It’s a fair question given the way most public opinion polls work, researchers say, but the answer is still yes.
By Nathan Collins
(Photo: Ashleigh Bennett/Flickr)
Support for gay marriage appears to be at an all-time high: A majority of Americans now say they support same-sex marriage, up from around one-third less than two decades ago. But, some worry, the key word might be “say”—maybe Americans say one thing because they think that’s what pollsters want to hear, not because they really support marriage equality. A new study finds that a majority of Americans really do support same-sex marriage, though there are indeed some basic difficulties with measuring public opinion.
Among the problems with conventional polls, the difficulties associated with measuring trends, and the strange effects of priming, the observation that even incidental things like where you answer questions can influence how you respond.
One particularly difficult problem is what political psychologists call “social desirability bias.” The easiest example to get your head around is race. It’s pretty hard to get somebody to come out and say they’re biased against African Americans—it’s a socially undesirable view—despite pretty good evidence that most white people are, to some degree, prejudiced. Where social desirability bias comes into play, therefore, conventional polls might inflate support for anti-discrimination policies.
There are basic difficulties with measuring public opinion.
Fortunately, there is a way to deal with social desirability bias, point out Columbia University political scientists Jeffrey Lax, Justin Phillips, and Alissa Stollwerk: the list experiment. While list experiments are far from perfect, they can avoid social desirability bias by presenting a list of positions and asking people not what point but rather how many points they agree with.
Here’s a completely hypothetical example, designed to assess true Republican voters’ support for Donald Trump. First, researchers present a control group with three candidates—Cruz, Kasich, and Rubio, say—and ask each person how many they’d be willing to vote for. Next, ask a second group — a treatment group — the same question, with Trump added to the list. Imagine members of the control group would support two of the three candidates on average—which two doesn’t matter. Now imagine that those in the treatment group support an average of three candidates. This suggests that, in addition to whichever two candidates people support, they’d also vote for Trump. More realistically, if the treatment group supports an average of 2.5 candidates, it suggests that 50 percent of Republicans would vote for Trump.
Using list experiments with a total of 1,878 participants, Lax, Phillips, and Stollwerk estimate that about 60 percent support allowing gays and lesbians to marry legally—if anything, slightly more than say they’d support same-sex marriage if they were asked directly, the researchers note.
“It is natural to ask whether some of the purported increase in support for same sex-marriage is due to the presence of social desirability bias, [but] we find no evidence of social desirability bias at the population level,” Lax, Phillips, and Stollwerk write in Public Opinion Quarterly, noting that their research “provides new evidence that there exists majority national support for extending marriage rights to same-sex couples.”