ensures obtaining a more representative proportion of the population. With quota sampling, which is equally scientific, there is a deliberate attempt to select specific percentages of various “types’ of people within a given population.
Factoring out Errors
According to George Gallup, Jr., a better effort in utilizing this method of polling could have been useful in the 1960 election, when John F. Kennedy ran against Richard Nixon. Even though the Gallup organization had been using random sampling methods, they did not factor in religion as a variable and this could have thrown off their results. If Zogby used random sampling in 2004, one wonders what factors he did not include.
Apparently, public opinion polling can provide accurate predictions within limitations. Although polling is a productive instrument in obtaining the opinions of the populace, it is difficult to factor out errors. Poorly worded questions and unrepresentative samples continually skew polls results. It is importatnt that the validity of survey results be questioned and poll watchers need to always be attentive to the source of the poll and to the methods used in polling.
Source:
Moore, David W. The Superpollsters: How They Measure and Manipulate Public Opinion in America. New York: Four Walls Eight Windows, 1995
More in History
Latest Articles