Medical Critique of the “Medical Critique of the Australian Medical Association’s Position Statement on Marriage Equality”

Dr Wei Hong

In response to the Australian Medical Association Position Statement on Marriage Equality1, the authors of the Medical Critique of the Australian Medical Association’s Position Statement on Marriage Equality2 (Critique) cite several studies to claim evidence of “poorer outcomes for children raised by same-sex couples.” The results of a study depend not only upon the reality being measured but also the methodology chosen, such that mere existence of a “peer-reviewed article” does not automatically constitute “putative evidence.” Health professionals should critically appraise studies in a sophisticated manner and avoid simply accepting their conclusions at face value.

At first glance, Sullins’ study of 207,007 children from the 1997-2013 US National Health Interview Survey, including 512 with same-sex parents, suggested a higher prevalence of emotional problems among children of same-sex parents compared to those of different-sex parents, after controlling for some confounders.3 However, no difference was found after additionally controlling for biological parentage. Indeed, after additional modelling, biological parentage fully explained the difference in emotional problems between same-sex and different-sex parents. Sullins nevertheless attributes the difference to parental sexuality, arguing that biological parentage “is not strictly speaking a proper explanatory variable” because “the absence of common biological parents” is “implicated in the definition” of same-sex parents. This extrapolation assumes that biological parentage is not itself confounded by another variable, overlooking that when compared to children living with two biological parents, children living without two biological parents are disproportionately more likely to have experienced a previous family breakdown, a known risk factor for adverse outcomes.

Allen used a 20% sample of the 2006 Canada census to identify approximately (Statistics Canada does not allow the sample sizes to be released) 2,000,000 children, including approximately 1,400 with same-sex parents, to conclude that children living with same-sex parents are less likely to graduate from high school compared to those living in different-sex marriage families, after controlling for some confounders, including whether the family moved within the past 1 year, as a marker of family stability.4 This definition is likely too narrow, as it is reasonable to expect family breakdown more than 1 year ago to have a substantial adverse impact on child school performance. Indeed, after controlling instead for whether the family moved within the past 5 years, no difference was found. Interestingly, this result is only apparent in a table that was relegated to the appendix, with minimal comment in the paper.

In the New Family Structures Study (NFSS), Regnerus interviewed 2988 young adults, including 236 raised by same-sex parents, from KnowledgePanel®, a representative dataset of the US population, concluding that children raised by same-sex parents fared worse in several out of “40 different social, emotional, and relational outcome variables” compared to those raised in intact biological families.5 However, despite noting that children raised by same-sex parents were disproportionately more likely to have experienced a previous family breakdown, with “just under half of such respondents” reporting that their same-sex parenting arrangement followed a previous “failed heterosexual union,” there was no attempt to control for this confounder. On the contrary, the confounding was amplified by the decision to exclude adopted, divorced, step, and single-parent families from the different-sex comparison group.

In their reanalysis of the NFSS data, Cheng and Powell identified numerous other methodological problems such as selection bias in screening questions, misclassification (for example, 58 out of 236 young adults “raised by same-sex parents” never lived with the parent or lived with them for only 1 year), lack of data validation for implausible or contradictory responses, and coding missing answers as “no” instead of as “missing.” Addressing some of these issues yielded very different results, suggesting that Regnerus’ findings were largely due to methodological choices rather than real differences between comparison groups. For example, the number of “statistically significant” outcome differences between lesbian mother and intact biological families was reduced from 24 to 6. Of the 6 outcomes, “family security,” “frequency of being arrested,” and “frequency of pleading guilty” were not robust on sensitivity analysis, appearing or disappearing depending on 1 or 2 highly influential cases. Of the remaining 3 outcomes, only “had affair while married/cohabiting” could be considered an adverse outcome.6 Out of over 30 questions (regarding outcomes that could be considered adverse), one or more spurious “significant differences” are likely to arise at the chosen “significance level” of p = 0.05 even if there are no differences between two comparison groups.

The Critique authors argue that Potter’s analysis of 19,043 children from the Early Childhood Study Kindergarten cohort, including 158 with same-sex parents, found that “children in same-sex parent families appeared to have lower baseline math scores” compared to those of two-biological parent families.7 They neglect to mention that this numerical difference was not “statistically significant” after controlling for socio-demographic confounders, and that after additionally controlling for family breakdown as measured by the number of transitions in family structure, the numerical difference actually favoured children with same-sex parents, although this advantage was not “statistically significant” (p = 0.486). This was despite the decision to exclude divorced, step, single-parent, cohabiting, widowed and “other” families from the different-sex comparison group.

In his reanalysis of 20,745 adolescents from US National Longitudinal Survey of Adolescent Health (Add Health) Wave I full sample, Sullins identified 17 with female same-sex parents and 6 with male same-sex parents, as opposed to the 44 with female same-sex parents originally analysed by Wainwright, Russel and Patterson8, concluding that “adolescents with same-sex parents experience significantly lower autonomy and higher anxiety, but also better school performance” compared to those with different-sex parents.9 While he criticises the methodological problems of the original study, Sullins’ own methodology is itself questionable, such as the decision to use multiple t-tests which, compared to the original ANOVA, inflates the family-wise error rate of false positives due to the multiple comparisons problem. Even assuming Sullins’ methodology were sound, his own results actually did not find a difference in anxiety after controlling for socio-demographic confounders, although the superior school performance of adolescents with same-sex parents remained “statistically” (p = 0.004) and clinically (0.430 higher grade point average) significant.

