Ms. Lake, the Democratic pollster, said Republicans could find it difficult to attack the wealth tax proposal in particular, because it appealed to voters’ values and emotions.
“It doesn’t require you to do math to figure it out,” she said.
Ms. Duppler, though, pointed to an internal Democratic math debate over the efficacy of the tax, noting criticisms from former Treasury Secretary Lawrence Summers over how much revenue it would raise to fund programs such as free college.
“Republicans are wise to let Democrats fight amongst themselves on this,” she said.
There are signs that the policies do not poll as a bloc — some voters like only a sampling of what the Democrats are offering.
Lonnie Shumate is a Marine veteran who works as a manager for an oil-change chain in western North Carolina. He earns less than $40,000 a year and lives mostly paycheck to paycheck. The American middle class, he said, is dying out. He considers himself conservative on economic issues, and generally votes for Republicans.
Yet Mr. Shumate, 35, said he was open to Medicare for all and noted that government-run health care systems worked in Europe. And he said he liked the idea of offering free college tuition, at least for two-year programs.
“I don’t mind the free college idea because college is way too expensive,” he said.
But while Mr. Shumate said he was concerned about inequality, he draws the line at the wealth tax. Some — but not all — of the assets subject to the tax would have already been taxed as income, which he said would not be fair.
“I don’t believe in taxing somebody on money that they’ve earned twice,” he said. “There are plenty of other places they could cut money from.”
About the survey: The data in this article came from an online survey of 2,662 adults conducted by the polling firm SurveyMonkey from July 1 to July 7. The company selected respondents at random from the nearly three million people who take surveys on its platform each day. Responses were weighted to match the demographic profile of the population of the United States. The survey has a modeled error estimate (similar to a margin of error in a standard telephone poll) of plus or minus three percentage points, so differences of less than that amount are statistically insignificant.