When people pair up with partners who share their traits, such as educational background or lifestyle habits, they can leave subtle footprints in the human genome that persist across generations. A new study of hundreds of thousands of people in the United Kingdom and the United States finds genetic signatures of this partner matching, known as assortative mating, with the signatures for education and smoking appearing stronger among more recently born Americans. The findings were published in Behavior Genetics.
Assortative mating is the non-random pairing of individuals who have similar characteristics. When people choose a partner, they frequently select someone with similar schooling, body measurements, or daily habits rather than pairing up by chance. Over multiple generations, this non-random pairing can concentrate specific traits within families and alter how inherited traits are distributed across a population.
When partners who share heritable traits have children together, genetic variants related to those traits tend to become linked across the genome in ways that would not occur under random mating. Geneticists refer to this genome-wide connection as gametic phase disequilibrium, which occurs when DNA variants on completely different chromosomes become statistically correlated.
In a 2018 study, researchers developed an approach to detect this pattern by comparing genetic predictors calculated separately from even-numbered and odd-numbered chromosomes within unrelated people. Under random mating, genes on separate chromosomes have no relationship to one another, so any positive correlation between these chromosome halves provides evidence of cumulative partner matching across past generations.
Partner matching matters beyond individual couples, because concentrating similar traits within families may widen social and health differences between households over time. Similarity between partners does not necessarily mean people choose each other for those traits, however. For instance, a study covered by PsyPost in 2025 indicated that educational similarity between partners arises mostly through shared social and family backgrounds rather than direct selection on education itself. Gretchen R. B. Saunders of the University of Minnesota conducted the new study to evaluate whether education, physical traits, and health behaviors show genetic signatures of partner matching, and whether those signatures have changed across twentieth-century birth cohorts.
To explore these dynamics, Saunders analyzed large genetic datasets from two major resources: the UK Biobank in the United Kingdom and the All of Us Research Program in the United States. The analysis included 82,398 likely spouse or long-term cohabiting opposite-sex pairs and 16,273 full sibling pairs from the UK Biobank, along with 333,491 unrelated individuals from the UK Biobank and 228,875 unrelated individuals from the All of Us cohort. Because existing genetic discovery models were developed primarily using participants of European descent, the researcher restricted all samples to individuals of European genetic ancestry to maintain statistical accuracy.
The investigation examined eight traits: educational attainment, height, body mass index, and five smoking and drinking behaviors. The substance use measures included whether an individual ever smoked regularly, their age when starting, the maximum cigarettes smoked per day, whether they quit smoking, and typical drinks consumed per week. For each participant, polygenic scores were constructed, which are numerical scores that sum up thousands of tiny DNA variations across the genome to estimate an individual’s genetic predisposition for a trait.
The author used several analytical strategies to track mating patterns across generations. First, the study measured direct similarities in physical traits and habits between spouses in the UK Biobank, as well as similarities in their polygenic scores. Next, polygenic score correlations among full siblings were examined to see if they exceeded the 0.50 level expected under random mating, which indicates assortative mating in the parents’ generation. Finally, the author calculated correlations between even-numbered and odd-numbered chromosomes in unrelated individuals to detect broad genomic signatures of assortment and tested whether these signatures changed across birth years.
The results indicated positive similarities between spouses for observable traits across all eight measures, evaluated on a correlation scale where 0 represents no resemblance and 1 represents a perfect match. Educational attainment showed the highest spousal correlation at 0.45, closely followed by weekly alcohol consumption at 0.42. Moderate spousal resemblances appeared for smoking cessation at 0.33, body mass index at 0.26, and smoking initiation at 0.23.
These patterns also pointed to two distinct trait clusters among couples. People with higher educational attainment tended to partner with individuals who were taller, had a lower body mass index, and were less likely to smoke. In contrast, individuals who engaged more heavily in smoking or drinking tended to pair with partners who shared similar substance use habits and higher body mass index values.
When examining polygenic scores, spousal genetic resemblances mirrored these observable traits but were smaller in magnitude. The strongest genetic pairing occurred for educational attainment, with a spousal genetic correlation of 0.16, followed by height at 0.08 and smoking initiation at 0.06. Sibling comparisons also supported the conclusion that partner matching occurred in previous generations, as full siblings shared more genetic similarities than the expected 0.50 benchmark for most analyzed traits.
On average, the genetic correlations between partners were about 80 percent smaller than the correlations for the traits themselves. Part of this gap is expected, because polygenic scores capture only a fraction of genetic influences, but it exceeded 90 percent for habits that can change during a relationship, such as drinks per week and cigarettes per day. Saunders interprets this as a sign that partners also come to resemble each other through shared settings and mutual influence, while traits largely set before partners meet, such as height and education, showed the smallest gaps.
