A new survey from the Public Policy Institute of California finds the races for governor and Senate just about where the Field Poll had them – with Jerry Brown and Meg Whitman in a dead heat for governor and Barbara Boxer leading Carly Fiorina in the race for U.S. Senate.
About the only significant shift PPIC found was a movement among those they identified as independents, who shifted 10 points in favor of Whitman, with Whitman now at 38% and Brown at 30% compared to July when Brown had 30% and Whitman had 28%. More on this odd finding later.*
In the PPIC poll, Brown is only winning 63-13% among Democrats while Whitman is holding 71-10% among Republicans. They are tied among men with Whitman leading among women by 2%. That gender split is at odds with historical patterns wherein the Democrat traditionally trails among men and leads among women.
PPIC also showed Brown leading Whitman just 32-25% among Latinos – a smaller margin than the USC/LA Times poll had (51-32%), but closer to what the Field Poll reported (43-40%). All those Latino numbers, however, were before Nicky Diaz told her story Wednesday about working for Whitman.
According to PPIC, seven in 10 liberals and a plurality of moderates prefer Brown while two-thirds of conservatives favor Whitman.
A couple of interesting crosstabs PPIC ran for Calbuzz that show some fault lines:
Brown voters lean 5-3 against Prop 23, which would suspend California’s law limiting greenhouse gases, while Whitman voters lean 4-3 in favor of the proposition.
On a question that gets to creating a path to citizenship for illegal workers, 61% of likely voters said “most illegal immigrants who have lived and worked in the United States for at least two years . . .? should be given a chance to keep their jobs and eventually apply for legal status” while just 35% said they should be deported back to their native country.”
Those favoring a path to citizenship lean 75-44% for Brown while those for deportation favor Whitman 52-20%. Also, Brown voters favor a path to citizenship over deportation by 46-21% while Whitman voters prefer deportation by 56-28%.
Here’s PPIC’s rundown on the Senate race:
Democratic incumbent Barbara Boxer holds a 7-point lead over Republican Carly Fiorina in the U.S. senate race, with 17 percent of likely voters undecided. In July, the race was closer (39% Boxer, 34% Fiorina, 22% undecided). Today, Democrats (72%) support Boxer at much the same level as they did in July (68%); Republican support for Fiorina is also consistent (72% today, 72% July). Independents are currently divided in their support for Fiorina (34%) and Boxer (32%); in July, independents were somewhat more likely to prefer Boxer (35%) over Fiorina (29%). Boxer receives overwhelming support from liberals (74%) while 66 percent of conservatives favor Fiorina. A plurality of moderates say they will vote for Boxer (46%) rather than Fiorina (25%).
PPIC: Sept. 19-26; 2,004 adults surveyed, including 1,563 registered voters and 1,104 likely voters. Margin of error for likely voters is ±3.6 percent.
Another poll just out:
From betting sitesTime/CNN: “Democrats Barbara Boxer and Jerry Brown have cemented leads over their GOP opponents… Boxer leads Fiorina 52% to 43% among likely voters. That’s a significant improvement from earlier this month when a CNN-TIME-Opinion Research poll found Boxer just edging past Fiorina amongst likely voters 48% to 44%. Likewise in the gubernatorial race, Brown leads former eBay CEO Meg Whitman 52% to 43% among likely voters, a reversal of fortunes for Brown who earlier this month was losing to Whitman 46% to 48% in a poll conducted Sept. 2-7. Brown and Boxer both benefit from moderates breaking for them: 59% for Boxer to Fiorina’s 32% and 59% for Brown to Whitman’s 36%…. 786 likely voters… margin of error of plus or minus 3.5%.”
Footnote for polling weedwhackers
*This would be a significant movement if it were clear that those defined as independents in the PPIC poll really are independents. But it’s not. Like many pollsters nationwide, PPIC uses random digit dialing (RDD) to sample the adult population of California and then, with a series of questions, identifies Democrats, Republicans and independents and from them, using other questions, PPIC culls a sample of likely voters.
PPIC’s total sample, based on respondents’ answers, was 45% Democrats, 31% Republicans and 23% independents – close to registration but a bit high on independents. Their likely voter sample (which was not in their public release) was 45% Democrats, 36% Republicans and 18% independents – very similar to the proportions other public pollsters are using.
Other pollsters working in California politics have, for the sake of certainty and cost, moved to using the Secretary of State’s voter list from which to draw a sample of actual registered voters and then, using their actual voting history (and sometimes supplemental questions) determine who should be counted as a likely voter. The voter list does not include unlisted phone numbers or numbers for people who chose not to list one when they registered to vote. But it does include actual registered voters and their cell phone number if that’s what they listed when they registered.
RDD sampling, on the other hand, has the advantage of ensuring that every residential phone number in California, listed or not, has an equal chance of being included in the survey. But it relies on people’s responses to determine what party they’re in and if they’re likely to vote. So someone who is a Democrat but somewhat pissed off at the Democrats might tell a pollster he’s an independent. Or an independent might say she’s a Democrat. There’s no way to really know what party, if any, they’re registered in and if they’re really a likely voter. (Moreover, pollster have to supplement with random cellphone calls for which actual home residency can be tricky.)
Mark Baldassare, who runs PPIC’s polling, is very good at what he does. And his findings are extremely close to what the Field Poll found and not that far from what the LA Times/USC survey found. But a 10-point movement among independents is an odd finding that seems hard to explain from the post-Labor Day course of the campaign.
It’s possible that this movement is a function of how likely voters are defined in the survey. PPIC’s likely voter screen includes native and foreign born US citizens who say they are registered to vote and who say they always or nearly always vote. They must also say they have a great deal or fair amount of interest in politics and have at least some college education and have lived at their current residence up to five years OR they describe their interest in politics as only a little but have lived at their current residence for five years or more. In the months before an election PPIC also uses voters’ professed intention to vote and their measure of interest in politics to winnow out unlikely voters.
Including people who are registered Decline to State and including them only if they’ve voted in previous elections sure would be more straightforward.