The “Schizophrenia is 80% Heritable” Fallacy: Deconstructing a 2003 Twin Study Meta-Analysis

The mainstream psychiatry literature contains the ubiquitous and mostly unchallenged claim that schizophrenia is roughly 80% heritable. A 2003 twin study meta-analysis by genetic researchers Patrick F. Sullivan, Kenneth S. Kendler, and Michael C. Neale is often cited in support of this claim. The author argues that the 81% heritability estimate Sullivan and colleagues produced is invalid because (1) schizophrenia twin studies should not be interpreted genetically because they are based on the unsupported assumption that reared-together MZ (identical) and same-sex DZ (fraternal) twin pairs grow up experiencing “equal environments”; (2) especially in the older studies, schizophrenia was not reliably identified (diagnosed) for research purposes; and (3) among other problems, heritability estimates do not indicate the “strength” or “magnitude” of supposed genetic influences. The author then discusses how Sullivan and colleagues performed their meta-analysis and their decisions on which twin studies to include or exclude. They failed to mention three schizophrenia twin studies, two of which were known to at least one of the co-authors. They decided to “relax” their original meta-analysis inclusion criteria because only four studies qualified. This decision led to their inclusion of mid-20th-century studies conducted by researchers with strong genetic confirmation biases, who did not define schizophrenia or make blindfolded diagnoses. The author concludes that Sullivan and colleagues’ 81% schizophrenia heritability estimate should be disregarded in the context of a proposed replication crisis era re-evaluation of behavioral twin studies and their underlying assumptions. More than six decades of failed DNA-based attempts to identify genes that cause schizophrenia support this conclusion.