The family you grew up in and the relationships you have now

In plain words

Many people believe that the family they grew up in sets the pattern for their adult relationships. This review checks five versions of that belief against studies that followed children for decades, compared adopted people with both sets of parents, compared the children of twins, or pooled national panels of about 20,000 people. Part of the belief holds. In one small study, the way a family argued when a child was 14 predicted how that child argued with a partner 17 years later, and divorce runs in families. But in a large Swedish adoption study much of the family resemblance in divorce travelled through genes, birth order shows no lasting effect on broad personality, and early care predicts adult relationships only weakly. These are findings about groups, and none says what any one person's upbringing did to them.

Why a family resemblance is hard to read

A resemblance between parents and their adult children in how they treat a partner can come from what the parents did, from the genes they passed on, or from circumstances such as income that shaped both generations. Studies that ask adults to recall their childhood add a further problem, since memories are coloured by present experience, which can inflate the link between an old family and a current marriage (Whitton et al., 2008).

The studies below handle these problems in different ways. Some observed families and then followed the children for decades. Others compared adopted people with the parents who raised them and the parents who gave them their genes, compared the children of identical and fraternal twins, or pooled panels large enough to show very small effects. Each answers a narrower question than the popular belief asks.

Do children repeat their parents' marriage?

The most direct evidence comes from Whitton et al. (2008). At 14, each of 47 participants was recorded on audiotape with their parents as the family tried to settle a disagreement, and 17 years later each was videotaped with a spouse or cohabiting partner. Hostility and positive engagement in the family predicted the same behaviour in the adult couple, and family patterns accounted for 14 to 20% of the variance in couple behaviour after psychopathology was controlled. For men, family hostility also predicted poorer marital adjustment. The authors are open about the limits, which include a small sample, over 30% of whom had been psychiatric inpatients as adolescents, little diversity of class or ethnicity, and the possibility that genetic similarity between parent and child accounts for part of the continuity.

D'Onofrio et al. (2007) studied 2,310 young adult children of twins in the Australian Twin Registry. Comparing cousins whose twin parents differed in whether they had divorced holds constant much of what the twin parents share, including their genes when the twins are identical. In the raw data, children of divorced parents had about 1.9 times the odds of marital instability, counted as divorce or the end of a long cohabitation early in adult life. After the cousin comparisons, about 66% of that association remained attributable to the parental divorce or to risks tied to it within the family. That figure rests on the comparison among the children of identical twins, whose estimate was not statistically significant (p = .27), and the authors note that limited statistical power also made their comparison of identical and fraternal twin families imprecise. They conclude that environmental causation and genetic selection both contribute.

The strongest evidence against a large rearing effect comes from Swedish national registers. Salvatore et al. (2018) found that 19,715 adoptees resembled in divorce history the biological parents who did not raise them (odds ratio 1.24 after adjustment) and not the adoptive parents who did (0.96, an interval that includes no effect). They also resembled biological siblings raised elsewhere and not their adoptive siblings. The authors estimate the heritability of divorce across generations at 0.13. Heritability describes how differences are spread across one population at one time and says nothing about why any particular marriage ended; D'Onofrio et al. (2007) add that genetic influence on divorce operates within a culture and cohort, while differences between cultures and eras are environmental. One Swedish analysis, of 82,698 offspring, found stronger resemblance to mothers they lived with than to fathers they did not, an indirect sign of rearing, and the authors note that screening adoptive parents for a stable home may have hidden rearing effects.

Divorce runs in families in both countries. Where the Swedish adoption design attributes the resemblance mostly to genes, the Australian cousin comparisons credit the divorce itself with a large but imprecisely estimated share, and the prospective evidence that conflict styles carry over rests on one small, unusual sample.

Does birth order shape character?

Bowen gave sibling position a place in family-systems theory (Miller et al., 2004), and Rohrer et al. (2015) trace the idea back to Galton in 1874 and to Adler, who expected firstborns to carry excess responsibility. Birth order has a documented effect in the intellectual domain. Firstborns score slightly higher on intelligence tests than later-born siblings, and Rohrer et al. (2015) confirmed this at about 1.5 IQ points per position; in two-child families the older sibling scored higher in 6 cases out of 10. Self-rated intellect also fell by about a tenth of a standard deviation per position, and the decline persisted with measured IQ controlled. Among 783 Indian university students, Narayanan and Sriram (2023) found that only children reported the least emotional cut-off, one part of differentiation of self, and middle-born children the most, an association from a single cross-sectional survey.

