Admirable as President Obama’s decision was to support same-sex marriage last week, its implications for his reelection chances are still being debated. Indeed, the latest polling has been less than reassuring. Both a New York Times/CBS News survey and a USA Today/Gallup poll released in the last week suggest that a quarter of Americans are less likely to back President Obama for re-election because of his announcement. Is this a sign that the Obama campaign misinterpreted the country’s support for gay rights? Is it time to conclude that the President will be politically punished for doing the right thing?
Not quite. Polling about voting behavior has become remarkably sophisticated in recent years. But on some topics—attitudes about minorities and moral values, in particular—polls are notoriously bad at predicting behavior. Election polls that center on the public’s attitude towards gay marriage, in other words, shouldn’t be used to predict their actions on Election Day.
Researchers have known for years about the difficulties of relying on such polling to anticipate behavior. Some of the earliest evidence came in the early 1930s, when Stanford researcher, R. T. Lapiere, traveled with two Chinese friends across America at a time of heightened anti-Chinese sentiment. The trio visited 250 hotels and restaurants across the country, asking to be served at each. To Lapiere’s surprise, all but one establishment admitted them as patrons. But when he contacted those places months later asking them if they served Chinese people, 92 percent said no. The researcher concluded that polls about social attitudes may be more likely to reflect how someone feels than what a person will do.
Again and again, research has shown such a gap between attitudes and behavior. That has especially been the case in my area of research, military values. In the 1970s, a significant percentage of officers said they would leave the service if women were admitted to West Point, reflecting widespread cultural resistance to women in the military. But when integration became a reality, there was no mass exodus; the opinions turned out to be just opinions.
The same was true when researchers surveyed foreign military members about allowing gay people to serve. When both Canada and Britain debated lifting their bans in the 1980s and 1990s, polls revealed fierce opposition. In a 1985 survey of 6,500 male soldiers, the Canadian Department of National Defence found that 62 percent of male service members would refuse to share showers, undress, or sleep in the same room as a gay soldier, and that 45 percent would refuse to work with gays. A 1996 survey of 13,500 British service members reported that more than two-thirds of male respondents would not willingly serve in the military if gays were allowed to serve. Yet when Canada and Britain subsequently lifted their bans, nothing of the kind transpired. At most two or three people resigned citing the policy changes (and reports suggested these people were planning to leave anyway).
Despite the prevailing knowledge among social scientists that surveying attitudes is a poor way to predict behavior, the very same argument about mass resignations was used recently to oppose ending “don’t ask, don’t tell” in the U.S. military. A 2008 poll by the Military Times found that 24 percent of active duty military members said they would leave or consider leaving the service if the ban were lifted. Citing the poll, an article in the National Review fretted that over half a million military members might give up their uniforms if President Obama carried through on his promise to end the ban. Senators subsequently delayed repeal, citing a widely publicized letter by over 1000 retired officers who warned that lifting the ban “would undermine recruiting and retention … and eventually break the All-Volunteer Force.”
The repeal did eventually go into effect on September 20, 2011, and this summer UCLA’s Palm Center will release a major study assessing the repeal’s impact. Our study is based on surveys and in-depth interviews with hundreds of currently serving military members, military and academic researchers, service academy and leadership training professors, government officials, as well as analysis of recruitment and retention figures and media and scholarly articles. While our study is not yet complete, our preliminary findings are that not a single departure resulted from repeal, despite the surveys predicting a substantial exodus. Some members of the chaplain corps considered leaving over the issue but decided against it. While it is difficult to prove a negative—someone somewhere may have declined re-enlistment in part because of opposition to the policy change—it is clear that the effect has been nothing like the exodus of thousands of troops that some predicted would “break” the armed forces.
Why is there such a persistent gap between the attitudes people express and the way they behave? Part of the explanation seems to be that people naturally treat surveys as an opportunity to register their feelings about values that are important to them. It’s an impulse that has probably only grown stronger in a new-media age when everyone craves, and even expects, to have a public platform to express themselves. But it may also be exacerbated by a sense among social conservatives and some moderates that their values are besieged and they need to push back. In any case, on the issue of gay rights, it’s clear that many people feel it’s important to continue expressing moral disapproval. But it would be a mistake to see that as an indication that those feelings will necessarily affect their vote.
Given what the data say about how unreliable attitudinal surveys are, Obama’s campaign probably needn’t worry about the latest poll. But it does raise the question of why Americans, despite their best intentions, use much less nuance when they express their values in words rather than deeds.
Nathaniel Frank, author of Unfriendly Fire and a visiting scholar at Columbia’s Center for Gender and Sexuality Law, is writing a book called The Anti-Gay Mind.