Today the US Supreme Court begins hearings on two same-sex marriage cases. Scores of organisations have presented “amicus briefs” to the Court as background to the legal arguments. Here we present an edited version of a submission by the Institute for Marriage and Public Policy and two distinguished scholars. Leon Kass, of the University of Chicago, and Harvey Mansfield, of Harvard University, are experts on the limits of the scientific method and on issues relevant to the appropriate structure of family life. This is a long but extremely valuable contribution to the debate over same-sex marriage.
* * * * *
This case should be decided on the basis of the law, without reliance on the social science studies and authorities that Respondents and their amici will undoubtedly put before the Court. The social and behavioral sciences have a long history of being shaped and driven by politics and ideology. This is partly because researchers often choose to study issues implicating controversial questions of public policy.
And it is partly because it is often impossible to perform the kind of objective observations and controlled experiments that are standard in the physical sciences. History is littered with notorious examples of false theories gaining wide acceptance among respected social and behavioral scientists, some of which supported pernicious public policies.
Although published academic studies typically contain caveats about the limitations of their methodology and of the data available to the researcher, those studies are frequently cited in litigation and in public debate for conclusions they cannot legitimately support. When organizations of social and behavioral scientists purport to speak for a professional consensus on controversial matters of public policy, special caution is warranted.
At one time, for example, psychiatrists almost universally considered homosexuality a mental disorder, and the American Psychiatric Association classified it as such in its Diagnostic and Statistical Manual of Mental Disorders (“DSM”). After a sustained political campaign against the Association, its members voted in 1973 to remove homosexuality from the DSM. The historical record shows that the change was not made because of new scientific findings, but rather in response to external political pressure and to political maneuvering within the Association.
Amici do not contend that the long-standing classification of homosexuality as a mental disorder was justified by reliable science, or that the alteration of the DSM resulted from scientific error. Our point, rather, is that science had little to do with the Association’s revision of the DSM, and that this episode illustrates why such organizations should not be taken for the voice of science. It would have been a mistake for this Court to rely on the official position of the American Psychiatric Association either before or after 1973.
It would also be a mistake to rely on briefs from this and similar organizations today. There is good reason to believe that the political climate has strongly influenced much of the existing research on issues raised in this case. That body of research, moreover, is radically inconclusive. Same-sex marriage is a very recent innovation, as is the practice of child rearing by same-sex couples. The effects of these new developments could certainly be significant. But only an advocate for social change could claim to know that the effects will be entirely or even largely benign.
Even if same-sex marriage and child rearing by same-sex couples were far more common than they now are, large amounts of data collected over decades would be required before any responsible researcher could make meaningful scientific estimates of the effects.
Social and behavioral scientists, moreover, have inadequate tools for measuring the effects of different family structures on children. Notwithstanding the patent weaknesses of the existing research, Respondents sought to persuade the courts below that there is a scientific basis for constitutionalizing same-sex marriage.
In fact, there is no such basis. There neither are nor could possibly be any scientifically valid studies from which to predict the effects of a family structure that is so new and so rare. The necessary data simply do not exist.
There could conceivably come a time when supporters of traditional marriage are compelled by scientific evidence to acknowledge that same-sex marriage is not harmful to children or to society at large. That day is not here, and there is not the slightest reason to think it is imminent. It is no less possible that scientific evidence will eventually show that redefining marriage to encompass unions of same-sex couples does have harmful effects on our society and its children. That day is also not yet here, but there is no basis for this or any other court to conclude that it will never arrive.
Now and for the foreseeable future, claims that science provides support for constitutionalizing a right to same-sex marriage must necessarily rest on ideology. Ideology may be pervasive in the social sciences, especially when controversial policy issues are at stake, but ideology is not science.
In recent decades, this Court has been inundated with arguments and evidence from social and behavioral scientists. Reliance on such briefing is no doubt sometimes appropriate. But the Court has frequently expressed its skepticism about such submissions, and for good reason. In this case, the relevant scientific evidence on which Respondents seek to rely is manifestly unreliable, and it should be given no weight at all. The case can and should be resolved on the basis of the law.
