The Swami is Gone; Can We Kill the “Bradley Effect”?
In some of the tributes to our late friend and mentor Mervin Field (we name no names), the old, discredited “Bradley Effect” raised its ugly head, as writers sought to give context to Field’s most spectacular blunder as a pollster: calling the 1982 governor’s race for Tom Bradley over George Deukmejian.
As every California school child knows, Bradley, the black Democratic former mayor of Los Angeles held a slight lead over former Republican Attorney General Deukmejian in the final Field Poll before the vote for governor in November 1982.
But in the end, Deukmejian squeaked by with 49.28% of the vote compared to 48.09% for Bradley – fewer than 100,000 votes out of 7.8 millions votes cast.
“Bradley Win Projected,” cried the San Francisco Chronicle’s first edition – a bulldog run of 60,000 then shipped to some outlying, conservative counties back when newspapers were willing to lose money to build statewide circulation. This, of course, was based on Field’s confident prediction on TV on election night that Bradley would win.
Which he did. In the ballots cast in precincts on election day, upon which Field had relied for both his final survey and his election-day exit poll.
What Field had not seen was that the Republicans and especially the National Rifle Association – which had spent $5 million fighting Proposition 15, a handgun registration measure on the ballot — had racked up big early absentee votes that were already in the can but hidden from plain view. (Prop. 15 lost 63-37%.)
After the election, but before it was understood what had happened, Field himself had suggested “race was a factor in the Bradley loss,” which it no doubt was. But there was no actual evidence to support what some analysts began calling the “Bradley Effect” – the false belief that voters had lied to pollsters before the election because they didn’t want to appear racist when being surveyed.
Lance Tarrance, Deukmejian’s Republican pollster, would later claim that his firm had the race much closer than the 7 points that Field had found in his final pre-election poll. But even Tarrance showed Bradley ahead 45-44% in his final tracking poll.
After Field — known among friends as “the Swami” — saw how he had failed to account for votes already cast, his polls from then on asked voters for whom they were going to vote or for whom they had already voted. And the Field Poll never made the same mistake again.
But the damn “Bradley Effect” concept refuses to die. It’s even in the Wiki on the 1982 election. It’s like a roach that refuses to be crushed underfoot.
Crunch.
Actually, the Field Poll was correct for polling of those that voted at the polls. On Election day, Bradley beat Deukmejian by 19,886 votes ( I still have the numbers). 93.5% of the ballots cast in that Governor race were cast at the polls and that is the voters that The Field Poll interviewed as they left the polls. With the 6.5% who voted by mail, Deukmejian beat Bradley by 113,231 (61.5% to 38.5%), thus, giving Deukmejian an overall victory margin of 93,345 votes. In the previous Governor election in 1978 it was only 4.4% that voted by mail so in 4 years it jumped 47.7%. VBM was 6.3% in the 1980 presidential election so that should of been a heads up.
That’s what the post says, Bob.
Yes, but I stated the actual numbers and a margin of only 19,886 on election day was not enough for the Field Poll to predict a Bradley win, so I believe there was a little “Bradley Effect” in voters answering the poll question outside the voting booths. And if the Field Poll had polled the mail voters, even with a “Bradley Effect” Field would have concluded that Deukmejian had won.