To the birth order page

To The Nurture Assumption home page

Why Can't Birth Order Account
for the Differences Between Siblings?

by Judith Rich Harris

It's one of the things that convinces people that my theory of development might not be as absurd as it appears at first glance. Here's how a behavioral neuroscientist expressed it recently in an online discussion group:

I think [Harris's] book got attention because its ideas were startling and yet somehow resonant with many people's experiences. For me, that resonance . . . had to do with my being one of three siblings who, despite seemingly similar parenting, are each very different (though each generally happy and successful). (Epstein, 2001)

Similar parenting but different kids -- a familiar story. How do we account for the differences between siblings? If children's personalities are formed by "nature" (their genes) and "nurture" (the way their parents bring them up), and if siblings have a similar genetic makeup (they inherited their genes from the same parents) and a similar home environment -- the same home environment, in most important respects -- how come they're not more alike? If being reared in poverty, or in luxury, or by a single parent, or by two parents who are constantly at each other's throats, or by two parents who aren't home much, or by strict parents, or by permissive parents -- if these things have effects on children, how come they don't have the same effects on all the children in the family?

It's not an illusion: siblings really do differ more than conventional views of  "nature and nurture" would lead you to expect. The results started appearing in the 1970s, reported by researchers in the field of behavioral genetics. The methods devised by these researchers enabled them, for the first time, to tease apart the contributions made by heredity and environment to the variations in personality that are so noticeable in our species. Though the methods were new, the questions these researchers were asking were familiar ones: Why is Justin more aggressive than Jared? Why is Claire shy and Alison outgoing? Why is Jamie cheerful and Taylor grouchy?

The behavioral genetic research produced two outcomes. The first surprised no one (or should have surprised no one): about half of the variation in personality is due to variations in genes. One of the reasons that siblings differ is that they have different genes. Just as one sibling might be born with straight hair and the other with curly hair, one might be born with a cheerful temperament and the other with a tendency to be irritable or gloomy.

But the second outcome produced by the behavioral genetic research was more like a question than an answer. The other half of the variation in personality -- the part that wasn't genetic so it had to be environmental -- couldn't be pinned on any of the usual suspects. Everyone had assumed that the nongenetic variation in adult personality was the result of differences in the circumstances in which individuals were reared. This would mean that two children reared in different homes by different parents should differ more in personality than two reared in the same home by the same parents. But that's not what the studies found. What they found was that two people reared in the same home by the same parents were not noticeably more alike than two people picked at random from the population, once you deducted the similarities due to shared genes. Full siblings, on average, share 50 percent of their genes. (For an explanation of this statement, click here.)

Siblings are alike in personality only to the extent that they share genes; if they do not share genes (if they are adoptive siblings or stepsiblings) they aren't alike at all. Growing up together -- going on the same trips to the museum or the ballpark, coming home to the same city apartment or house in the suburbs, living for 18 years with the same parent or parents -- does not make them more alike.

Why Are the Differences Between Siblings a Problem for Developmental Psychologists?

Why does it matter that being reared in the same home and the same family doesn't make siblings more alike? Why should we expect that being reared in the same home would make people more alike?

The answer is that people who do research in developmental psychology (my field) would like to think that the home environment has predictable effects on children, and hence predictable effects on what they will be like as adults. Consider, for example, the personality trait psychologists call "conscientiousness" -- the tendency to be responsible and well-organized. Wouldn't you think that someone who was reared in a well-run home by conscientious parents would tend to be more conscientious as an adult than someone who was reared by happy-go-lucky slobs? Well, if this were the case, then two people reared by conscientious parents should, on average, both be more conscientious than two reared by slobs, which means that two people reared by conscientious parents (or two reared by slobs) should, on average, be more alike in conscientiousness than one reared by a conscientious parent and one reared by a slob. But that is exactly what the behavioral geneticists did not find. What they found was that two adoptive siblings reared in the same family are no more alike in conscientiousness than two adoptees chosen at random from different families, and that a pair of identical twins reared in the same family are no more alike in conscientiousness than a pair separated at birth and reared apart. These results show that being reared by conscientious parents does not, on average, make people more conscientious; nor does it make them, on average, less conscientious (Bouchard, 1994; Bouchard et al., 1990; Rowe, 1994).