Personality is where the evidence has been contested. Rohrer et al. (2015) review earlier studies, some of which confirmed Sulloway's predictions about personality, built on the idea that siblings take on divergent roles in the family, while many others supported only part of them or found no birth-order effect at all. Their own analysis combined national panels from the United States (5,240 people), Great Britain (4,489) and Germany (10,457), compared siblings within and between families, and found that birth order had no effect on extraversion, agreeableness, conscientiousness, emotional stability or imagination. The samples could detect very small effects, and the result held in each panel and both designs. Miller et al. (2004) likewise found little empirical support for Bowen's theory of sibling position. The personality measures were self-reports, which may miss parts of personality that people cannot see in themselves, and Rohrer et al. (2015) mention a proposal that birth-order effects might show inside the family without carrying over to relationships outside it. All three panels came from industrialised Western countries. What survives is a small advantage for earlier-borns on intelligence tests and self-rated intellect, no detectable effect on broad personality in samples built to find small effects, and, among the studies read here, no personality test of this scale outside the industrialised West.

Does growing up in one home make siblings alike?

Behaviour genetics divides differences among people into a part linked to genes, a shared environment (whatever makes children raised together more alike than their genes predict) and a non-shared environment (whatever makes them differ, plus measurement error). The cleanest estimate of the shared part is the similarity of unrelated children adopted into one home. Plomin and Daniels (2011), a reprint of a 1987 article, reported that such siblings correlate close to zero on most psychological characteristics, and concluded that the environmental influences that matter leave two children in one family about as different as two children drawn at random. Plomin (2011), commenting on the reprint, says the finding has not been seriously challenged since.

The same commentary carries the strongest form of the opposing view. Most environmental influence on most traits is non-shared, Plomin (2011) writes, but for antisocial behaviour in adolescence the shared environment accounts for about 15% of the variance (against about 40% non-shared), and academic achievement and childhood IQ show a shared component too. Burt (2009), a meta-analysis of 490 twin and adoption studies of young people, read here as an abstract only, found that the shared environment accounted for 10 to 19% of the variance in conduct problems, anxiety, depression and broad internalising and externalising problems, and generally 10 to 30% once age, informant and sex were considered; attention-deficit/hyperactivity disorder was the exception. All of these outcomes were measured before adulthood.

A small shared environment does not mean parents have no effect, and Plomin and Daniels (2011) call their data descriptive. Plomin (2011) argues that environments reach each child separately: parents describe treating their children only modestly differently (a sibling correlation of about 0.70), while the children describe treatment so different (about 0.25) that they might have grown up in different families. A 2000 meta-analysis of 43 papers that Plomin (2011) cites judged that measured sibling differences explained little, and his more optimistic reading credits them with 13% of the total variance. For most traits, siblings resemble each other mainly through shared genes, while a shared home accounts for a modest but real share of some childhood and adolescent problems, and neither finding says that what parents do is unimportant.

Does separateness from one's family predict how a person loves?

Differentiation of self is the central idea in Murray Bowen's family-systems theory. Calatrava et al. (2022) define it as a capacity for emotional self-regulation with two sides, telling one's thinking apart from one's feelings and keeping one's own position while staying intimately connected to others. In Bowen's account, as Narayanan and Sriram (2023) summarise it, a person's basic level is set by adolescence and stays fixed; the theory also holds that partners marry at similar levels and that the level passes down the generations.

The associations are well documented. Calatrava et al. (2022), a scoping review, found 295 eligible primary studies up to 31 July 2020, and of the 56 that tested whether higher differentiation goes with better couple relationships, 39 supported it, 12 gave mixed results and 5 did not. Miller et al. (2004) reported support for links with chronic anxiety, marital satisfaction and psychological distress.

