A new report is shedding light on the effectiveness of statewide campaigns against same-sex marriage, although the findings are raising additional questions.

The report — which examines the trend of public opinion on same-sex marriage in 33 states that have had the issue on the ballot — found efforts during campaign periods had very little impact on moving voters to oppose same-sex marriage bans on Election Day.

Additionally, the report found polling data gathered during campaigns on marriage initiatives is misleading because a greater percentage of people vote in favor of same-sex marriage bans than the percentage who tell pollsters they will support the ban.

Patrick Egan, author of the report and a gay political science professor at New York University, made the findings public Tuesday.

He said that he had limited explanations for what caused this behavior among voters. But at a press event in San Francisco, Egan explained that his report dismisses a number of theories popularly used to explain why polling data for marriage ballot questions doesn’t accurately reflect election results.

One theory that Egan advances in his report — but says he finds no evidence to support — is the idea that responders are lying to pollsters when they say they’ll vote against a same-sex marriage ban so that they seem more tolerant.

Such a phenomenon would be similar to the “Bradley effect,” a theory that polling participants would lie to pollsters by saying they’ll vote for a non-white person in an election and instead vote for a white candidate at the polls.

Egan dismissed this theory with regard to marriage initiatives after looking at several contexts in which voters may feel more social pressure to vote in opposition to bans on same-sex marriage, such as in states with a greater population of openly gay, lesbian and bisexual people, or polls conducted by live interviewers as opposed to automated pollsters.

In all these contexts, Egan said he could find “no discrepancy” in voters being more truthful about what they’re telling pollsters in certain states or in certain situations.

“All of the findings here just show that voters do not appear to be lying to public opinion pollsters when they are asked about their support for same-sex marriage bans,” Egan said.

Another theory that Egan refutes with regard to the discrepancy between polls and election results is that voters are confused about what a “yes” vote and a “no” vote entails on an initiative. Egan said this theory doesn’t hold up because polling information is as unreliable at the start of the campaign — before voters have been educated on the subject — as it is closer to Election Day.

“The gap does not become smaller over the course of the campaigns, so polls are just as accurate on the night before Election Day as they are six months out — just as inaccurate, I should say,” he said.

Egan said this theory is shown to be invalid when comparing polling data and election results from states with more educated voters to states with less educated voters.

“Even in states where voters are informed — that is, we know from other data that state residents tend to be more interested, engaged and informed about politics — we are not seeing that gap become any smaller than in states where voters don’t pay too much attention to politics at all,” Egan said.

In an attempt to determine why polling data on the marriage issue is unreliable, Egan said his answer as a political scientist is “more research is needed,” but also speculated it may relate to how pollsters determine likely voters.

Noting that most of the surveys in his report are of likely voters, Egan said pollsters could be screening out people who would vote for same-sex marriage bans on Election Day.

“That would help explain the difference we see between polling and election results, and why it’s so consistent over time,” he said.

A number of LGBT civil rights leaders at the San Francisco press conference said they intend to use the report to guide strategy for future ballot initiatives on marriage. Activists in California, where Proposition 8 ended gay nuptials in 2008, are looking to bring the issue of same-sex marriage back to the ballot to reverse the initiative in 2012.

Geoff Kors, executive director of Equality California, said the findings show voters are “at their least persuadable” during the course of a campaign.

“But when we look over the last decade at the amazing movement we’ve seen on what is one of the most challenging social issues to move people on, we’ve seen that the movement happens not during the campaign, but away from the campaign,” Kors said.

He noted that California in 2000 passed Prop 22, a statutory ban on same-sex marriage by 23 points, and in 2008 passed Prop 8, the constitutional ban, by four points.

“All that movement happened not in the couple months before Prop 8, but in the years between those elections,” he said.

Kors said the process is continuing in California with recent public polls showing a 50 percent or majority support for same-sex marriage.

Kate Kendell, executive director for the National Center for Lesbian Rights, also said the study demonstrates efforts to change the hearts and minds of voters must be made before a campaign begins.

“In the midst of a campaign, voters are perhaps least likely to have their views changed — particularly on an issue like marriage, an issue they feel like they understand and know,” she said.

Kendell said “it’s absolutely clear” in the fight for same-sex marriage that conversations “need to happen now about who we are, our lives, our families, our children, our hopes and dreams.”