Many opponents of same sex marriage use as their bedrock argument, the claim that the children of same sex relationships fair poorly when compared to those from stable, heterosexual marriages. I do not think that is the case. In the USA, the children of African-American couples fair more poorly than do the children of same sex relationships, as do the children of people having an income below US$25,000 a year (Rosenfeld 2010). In both cases the performance is significantly less than is the average for the children of heterosexual married couples (significant to the 99.9% level). Yet we do not on that basis argue that the marriage act should be amended so as to forbid the marriage of African-Americans or of the relatively poor within our society. The reason we do not apply a parallel argument to that used against same sex marriage, even though the evidence supporting the argument is stronger, is that we do not think that race or wealth are morally relevant criteria on which to make that distinction. Not only that, we do not think that the comparatively poor outcomes for the children of African-American or impoverished couples are sufficient basis to make them distinct moral categories with respect to marriage. I think we can be stronger than that. For most of us, including me, we think it would be morally offensive to argue for a ban on the marriage African-American or impoverished couples on that basis.
It follows from that, and that for some people they consider the outcomes of children relevant to the same sex marriage debate that they have already included the moral distinctness with respect to marriage of same sex couples as a premise in their argument. Had they not already included that distinction, then the evidence with regard to children would be as irrelevant as it is in the case of race or poverty. So, rather than being an argument from the moral distinctness of same sex relationships when it comes to marriage (as it purports to be), the argument from the welfare of children already assumes its conclusion in it premises. It acts as an apologia to reinforce prejudice rather than as a reason that stands on its own.
Opponents of same sex marriage may consider that an unfair conclusion. They might urge in response that heterosexual couples can have children naturally through birth, an option that is not generally available to same sex couples. It follows, they might claim, that the moral distinction in inherent in that biological fact. But it is far from the case that same sex couples cannot have children. Most importantly, they may have children from prior relationships (with at least one of the couple being a step parent to some or all of the children in the family). If the children of de facto heterosexual couples do more poorly than those of married heterosexual couples (as the data shows), the presumption must be that the same applies with homosexual couples. The increased stability, commitment and social acceptability implied by marriage will be beneficial to children, regardless of whether or not the couple they are living with are heterosexual or same sex. That being the case, this difference in fact points to the necessity of allowing same sex marriage rather than prohibiting it. By allowing it, we also allow that for at least some children of people in same sex relationships will benefit from the extra stability implied by marriage.
So much for the logic of the case. What is worse for the argument that same sex relationships among their parents are worse for children is that it is certainly unproven, and unlikely to be true. To be very clear about this, it does appear to be true that the children of same sex relationships do perform more poorly than those of heterosexual marriages in general; but they do not perform more poorly than the children of de facto marriages. Nor do the perform more poorly than the children of heterosexual marriages in which one of their parents is a step parent. Given that for the children of a homosexual couple, in nearly all cases the relationship will by law by a de facto relationship (or only recently have been permitted to be a marital relationship), and that in almost all cases at least one of the parents will be a step parent - that the performance of children of same sex relationships perform as well as those of de facto and/or relationships involving a step parent strongly suggests that the homosexuality of the parents is not a significant contributor to those children's disadvantage.
In a way that is surprising. We would expect the children of same sex relationships to perform, relatively more poorly. There are high levels of bullying applied to lgbti children in the playground, with 61% of them experiencing verbal abuse, and 18% experiencing physical abuse because of their sexual orientation. The high levels of homophobia implied by these statistics suggest (and anecdote confirms) that the children of same sex parents will face similar levels of bullying in the playground. Given that, it would be unsurprising if those children had no adverse consequences from that bullying. In fact, if the children of same sex couples faired more poorly than those from the families of equivalent heterosexual couples, that might well be the result of such bullying alone. In that case the argument from the effects on children would be an argument to cement lgbti disadvantage into the law because they have been insufficiently protected by the law (indeed, persecuted by the law until quite recently) in the past. Of course, the children of same sex relationships do not in general fair worse than those of equivalent heterosexual relationships, as already noted.