Newswise — During this year's Presidential campaign, many political experts expressed concerns whether polling data would accurately predict the actual voting outcome of the recent Presidential election.
The uncertainty of race being interjected into politics has long been an issue for candidates and pollsters. Many have suggested that white voters would tell pollsters their intent to vote with one candidate and changed their mind once they entered the voting booth.
This concept had even been given a name — Bradley Effect. It was named after Tom Bradley, the former African-American mayor of Los Angeles who ran for governor of California in 1982. Surveys conducted after voters had cast their ballots showed him leading the race by a large margin. However, when the votes were counted, he lost to Republican George Deukmejian.
That election established in many pundit's minds that some voters were secretly prejudiced but were embarrassed to tell pollsters.
The election of Barack Obama dispelled the Bradley Effect, at least in this election, according to Wes Colley, a researcher at The University of Alabama in Huntsville.
"The polls were right," said Colley, who works in the university's Center for Modeling, Simulation and Analysis. "The answer in my mind is there was almost no Bradley effect. There was no last-minute shift from polling data to actual voting. It does not appear there was a racial component."
Dr. Colley is a scientist but not a political scientist. He has not taken a single course in political science. Actually, he has little interest in political philosophy. But, it is his research in mathematics and statistics where he has devised models to predict presidential races and rank college football teams.
Colley and his graduate thesis adviser, J. Richard Gott III at Princeton University, have devised what they describe as a very simple, but surprisingly effective means to predict the outcome of the U.S. Presidential election — median statistics.
Their methodology uses the margins of victory for each candidate in each poll during the past month. Those margins are then ranked from the largest margin to the smallest. The middle number, or the median, is used as the candidate's score for the individual states.
He said the method appears particularly well adapted to U.S. Presidential elections where the candidates are chosen well in advance, and where outcomes in individual states determine the winner.
A first attempt was made to predict the 2004 election. Their method correctly predicted 49 states, missing only Hawaii that year.
This year, the Colley Rankings missed three states as a result of a lack of polling data in North Dakota and the extremely close polling results for Indiana and Missouri.
Despite the three misses, their statistical model was extremely close on predicting the Electoral College vote. The day of the election, their model revealed that Obama would get 367 votes. The latest count revealed that he won 364 electoral votes.
Colley is nationally known for his use of a computer modeling system in predicting college football rankings. Colley's system is one of six that is used by the NCAA to determine rankings for the Bowl Championship Series poll.
RSS