BOSTON (AP) — The leader of a state group that opposes gay marriage acknowledged it did not release portions of a poll that indicated voters are deeply divided on whether to ban same-sex marriage.

Ron Crews of the Massachusetts Family Institute said he regretted downplaying the omitted survey results as irrelevant.

“I want to apologize,” Crews said. “I misspoke. I misspoke primarily out of ignorance, but that does not excuse misspeaking. There were other questions, and we are ... going to release those other questions.”

At a recent rally, the group touted Zogby poll results that indicated 69 percent of respondents wanted a chance to vote on a constitutional amendment to ban gay marriage.

The group also highlighted a question that showed 52 percent said that “only marriage between one man and one woman should be legal,” with 42 percent disagreeing.

The group didn’t release information that poll respondents opposed the constitutional amendment, by a split of 49-48 percent. It also didn’t mention that poll respondents, by a margin of 48-46, did not want lawmakers to prevent marriage licenses from being issued to homosexual couples in May, when the Supreme Judicial Court decision legalizing gay marriage takes effect.

Pollster John Zogby said it is standard practice for advocacy groups to issue entire polls for public examination. “All of the questions are important,” said Zogby, whose firm is based in Utica, N.Y.

The poll was taken among 601 likely voters. It had a margin of error of plus or minus 4.1 percentage points.

In a press release, Crews said that polls are “proprietary, important and relevant to those who commissioned the poll” and that the portions released were those most relevant to the issue before the Legislature.

Zogby had released the poll questions that were not disclosed by the group.

Crews said the Coalition for Marriage, a group of statewide and national organizations that commissioned the poll, is in legal discussions with Zogby International about the “unauthorized disclosure” of information in the poll.