The implication was unsettling. If being reared by conscientious parents has any effect at all on children, then it must be a random one. It must make some children more conscientious and others less so, unpredictably. And randomness and unpredictability are not what developmental psychologists expect to find. They believe that how a child is reared has predictable effects -- that there are good ways and bad ways to rear kids, good home environments and bad ones. They believe that good home environments, on the whole, produce better kids.

Many aspects of personality and behavior have now been studied with behavioral genetic techniques. In stark contrast to the here-today-gone-tomorrow results so common in psychological research, the behavioral genetic results have been resoundingly consistent. Despite differences in the ways personality is measured (standard personality tests, judgments by parents or teachers, direct observations of behavior, etc.), and differences in the kinds of subject pairs who participate (adoptive siblings, biological siblings or half-siblings, identical and fraternal twins, reared together or apart), the conclusions are almost invariably the same. Siblings are alike only to the extent that they share genes. Genes make biological siblings more alike in personality; growing up together does not.

The results are still coming in, but the handwriting was on the wall by 1987. That was the year when behavioral geneticists Robert Plomin and Denise Daniels published an article titled "Why Are Children in the Same Family So Different from One Another?" The overview Plomin and Daniels gave of the behavioral genetic findings, and their explanation of why these findings are a problem for developmental psychologists, have stood up well. The results they summarized have been backed up by later research, and they are just as much a problem for developmental psychologists today as they were in 1987.

How Did the Developmental Psychologists React?

The first real attempt by a developmental psychologist to face up to the implications of the behavioral genetic findings was made in 1991 by Lois Hoffman. Hoffman began by questioning whether siblings are really as different as the behavioral genetic evidence suggested; then she said that sibling differences are what you would expect, given the way that families work.

Now, at last, we get to birth order. Birth order (which she called "ordinal position") is one of the reasons, according to Hoffman, why siblings are different:

First, family roles and sibling interactions are very much affected by ordinal position. Firstborn children are treated differently by parents throughout life than are subsequent children. . . . In addition, it is a different experience to be an older sibling than it is to have an older sibling. (Hoffman, 1991, p. 193)

Hoffman cited studies showing that parents treat firstborns and laterborns differently: parents show "a special involvement" with firstborns, they give firstborns more responsibility and attention, they have higher expectations for the firstborn, and so on.

Unfortunately, as Hoffman admitted, Plomin and Daniels (1987) had already considered and rejected the possibility that birth order could account for the differences between siblings. Their rejection was based on plenty of evidence, because psychologists had been studying birth order for decades. The evidence (much of which had been summarized in 1983 by the Swiss researchers Cécile Ernst and Jules Angst) indicated that birth order has little or no effect on personality: well-done studies generally find that firstborns and laterborns are indistinguishable in personality. A 1990 book on sibling differences written by Plomin and his wife, developmental psychologist Judy Dunn, dismissed the idea that birth order could get developmental psychology out of its pickle. Dunn and Plomin agreed that parents do treat firstborns and laterborns differently, but they carefully explained why this differential treatment can't answer the question of why siblings differ in personality:

If there are no systematic differences in personality according to birth order, then any differences in parental behavior that are associated with birth order cannot be very significant for later developmental outcome. (Dunn & Plomin, 1990, p. 85)

Hoffman acknowledged that the behavioral geneticists had dismissed birth order as "unimportant," but conjectured that "personality outcomes are affected by a multiplicity of interacting environmental influences, and any given one is unlikely to explain much variance" (1991, p. 194).

Explaining the Variance

When Hoffman spoke of explaining the variance, she was talking about something developmentalists do -- or try to do -- all the time. Modern developmental psychology is all about explaining the differences among children (or differences among the adults the children become). The aim is to account for the variance -- the variation from one individual to another -- in whatever characteristic or behavior the developmentalists are interested in. The term variance has a precise mathematical meaning, and accounting for it involves a good deal of mathematical calculation, but the important point is that the researchers are trying to explain why some children score high and others score low on whatever it is they are measuring.

