The paper has a straightforward title: "Emotional Problems Among Children With Same-Sex Parents: Difference by Definition." It's not hard to guess the conclusion of this research, released last month by Catholic University professor Paul Sullins: Kids with gay parents have more emotional problems than kids with straight parents, and it's because their parents are gay.

This is not a new argument. Especially in the past decade, as gay marriage has been legally recognized in many states, a small number of scholars have claimed that kids of same-sex parents are exposed to more potential harms than kids of straight parents. This, in turn, has been used to argue against gay adoption and marriage.

But just because some studies support this finding doesn't mean it's true. In fact, many, many more studies reached opposite conclusions. "Research ... has developed a scholarly consensus that shows that children raised by same-sex couples are at no important disadvantage," wrote Stanford University sociologist Michael Rosenfeld in an email. "There is a noisy fringe of academics who claim that children raised by same-sex couples are in disastrous peril," a viewpoint which "has little or no credibility within academia."

But in these last few months before the Supreme Court issues a decision on gay marriage, that academic consensus might matter less than how research like Sullins's is received by the courts—and by regular people. Already, the conservative Witherspoon Institute has published a post on his paper, which makes an argument against the legalization of gay marriage. This has been shared more than 17,000 times on Facebook and 2,000 times on Twitter, and that's just one article—other, mostly conservative or religious websites have also circulated the findings.

Sullins's paper is not just any argument against gay marriage. It's an argument presented in the form of science, complete with academic citations, hypothesis testing, and statistical evidence. This is not simply a matter of ideology; it's a question of how social science is used to further ideological goals, and the unique power that has in the public sphere.

* * *

From an academic perspective, there are a number of flaws in the design of Sullins's research. To his credit, he used a large sample of data compiled by the CDC to test his hypothesis, looking at kids who were living with same-sex parents at the time of various surveys taken between 1997 and 2013. But "what Sullins's paper does not show is that these children were actually raised by the same-sex couple," wrote Rosenfeld in an email.

Reading the paper, it's impossible to say whether the kids in question spent most of their lives with heterosexual parents who then got divorced, for example, or a single parent who had multiple partners over time. This family history matters: "We have decades of research showing that family instability and divorce takes a toll on children," Rosenfeld wrote. Because of this constraint, he said, the paper cannot speak to the way being raised by same-sex parents affects the well-being of children. In an email, Sullins disputed this criticism, pointing to other widely accepted studies on emotional well-being and family structure that rely on the same data.

But there are other objections. In an interview, Abbie Goldberg, a psychology professor at Clark University, pointed out that the situation of gay couples in America has changed a lot since 1997, when social acceptance of homosexuality was significantly lower; kids surveyed at that time were probably more likely to have had a gay parent who divorced his or her opposite-sex partner. Scholars must pay to be published in the journal which accepted Sullins's paper, the British Journal of Education, Society & Behavioural Science, which is run by a for-profit company and not affiliated with any academic society. And although the paper ostensibly went through an "open-access" peer-review process, as University of Maryland professor Philip Cohen pointed out in a blog post, that process is pretty thin.

"I don’t want to imply that three journals are illegitimate just because they are run for profit by low-status academics from developing countries," Cohen wrote about this and two other papers published by Sullins. "But looking at the evidence so far I think it’s fair to call these journals bogus." The company that publishes these journals is also listed on the leading index of "potential, possible, or probably predatory scholarly open-access publishers," a sign of those journals' lack of credibility. Sullins defended his choice of venue for publication in an email, pointing out that it's free and available for anyone to read, unlike other journals; he also said his paper was reviewed by four people, who prompted considerable revisions before publication.

In his paper, Sullins concludes that the biological connection between parents and their children is "necessary and sufficient" to explain why kids with straight parents are less likely to develop emotional issues than kids with gay parents. In an email, he suggested that this is connected to other research about family instability. "Every child who is the biological child of a same-sex parent also has an absentee parent somewhere," he wrote. "[A]s the child comes to understand where babies come from, it is inevitable that she will wonder about her own origins, and may experience rejection or stress at the relative absence of her other parent."

But as he himself acknowledges, this conclusion directly contradicts a large body of research on this topic, which suggests that there are no differences between kids raised in stable households by gay or straight parents. In an email, Sullins argued that this "entire body of small-sample research is mistaken and highly misleading," pointing to biased methodology. But a number of nationally representative, large-sample surveys have consistently found that kids from stable gay households fare the same as kids from stable heterosexual households.

There's a certain back-and-forth logic to all of this: A scholar points to methodological flaws in a study, the author shoots back a counterpoint citation, and on and on. This is the normal way that academic debates work. But Rosenfeld argued that this is anything but a normal academic debate. "I don't think you can characterize the research and debates on children raised by same-sex couples as 'a back and forth,' any more than the debate over global warming or the debate over whether smoking causes cancer," he said.

The scientific community has not always been in such firm agreement on this subject. "Maybe 20 years ago, the research on children raised by same-sex couples was scanty enough that it was difficult to establish a consensus," Rosenfeld said in an interview. Over time, he said, more and more scholars around the world have taken up this research question, consistently reproducing this "no-difference" finding.

It's because of this broad consensus that the "no-difference" finding has gained increasing acceptance from the courts. In its 2014 decision in the same-sex marriage case, DeBoer v. Snyder, the Sixth Circuit Court of Appeals dismissed out of hand three social-science studies presented by expert witnesses as evidence that children of same-sex parents face heightened risks. Even in her dissent to the ruling, Judge Martha Craig Daughtrey pointed out that, "with the admitted biases and methodological shortcomings prevalent in the studies performed by the defendant’s experts, the district court found those witnesses 'largely unbelievable' and not credible." And in 2013, the American Sociological Association filed a brief in the Supreme Court case United States v. Windsor, pushing back against claims made by the defense that "same-sex parents produce less positive child outcomes than opposite-sex parents—either because such families lack both a male and female parent or because both parents are not the biological parents of their children." This argument, the brief said, "contradicts abundant social science research."

As the Supreme Court considers the question of same-sex marriage, it's unclear that child welfare will even play any substantial role in its ruling. As Judge Jeffrey Sutton wrote in his opinion in the Sixth Circuit case, "Over time, marriage has come to serve another value—to solemnize relationships characterized by love, affection, and commitment. Gay couples, no less than straight couples, are capable of sharing such relationships."

In an email, Sullins wrote that he hopes his research will help shape the same-sex marriage debate. "We define it as a question about marriage, but it is really a question about families. Same-sex persons say they have certain rights regarding marriage not involving a person of the opposite sex (and who am I to judge?)," he wrote in apparent reference to Pope Francis's now-infamous 2013 remark on homosexuality. (Sullins is a Catholic priest and professor at Catholic University.)

"But don't children have rights too?" he continued. "Is it justice to solemnize as marriage a social arrangement that says to every biological (i.e., non-adopted) child involved, you can never have the love and care of both your mother and your father?"

Justice: This is an important tell. At times, social science may be used in the service of justice, influencing the decisions of lawmakers or judges or everyday citizens. But this is problematic when the veneer of science is used to disguise an argument for a deeply help moral or religious belief. By design, social-scientific research is meant to be transparent, testable, and reproducible at every step; but in reality, regular people don't necessarily have the skills or interest or time to engage in that kind of process. Instead, as the large audience for the Witherspoon Institute's post shows, people may be drawn to articles that confirm what they already think.

"This is part of the nature of politics and science in the public realm," said Rosenfeld in an interview. "Science competes with all sorts of other pseudoscience." It also competes with arguments based on religious and moral beliefs, as it should; but when those beliefs come packaged as data-based findings, that's a problem.