New research challenges the assumption that family breakdown is purely a parental story. Genomic data from 40,000 children tells a more complicated picture
Family research has an implicit assumption built into most of its models. Parents and environments are the inputs. Children are the outputs. The question researchers typically ask is how parenting style, household income, family stability, and neighborhood quality shape what a child becomes. The direction of influence is understood to flow from parent to child.
A new study published in PNAS, the Proceedings of the National Academy of Sciences, has tested whether that assumption captures the whole picture. Using genomic data from nearly 40,000 Norwegian families, researchers measured genetic propensities in children at or near birth and tracked what happened to the parents in the years that followed. The question was whether something about the child, encoded in their DNA before they had any chance to act on it, could predict what the parents would do next.
For most traits, the answer was no. But for one, it was striking.
Children with a higher genetic propensity for ADHD were significantly more likely to be followed by their parents separating. The effect held even after accounting for the parents’ own genetics. It was not simply that ADHD traits run in families and parents with ADHD genes are also more likely to split up. When the researchers statistically removed the parents’ genetic contribution, the child’s own genetic profile still predicted parental separation independently.
The study was led by researchers at the Norwegian Institute of Public Health and collaborating institutions, using data from the Norwegian Mother, Father and Child Cohort Study, a population-level registry that has tracked tens of thousands of families across decades.
What the genetic data actually measured
The researchers used a tool called a polygenic index, a score calculated from hundreds of thousands of genetic variants that together predict a trait across a population. Polygenic indices for ADHD have been developed from large genome-wide association studies and can estimate the degree to which a person’s genome predisposes them toward ADHD traits, even before any diagnosis has been made or any symptoms have appeared.
By calculating polygenic indices for the children in the cohort and then tracking family outcomes over the following years, the team could ask whether a child’s genetic predisposition, present from conception, correlated with what happened to the family structure around them.
This approach avoids a central problem in family research: the difficulty of separating what parents do from who they genetically are. When researchers find that children with ADHD are more likely to come from broken homes, it is hard to know whether the family breakdown caused the ADHD, whether ADHD-prone genetics in parents drove both the divorce and the child’s traits, or whether something about the child’s own traits shaped the family’s trajectory. By focusing on children’s polygenic indices while controlling for parents’ polygenic indices, this study gets closer to isolating the child’s independent contribution.
The finding that emerged was specific. Across traits including educational attainment, cognitive ability, conscientiousness, and neuroticism, the children’s genetic propensities showed limited ability to predict what happened to the parents. The exception was ADHD. A one standard deviation increase in the child’s ADHD polygenic index was associated with a 7% higher likelihood of parental separation in the years following the child’s birth.
The groups where the effect was strongest
Two patterns in the data stand out. The first is about gender. The association between a child’s ADHD genetic propensity and parental separation was substantially stronger for daughters than for sons. Among daughters, a one standard deviation increase in ADHD polygenic index was associated with an 11% higher likelihood of parental separation, compared to around 4% for sons. Why ADHD-prone daughters would place more strain on a parental relationship than ADHD-prone sons is not explained by the data, though the researchers note it may relate to how ADHD presents differently across genders and how those presentations interact with parental and social expectations.
The second pattern is more counterintuitive. The effect was strongest in high-socioeconomic-status families. Among families in the highest third of socioeconomic standing, a one standard deviation increase in the child’s ADHD polygenic index was associated with a 15% higher likelihood of parental separation.
This runs against what most people would assume. Low-income families face more stressors that might amplify the impact of a child with ADHD traits. But the data point in the opposite direction. One possible explanation is that high-SES families have higher expectations for developmental milestones and behavioral norms, and a child whose traits fall outside those expectations creates more visible tension. Another is that high-SES couples have more resources and independence to separate when a relationship becomes difficult, while low-SES couples may stay together for practical reasons regardless of relational strain.
The researchers do not claim to have identified the mechanism. What they documented is the pattern.
Why this reframes the standard model
The dominant model in family studies treats divorce and separation primarily as adult phenomena with downstream consequences for children. Decades of research have documented what separation does to children’s educational outcomes, mental health, and development. The policy conversation has focused on how to protect children from parental conflict and how to support them through family breakdown.
What this study adds is a causal arrow pointing in the other direction. Children are not only recipients of family circumstances. Their traits, present from birth, can shape the circumstances themselves.
This is not a new idea theoretically. Behavioral geneticists have described what they call evocative gene-environment correlations for decades: the phenomenon by which a child’s inherited traits evoke responses from the environment around them. A child who is impulsive, inattentive, and difficult to manage will evoke different parenting, different school experiences, and different family dynamics than a child who is not. This study provides population-level genomic evidence that the evocative effect extends to parental relationship stability.
“Family research has long emphasized how parents shape their children,” the researchers write, “but children may also influence their parents.”
What this does not show
The study is observational and specific to Norway, a country with high social support systems, generous parental leave, and relatively high rates of both cohabitation and separation compared to many other countries. Whether the same pattern holds in countries with different family structures, different ADHD awareness, or different economic pressures is unknown.
The polygenic index for ADHD predicts group-level tendencies, not individual outcomes. A high ADHD polygenic score does not mean a child has ADHD, and most children with high scores will not develop the disorder. The score captures a genetic predisposition that interacts with environment in complex ways.
The study also cannot tell parents what to do with this information. Knowing that a child’s genetic profile statistically predicts elevated separation risk does not provide guidance for individual families, and the researchers are clear that their findings should not be used to make predictions about specific children or couples.
What the study does establish, across 40,000 families with genomic data tracked over years, is that the influence flowing through a family is bidirectional. Children are not only shaped by the families they are born into. From the first years of life, they also shape those families in return.
The study “When children shape their families: Evocative effects analysis of 40,000 trios from the Norwegian Mother, Father and Child Cohort Study” was published in PNAS in July 2026.
Source: Norwegian Institute of Public Health / PNAS. DOI: 10.1073/pnas.2532642123
