"A government big enough to give you everything you want, is big enough to take away everything you have"
Thomas Jefferson

Tuesday, May 20, 2008

Polling Data

The polling data for the Presidential election this year might be considered suspect through out this election cycle.

The polling data collected and disseminated this year is possibly going to be skewed and it has nothing to do with the pollsters or methodology and everything to do with respondents and what is referred to as the "Bradley effect."

The Bradley effect is a so called tendency of white poll respondents to overstate their support for a black candidate thereby skewing the pre-election polling. The Bradley effect is a documented effect on polling in biracial contests that has occurred in previous election contests. It received its tag after a gubernatorial election in California, which Los Angeles Mayor Tom Bradley, an African American, was leading in polls just days before the election and lost.

The polling anomaly occurred again in the Virginia Governor's race in which L. Douglas Wilder, an African American, barely won despite a large double digit lead. In addition, it has already occurred this election cycle in the New Hampshire primary. Obama led in polls leading up to the primary and was projected to win by 8 points, but lost by 3 points while the undecided voters only went for Clinton by a slight margin, not enough to alter the margin to the extent it changed between polling and outcome.

The Bradley effect was suspected in the Rhode Island primary where Clinton had 9 point lead in the polls prior to the primary and she won by 18 points.

Another factor in forecasting this year's election is the large increase in voter registrations and large projected turn out which will make it difficult to model the actual voters.

Some will doubt and dispute the effect on polling results; nevertheless, it is a phenomenon that has to be considered and will leave suspicion in the minds of those that attempt to predict the results.

No comments:

Live Blogging