In depth and fascinating article about this issue.
Has there been a meta-analysis or review of the evidence that checks for publication bias? Another argument against the "no effects" hypothesis could be that; due to the political nature of this topic, only studies finding neutral or positive effects are more likely to be published.
Unfortunately, I wasn't able to assess the degree of publication bias on this topic. However, I tend to think that publication bias is not driving the "no differences" hypothesis for a few reasons:
1. Kabatek and Perales (2021) and Mazrekaj, DeWitte, and Cabus (2020) both analyze the same dataset using different empirical strategies (controlling for number of family transitions vs restricting comparison to children born to same-sex couples) and find effects w/similar magnitude and direction.
2. The peer-reviewed literature which does claim to contradict the "no differences" hypothesis is of a fairly low quality (as shown in the section on contrarian studies). Because even these low-quality papers managed to make it to publication, I have a hard time imagining a scenario in which a high-quality paper which contradicts the "no differences" hypothesis was unable to be published.
When one looks at the marginal effects in the third column (which includes all of the controls), the effects of same-sex parents are no longer statistically significant.
The numbers in the parentheses are z-scores, not standard errors, as pointed out in the text underneath the table. But, for the sake of argument, let's assume that this insignificance occurs because of large standard errors. This would simply mean that the dataset lacks sufficient power to make any kinds of strong claims about the effects of same-sex parenting and more credible evidence is needed.
The good news is that the other sections of the essay cover more credible evidence. Mazrekaj, DeWitte, and Cabus (2020) and Boertien and Bernardi (2019) use much larger samples and find no negative effects of same-sex parentings on academic progress. Thus, the weight of the evidence stills supports the "no differences" hypothesis.
In depth and fascinating article about this issue.
Has there been a meta-analysis or review of the evidence that checks for publication bias? Another argument against the "no effects" hypothesis could be that; due to the political nature of this topic, only studies finding neutral or positive effects are more likely to be published.
Unfortunately, I wasn't able to assess the degree of publication bias on this topic. However, I tend to think that publication bias is not driving the "no differences" hypothesis for a few reasons:
1. Kabatek and Perales (2021) and Mazrekaj, DeWitte, and Cabus (2020) both analyze the same dataset using different empirical strategies (controlling for number of family transitions vs restricting comparison to children born to same-sex couples) and find effects w/similar magnitude and direction.
2. The peer-reviewed literature which does claim to contradict the "no differences" hypothesis is of a fairly low quality (as shown in the section on contrarian studies). Because even these low-quality papers managed to make it to publication, I have a hard time imagining a scenario in which a high-quality paper which contradicts the "no differences" hypothesis was unable to be published.
I hope that answers your question!
This makes sense. Thank you for your detailed reply!
The results from the appendix are publicly available at this link: https://link.springer.com/article/10.1007/s11150-013-9220-y/tables/8
When one looks at the marginal effects in the third column (which includes all of the controls), the effects of same-sex parents are no longer statistically significant.
The numbers in the parentheses are z-scores, not standard errors, as pointed out in the text underneath the table. But, for the sake of argument, let's assume that this insignificance occurs because of large standard errors. This would simply mean that the dataset lacks sufficient power to make any kinds of strong claims about the effects of same-sex parenting and more credible evidence is needed.
The good news is that the other sections of the essay cover more credible evidence. Mazrekaj, DeWitte, and Cabus (2020) and Boertien and Bernardi (2019) use much larger samples and find no negative effects of same-sex parentings on academic progress. Thus, the weight of the evidence stills supports the "no differences" hypothesis.
Thank for your response