Spring Break
The only important poll is the election itself
Michael Harte
Issue date: 10/24/08 Section: Ed-Op
| |
|
Every significant news media outlet bases its predictions on Gallup and other polls as if the information they provide is a guaranteed foretelling of the future.
Gallup boasts accurate polls by using what they like to call "likely voter models" to point out periodically the difference between their general registered voter polling and what they believe is a more accurate model of Election Day outcome. While their methods have been refined over decades, there is no way to be sure that their results are not affected by outside bias or confounding variables. For example, Gallup states that since "Barack Obama is the first major-party black candidate in U.S. history, it would not be unexpected to see unusually high turnout among black voters."
This is an incredibly biased assumption. Gallup assumes that more black people will vote simply because we have a black presidential candidate. They do not assume, however, that more women will vote due to the female vice-presidential candidate. It appears that Gallup's seemingly accurate voter model has some significant flaws.
Additionally, Gallup and other polling services have no way to take confounding variables into account when conducting their stratified random samples. The possibility of last minute statements about global events by candidates can often sway the vote outside of the prediction, and statements made around the time of polling can distort voters' actual preferences. The process of conducting stratified random samples takes large populations and divides them into homogeneous sub-groups and then asks individuals questions based on the sub-groups the individuals have been associated with. This method of sampling is subject to bias and other influences.
Ideally, Gallup would use simple random sampling when conducting their polls; however, a simple random sample of the entire United States would be very difficult to conduct, made even more difficult by the frequency at which Gallup conducts their polls.
History can show us that Gallup polls are not always accurate. In 1980, Gallup polls in late October were showing incumbent Jimmy Carter with a significant lead over former California Governor Ronald Reagan. Just ten days later, Reagan defeated Carter by a significant margin.
Even if polls had demonstrated accuracy in the past, Gallup themselves admit that this election will likely be different than any other in recent history. The ability of the Obama campaign to register and mobilize groups which traditionally have low voter turnout may have a significant impact on the results of the election, and certainly on Gallup's definition of a "likely voter."
Both major candidates seem equally as unsure about poll results, with Obama urging his supporters not to get "cocky" and McCain promising Pennsylvania voters he will turn the polls around by Election Day. Given that both major candidates express their disapproval of Gallup polls in this way, it is safe to assume that the polls are in no way accurate indicators of the election outcome.
With the media coverage that this election is receiving, it is important not to confuse voters by telling them the polls say one thing over another. Since polls can be biased, influenced by confounding variables and historically prone to inaccuracy, it is unfair to voters for the media to portray poll results as concrete predictions. Both candidates' uncertainty should show the American people that they should not put all of their faith in the polls. The constant fluidity of the polls will continue to change up until Election Day, and even then they are not always accurate - the candidates, the evidence and the history agree.
Michael Harte is a freshman majoring in marketing. He can be reached at op-ed@thetriangle.org.
Be the first to comment on this story