1. This court has recognized that unreliable expert opinions are a serious threat to the integrity of the legal system.
Modern science advances our understanding of the world by testing potentially falsifiable hypotheses against observable and measurable data. Because it is seldom if ever possible for all relevant data to be accounted for, and thus for all but one of the logically possible alternatives to be falsified, scientific theories are in principle always subject to revision on the basis of new data or better measurements.
Our legal system, of course, cannot treat all scientific conclusions as tentative or inadmissible. It must therefore often rely on expert testimony or on the consensus of scientific authorities. Cases in the Daubert line frequently involve characteristically scientific issues about causation in the physical world.
Even here, experts frequently overstate the reliability of their conclusions, for a variety of reasons including the incentives they may have to favor one party or another in litigation. Accordingly, this Court has recognized that reliance on such opinion evidence is often perilous.
The Court’s deep concern about the use of unreliable evidence in the context of physical causation should be magnified a thousand-fold in a case like this one.
Unlike a tort case, this litigation raises elusive and contentious issues about the nature of homosexuality and the personal and social effects of alternative family structures. A decision constitutionalizing a right to same-sex marriage, moreover, would have social implications far beyond any that might arise from a mistake in a product liability case.
Academic studies of the issues raised in this case, like many others in the various fields of social science, are subject to severe constraints arising from limited data and from a dearth of the kind of controlled and replicable experiments that are characteristic of the physical sciences. This Court should not rely on the social science research that will undoubtedly be cited by Respondents and their amici.
2. Social and behavioral science is frequently shaped and driven by politics and ideology.
Even in the physical sciences, research is often tainted by the bias of the researchers. These biases can arise from a multitude of causes, frequently invisible to the researchers themselves, including the researcher’s policy preferences, unquestioning acceptance of conventional wisdom, personal ambition, and ideology.
The effort in recent years to close off debate about issues related to global warming provides one example, and a striking one because it has come close on the heels of warnings from scientists about the possibility of a new ice age caused by global cooling.
This Court has not rushed to embrace an end to the debates, and for good reason. It can sometimes take a very long time for a genuinely settled consensus based on scientifically valid studies to arise. The debate in astronomy over geocentric theory, for example, remained open for hundreds of years after Copernicus.
Only in the 19th century did new technology finally permit observations conclusively demonstrating that the earth does move in relation to what were once called the “fixed stars.” The social sciences are far more prone to biased research than the physical sciences. That is partly because such research frequently addresses questions with immediate implications for controversial issues of public policy. And it is partly because it is inherently much more difficult—and often impossible—to perform the kind of objective observations and replicable experiments that are the staple of the physical sciences.
It is therefore often difficult to definitively disprove theories that have little or no basis. History is littered with notorious illustrations, including phrenology, Marxist economics, and so-called scientific racism, all of which were once widely accepted by respected social and behavioral scientists.
The late Senator Daniel Patrick Moynihan, himself a distinguished social scientist, acutely diagnosed the susceptibility of social science to politicization:
“[S]ocial science is rarely dispassionate, and social scientists are frequently caught up in the politics which their work necessarily involves…
Moreover, there is a distinct social and political bias among social scientists. In all fairness, it should be said that this is a matter which social scientists are quick to acknowledge, and have studied to some purpose. It all has to do, one suspects, with the orientation of the discipline toward the future: It attracts persons whose interests are in shaping the future rather than preserving the past. In any event, the pronounced ‘liberal’ orientation of sociology, psychology, political science, and similar fields is well established.”
When Senator Moynihan wrote this in 1979, the “‘liberal’ orientation” in these fields was indeed well established by surveys of university faculties. More recent surveys indicate that this orientation has become considerably more pronounced in recent decades, and that it is stronger in the realm of “social or ‘lifestyle’ liberalism than it is in economic liberalism.”
Multiple-regression analysis has provided preliminary results consistent with the hypothesis that when academic achievement is controlled for, academics who do not hold progressive political views experience negative effects on their professional advancement. If confirmed by further research, these results might be explained in part by the dynamics of group psychology. These dynamics might also help to explain why research in certain fields can consistently and for reasonably long periods of time support conclusions that are eventually proven false.
When organizations of social and behavioral scientists purport to represent a consensus of their professions, special caution is warranted. A telling illustration is provided by the history of classifying homosexuality in the American Psychiatric Association’s Diagnostic and Statistical Manual of Mental Disorders (“DSM”). As recently as the 1960s, there was an overwhelming consensus in the psychiatric profession that homosexuality should be classified as a mental disorder.