In addition, comparisons between spouse pairs and sibling pairs suggested that genetic assortment was either stable or expanding from the parental generation to the offspring generation. Traits such as educational attainment, height, and smoking initiation displayed increases in genetic resemblance between partners compared to the excess similarity observed between siblings.
In the unrelated samples, the even-odd chromosome method provided evidence of persistent genetic assortment across the genome. Both the UK Biobank and All of Us cohorts showed an even-odd chromosome correlation of roughly 0.07 for educational attainment and 0.04 for height and smoking initiation. Under random mating, this correlation would be expected to be zero. Cross-trait analyses also indicated that genetic predispositions toward higher education were negatively correlated with genetic predispositions toward smoking across separate chromosomes, though the author notes such patterns could also partly reflect shared biology or subtle population differences.
Birth-cohort analyses indicated differences in how these patterns shifted over the twentieth century between the two countries. In the UK Biobank, which included people born between 1936 and 1970, the even-odd measure remained largely stable across decades, showing an increase only for smoking cessation. In the American All of Us cohort, which spanned birth years from 1900 to 2005, genetic assortment showed notable increases across generations for smoking initiation, smoking cessation, and educational attainment. Specifically, the genetic signature for educational assortment in the United States rose from approximately 0.04 among those born in the 1930s to 0.09 among those born around 2000.
Saunders notes that the stronger U.S. trends may partly reflect the All of Us sample’s much wider range of birth years, which makes small changes easier to detect. She also points to differences between the two countries, including a faster and more varied expansion of American higher education and declines in U.S. smoking that were more sharply divided along social lines.
“The main takeaway is that patterns of similarity between partners can leave detectable genetic signatures across generations,” said Norbert Meskó, a full professor in the Department of Cognitive and Evolutionary Psychology at the University of Pécs and director of the Evolutionary Psychology Lab, who was not involved in the research. “People resemble their partners in education, smoking, and alcohol consumption, and this study also finds smaller but consistent similarities in genetic scores associated with these characteristics.”
“The finding that deserves particular attention is that some of these genetic signatures were stronger in more recent U.S. birth cohorts, especially for educational attainment and smoking-related traits,” Meskó told PsyPost. “This suggests that the accumulated genetic consequences of assortative mating may differ across generations. However, I would distinguish this finding from direct evidence that people today are choosing more similar partners: the study primarily infers historical mating patterns from genetic data.”
There are a few things to keep in mind when interpreting these findings. The study focused exclusively on participants of European genetic ancestry, which means the results cannot be assumed to apply to other ancestral or cultural backgrounds where social dynamics and mating customs might differ.
“I find the broad evidence for assortative mating more convincing than the evidence that it has become stronger over time,” Meskó explained. “These are comparisons between people born in different periods, rather than direct observations of partner formation across successive generations. Differences in recruitment, participation, survival, and population composition could influence those comparisons. Large samples improve statistical precision, but they do not automatically eliminate systematic bias.”
Additionally, the UK Biobank does not formally record marital status through official registries, meaning spouses were identified using shared addresses and household characteristics. While strict checks were applied, minor misclassifications could have occurred, which would tend to make the estimated partner similarities smaller than they truly are.
“The clearest potential misunderstanding would be to conclude that our genes determine whom we fall in love with, or that people somehow recognize and select partners with similar genetic profiles,” Meskó warned. “Neither conclusion follows from this study. People meet within schools, workplaces, neighborhoods, and social networks that already structure their opportunities to encounter potential partners. Similarity arising through those settings can also produce genetic similarities between partners.”
“Likewise, a polygenic score associated with education is not a measure of someone’s fixed educational potential,” Meskó added. “It is a statistical predictor derived from associations in particular populations and environments. Finally, the implications for inequality should be treated as plausible consequences rather than outcomes directly demonstrated here. The study does not establish how much these mating patterns have increased socioeconomic or health inequalities.”
It is also worth noting that non-random mating cannot be separated entirely from other social processes, such as partners adopting each other’s habits over time or meeting simply because they share the same neighborhood or workplace.
“I would particularly welcome longitudinal research that measures people before they form a relationship and follows both partners over time,” Meskó said regarding next steps. “This would help distinguish three processes: choosing someone who is already similar, meeting someone through a shared social environment, and becoming more similar through life together. Combining that design with genetic data and information across generations would make the historical interpretation stronger.”
“Overall, I see this as a useful contribution to understanding how social patterns of partnering become reflected in genetic data,” Meskó concluded. “Its strongest message is that partner similarity matters for interpreting population genetics; its more tentative message concerns how that similarity has changed over time.”
The study, “Changing Patterns of Assortative Mating for Educational, Substance Use, and Anthropometric Traits,” was authored by Gretchen R. B. Saunders.
Leave a comment
You must be logged in to post a comment.