Thursday, November 3, 2016

Bradley Effect for Trump?

In 1982, the popular African American mayor of Los Angeles, Tom Bradley, was running for Governor of California.  In the month prior to the election, polls were showing him ahead of his Republican opponent by 5%.  When the votes were counted he lost by 1% of the vote.  The exit polls predicted that he would win.  Looking for an explanation as to why this was so social scientists theorized that a small segment of mostly white voters did not want to admit to pollsters that they did not want to admit to pollsters that they did not want to vote for an African American.  They termed this phenomenon The Bradley Effect.  Similar polling discrepancies were seen in large races with African American Candidates.  This effect was not seen in the presidential elections of Barack Obama in 2008 and 2012 and in other more recent elections with African American Candidates.  The polls were accurate both times.  

For this election cycle, it has been speculated that a similar effect may happen in elections where there is no African American on the ballot.  With Donald Trump warning of a rigged election and a "Reverse Bradley Effect."  Some political bloggers and talk show hosts have fanned the flames of this theory. 

Blogger and right wing talk show host Douglas V. Gibbs has fanned the flames of this theory at his blog The Political Pistachio.  He states that he has spoken with Trump supporters who are afraid to put Trump bumper stickers on their cars for fear of vandalism.  This is the only evidence he presents in support of his theory.  He does claim that there is evidence from the Podesta emails that Democrats are over sampled in polls and that polls are only targeting likely voters but these are issues with polling and sampling methodology not with the survey respondents as in the Bradley effect.  There is little evidence of this effect in the primaries when Trump won handily.  
The Gap Between Poll Numbers and Election Results for African American Candidates Over Time

There were fears among Democrats that the Bradley effect would come into play in the 2008.  As I stated before that didn't happen.  The effect is thought by some to be a thing of the past which may or may not be true.  The effect may be minimized with more advanced polling methodologies such as online surveys or phone surveys with automated rather than live questioners.  It may be more comfortable for respondents to speak openly to a machine than to a human being.

Of course it may be just wishful thinking on the part of Trump and Gibbs for a reverse Bradley effect.  No one really knows until the votes are counted.  The general election is a different beast than the primaries.  I haven't heard any discussion of this effect when there is a woman on the ballot.  The reverse Bradley effect refers to voters not being willing to tell pollsters that they support a candidate who is openly supported by white supremacist leaders like David Duke.  Time will tell.

**Related Posts**

More Hate Groups in States Where Trump and Clinton Win (and in DC Where He Lost)


SPLC Hate Group Update: Washington, DC has an Increase in Activity


A Wave of Hate Groups in California? No in Washington, DC


Terrorism Boost for Trump