Franklin & Marshall College Franklin & Marshall College

    • u-h-419733c4a47a-jpg
    • u-h-9af179e6f4f3-jpg
    • u-h-c8018e0f0ae5-jpg

Is the Bradley Effect a Myth?

October 30, 2008

by Dr. G. Terry Madonna and Dr. Michael Young

Is it bona fide or is it bogus? No question has been asked more frequently nor debated more fiercely during the latter stages of this presidential election than the so called Bradley effect – the “theory” that black candidates under perform in elections when compared to pre-election polls in part because some white voters will lie to pollsters and indicate they will vote for an African American candidate when in reality they will not.

The Bradley effect was named for the former black Los Angeles Mayor Tom Bradley after polls showed him far ahead in a 1982 gubernatorial race he subsequently lost. It was similarly noted in several high profile races with black candidates running against white opponents including Douglas Wilder’s 1989 Virginia gubernatorial race and David Dinkin’s 1989 mayoralty race in New York.

Those who subscribe to the notion of a Bradley effect believe that some white voters mislead pollsters about their real voting intentions by saying they are undecided or intend to vote for the black candidate. Then in the privacy of the voting booth they vote their prejudices.

The questions raised by the Bradley effect as election day approaches are critical ones. Assuming that Obama is comfortably ahead in the polls, are those polls reliable; is Obama’s lead large enough to overcome the Bradley effect; and most crucially, is it large enough in the close battleground states where the election will be won or lost?

In no state have these questions been raised more often than in Pennsylvania. There are three main reasons that Bradley is being discussed in Pennsylvania: one, it fits stereotypes of the Pennsylvania voter as redneck and backwards; two, Pennsylvania is seen as the quintessential battleground state that may decide the election; and three, predictions of a Bradley effect election have been fed by recent controversial remarks from two leading state political figures.

The latter note deserves some elaboration. Two of the state’s most important Democratic office holders have raised the specter of Bradley effect voting in the Commonwealth. First, Governor Ed Rendell, when campaigning for Senator Clinton, publicly raised the issue by suggesting that perhaps as many as five percent of the state’s voters were Bradley effect voters. Rendell insisted that he would have done five percent worse in his 2006 reelection had his opponent been white instead of black.

And the issue has been renewed even more vividly and provocatively by Congressman Jack Murtha’s recent comments that folks in western Pennsylvania are racists.

Searchers for evidence of the Bradley effect in Pennsylvania argue it will be found in two of the state’s key voting groups: working class Democrats and rural white Republicans. Both groups require brief description.

  • Working class Democrats – The first of these are culturally conservative, Democratic, blue collar working class, Catholic voters – many living in the southwestern part of the state or in the Scranton and Wilkes Barre region in the northeast. They voted for Hillary Clinton in the heavily contested seven week primary against Barack Obama.
  • Rural white Republicans – The second group consists of conservative, rural, white Republican voters living in the middle of the state. This is Pennsylvania’s celebrated "T", made infamous by James Carville in his widely reported assertion characterizing Pennsylvania as Philadelphia, Pittsburgh, and Alabama in between.

So the chain of reasoning is clear enough as is its conclusion: some significant numbers of Pennsylvania voters are going to be casting racially motivated ballots. And it could determine the election.

But not clear is what evidence really exists to support such a contention. Are respected politicians like Rendell and Murtha right? Will there be a Bradley effect in Pennsylvania’s vote November 4th?

We argue that they are not right, that there is no evidence to support their conclusion, and that the so called Bradley effect in Pennsylvania is largely a phantom fear without substance or effect. Moreover we believe Obama is as likely to out-perform as he is to under-perform his polls on Election Day.

Congressman Murtha’s assertions seem easiest to dismiss. They are based on no real evidence and he links them to anecdote and his "experiences." In fact, he has backed away from them now and seems satisfied to relegate his constituents to the status of redneck rather than racist.

Rendell’s conclusion fares no better. In 2006 he had as his opponent an African American and former football legend, Pittsburgh Steeler wide receiver Lynn Swann. Rendell won by 20 percentage points, while the Real Clear Politics final average had him ahead by 21.8 percent. No Bradley effect apparent there. Indeed if anyone was "Bradlied" it was Rendell himself.

The 2006 network exit poll in the Rendell/Swann Pennsylvania contest offers some additional evidence against the Bradley effect. Regionally and demographically Swann did very well with those very groups most suspect as Bradley effect voters. In Carville’s "Alabama," Swann won small city and rural Pennsylvania 52% to 47%, representing about 16% of the state’s voters. And he won the born-again or evangelical Christian segments of the electorate 60% to 40%.

Moreover in southwestern Pennsylvania – widely touted as a hotbed of Bradley effect voters – Lynn Swann consistently outpaced the vote for fellow Republican, white U.S. Senator Rick Santorum, among the blue collar working class electorate there. In fact, of the seven counties in the region, only in Westmoreland did Santorum do better than Swann and then only by a single percentage point.

Finally, the broader national analysis done on the Bradley effect has concluded it is largely a myth or at worst a real once but now defunct phenomenon. One recent Harvard study of 133 elections from 1989 to 2006 showed some Bradley effect (about 3 points) until 1996 – but thereafter none or perhaps a slight advantage to black candidates.

A more recent analysis of five state elections involving black candidates shows that African Americans did better than the polls suggested in two states and worse in three.

Summing up, it is difficult if not impossible to make the case today that any significant Bradley effect is likely for this election. Based on what we have seen in earlier Pennsylvania elections, together with national studies on the question, it is as likely that Obama will out-perform the polls as under-perform them. This is not good news for John McCain, but it is good news for those who worry that a new round of virulent racism looms in America’s electoral politics.

Politically Uncorrected™ is published twice monthly. Dr. G. Terry Madonna is a Professor of Public Affairs at Franklin & Marshall College, and Dr. Michael Young is a former Professor of Politics and Public Affairs at Penn State University and Managing Partner at Michael Young Strategic Research. The opinions expressed in this article are solely those of the authors and do not necessarily reflect the opinions of any institution or organization with which they are affiliated. This article may be used in whole or part only with appropriate attribution. Copyright © 2008 Terry Madonna and Michael Young.