Polls undercount support for same-sex marriage ban
Polls undercount support for same-sex marriage ban
Measures on 5 state ballots likely to pass despite survey results
- Wyatt Buchanan, Chronicle Staff Writer
Friday, October 27, 2006
Proponents of measures to prohibit same-sex marriage say New Jersey's Supreme Court ruling that committed same-sex couples deserve the same rights as heterosexual couples will motivate voters to pass constitutional bans.
But most of the measures on the Nov. 7 ballot in eight other states already have strong voter support. In fact, they may be even farther ahead than they appear, because polling on the issue has been consistently and inexplicably inaccurate.
Same-sex marriage ban supporters and opponents agree that pre-election polls often undercount support for the measures.
Voters in 15 states have approved such bans since August 2004, and polls conducted before elections in seven of them underestimated the yes vote. (No polls were published in three of the states, and poll results in the other five were within the margin of error.)
Polls that underestimated support for the bans were off by as much as 19 percentage points in North Dakota and 7 to 16 percentage points in six other states.
"What it means is that if history is any guide, which I think it is, you have to subtract at least four percentage points from pre-election polls to get a more accurate reading of what the results are going to be on election day," said Matt Foreman, executive director of the National Gay and Lesbian Task Force, a gay rights group working in opposition to the amendments.
Bans are expected to pass Nov. 7 in Idaho, Virginia, South Carolina, South Dakota and Tennessee. The races still appear close in Colorado, Arizona and Wisconsin.
Polls have been published on the proposed constitutional amendments in six of those states, and in all six, the most recent survey showed the bans passing -- in Arizona by a margin of 9 percentage points, and in Tennessee by 53 percentage points.
A poll of likely voters that gay-rights supporters in Arizona conducted for their internal use shows their state's measure failing 48 to 41 percent, with 11 percent of respondents undecided.
Marty Rouse, national field director of the Human Rights Campaign, another gay rights organization opposing the amendments, estimates most polls are off by five percentage points.
"On the undecided factor, they're not undecided," he said. "I put them in the 'anti' (same-sex marriage) column. People have very firm feelings of where they are on this issue."
Ban supporters also account for the consistent polling error in their strategies.
"We've seen it, I think, in every single case, that it is underpolled every single time," said Tom McClusky, vice president of government affairs at the conservative Family Research Council. "I've seen higher, but normally we would add 5 to 10 percentage points to any polling."
Gay rights supporters blame people's unwillingness to express an anti-gay opinion to a pollster for the discrepancy between polls and the ballot box, and McClusky agrees. But public opinion experts who study the phenomenon of "social desirability," which leads people to lie to pollsters on issues like church attendance or whether they would vote for an African American candidate, think other factors may be more significant.
Scott Keeter, director of survey research for the Pew Center in Washington, said he doubts there is stigma about opposing same-sex marriage, because 19 states have passed bans by wide margins, and many Democrats who support gay rights also oppose same-sex marriage.
John Krosnick, a Stanford University professor who studies social desirability in public opinion polling, said how pollsters ask the question and decide who is a likely voter may skew the numbers.
"It is completely inappropriate just to assume that there is social-desirability bias or intentional lying in a survey," he said.
Another professor who studies political polls said anti-gay sentiments might affect them, though.
"People are more likely to say they support individual or group rights even when in some cases they don't," said Michael Traugott of the University of Michigan.
E-mail Wyatt Buchanan at
[email protected].