The claims about how differentiation develops and passes between people fare worse. Of the 295 studies, 250 were cross-sectional and 33 longitudinal, one used a randomised design (with differentiation not its main outcome), and all used convenience samples; Calatrava et al. (2022) note that the cross-sectional designs limit what can be said about whether differentiation leads to satisfaction in a couple. The idea that people marry at their own level was not supported by 9 of 16 studies and supported by 2, and Miller et al. (2004) had found no support for it. Stability over time had been tested by one study with mixed results, and transmission across generations by five, two supportive, two not and one mixed. Miller et al. (2004) also found little support for triangulation and called for tests of Bowen's claim that the theory holds in every culture.

The question of culture bears on India in particular. Calatrava et al. (2022) report that the studies over-sample White, heterosexual participants from the United States. Narayanan and Sriram (2023), who found no earlier measurement of differentiation in Indian youth, surveyed 783 students aged 19 to 25 in Mumbai, Thane and Pune online in November 2020, during the pandemic. Overall scores were average, and scores varied with parents' education, with being in a relationship, with birth order and with gender, which the authors read as contrary to Bowen's claim of universality. They note that the questionnaire was normed on mostly white samples and that closeness counted as healthy in one group may be scored as fusion in another. The sample was urban and English-speaking, and apart from recording whether participants were in a relationship, the survey did not measure their relationships.

Differentiation is therefore a dependable correlate of distress and of reported satisfaction with a partner, mostly measured at a single point in time, while the claims that make it a family inheritance (stability, transmission and matched partners) are unsupported or barely tested.

Does early care settle adult relationships?

Attachment research gives this belief its strongest form. In a Minnesota cohort whose mothers were recruited during pregnancy in 1975 to 1977 while living below the poverty line, the quality of care was observed during the child's first three and a half years. Raby et al. (2015) analysed 243 of the children with outcomes through age 32, with about 30% of outcome data missing on average. Early maternal sensitivity predicted social competence, rated by teachers in childhood and adolescence and from interviews about romantic relationships at 23 and 32; with the teacher ratings, the correlation stayed roughly constant across the years (an average of .17).

In the same study that link proved fragile. Once child gender, ethnicity, maternal education and socioeconomic status were added, the lasting effect of early sensitivity on social competence was no longer distinguishable from zero (a path of .04), although its link with academic outcomes survived, and the authors suggest early socioeconomic factors may account for part of the social association. Fraley et al. (2013) used the NICHD cohort of 1,364 families recruited in 1991, of whom 707 completed attachment questionnaires at 18. Early maternal sensitivity predicted lower general avoidance at 18, but so did relative increases in sensitivity over the years, the child's social competence and the quality of a best friendship, while romantic avoidance was mostly unrelated to early care and attachment anxiety had few antecedents. They summarise earlier longitudinal studies as finding a correlation of about .10 between how good the parent-child relationship was and adult attachment.

Attachment also moves in adulthood. In five longitudinal data sets totalling 628 people, with attachment scored from personality ratings that study staff made from interviews and observed tasks between ages 13 and 72, Chopik et al. (2019) found that attachment anxiety fell on average with age, avoidance declined steadily, and being in a relationship went with lower levels of both. Early care, on this evidence, predicts later relationships weakly and does so alongside socioeconomic circumstances, later care, friendship and adult experience, which makes it an early influence that falls well short of a settled outcome.

What is not yet known

• Why the Swedish adoption design and the Australian children-of-twins design disagree about how much a parental divorce itself contributes, and whether country, cohort, the screening of adoptive homes or the definition of a break-up explains it.
• Whether the continuity in conflict style seen in 47 people by Whitton et al. (2008) holds in a larger, representative sample, as the authors ask.
• Whether differentiation of self leads to better couple relationships, follows from them, or shares a cause with them, and whether a person's level is stable across adulthood, which one longitudinal study has examined.
• How family of origin relates to adult relationships in regions little represented in these studies, such as South Asia. Among the sources read for this review, the only Indian evidence is one cross-sectional student survey.
• Whether birth order affects narrower traits such as risk-taking, or personality in larger families and non-Western countries, all of which Rohrer et al. (2015) list as open.
• Which experiences that siblings do not share account for their differences, where Plomin (2011) reports slow progress.

