A large-scale study using Norwegian registry data provides evidence that the similarity in education levels between partners is mainly the result of shared family environments and associated social factors, not simply individual educational attainment. The research, published in Nature Communications, introduces a new framework for studying how people end up with partners who are similar to them, and how these patterns influence the transmission of education from parents to children.
People often marry someone with a similar level of education. Researchers have long observed this trend and debated what drives it. One possibility is that individuals deliberately choose partners who have the same educational background. Another is that people who come from similar environments—such as the same neighborhoods, social classes, or family cultures—are more likely to meet and form relationships. These ideas are not mutually exclusive, and both could influence how couples form.
Understanding these processes matters because partner similarity can have long-term consequences. When highly educated people pair up, their children tend to benefit from both genetic and environmental advantages. This can reinforce patterns of social inequality over generations. However, earlier research has not always agreed on why these partner similarities exist or how they influence children’s outcomes. The definitions and methods used in the past have varied widely, leading to inconsistent results.
“Past research often reported conflicting findings about why partners tend to have similar levels of education. Some studies suggested that this similarity was largely due to genetic or constitutional factors, while others argued it was mainly due to environmental factors such as family background and social environment,” said study author Hans Fredrik Sunde, a postdoctoral researcher at the Centre for Fertility and Health at the Norwegian Institute of Public Health.
“Because definitions and methods varied widely, the field lacked a clear framework, making it difficult to interpret and compare results and understand what partner similarity means for intergenerational transmission of education. Crucially, earlier studies reached wildly different conclusions about why children attained similar education as their parents depending on what they assumed about partner similarity. Some concluded that genetic transmission was a sufficient explanation, others that environmental effects were very important.”
“We wanted to bring clarity to the study of partner similarity, and see if we could reconcile the discrepant findings in the literature. Ultimately, we want to understand (1) why some people end up together, and (2) why education runs in families.”
To explore these questions, the researchers developed a new way of thinking about partner similarity. Their framework separates the idea of “assortative mating”—when people pair up based on shared traits—from other sources of similarity, like growing up in the same type of environment. Importantly, they distinguish between different types of similarity: genetic, social, and personal (or idiosyncratic).
They also introduced the concept of a “sorting factor,” which represents the combination of traits that people are actually selecting for in partners. This might not be identical to formal education levels. For example, someone might value ambition, family background, or cultural compatibility—all factors that tend to be linked with education but are not exactly the same.
In this study, the researchers applied their model to more than 1.5 million individuals from 212,070 extended families in Norway. These included monozygotic twins, dizygotic twins, and full siblings. Using statistical models that considered partners, twins, and their children, they tried to tease apart how much of the similarity between partners was due to genetics, shared family environments, or other factors.
The team used data from the Norwegian population registry, which includes detailed records on births, education, and family relationships. They focused on Norwegian-born families with children born between 1975 and 1995, giving them access to both parent and offspring generations.
To model how people select partners, they looked at pairs of twins and their partners. This allowed them to compare how similar someone is to their sibling’s partner—a relationship that reveals whether partner similarity is due to shared family background or specific personal characteristics. If partner choice were based purely on individual achievement, these in-law relationships would not show strong similarities.
The researchers used two main models: one that focused on partner similarity (called iAM-ACE) and another that also considered children’s outcomes (iAM-COTS). These models separated out the influence of genetics, shared environments, and unique personal experiences.
The results indicate that people do not primarily choose partners based on their actual education levels. Instead, they tend to match with people who come from similar family backgrounds and environments that are conducive to achieving higher education. This kind of partner selection is referred to as “social homogamy.”
The estimated similarity in the underlying traits that influence partner selection was stronger than the similarity in observed education. Based on their models, the researchers found that only about 38 percent of the correlation between partners in education was due to genetic similarity. A much larger portion was explained by shared family environments, particularly those shared between siblings.
“We were struck by how much of the educational similarity between partners could be explained by social background rather than matching on heritable traits,” Sunde told PsyPost. “Earlier studies had convincingly shown that partners are more genetically similar than expected under direct matching on education, leading many to believe that partners must be matching on highly heritable traits associated with education. This was what we expected we would find when we started the project.”
“And while we did find that matching on heritable traits must play a role, we also found that partners must be primarily matching on social background, and that the greater-than-expected genetic similarity between partners can mainly be attributed to gene-environment correlations.”
Notably, when models assumed that partners chose each other directly based on education levels—rather than on deeper traits or backgrounds—they underestimated the importance of environmental influences. These models also failed to fit the data well.
“Partners are strikingly similar in their education, roughly as similar as full siblings (meaning very similar),” Sunde explained. “While other conclusions are harder to frame in terms of simple effect sizes, one point is clear: partners are not matching directly or solely on observed education. Instead, social background appears to be far more influential: around five times more important for partner similarity in education than education itself. The practical significance lies in recognizing that partner similarity in education shapes broader patterns of intergenerational transmission and social mobility.”
The researchers found that 62 percent of the similarity in education between parents and their children could be explained by genetic transmission. However, the remaining 38 percent was due to environmental factors—both the direct influence of parents on their children and environmental influences shared across the extended family.
“A key takeaway is not just the technical details of this study, but the broader fact that partners tend to be very similar in their education and social background,” Sunde said. “This has important consequences for society: resourceful parents often end up with other resourceful parents, and those advantages accumulate in their children.”
“Likewise, parents with fewer resources tend to partner with each other, reinforcing disadvantage. In practical terms, this means that if you have one highly educated parent, you are very likely to have two highly educated parents—and the same pattern holds in the other direction. This dynamic helps explain why social differences can persist or even widen across generations.”
The authors acknowledge some limitations. The models require very large datasets, particularly ones that include detailed family structures like twins and their partners. Additionally, while the models can estimate how much influence comes from genetics or environment, they do not reveal the exact mechanisms. For instance, the environmental effects could come from parenting style, cultural values, neighborhood characteristics, or other factors shared across families.
“A common misunderstanding is to treat assortative mating as synonymous with partner similarity, or as a single process with a single cause,” Sunde told PsyPost. “In our paper, we take great care to separate the concept of assortative mating (i.e., matching on similar traits) from the broader observation of partner similarity, and to distinguish the different possible causes, because these have often been conflated in the past. For example, social homogamy and genetic homogamy are both examples of assortative mating, and can both be involved in creating partner similarity for the same trait.”
“Another point worth stressing is that the family models we use—like all twin and extended-family models—are highly agnostic about the specific mechanisms. It essentially reframes the observation that siblings-in-law and piblings-niblings are much more similar than would be implied by genetic transmission alone. We do not know what the precise mechanisms are. That is for future research to identify.”
“I hope this paper can serve as a go-to source for researchers thinking about the different possible explanations for why partners are similar,” Sunde added. “The aim was to provide clarity in a field where concepts have often been muddled, and to give researchers a framework they can build on. For psychologists, the value lies in using the paper as a reference point: when they see partner similarity in their own research, they can use our distinctions and explanations to ask more precise questions about what might be driving it. The broader importance is that understanding these processes helps us see how private relationships and public inequalities are linked.”
The study, “Understanding indirect assortative mating and its intergenerational consequences for educational attainment,” was authored by Hans Fredrik Sunde, Espen Moen Eilertsen, and Fartein Ask Torvik.