This consensus was reinforced by an in-depth study comparing 106 male homosexuals and 100 male heterosexuals under the care of members of the Society of Medical Psychoanalysts. The research was carried out over a period of ten years, and the results were reported in a massive volume signed by Irving Bieber and nine co-authors. Even those who did not adhere to the dominant psychoanalytic approach in psychiatry agreed that homosexuality should be considered an abnormality.
Doubts about the validity of this diagnosis were raised by research from outside psychiatry, including that of Alfred Kinsey and students of comparative anthropology and primatology. That research, however, was subject to various interpretations, and psychiatrists disagreed among themselves primarily about the etiology and treatment of what they agreed was a disorder. Beginning in 1970, the American Psychiatric Association came under sustained attack from an organized political movement determined to force the Association to remove homosexuality from the DSM.
Within the short space of three years, this attack succeeded. As a detailed (and by no means unsympathetic) history of this political struggle has demonstrated, the change in the Association’s position was not the result of scientific advances. Rather, it was a response to political tactics that included public denunciations of the profession and disruption of scholarly conferences. The intricate maneuvering for change within the Association was not led by experts on homosexuality; those who resisted the proposed change, moreover, alleged that some of its public supporters privately acknowledged that they considered homosexuality a pathological condition, but were afraid to say so publicly. Eventually a referendum was held, and the deletion of homosexuality from the DSM was approved, though only by 58 percent of the Association’s members.
Amici do not contend that the long-standing classification of homosexuality as a mental disorder was justified by reliable science, or that the alteration of the DSM in 1973 resulted from scientific error. Our point, rather, is that science had little to do with what happened, and that this episode illustrates why organizations of social and behavioral scientists should not be taken for the voice of science. The American Psychiatric Association’s treatment of homosexuality in the DSM was not based on settled science either before or after its political decision to alter its position. It would have been a mistake for this Court to rely on the classification of homosexuality in either version of the DSM.
It would also be a mistake to rely on briefs or official statements from this and similar organizations today.
There is good reason to believe that the political climate has strongly influenced much of the existing research on issues raised in this case. Norval Glenn of the University of Texas, for example, has written: “Given the widespread support for same-sex marriage among social and behavioral scientists, it is becoming politically incorrect in academic circles even to suggest that arguments being used in support of same-sex marriage might be wrong.”
Similarly, two strong opponents of what they call “heterosexism” have attacked the scholarship of those who support traditional marriage, but have also said, “We wish to acknowledge that the political stakes of this body of research are so high that ideological ‘family values’ of scholars play a greater part than usual in how they design, conduct, and interpret their studies.” They have also suggested that many psychologists sympathetic to parenting by homosexuals are apt to “downplay the significance of any findings of differences.”
The record in this case offers some revealing examples. Some of Respondents’ own expert witnesses have acknowledged that opinions (including their own) about whether homosexuality is a psychological disorder are not scientific judgments.
At trial, Respondents’ expert on child development, Dr Michael Lamb, read the following statement by the American Academy of Child and Adolescent Psychiatry: “Lesbian, gay, bisexual or transgender people have faced more rigorous scrutiny than heterosexual people regarding their rights to be or become parents.” Under cross examination, Dr. Lamb was unable to dispute the proposition that “this statement is not based in empirics, but, rather, in politics.”
Dr Lamb was cited by the district court for the broad and unqualified conclusions that the “gender of a child’s parent is not a factor in the child’s adjustment” and that “having both a male and female parent does not increase the likelihood that a child will be well-adjusted.” At trial, however, Dr. Lamb had conceded that his own published research concluded that growing up without fathers had significant negative effects on boys, and that there is data indicating that there are significant differences between men and women in their parental behavior.
At trial, he also conceded that there is considerable research indicating that traditional opposite-sex biological parents appear in general to produce better outcomes for their children than other family structures do. The district court’s opinion contains numerous other conclusions that in fact are not and cannot possibly be established by the evidence on which that court relied. The district court’s citations to the testimony of Respondents’ witnesses and to a statement by the American Psychological Association should not be mistaken for reliance on credible science.