References

1. Burt, S. A. (2009). Rethinking environmental contributions to child and adolescent psychopathology: A meta-analysis of shared environmental influences. Psychological Bulletin, 135(4), 608-637. Abstract read via PubMed. https://pubmed.ncbi.nlm.nih.gov/19586164/
2. Calatrava, M., Martins, M. V., Schweer-Collins, M., Duch-Ceballos, C., and Rodríguez-González, M. (2022). Differentiation of self: A scoping review of Bowen Family Systems Theory's core construct. Clinical Psychology Review, 91, 102101. https://repositorio-aberto.up.pt/bitstream/10216/139264/2/526410.pdf
3. Chopik, W. J., Edelstein, R. S., and Grimm, K. J. (2019). Longitudinal changes in attachment orientation over a 59-year period. Journal of Personality and Social Psychology, 116(4), 598-611. Read as the authors' preprint. https://osf.io/3xhe4
4. D'Onofrio, B. M., Turkheimer, E., Emery, R. E., Harden, K. P., Slutske, W. S., Heath, A. C., Madden, P. A., and Martin, N. G. (2007). A genetically informed study of the intergenerational transmission of marital instability. Journal of Marriage and the Family, 69(3), 793-809. https://pmc.ncbi.nlm.nih.gov/articles/PMC2904988/
5. Fraley, R. C., Roisman, G. I., Booth-LaForce, C., Owen, M. T., and Holland, A. S. (2013). Interpersonal and genetic origins of adult attachment styles: A longitudinal study from infancy to early adulthood. Journal of Personality and Social Psychology, 104(5), 817-838. https://pmc.ncbi.nlm.nih.gov/articles/PMC3885143/
6. Miller, R. B., Anderson, S., and Keala, D. K. (2004). Is Bowen theory valid? A review of basic research. Journal of Marital and Family Therapy, 30(4), 453-466. Abstract read via PubMed. https://pubmed.ncbi.nlm.nih.gov/15532253/
7. Narayanan, M., and Sriram, S. (2023). An exploration of differentiation of self in Indian youth. Human Arenas, published online 14 April 2023, 1-28. https://pmc.ncbi.nlm.nih.gov/articles/PMC10103032/
8. Plomin, R. (2011). Commentary: Why are children in the same family so different? Non-shared environment three decades later. International Journal of Epidemiology, 40(3), 582-592. https://pmc.ncbi.nlm.nih.gov/articles/PMC3147062/
9. Plomin, R., and Daniels, D. (2011). Why are children in the same family so different from one another? International Journal of Epidemiology, 40(3), 563-582. Reprint of the 1987 article in Behavioral and Brain Sciences. https://pmc.ncbi.nlm.nih.gov/articles/PMC3147063/
10. Raby, K. L., Roisman, G. I., Fraley, R. C., and Simpson, J. A. (2015). The enduring predictive significance of early maternal sensitivity: Social and academic competence through age 32 years. Child Development, 86(3), 695-708. https://pmc.ncbi.nlm.nih.gov/articles/PMC4428971/
11. Rohrer, J. M., Egloff, B., and Schmukle, S. C. (2015). Examining the effects of birth order on personality. Proceedings of the National Academy of Sciences of the United States of America, 112(46), 14224-14229. https://pmc.ncbi.nlm.nih.gov/articles/PMC4655522/
12. Salvatore, J. E., Larsson Lönn, S., Sundquist, J., Sundquist, K., and Kendler, K. S. (2018). Genetics, the rearing environment, and the intergenerational transmission of divorce: A Swedish national adoption study. Psychological Science, 29(3), 370-378. https://pmc.ncbi.nlm.nih.gov/articles/PMC5854499/
13. Whitton, S. W., Waldinger, R. J., Schulz, M. S., Allen, J. P., Crowell, J. A., and Hauser, S. T. (2008). Prospective associations from family-of-origin interactions to adult marital interactions and relationship adjustment. Journal of Family Psychology, 22(2), 274-286. https://pmc.ncbi.nlm.nih.gov/articles/PMC3072794/