Let's say, for instance, that the researchers are interested in aggressive behavior. They get a numerical estimate of the aggressiveness of each child in the study, perhaps by asking the child's teachers or peers to indicate, on a scale of 1 to 10, how aggressive this child is compared to others of the same age. Because the researchers believe that children's aggressiveness is related to their parents' child-rearing style, they also interview the parents and get an estimate of how often each child is spanked at home. Then the researchers put the two sets of numbers together. They find that there is a correlation between the measure of the children's aggressiveness and the measure of how often they are spanked -- children who receive more spankings are, on average, judged to be more aggressive -- and they interpret this correlation as evidence that some of the variation in aggressiveness can be blamed on the spankings. The square of the correlation gives the proportion of the variance that has been "accounted for." If the correlation is .30 the amount of variance accounted for is .09, which means that spankings account for 9 percent of the variance in aggressiveness. Though it doesn't sound like much, developmentalists break open the champagne if they can do that well.

But the study I've just described -- a typical study in developmental psychology -- has a serious flaw: it didn't provide any controls for genetic effects. Whenever children are found to resemble their parents in some respect -- in this case, in aggressiveness (parents who spank are behaving aggressively) -- it's necessary to consider the possibility that some or all of the resemblance is due to the fact that children are genetically similar to their biological parents. That interpretation is supported by the results of studies that do provide controls for genetic effects. A study of adopted and nonadopted children by Kirby Deater-Deckard and Robert Plomin (1999) showed that genetic influences on behavior accounted for about half of the variance in the children's aggressiveness, when their aggressiveness was judged by their teachers (each child was assessed by five different teachers over a period of several years). In contrast, being reared in a given home accounted for zero percent of the variance. According to the teachers' judgments, adoptive siblings who were being reared in the same home by the same parents showed no resemblance at all in how aggressive they were in the classroom or the playground. Biological siblings did show some resemblance, but the resemblance was entirely due to their shared genes.

Deater-Deckard and Plomin's results, which are typical of behavioral genetic studies, leave half of the variance unaccounted for. Only half of the variation in aggressiveness was accounted for by the effects of genes. What about the other half? The other half of the variance -- the nongenetic differences in aggressiveness among the children in this study -- could not be attributed to any aspect of the home environment that siblings share. Many aspects of the home environment are shared by siblings: whether the parent or parents are harsh and irritable or warm and mellow; whether they do or do not believe in the use of physical punishment. The results showed that factors of this sort, which are experienced by all the children in the family, have no consistent effects on the children's aggressiveness. In other words, being reared by harsh, irritable parents who use spanking as a routine way of enforcing obedience does not, on average, make children more (or less) aggressive. And yet something other than genes makes them more (or less) aggressive.

Since aspects of the home environment shared by siblings couldn't explain the variation, the next question was obvious: What about aspects of the home environment not shared by siblings? For example, what if the parents were nicer to one sibling than to the other? What if one sibling got more affection, the other got more spankings? Couldn't that explain why they differ in aggressiveness?

As everybody knows, parents don't treat their kids all alike. And one of the things that makes parents treat their children differently -- this is something that Lois Hoffman and Dunn and Plomin agreed on -- is birth order. Firstborns and laterborns have different experiences in the home, right from the start. Firstborns are born to inexperienced and anxious parents, laterborns to veterans. Firstborns have their parents all to themselves for a while and then are abruptly dethroned by a rival; laterborns always have to compete for their parents' attention. Parents give firstborns more responsibility; they give laterborns more affection.

That's right: it's the younger child who gets more affection. Two studies (Dunn & Plomin, 1990; McHale et al., 1995) have shown that at least half of parents with two children admit to loving one better than the other, and a large majority of these parents -- more than 80 percent -- say they love their younger child best. These are big differences in parental affection. If being loved more by their parents made children less aggressive (or more aggressive), then we should see birth order effects on aggressiveness. But the teachers in Deater-Deckard and Plomin's (1999) study did not judge younger siblings to be any less aggressive (or more aggressive) than older siblings.

