Age and race, meanwhile, were not as strong factors as assumed. According to the poll, 56 percent of voters over age 55 and 57 percent of nonwhite voters cast a yes ballot for the gay marriage ban.
People who identified themselves as practicing Christians were highly likely to support the constitutional amendment, with 85 percent of evangelical Christians, 66 percent of Protestants and 60 percent of Roman Catholics favoring it.
The poll also showed that the measure got strong backing from voters who did not attend college (69 percent), voters who earned less than $40,000 a year (63 percent) and Latinos (61 percent).
The proposition, which passed with 52 percent of the vote, overturned the state Supreme Court's May decision legalizing gay marriage in California. The measure inserts language into the constitution limiting marriage to one man and one woman.
The poll found that, overall, 48 percent of voters oppose the idea of making gay marriage legal. Forty-seven percent support it, while 5 percent are undecided.
The results mirror previous PPIC polls from the last three years, suggesting that the $73 million spent for and against the measure did not do much to change public attitudes on allowing gay couples to wed, said survey director Mark Baldassare.
"At no point in time, before or after the election, did we have a majority of Californians saying they supported gay marriage," Baldassare said. "My takeaway from this is that until there is a major shift in public opinion one way or another, it's going to be another issue where voters are deeply divided."
Geoffrey Kors, executive director of the gay rights group Equality California, said the PPIC poll demonstrates that same-sex marriage advocates "need to make inroads in every category. If 2 percent of voters had voted differently, we would have had a different result," he said.
The poll was based on a phone survey of 2,003 California voters in the Nov. 4 election who were interviewed from Nov. 5-6. The sampling error was plus or minus 2 percentage points.
On the Net: