Unreviewed research platforms are inadvertently hosting debunked theories about race and genetics. A recent analysis reveals that individuals promoting scientific racism use mainstream open science archives to distribute flawed comparisons of human behavior. The study was published in Behavior Genetics.
Over the last decade, the open science movement has drastically changed how academics share information. Researchers frequently upload early drafts of their manuscripts, known as preprints, to internet archives. This allows scientists to bypass the traditional peer review process and share their data immediately.
Peer review usually involves independent experts evaluating a study for methodological errors before it is accepted by an academic journal. By skipping this step, open science platforms accelerate scientific discovery. They also remove the editorial gatekeeping that typically prevents poorly conducted or pseudoscientific research from reaching the public.
Eugenics is a historically debunked movement claiming that human populations can be improved through selective breeding. Related concepts of scientific racism falsely suggest that there are fundamental biological differences in cognitive ability or behavior among socially defined racial groups. These ideas were largely pushed out of the academic mainstream following the mid-twentieth century.
Proponents of these discredited concepts instead relied on obscure fringe publications to share their ideas. Foremost among these is a publication called Mankind Quarterly, which was founded by segregationists and routinely publishes racist pseudoscience. Because mainstream scientists quickly learned to ignore these specialized outlets, the authors struggled to reach broader audiences.
To appear more credible, some of these individuals have recently changed their tactics. They have begun uploading their papers to popular preprint servers right alongside legitimate scientific studies.
Evan J. Giangrande, a researcher at the Broad Institute of MIT and Harvard, wanted to quantify the scope of this issue. He sought to understand how often flawed genetic studies regarding group differences appear on modern preprint platforms. He also aimed to categorize the common tactics used in these papers.
Giangrande started his investigation by identifying all authors who published an article in Mankind Quarterly between 2014 and the spring of 2024. He then searched for those specific authors across four major open science archives. These online archives included PsyArXiv, bioRxiv, SocArXiv, and OSF Preprints.
The search returned 79 distinct preprints tied to these authors. Giangrande narrowed the list to 42 papers focused specifically on behavioral differences among human groups. He categorized each paper by the traits it examined, the groups it compared, and the methods it employed.
These preprints attracted thousands of online views and document downloads. A majority of the papers categorized humans into socially defined racial, ethnic, or national groups. The author noted that most of the group comparisons focused on intelligence, cognitive ability, or educational attainment.
Out of the 42 papers focused on group disparities, 31 discussed genetics and 13 performed new genetic analyses. Giangrande wrote that these analyses routinely relied on elementary misunderstandings of biology. The papers used these misunderstandings to falsely argue that social disparities are caused by innate genetic gaps rather than environmental factors.
Many of the preprints incorrectly assumed that estimates of heritability within one specific population can explain differences between completely separate populations. Heritability is a measure of how much genetic variation accounts for a trait within a specific group of people at a specific time. Mainstream geneticists agree that this metric cannot be used to explain the average differences seen between distinct, separate groups.
Other papers included in the sample misused polygenic scores in an effort to establish fundamental genetic differences. Polygenic scores are mathematical estimates of a person’s genetic likelihood for a certain trait based on thousands of tiny DNA variations. Comparing these scores across different ancestral groups is notoriously unreliable because the variations do not map equally across populations.
A few preprints relied on a technique called admixture regression, assessing individuals with mixed ancestry to separate genetic impacts from environmental ones. Biologists heavily criticize this method for its inability to accurately account for social confounds. Despite these basic mathematical errors, the preprints presented their findings as definitive proof of hereditary differences in intellect.
The authors of these preprints often utilized publicly available data or federally funded databases to perform their analyses. They interpreted this data through the lens of thoroughly debunked evolutionary frameworks. One prominent example was the cold winters theory, which baselessly argues that humans who migrated to colder climates evolved higher intelligence to survive harsh environments.
Another common theme identified in the preprints was differential-K theory. This racist biological concept suggests that certain minority populations evolved to have high fertility rates and low investment in offspring while majority groups evolved the opposite traits. The papers also pushed dysgenics, a false eugenic concept suggesting that a population’s genetic quality is deteriorating due to the reproductive rates of marginalized groups.
By using complicated methods, statistical software, and academic formatting, these papers create a false sense of authority. Giangrande reasoned that unfamiliar readers might struggle to distinguish this content from legitimate scientific inquiry. Unsuspecting public audiences could then cite these preprints as proven facts in online discussions.
The study features a few limitations. The inclusion criteria required a prior publication in Mankind Quarterly for an author to be tracked in the search. This strict rule means the calculated number of problematic preprints is likely an underestimation of the actual volume of pseudoscience online.
The research also focused explicitly on four platforms with relevance to behavioral genetics. It did not catalog other general research clearinghouses or newer artificial intelligence platforms. The issue of weaponized science extends beyond preprints to social media, blogs, and occasionally mainstream academic journals.
Giangrande outlined several potential responses for the scientific community to handle this recurring issue. He proposed that scientists must be extremely precise in their own language to prevent their legitimate data from being misconstrued. Vague terms regarding race and genetics invite bad actors to exploit the remaining confusion.
Improving public education regarding human genetics will also help everyday readers spot logical errors in pseudoscientific work. Individual academics can actively format rebuttals and label flawed research when they encounter it online to provide context for casual readers. Finally, the study recommends that open science platforms could provide explicit warnings that unreviewed content may not meet basic scientific standards.
The study, “The Preprint Problem: Fringe, Genetically Informed Studies of Group Differences in Behavior Housed on Open Science Platforms,” was authored by Evan J. Giangrande.