The goal here is to account for 50 percent of the variance in personality -- the 50 percent that is not genetic. If the differences between siblings are due in part to the effects of birth order, then birth order should account for some non-negligible portion of that variance. But it doesn't. In studies of adults, personality is usually measured with self-report personality tests, the kind in which subjects check off their agreement or disagreement with a list of self-descriptive statements. Birth order accounts for little or none of the variance in scores on these tests (Ernst & Angst, 1983; Sulloway, 1999; Jefferson et al., 1998).

Reluctant to give up their belief in birth order, some theorists have instead given up their faith in standard self-report personality tests (see "Why Did Sulloway's Results Differ From Those of Ernst and Angst?" forthcoming on this website). They've claimed that these tests are invalid -- that they are inaccurate or insensitive measures of personality (Kagan, 1998; Sulloway, 1998, 1999). This is like blaming the yardstick when one's theory of why some kids are taller than others fails to be confirmed. Standard self-report personality tests are the yardstick that produced the results we are trying to account for!

This is an important point so I will take the risk of belaboring it. We are trying to account for the unexplained differences in personality between siblings. How did researchers discover these perplexing differences between siblings? By giving them standard self-report personality tests. The variance we are trying to account for is, for the most part, variance in scores on standard self-report personality tests. So if birth order can't account for a visible portion of the variance in those scores, then birth order isn't the answer -- isn't even one of the answers -- we are looking for.

The validity of self-report personality tests has been demonstrated over the years in two different ways: by their agreement with other ways of measuring personality, such as judgments by other people (Jefferson et al., 1998); and by their agreement with other ways of measuring the effects of genetic and environmental influences. Behavioral genetic studies using self-report personality tests produce results that are consistent with those of studies that measure other things: aggressive behavior on the playground, the likelihood of getting divorced, and so on. Regardless of what is measured and how it is measured, the results are almost invariably the same: genetic factors account for 40 to 60 percent of the variance; shared home environment accounts for little or none.

The failure to find systematic differences between firstborns and laterborns on standard self-report personality tests is like an elephant in the living room: you can't ignore it and you can't get around it.

A Spoonful of Evolutionary Psychology

Some theorists have proposed that the differences between siblings result from the effects they have on each other, rather than from the way they are treated by their parents. Here, for example, is an explanation for sibling differences given by developmentalist Jerome Kagan:

The child holds a set of standards regarding proper behavior and has conceptualized his role in relation to authority. . . . Now let us suppose that another enters his life space whom he cannot ignore because the other is a competitor for some of the same resources the child desires. If the other is a sibling, the child perceives the other to share parental attention. The child now is pushed to differentiate himself from the other, toward the values of the parents, in order to retain the favored position. . . . Hence, first-borns with younger siblings are more likely to adopt the values of the parents and to be rigid about adherence to those early standards than younger siblings. (Kagan, 1976, p. 170; italics in the original)

A similar view was expressed twenty years later by Frank Sulloway in his book Born to Rebel (which he dedicated to Kagan):

Siblings compete with one another in an effort to secure physical, emotional, and intellectual resources from parents. . . . Siblings create different roles for themselves within the family system. These differing roles in turn lead to disparate ways of currying parental favor. Eldest children, for example, are likely to seek parental favor by acting as surrogate parents toward their younger siblings. Younger siblings are not in a position to ingratiate themseves with parents in the same manner. Their niche is typically less parent identified, less driven by conscientious behavior. . . . In particular, eldest children tend to identify more closely with parents and authority. (Sulloway, 1996, p. 21)

What Sulloway added to Kagan's formulation was a spoonful of evolutionary psychology. Siblings differ, according to Sulloway's theory, because they are engaged in a Darwinian struggle for survival -- a competition for parental favors that, in ancestral times, could have proved fatal for the loser, either because the war between siblings turned bloody (Cain and Abel are mentioned more than once in Sulloway's book) or, more commonly, because the loser's share of parental favors was inadequate to keep him or her alive.

