
SAN FRANCISCO — California Democrats would take Gavin Newsom over Kamala Harris in a 2028 presidential primary, according to a new poll from POLITICO and its partners.
Between the two high-profile Californians, the governor leads the former vice president 25 percent to 19 percent among the state’s registered Democratic voters and Democratic-leaning independents, the latest POLITICO-Citrin Center-Possibility Lab survey found.
The finding reflects, in part, success Newsom has had asserting himself as a leader of Democrats’ resistance to President Donald Trump’s agenda — from the governor’s plan to counter gerrymandering in Texas to his opposition to immigration raids and lawsuits challenging Trump’s tariffs.
Harris, by contrast, may be older news. Largely avoiding the spotlight since losing the White House to Trump last year, she recently passed on running for California governor. Harris said she doesn’t “want to go back in the system” for now.
“There’s affection for her, but maybe less confidence that she would be a strong candidate,” said Jack Citrin, a veteran political science professor at UC Berkeley and partner on the poll. “(Newsom is) in the news everyday. If you think someone is running, you’re more likely to support them.”
The survey found that 75 percent of registered Democratic voters said they are “excited” about the prospect of Newsom running, compared with 67 percent who are enthusiastic about Harris.
Both Newsom, who is likely to run, and Harris, who is weighing a campaign, would be leading contenders in the California primary, according to the survey. But the more lukewarm Democratic support for Harris in her home state suggests the uphill battle she could face tapping into her party’s base.
Behind Newsom and Harris, the survey had former U.S. Transportation Secretary Pete Buttigieg polling in third place, with the support of 13 percent of Democrats, followed by New York Rep. Alexandria Ocasio-Cortez at 10 percent.
Other potential Democratic contenders — including Minnesota Gov. Tim Walz, New Jersey Sen. Cory Booker, Silicon Valley Rep. Ro Khanna, Pennsylvania Gov. Josh Shapiro and Michigan Gov. Gretchen Whitmer — were all trailing in the single digits.
The survey also looked at the opinion of policy influencers, including a group of roughly 400 Democratic POLITICO Pro subscribers who are deeply versed in the state’s political landscape. They were less enthusiastic about both Harris and Newsom at the top of the ballot than Democratic voters at large.
Among those influencers, only 2 percent said they would support Harris, and 14 percent support said they would back Newsom. Buttigieg led in that survey, with 19 percent support.
Citrin said the gap between the public survey and poll of policy influencers suggests that party insiders are evaluating which Democrat is best positioned to defeat Vice President JD Vance or another Republican in the November 2028 election. The pollster said Newsom and Harris are often seen as more liberal than the rest of the Democratic bench, a problem that has already cropped up for Newsom outside of the confines of his coastal state.
He added, “Getting the nomination is one thing — and then winning is a second.”
The survey consists of two separate opinion studies of the California electorate and policy influencers in the state, fielded by TrueDot, the artificial intelligence-accelerated research platform, in collaboration with the Citrin Center and Possibility Lab at UC Berkeley and POLITICO. The public opinion study, made possible in part with support from the California Constitution Center, was conducted in the field between July 28 and Aug. 12.
The sample of 1,445 total registered voters, including 807 registered Democrats and Democratic-leaning independents, was selected at random by Verasight, with interviews conducted in English and Spanish, and includes an oversample of Hispanic voters. The modeled error estimate for the full sample is plus/minus 2.6 percent. The policy influencer study was conducted from July 30 to Aug. 11, among 512 subscribers to POLITICO Pro, and the modeled error estimate is plus/minus 3.7 percent.
Comments