There could conceivably come a time when supporters of traditional marriage are compelled by scientific evidence to acknowledge that same-sex marriage is not harmful to children or to society at large. That day is not here, and there is not the slightest reason to think it is imminent. It is no less possible that scientific evidence will eventually show that redefining marriage to encompass unions of same-sex couples does have harmful effects on our society and its children. That day is also not yet here, but there is no basis for this or any other court to conclude that it will never arrive. Now and for the foreseeable future, claims that science provides support for constitutionalizing a right to same-sex marriage must necessarily rest on ideology. Ideology may be pervasive in the social sciences, especially when controversial policy issues are at stake, but ideology is not science.
3. The effects of same-sex marriage on family life are unknown, and currently unknowable
Same-sex marriage is a very recent innovation, as is the practice of child rearing by same-sex couples. The effects of these new developments certainly could be quite significant for same-sex partners, for children raised by same-sex couples, and for our society.
But only an advocate for the cause of same-sex marriage could claim to know that the effects will be entirely or even largely benign. Such claims can be based only on conjecture or faith, not science. Even if same-sex marriage and child rearing by same-sex couples were far more common than they now are, large amounts of data collected over decades would be required before any responsible researcher could make meaningful scientific estimates of the effects.
Social and behavioral scientists, moreover, do not have adequate tools for measuring the effects of different family structures on children. Typical measures include educational attainments and rates of social deviance (using criteria such as drug use and other forms of delinquency). But these can hardly begin to assess the success of children (or adults for that matter) as human beings, let alone how happy they are.
Accordingly, the statements that one encounters in the existing research literature typically amount at best to claims that “no evidence exists” of bad effects from same-sex marriage or from child rearing by same-sex couples. Such conclusions should hardly be surprising inasmuch as there is manifestly too little evidence from which to draw any reliable conclusions.
Thus, one could just as easily say that there is no reliable evidence that such practices are beneficial or harmless. But that is something one rarely if ever hears from proponents of legalizing same-sex marriage.
Instead, researchers and social science advocacy organizations have promoted the myth that their failure to find evidence of bad effects implies or strongly suggests that such bad effects will not ensue.
A brief filed in the court below by several organizations—including the American Psychological Association, the American Psychiatric Association, and the American Association for Marriage and Family Therapy—provides a revealing illustration.
Much of this brief to the Ninth Circuit argued for conclusions that are only peripherally relevant at best, such as the proposition that some homosexuals form long-lasting relationships, or are noncontroversial, such as the proposition that married heterosexuals are statistically more likely than unmarried heterosexuals to exhibit certain indicia of physical and psychological health. On the issues that might be thought central, however, the brief offered only a mélange of weak and unreliable evidence from which unjustified inferences were drawn or suggested.
Consider, for example, Section III.A of the brief, titled “Gay Men and Lesbians Form Stable, Committed Relationships That Are Equivalent to Heterosexual Relationships in Essential Aspects.” In support of this conclusion, the brief cited several studies, based on non-representative samples, for the proposition that a significant fraction of gay men and lesbians are or have been in a “committed relationship.” What the brief called “more representative samples” were said to support this finding. So far as amici have been able to determine, no studies using the scientific standard of comparing large random samples with appropriate control samples were cited here or anywhere else in the brief.
Instead, the brief drew the following conclusion: “Based on the empirical research findings, the American Psychological Association has concluded that ‘[p]sychological research on relationships and couples provides no evidence to justify discrimination against same-sex couples.’” To the extent that this might be thought to support the brief’s ultimate conclusion that “[t]here is no scientific basis for distinguishing between same-sex couples and heterosexual couples with respect to… civil marriage,”, one could just as well say that there is no scientific basis for denying that such couples should be distinguished. This corollary was conveniently ignored.
Another example of misleading argumentation appeared in the brief’s effort to argue that California’s laws on marriage deny important social or psychological benefits to same-sex couples. First, the brief cited no studies at all that attempt to compare the benefits of marriage with the benefits of California’s same-sex civil unions. Apart from that rather glaring problem, the brief acknowledged that no empirical studies have systematically compared married same-sex couples with unmarried same-sex couples.
Remarkably, however, the brief purported to rely on its signatories’ “scientific and clinical expertise” for the proposition that it is appropriate to extrapolate from research on heterosexual couples to predict the effects of legalizing same-sex marriage. Whatever this “scientific and clinical expertise” may amount to, the brief offered no evidence that such extrapolation is justified by the application of scientific methods to appropriate bodies of data.
