Polls accurately predict big Obama victory, but prop. predictions more mixed

The election of Barack Obama may once and for all put to rest the notion of a race-based “Bradley effect” – the name given to the phenomenon of poll takers lying to pollsters about supporting a black candidate for fear of appearing racist.

But as pollsters accurately predicted the big Obama victory, the results were more mixed when it came to California ballot initiatives. Polls failed to accurately predict the outcome of a number of initiatives, including a Constitutional ban on gay marriage, a measure that would have required doctors to notify parents of minors seeking an abortion and some environmental and public safety measures.
Perhaps the most notable difference between the polls and Tuesday’s vote came with Proposition 8, in which voters approved a constitutional ban on same-sex marriages. Earlier, the California Supreme Court ruled that gay couples were legally entitled to marry – a decision that overturned an earlier voter-approved ban on gay marriage and triggered the Prop. 8 fight.

Polls did show a massive tightening of the Yes on 8 campaign – the measure initially trailed by wide margins in early surveys. But no poll showed the measure actually passing.

As of Wednesday morning, Prop. 8 appeared to have passed, 52 percent to 48 percent. In a conference call, opponents of Prop. 8 said three to four million absentee and provisional ballots remained to be counted. They face a 428,000 vote deficit.

Opponents of Prop. 8 continued to cling to a faint hope that uncounted ballots would reverse the ballot result. “Depending on turnout model, we’re looking at millions of votes yet to be counted,” said Kate Kendall, executive director of the National Center for Lesbian Rights. “This election is too close to call.”

Even if the No side pulls out an unlikely victory, it is clear they did not get the support polls suggested they could.
In a poll released on Oct. 22, the Public Policy Institute of California showed Prop.8 losing, 52 percent to 44 percent. A Field Poll released on Halloween showed the measure failing, 49 percent to 44 percent.

Polling at less than 50 percent “yes” is usually a death knell for a proposition. But the Yes on 8 campaign had repeatedly claimed they would outperform their polling numbers. This was the case in 2000, when voters passed Prop. 22, a non-constitutional effort to bar gay marriage. Prop. 22 got 61 percent of the vote, eight points better than predicted by a Field Poll released days before that election.
After a Field Poll released on Sept. 18 this year showed Prop. 8 losing 55 percent to 38 percent, Yes on 8 campaign manager Frank Schubert actually cited the Bradley Effect in relation to the race. He said that respondents would want to be “politically correct” and not want to look “intolerant” to pollsters.

The Yes side also touted other polls they claimed were more accurate. On Oct. 21, they announced the results of a poll by the Marist College Institute of Public Opinion between Sept. 28 and Oct. 5. Commissioned by the Knights of Columbus, a conservative Christian group that gave $250,000 to the Yes campaign, it showed the measure passing 52 percent to 43 percent.

A SurveyUSA poll announced on Oct. 6 showed Prop. 8 leading 47 percent to 42 percent. This was an automated poll, in which respondents did not have to speak directly to a pollster.

It is also possible that some voters decided or even changed their minds late in the campaign. With $73 million spent on both sides, voters were hit with numerous conflicting commercials on Prop. 8. The Yes side made a late push to emphasize that young children would be taught gay marriage in schools — a factor that is not directly addressed by anything in the initiative.

Independent groups made a late push on the No side. This included a commercial from the Courage Campaign Issues Committee, widely condemned by religious groups, which showed a pair of Mormon missionaries barging into a lesbian couple’s home and tearing up their marriage certificate.

The Yes and No sides seesawed back and forth in their fund-raising battle, with neither side ever getting a lead of more than a few million dollars. The donations came in right to the very end of the campaign. One donor, Claire Reiss of La Jolla, gave $500,000 to the Yes campaign on Oct. 31.

Meanwhile, the first black president-elect showed little sign of suffering from any Bradley Effect. Obama clobbered Republican John McCain in California, 61 percent to 37 percent. McCain did beat the 35 percent predicted by Electoral-Vote.com, a website the averages several top polling operations, but by a statistically insignificant two points.

An Oct. 30 Field Poll showed Obama leading, 55 percent to 33 percent, while the Oct. 22 PPIC survey showed almost exactly the same, indicating that Obama received just over half of the late-breaking undecideds.

The Bradley Effect refers to the 1982 governor’s race in California in which Los Angeles Mayor Tom Bradley, a Democrat and African-American, was defeated by Republican state Attorney General George Deukmejian, even though pre-election polls showed Bradley leading.

The extent of this effect has been questioned, however, by those who note that Bradley had only a small lead that was narrowing in the days before the election.

If there was a conservative-backed measure that underperformed, it was Proposition 4. This measure to require parental notification before a girl under 18 years old could receive an abortion failed by 4.6 points at the polls.

A Nov. 1 Field Poll showed Prop. 4 leading 45 percent to 43 percent. The Oct. 22 PPIC survey showed it leading 46 percent to 44 percent. The actual election results, however, were in line with the conventional wisdom on polls, which says most measures lose if they don’t have above 50 percent support before election day.