The conventional wisdom about the special election is that the unions won
and the governor lost. But the pollsters also won: They predicted the
defeats of most of the propositions with a high degree of accuracy, despite
criticism throughout the campaign from the governor’s team.
Of four major polling organizations–Field Poll, Los Angeles Times, Public
Policy Institute of California, and Hoover Institution/Knowledge
Networks–none was wrong on more than one proposition. Field predicted the
defeat of all eight, and was almost dead on in half the races.
“It’s really hard to poll on initiatives,” said Mark Blumenthal, a
Washington, D.C,-based consultant and author of the Mystery Pollster blog.
“It’s hard to predict turnout.”
These difficulties are magnified in a proposition-only election, Blumenthal
said. Initiatives can be hard to understand–they are sometimes intentionally
written that way–leading even people who bother to make the trip to the
polls to cast votes on some initiatives but skip others.
Even so, he said, the major telephone polling organizations–Field, PPIC and
LA Times–had all pretty much predicted the outcomes two months before the
election, despite a large number of undecided voters.
Republicans did get some ammunition at the end of October, when the second
of three Hoover/Knowledge Network polls showed slights leads for
Propositions 74 (teacher tenure) and 77 (redistricting), and a daunting 64
percent-to-36 percent lead for Proposition 75 (union dues).
On Nov. 7, however, Hoover put out a third poll. This one showed only a tiny
lead for Prop. 75, with all others trailing. It even showed higher levels of
support for the two Democratic-supported offerings–Propositions 79 (consumer
drug prices) and 80 (electric service regulation–than several other polls.
What changed? According to Morris Fiorina, a political science professor at
Stanford University and a senior fellow at Hoover, the second poll got an
unusual sample. The online poll sampled for likely voters. In that batch, he
said, Republican respondents were far more likely to indicate they were
likely voters, raising the possibility that many Democrats might not vote.
“You can’t reject your data,” Fiorina said “We thought it was kind of
strange, but we though maybe Republicans had fallen in behind the governor.”
Blumenthal said there were a lot of undecideds until late in the
election–but they coalesced into “no” voters. This quickly made a mockery of
the 44 percent to 34 percent Republican advantage among likely voters
indicated in the second Hoover poll.
“The idea that likely voters in California were net GOP is pretty
far-fetched,” Blumenthal said.
Mark DiCamillo, director of the Field Poll, said the large number of
undecideds on some propositions made life hard for pollsters. In one October
poll, only 36 percent of respondents had an opinion on Proposition 80–one of
the lowest levels of awareness around a proposition that Field had found in
60 years of polling.
On the other hand, people showed an above-average awareness of the
Governor’s four propositions–74 through 77–DiCamillio said. This turned out
to be bad news for the governor, and maybe for the other four measures on
the ballot as well, without opposition to all hardening in the final days.
“Many voters were voting in a block against them,” DiCamillo said. “They had
already made up their minds by the time we called.”
It became clear that the writing was on the wall for the governor in the
final weeks, said Mark Baldassare, director of research at the PPIC.
Schwarzenegger repeatedly made efforts aimed at bucking up the poll numbers,
he said: announcing his reelection campaign, holding town hall meetings,
finally the television commercials where he humbly admitted his mistakes. In
each case, the polls didn’t budge or actually got worse.
“It didn’t matter what he did, the basic results didn’t change in his
favor,” Baldassare said. “That really said it all.”