COLUMBUS, Ohio — Here in the state that handed President George W. Bush his narrow re-election victory two years ago, it has become almost an article of faith that the placement of Issue 1 — a ban on same-sex marriage — on the November 2004 ballot drove conservative and evangelical voters to the polls in unprecedented numbers, sealing a second term for the president.
But new research suggests that Issue 1 — which passed, 62 percent to 38 percent — may not have been such a turnout engine for Bush in the Buckeye State after all. Rather, the ballot measure worked in a much more indirect way, influencing smaller — and, apparently, non-evangelical — segments of Ohio’s electorate to think about the issues in ways that drove them into the Bush camp.
The issue continues to be relevant this year, as eight more states prepare to ask voters to weigh a constitutional amendment that defines marriage as a union between a man and a woman. The states with pending votes are Arizona, Colorado, Idaho, South Carolina, South Dakota, Tennessee, Virginia and Wisconsin. All told, 20 states have so far amended their constitutions to ban gay marriage.
Liberals have picked up the strategy of using ballot measures to drive turnout, with such issues as the minimum wage and stem-cell research. So an understanding of how ballot measures influence voter behavior could be a crucial piece of information in the 2006 campaign cycle and beyond.
Daniel Smith, a University of Florida political scientist, has co-authored two papers that address the question of how the 11 statewide anti-gay-marriage initiatives in 2004 — and particularly the one in Ohio — shaped the results of other election contests that day.
Most of the states that had gay marriage on the 2004 ballot were solidly Republican, so the measure’s easy passage in those states was not a surprise. The only three politically competitive states to weigh in on the issue on Election Day 2004 were Michigan, Oregon and Ohio. And while the measure passed in each of those states, Ohio was clearly the most pivotal of the three.
For months before Election Day, journalists had been scrutinizing the two parties’ voter-mobilization efforts in Ohio, and ultimately, the election was indeed decided in Ohio — and by fewer than 120,000 votes out of the 5.6 million cast by Ohioans.
Judging by turnout figures, the get-out-the-vote effort for Sen. John Kerry (D-Mass.) met its goals and then some. But Bush’s did even better — especially in lightly populated rural areas that few in the national media (to say nothing of Democrats) had been paying attention to.
As Columbus Dispatch politics editor Joe Hallett put it during a session with Smith and others at an August conference of statehouse reporters here, “If you had not known who had won, you could have looked at the results in three counties and said that we elected President Kerry. The formula for Democratic success was that you have to carry Cuyahoga County [Cleveland] by more than 150,000 votes; Kerry won by 227,000. You have to hold the Republican margin in Hamilton County [Cincinnati] to below 60,000 votes; Kerry held it to 23,000. And here in Franklin County, you have to battle even. Kerry won Franklin County by 48,000 votes. You have to look at all those factors and ask, ‘How did this guy not win?’”
The answer, to many observers — and especially to social conservatives who had supported the gay-marriage ban — was that the presence of Issue 1 drove voters to the polls. Right after the election, for instance, Family Research Council president Tony Perkins boasted to The Washington Post that gay marriage “was the hood ornament on the family values wagon that carried the president to a second term.”
Exit polls that cited voter concern over moral issues — combined with massive levels of support for Bush among such voters — seemed to provide evidence for that hypothesis.
But the studies that Smith co-authored, one with his graduate students Matthew DeSantis and Jason Kassel that was published in the journal State and Local Government Review, and the other with Todd Donovan and Caroline Tolbert that was presented at a meeting of the American Political Science Association, suggest that the gay marriage measure had little if any direct impact on turnout rates in Ohio. Instead, the ballot measure may have acted in other ways to tip the presidential race to Bush.
Smith, DeSantis and Kassel used aggregate county-level data in Ohio (and, for comparison’s sake, Michigan) to gauge turnout and voting patterns. When they looked to see whether turnout increased from 2000 to 2004, they found that a county’s level of support for Issue 1 had no effect on increasing turnout in 2004.
In other words, the gay marriage issue didn’t boost turnout. Instead, most of Bush’s county-by-county support in Ohio in 2004 can be explained by his level of support in that county in 2000, the authors found. Once you control for all the relevant variables, Smith said in an interview, “the counties that came out in 2000 came out again in 2004.”
The Ohio county-level findings jibe with studies that found that turnout levels in the 11 states with gay marriage bans on the ballot were similar in 2000 and 2004, once you control for factors such as the presence of a Senate race or a competitive presidential race on that state’s ballot.
Instead, the ballot measure did appear to have an effect in “priming” the electorate to frame their presidential choice through the lens of gay marriage, Smith said.
Using national data from Pew polls and state-level figures from the Ohio Poll, Smith, Donovan and Tolbert found that the existence of a gay-marriage ballot measure made a difference when voters were asked whether gay marriage was important for them in making a presidential decision.
Nationally, 32 percent of voters said that gay marriage was a very important factor in making their presidential decision. But non-evangelical Christians in the initiative states were 9 points more likely than their demographically similar peers in non-initiative states to place the same level of importance on gay marriage.
“It sent a signal to non evangelicals — not to evangelicals — that gay marriage was an important issue,” Smith said.
The effect, he said, could be seen most clearly with white, independent, Protestant women in Ohio who had average incomes and education levels. Such voters who said that gay marriage was not a very important factor in their presidential decision sided slightly with Kerry. But voters in that demographic who thought that gay marriage was an important factor gave Bush 69 percent support.
“That’s a small slice of voters, but it’s indicative of how an issue can affect a targeted group,” Smith said. “In this case, that issue really changes their vote. They may not have gotten that voter out to polls, but once she was there, she was thinking about the candidates differently.”
At the Columbus conference, some expressed surprise at the notion that the ballot measure didn’t directly affect turnout. But most agreed that the measure did have some bearing on the outcome.
“From covering the election, it felt like Issue 1 had an impact,” Hallett said. “In Appalachia, the voting goes back and forth between the parties because their economic lot never improves. Between 1996 and 2000, 16 of those counties went from Clinton to Bush. Bush kept those counties in 2004 and increased his vote there, even though unemployment rates were higher in 2004 than they were in 2000. Those are culturally conservative areas, and I can’t figure out why they stayed with President Bush. I think one thing may have been guns, and the other thing was gay marriage.”
Bob Paduchik, a key strategist for Bush’s 2004 Ohio campaign, told the conference that while “values issues” constituted an important message for the campaign, he and his colleagues did not spend an inordinate amount of time thinking about the ballot measure and its potential effect on Bush’s chances.
Rather, he said, the campaign focused on traditional get-out-the-vote tactics, such as making 450,000 calls to voters on Election Day.
“I see ballot measures as a secondary tactical activity — not something you put a lot of resources behind,” said Paduchik, who now works for the DCI Group in Washington, D.C. “In a political campaign, there are 1,000 things you can do, of which 100 will help get your candidate elected. The challenge is keeping the staff focused on the narrow job of getting those things done.”