Finally, the brief cited numerous studies purporting to support the inference that homosexual parents are indistinguishable from heterosexual parents in their effects on children. Once again, the brief cited no studies based on large, randomized samples.
Once again, the brief could actually claim at most that studies using severely limited data have failed to prove that children raised by homosexual parents fare less well than children raised by heterosexual parents. And once again, we can say that it is equally true that the studies do not prove that children do fare as well with the one as with the other.
Apart from the fact that this brief proves on close examination to have been misleading on its face, its statements about “no evidence” were false. The brief simply ignored research that found, among other things, that the children of homosexual parents had higher levels of problematic behavior (such as excessive drinking, drug use, and lower assessments of educational performance and socialization) than the children of heterosexual parents. This work may not be more reliable than the research relied on by Respondents, but it is evidence in the same sense as the research that Respondents cited.
What is more, there is now a body of new evidence—based for once on recognizably scientific methods—arising from a study using a large randomized sample, objective measures of well-being, and reports of grown children rather than their parents. This study found that children raised in a household where a parent was involved in a same-sex romantic relationship were at a significant disadvantage on several objective measures of wellbeing. This obviously implies nothing conclusive about the effects of same-sex marriage, about which there is too little data from which to draw any clear inferences at all. But neither can its possible implications be dismissed, especially in light of the weaknesses of the earlier research that tended to find little or no difference in the outcomes for children raised by same-sex couples. The earlier research was based on severely biased data. One prominent study, for example, relied on a sample recruited entirely at lesbian events, in women’s bookstores, and in lesbian newspapers. Others relied on samples as small as 18 or 33 or 44 cases. And most of them relied heavily on reports by parents about their children’s well-being while the children were still under their own care.
This is hardly the stuff from which scientifically valid conclusions could possibly be drawn.
Not surprisingly, a detailed re-analysis of 59 studies cited by the American Psychological Association in a 2005 publication showed serious flaws in the research, and concluded that “strong, generalized assertions, including those made by the APA [publication], were not empirically warranted.”
The new research cited above, which suggests that being raised by homosexual parents may have adverse effects on children, is the most scientific of the studies now available, but it certainly is not the last word on the subject. Its author, Professor Mark Regnerus, freely acknowledges that his work is only the beginning of a long-term scientific project. He has, moreover, specifically cautioned against drawing conclusions about causality from his findings, and has warned against basing legal decisions on his preliminary research.
Amici agree that the outcome of this case should not be determined by Professor Regnerus’ research, any more than it should be affected by the less scientific studies that preceded his. But it is now undeniably false to say that all the scientific evidence points toward an equivalence of outcomes for children raised by homosexual and heterosexual parents.
The simple fact is that nobody knows, or could possibly know, what the effects of legalizing same-sex marriage will be. Human well-being is an extraordinarily complex phenomenon, which is affected by an extremely large and diverse number of causal factors. Decades from now, it may be possible for researchers using scientific methods to provide meaningful measures of the effects of same-sex marriage on individuals and society. Today it is not.
4. Inconclusive studies are often used to argue that controversial policies are scientifically supported.
Studies conducted by social and behavioral scientists are frequently cited to support policy decisions for which the studies themselves offer little or no support. While the results published in academic journals typically contain caveats about the data and methodology used by the researcher, the studies are often cited for propositions far beyond what the research can legitimately support. Journalists, activists, litigants, and interested amici are especially prone to such overstatements, but government officials are not immune and neither are social and behavioral scientists themselves.
A revealing example is provided by two nearly simultaneous commissions that studied the effects of popular media on viewers. The National Commission on the Causes and Prevention of Violence found that “[t]he preponderance of available research evidence strongly suggests… that violence in television programs can and does have adverse effects upon audiences—particularly child audiences,” and that broadcasters should accept “the burden of proof that such programs are not harmful to the public interest.”
The President’s Commission on Obscenity and Pornography found that “extensive empirical investigation… provides no evidence that exposure to or use of explicit sexual materials plays a significant role in the causation of social or individual harms such as crime, delinquency, sexual or nonsexual deviancy or severe disturbances.” The contrast is arresting, as is the fact that at least one academic participated in both commissions and managed to provide support for both.
