2026 · bioRxiv (Cold Spring Harbor Laboratory)
Genomics of Educational Attainment Across 80 Years of Social and Political Transformation in Germany
Verdict
UNSCIENTIFIC
Neither the paradigm nor the study inside it passes. The verdict is unscientific.
1·Study methodology · inside the paradigm
UNSCIENTIFIC
Does not follow the scientific method within Genomics / DNA Model.
2·Paradigm · Genomics / DNA Model
UNSCIENTIFIC
Fails the tests. The worse layer decides.
How the verdict is decided
The verdict grades both layers: the paradigm this study assumes, and how the study was carried out inside it. Both must be scientific for the whole thing to be scientific. A clean method inside an unscientific paradigm is unscientific. An unscientific method inside a better paradigm is unscientific. A split on either layer is mixed. The worse layer decides.
1·Study methodology
Did this study test its claim with methods that are independent, falsifiable, and non-circular?
UNSCIENTIFIC
This study claims that DNA-based scores for educational attainment, called polygenic indices, are associated with years of education and intergenerational educational mobility across 80 years of German history, and that these associations vary modestly by birth cohort, region, and gender. All three tests fail.
Independently Verifiable
Fail
The predictor only exists as the model's interpretation of chip fluorescence weighted by model-derived numbers; a stranger sees light, not a genetic score for education.
Falsifiable
Fail
A null association would be called insufficient predictive power or missing heritability, not evidence that genetic variants do not influence education.
Non-Circular
Fail
The arrays, reference panels, GWAS weights, and ancestry filters were all built from the DNA model the study claims to validate.
Why
They took blood or saliva from 13,049 people and ran it through chips that lit up in patterns. The model says each pattern marks a genetic variant. They weighted those patterns using numbers from a prior study of three million people that had already assumed the patterns are real variants shaping education. They added the weighted readings into a total they call a polygenic index. Then they checked whether people with higher totals completed more years of schooling. The years of schooling are real. A person either finished university or did not. The polygenic index is not real in the same way. It exists only after the model converts light into variants, variants into weights, and weights into a score. Nobody who rejects the DNA model can see that score. They see fluorescence on a chip. The study also filtered every participant through ancestry principal components computed from reference panels built by the same model, so even the group definitions presuppose the framework. If the score had shown zero association with education, the field would have called it insufficient power or missing heritability, not evidence that genes do not shape schooling. The null was available on paper but structurally blocked from counting as a disproof.
2·Paradigm · Genomics / DNA Model
Does the framework this study assumes pass the three tests?
UNSCIENTIFIC
Independently Verifiable
Fail
Nobody has seen a gene or DNA inside a living cell doing what the model says it does; the array produces light and the model supplies the meaning.
Falsifiable
Fail
Every null result in genomics gets absorbed as underpowered design, missing heritability, or complex architecture rather than evidence the model is wrong.
Non-Circular
Fail
The reference genome, the arrays, the imputation panels, and the GWAS weights were all built from the DNA model they claim to validate.
Why
The genotyping array produces fluorescence. The model says the fluorescence marks genetic variants. A stranger who has never been taught the model sees light on a screen. They do not see genes. They do not see a hereditary code. They do not see variants that shape how long someone stays in school. The model converts the light into all of those things, and the field calls that conversion discovery. The reference panels used to impute the readings were assembled from sequencing that assumed the model. The arrays were designed to detect what the model already named. The GWAS that produced the weights assumed the readings are real variants with real effects. Every tool in the chain was built from the conclusion it claims to confirm. When a polygenic index predicts almost nothing, the field calls it missing heritability and says more variants are needed. When a CRISPR cut produces no phenotype, the field says the guide was poorly designed. No result kills the paradigm. The null is always a technical limitation, never a disproof. The model treats a tagged signal as the cause of a visible outcome rather than asking what actually shaped the person's life. Suppressing or celebrating the signal does not explain why the person finished school or did not.
From the study
The focal polygenic index was PGI-Education, based on the EA4 GWAS of educational attainment in approximately three million individuals of European genetic ancestries (Okbay et al., 2022). Posterior SNP weights were estimated with SBayesR (Lloyd-Jones et al., 2019) and applied with allele-aware PLINK2 scoring (Chang et al., 2015).
Analyses were restricted to respondents classified as genetically similar to European reference populations, matching the ancestry composition of the discovery genome-wide association studies (GWAS) (Price et al., 2006). Respondents were projected onto principal components estimated in the 1000 Genomes phase 3 reference panel
Genotyping used Illumina Global Screening Arrays in SOEP-G and TwinLife, Affymetrix Genome-Wide Human SNP Array 6.0 in BASE-II and SHIP-START, and Illumina HumanOmni2.5 and Global Screening Array batches in SHIP-TREND.
We discuss the methodological limitations of polygenic indices for examining gene-environment interactions, and emphasize that evaluating whether these differences represent progress towards equal educational opportunity requires a normative framework that social science genetics alone cannot supply.