Abstract
Political polling was a hot topic in 2008 with an extended presidential primary election season and a historical presidential election in November. This research examines the quality of political polling during the presidential primary election season in an effort to predict which political polls would be most accurate in the general presidential election. This analysis was done using thousands of political polls from across the United States that were released to the public from a variety of sources. The statewide primary election polls themselves differed in a variety of ways and the data was stratified by many factors including: which polling company conducted the poll; was the state a "blue" state or "red" state; whether a caucus or an election was performed; was the primary an "open" or "closed" primary; how far prior to the election did the poll take place; what month the primary election took place. The research examined if the poll was accurate in whether the correct winner was predicted and whether the actual electoral results were within the margin of error of the poll. The results of our analysis showed that polls were statistically inaccurate in most of the "deep south" in primary elections as candidate Obama's support was underestimated when compared with candidate Clinton. But, these polling companies corrected themselves in time for the general election. The data also showed that time was a major factor in these polls as polls conducted even three days before an election were obsolete in terms of prediction.
Faculty Sponsors
Dr. Jason Gershman
Project Type
Event
Location
Alvin Sherman Library
Start Date
4-3-2009 12:00 AM
End Date
4-3-2009 12:00 AM
Math and Politics: Analyzing Political Polling in 2008
Alvin Sherman Library
Political polling was a hot topic in 2008 with an extended presidential primary election season and a historical presidential election in November. This research examines the quality of political polling during the presidential primary election season in an effort to predict which political polls would be most accurate in the general presidential election. This analysis was done using thousands of political polls from across the United States that were released to the public from a variety of sources. The statewide primary election polls themselves differed in a variety of ways and the data was stratified by many factors including: which polling company conducted the poll; was the state a "blue" state or "red" state; whether a caucus or an election was performed; was the primary an "open" or "closed" primary; how far prior to the election did the poll take place; what month the primary election took place. The research examined if the poll was accurate in whether the correct winner was predicted and whether the actual electoral results were within the margin of error of the poll. The results of our analysis showed that polls were statistically inaccurate in most of the "deep south" in primary elections as candidate Obama's support was underestimated when compared with candidate Clinton. But, these polling companies corrected themselves in time for the general election. The data also showed that time was a major factor in these polls as polls conducted even three days before an election were obsolete in terms of prediction.
