The real disagreement is seldom over what the data reveal. It’s how scholars present and interpret the data that differs profoundly. You can make the children of same-sex households appear to fare fine (if not better), on average, if you control for a series of documented factors more apt to plague same-sex relationships and households: relationship instability, residential instability, health and emotional challenges, greater economic struggle (among female couples), and—perhaps most significantly—the lack of two biological connections to the child. If you control for these, you will indeed find “no differences” left over. Doing this gives the impression that “the kids are fine” at a time when it is politically expedient to do so.

This analytic tendency reflects a common pattern in social science research to search for ‘‘independent’’ effects of variables, thereby overlooking—or perhaps ignoring—the pathways that explain how social phenomena actually operate in the real world. By way of a helpful comparison, I can state with confidence that after controlling for home ownership, residential instability, single parenthood, and neighborhood employment levels, there is no association between household poverty and child educational achievement. But it would be misleading to say this unless I made it clear that these were the pathways by which poverty hurts educational futures—because we know it does.

The academy so privileges arguments in favor of same-sex marriage and parenting that every view other than resounding support—including research conclusions—has been formally or informally scolded. I should know. The explosive reaction to my 2012 research about parental same-sex relationships and child outcomes demonstrates that far more is at work than seeking answers to empirical research questions. Such reactions call into question the purpose and relevance of social science. Indeed, at least one sociologist holds that social science is designed “to identify and understand the various underlying causal mechanisms that produce identifiable outcomes and events of interest.” That this has not been the case with the study of same-sex households raises a more basic question.