The “no evidence” conclusion of the pornography commission should have come as no surprise, given the obstacles to obtaining reliable scientific evidence that such effects either do or do not exist. Yet the commission went on to make recommendations about public policies based in significant part on research finding “no evidence” of harmful effects. The violence commission’s Task Force on Mass Media and Violence, for its part, relied on research that manifestly did not support its conclusions. This led a leading social scientist, Harvard’s James Q. Wilson, to say: “The blunt truth is there is almost no scientific evidence whatsoever to support the conclusions of either the Task Force or the Commission… unless what one means by ‘violent behavior’ is a willingness to engage in certain forms of harmless play.”
Professor Wilson went on to lament one feature of the commission’s report in particular: “Perhaps the most distressing aspect of the entire enterprise is the tone of advocacy that pervades some of the chapters written by social scientists who seem more interested in finding any data, however badly interpreted, that will support their policy conclusions.”
Professor Wilson, we should stress, was a strong proponent of modern social science, who believed that it can discover evidence that may have implications for public policy.
But “[w]hen social scientists are asked to measure consequences in terms of a badly conceptualized or hard-to-measure ‘effect’ of one among many highly interrelated ‘causes,’ all of which operate (if at all) over long periods of time, they tend to discover that there is no relationship or at best a weak and contingent one.” Accordingly, he did not invoke science to claim that exposure to media violence is harmless or that pornography is harmful. Rather, his analysis showed that the kind of social science relied on by these two commissions cannot answer—or even meaningfully contribute to answering—the public policy questions they addressed. The same is true of the research that Respondents and their amici have urged upon the courts in this case.
Like Professor Wilson, this Court has frequently been skeptical about the findings of social and behavioral scientists, especially in the area of human psychology. This has led to some frustration among academics. One commentator, for example, castigated the Court at length for its resistance to using the results of psychological research in decisions about trial process.
In the course of his critique, the commentator announced without reservation that “psychologists agree that eyewitness identification of strangers is unreliable,” citing as an authority a 1985 publication by Professor Gary Wells. At the time, such a consensus may have existed among researchers in this area. Subsequently, however, Wells himself, and other researchers as well, have concluded that such broad statements are not supportable. The Court has been right in refusing to change the law on eyewitness identification in response to preliminary research by social scientists. There is at least as much reason not to change the existing law in this case.
5. The court should rely for its decision in this case on the law, rather than on speculation and ideology masquerading as science.
Beginning with the development of “Brandeis Briefs” early in the last century, and increasingly in recent decades, it is fair to say that this Court has been inundated with arguments and evidence from social and behavioral scientists. There undoubtedly are areas where social science can offer meaningful assistance to policymakers and to courts. This Court has found guidance, for example, from economics in the field of antitrust law and from statistical studies in the field of employment discrimination. No such meaningful assistance can possibly be drawn from the kind of studies that Respondents and their amici cited in the courts below, and which they presumably will cite again in this Court. The research they offer cannot possibly confirm that the effects of same-sex marriage will be harmless or beneficial. The scientific evidence cited to support this change in social policy is manifestly inconclusive, and there is no good reason to give it any weight at all. The social and behavioral scientists who make rosy predictions are using their academic credentials to advance a policy they prefer for reasons outside their fields of expertise.
This case can and should be resolved on the basis of existing law, which should not be altered in response to advocacy posing as science.
For all the foregoing reasons, and for the reasons set forth by Petitioners, the Court should decline to remove decisions about legalizing same-sex marriage from the democratic process.
The original amicus brief, with its supporting documentation, can be found here.
Leon R. Kass is the Madden-Jewett Scholar at the American Enterprise Institute, and Addie Clark Harding Professor Emeritus in the Committee on Social Thought at the University of Chicago. Originally trained in medicine (M.D., Chicago, 1962) and biochemistry (Ph.D., Harvard, 1967), he is the author of numerous books and articles about the relation between science and society.
Harvey C. Mansfield is the William R. Kenan, Jr., Professor of Government at Harvard University. His work in political science has included studies of the nature of modern political science and its effects on public policy, as well as studies of the relationship between the sexes in the light of science and philosophy.
The Institute for Marriage and Public Policy is a nonprofit, nonpartisan organization dedicated to strengthening marriage as a social institution.