The Critique authors furthermore claim that Sullins’ reanalysis found the (supposedly) “statistically significant increase in anxiety” to be “worse for children of legally married same-sex couples” (emphasis in original). This is a gross falsehood. At the time of Add Health Wave I, same-sex marriage was not legal in any part of the US. Rather, as Sullins himself admits, the comparison was between same-sex parents who were legally unmarried but answered “married” nevertheless, versus those who did not. Hence, this finding is completely irrelevant to the separate questions of whether same-sex parenting or legalised same-sex marriage have an adverse effect on children.

A requirement of sound science is to distinguish the effect of the exposure of interest from those of confounding variables. The Critique authors cited several studies as “evidence of harm” for children of same-sex parenting while arguing that methodological limitations of the ACHESS study10 preclude its use as evidence for claim that “same-sex parenting is neutral or beneficial for children.” Upon closer inspection, not one of the studies claimed as “evidence of harm” found poorer outcomes attributable to parental sexuality. Rather, any ostensible disadvantages appear to be the result of other confounders such as family breakdown. The NFSS found disadvantages using a methodology that actively inflated this confounder, and even these findings were not robust upon closer analysis. The remaining four larger studies did not actually find any disadvantages after controlling for confounders. On the contrary, with their very large “random, representative sample[s]… of sufficient size,” these studies constitute some of the strongest evidence yet that same-sex parenting is as good as different-sex parenting. Indeed, these consistent “neutral or beneficial” findings were in spite of very large sample sizes and authors known for political activism against same-sex marriage.11, 12

Overall, the Critique’s analysis of the scientific literature on same-sex parenting consists of gross misrepresentations and outright falsehoods, misleading the Australian public and medical community. It should be retracted by its authors and medical practitioners who have signed the petition in its support should withdraw their signatures.



  1. Australian Medical Association. Position Statement on Marriage Equality 2017 [Internet]. Barton ACT: AMA; 2017 [cited 2017 August 17]. Available from:
  2. Middleton C, Pollnitz R, Dunjey L, Liangas G, van Gend D, Kafataris C. Medical Critique of the Australian Medical Association’s Position Statement on Marriage Equality [Internet]. 2017 [cited 2017 August 17]. Available from:
  3. Sullins DP. Emotional problems among children with same-sex parents: Difference by definition. British Journal of Education, Society and Behavioural Science. 2015;7(2):99-120.
  4. Allen DW. High school graduation rates among children of same-sex households. Review of Economics of the Household. 2013;11(4):635-658.
  5. Regnerus MD. How different are the adult children of parents who have same-sex relationships? Findings from the New Family Structures Study. Social Science Research. 2012;41(4):752-770.
  6. Cheng S, Powell B. Measurement, methods, and divergent patterns: Reassessing the effects of same-sex parents. Social Science Research. 2015;52:615-626.
  7. Potter D. Same‐sex parent families and children’s academic achievement. Journal of Marriage and Family. 2012;74(3):556-571.
  8. Wainwright JL, Russell ST, Patterson CJ. Psychosocial adjustment, school outcomes, and romantic relationships of adolescents with same‐sex parents. Child development. 2004;75(6):1886-1898.
  9. Sullins DP. The unexpected harm of same-sex marriage: a critical appraisal, replication and re-analysis of Wainright and Patterson’s studies of adolescents with same-sex parents. British Journal of Education, Society & Behavioural Science. 2015;11(2):1-22.
  10. Crouch SR, Waters E, McNair R, Power J, Davis E. ACHESS–the Australian study of child health in same-sex families: background research, design and methodology. BMC Public Health. 2012;12(1):646.
  11. Allen DW, Araujo RJ, Basset UC, Beckwith FJ, Benne RD, Bleich JD, et al. Brief of Amici Curiae 76 scholars of marriage supporting review and affirmance re April DeBoer, et al., v. Richard Snyder, et al [Internet]. SCOTUSblog; 2014 [cited 2017 August 17]. Available from:
  12. American College of Pediatricians*, Family Watch International, Marks LD, Regnerus MD, Sullins DP. Brief of Amici Curiae American College of Pediatricians, Family Watch International, Loren D. Marks, Mark D. Regnerus and Donald Paul Sullins in support of Respondents re James Obergefell and Brittani Henry, et al., v. Richard Hodges [Internet]. Supreme Court Review; 2015 [cited 2017 August 17]. Available from:
  13. Schlatter E. 18 Anti-Gay Groups and Their Propaganda. Southern Poverty Law Center Intelligence Report. 2010;140.

* The American College of Pediatricians is a fringe group estimated to represent < 1% of American paediatricians13 and should not be confused with the American Academy of Pediatrics.