The fraternal birth-order effect as a statistical artefact: convergent evidence from probability calculus, simulated data, and multiverse meta-analysis.
Level 3 - non-randomized controlled study
Meta-analysis of observational/cohort and case-control studies combined with statistical simulations
PubMed 37609443 · doi:10.7717/peerj.15623
What was done
The authors investigated the empirical and statistical validity of the fraternal birth-order effect (FBOE) in sexual orientation using three complementary approaches: mathematical analysis using probability calculus to evaluate assumptions surrounding family-size adjustments and ratio variables; data simulations assessing error rates in common statistical practices; and a specification-curve and multiverse meta-analysis re-analyzing empirical evidence across 81 samples (64 male and 17 female samples, total N = 2,778,998).
What was found
The authors demonstrated that standard statistical methods in the FBOE literature rely on flawed assumptions regarding ratio variables and family-size adjustments, frequently generating false conclusions in simulations. In the multiverse meta-analysis (N = 2,778,998), the specific association between older brothers and homosexual orientation was found to be small, heterogeneous, not specific to men, and inflated by small-study effects. The abstract reports no specific numerical effect sizes or confidence intervals.
Why it matters
This study directly challenges a prominent biological hypothesis in sexual orientation research by showing that the fraternal birth-order effect is largely attributable to methodological artifacts and small-study bias.
Limits
The abstract does not report exact numerical effect sizes, confidence intervals, or specific heterogeneity statistics. The analysis is also constrained by the quality and reporting limitations of the underlying primary observational datasets.