Users    RSS Feed    Tags    Media & Public Appearances    Partners    Contact Us   Contribute    About Us   

As many readers will be aware, the "Bradley Effect" is a term that indicates errors in polling that occur when those being polled have an unconscious inclination to say something they don't mean.  It is named for former Los Angeles Mayor Tom Bradley, a prominent black political figure, who narrowly lost a 1982 gubernatorial race in California despite sizable leads in the polls throughout the campaign.  The logic followed that nobody wanted to appear to be racist--not even to an anonymous phone bank interviewer gathering data for the poll. 

The Wall Street Journal provides a pretty thorough analysis of what the Bradley Effect is, and how it might be playing a role in this election. Kate Zernike of the New York Times has also recently taken on this idea of how race, and how people perceive themselves on the issue of race, might be impacting the results of so many polls.

It might be true, in this election, that the Bradley Effect is counter-balanced by the failure of polls to reflect young voters, who we have every reason to expect will be turning out in record numbers for this presidential election.  Under current polling methods, the responses of young people are typically weighted less heavily than responses for older demographics because of the historically-based yet no longer accurate expectation that the 18-34 segment of the population will provide the lowest portion of the overall electorate.  Mark Blumenthal of National Journal discusses the idea of age weighting in the current polls, and what impact that might have on Obama's prospects.  A more math-geeky analysis is provided by Nate Silver on his statistics-analysis website, www.fivethirtyeight.com. 

Beyond just the matter of how polls are weighted within certain age groups, is the possibility that cell phone-only households are being left out entirely.  A Pew Research study breaks down this unmeasured voting block--cell phone users.  As one might expect, cell-phone only household are predominantly young people who, according to the Pew Research study, are supporting Obama over McCain at a 59% to 32% margin.  What's more is that young people who only use cell phones demonstrate an even greater disparity of 62% to 27%.  The omission of cell phone-only households is clearly leaning the polls toward McCain, though Pew concludes that the total number of such voters is still small enough that the distortion should not be severe.  Although there are others who actually think its a pretty big deal.

Yet blogs are still buzzing about the potential of a reverse Bradley Effect, where, wishing to appear unmotivated by racial considerations, or motivated by localized social pressure to support only "white" candidates, whites might under-report their likelihood of voting for the black candidate. At the University of Washington, political scientist Bethany Albertson and psychologist Anthony Greenwald have presented data analysis indicating that throughout the primary election, Obama experienced a reverse Bradley Effect.

It is also perhaps true that, for a variety of reasons, the Bradley Effect is decreasingly important as time goes by, and other issues emerge as undeniable considerations for respondents of polls.  In fact, Daniel Hopkins of Harvard argues in a report that the Bradley Effect hasn't been significant in any election since the middle of the 1990's.

In the end, the idea that the Bradley Effect could account for a 3-7% discrepancy between the polls and Election Day results, as posited by commentators on Fox and elsewhere, does not seem credible, especially when compared with countervailing pressures on polling accuracy such as under-polling, poll weighting, and reverse effects.

Areas of Focus:

Democratizing Elections, Polls & Media Coverage (No More Stolen Elections)

Tags:

---

---

User Comments

No Comments.

Please login at the top of the page or register as a Democracy Square member if you would like to comment.