The principles of evolutionary psychology that Sulloway makes use of were first proposed by William Hamilton (1964) and Robert Trivers (1985). Hamilton explained why it makes sense, in evolutionary terms, for humans to share food or put themselves in danger in order to help close relatives, such as their children. Close relatives share genes, so a person who increases the survival chances of her relative is also increasing the survival chances of the genes that she and the relative have in common (see Dawkins, 1976; Pinker, 1997).

Trivers explained why it makes sense, in evolutionary terms, for a person to put her own interests above the relative's interests: because they don't share all of their genes, only some of their genes. A parent and a child, for example, have 50 percent of their genes in common, not 100 percent, which means that the parent's interests don't always coincide with the child's. The result is what Trivers called "parent-offspring conflict": for example, the mother wants to wean the infant, the infant doesn't want to be weaned.

Like parents and offspring, siblings share 50 percent of their genes. Thus, siblings should help each other, just as parents help their offspring. But Sulloway's theory puts more emphasis on the 50 percent of their genes that siblings don't share. Some see the cup shared by siblings as half full; Sulloway sees it as half empty. It's the fact that siblings don't share all their genes that leads to "sibling conflict" (1996, p. 60). But the conflict is an uneven match, because the firstborn got there first and is bigger and stronger. Thus, firstborns and laterborns must devise different strategies in their competition for family resources. Differences in personality between siblings, according to Sulloway, are the outcome of these different strategies. Firstborns develop patterns of behavior in their attempts to monopolize their parents' attention and to dominate their younger siblings. Laterborns develop other patterns of behavior in their attempts to win some of their parents' attention and to placate their older siblings.

I do not doubt that children vie for their parents' attention and use different strategies to win it. But Sulloway is ignoring an important part of Trivers' theory. Here's how the evolutionary psychologist Steven Pinker explained it:

Trivers reasoned that, according to the theory of parent-offspring conflict, parents should not necessarily have their children's interests at heart when they try to socialize them. Just as parents often act against a child's interests, they may try to train the child to act against its own interests. . . . So even if children acquiesce to a parent's rewards, punishments, examples, and exhortations for the time being because they are smaller and have no choice, they should not, according to the theory, allow their personalities to be shaped by these tactics. (Pinker, 1997, pp. 447-448)

In other words, children should accommodate themselves to their situation in the family by developing temporary strategies -- strategies that will enable them to get by until the time comes when they are no longer dependent on their parents. They shouldn't allow their parents to have permanent effects on their behavior because the behaviors desired by the parents might not serve them well in their adult lives.

Trivers' reasoning is supported by the evidence, which shows that the patterns of behavior children acquire at home -- bossing around younger siblings, deferring to older ones, whining for their mother's attention, and so on -- are not carried to the schoolroom or the playground. Though these patterns of learned behavior may affect relationships with parents and siblings even in adulthood, they don't affect relationships with other people -- with peers, for example -- even in early childhood (Abramovitch et al., 1986; Stocker & Dunn, 1990). As I pointed out in my previous essay on this website (see "Why Are Birth Order Effects Dependent on Context?"), it wouldn't make evolutionary sense for patterns of behavior acquired within the family to generalize to social contexts outside the family, because these patterns of behaviors are likely to be counterproductive in other contexts. Evolution designed children wisely. Childhood is preparation for a later stage of life -- a stage on which the most important characters will probably not be labeled "Mom," "Dad," "little sister," or "big brother."

It is also inconsistent with the principles of evolutionary psychology to expect people to behave the same way with people outside the family as they do with their parents and siblings. Most mammals, including humans, behave differently toward kin and nonkin. The ability to make such distinctions is widespread in the animal kingdom (Mock & Parker, 1998; Pfennig & Sherman, 1995).

The theory that sibling differences are the result of competition between siblings is based on a particular pattern of family life: the nuclear, child-centered family, common today in industrialized societies. Observations of hunter-gatherer and tribal societies (e.g., by Eibl-Eibesfeldt, 1989) indicate that the kind of intense, prolonged parental care we give our children, and the intense, prolonged sibling rivalry that often accompanies it, are recent innovations in the history of our species. In traditional societies, intense parental (that is, maternal) care lasts only about three years. After that, children spend most of their time in the local play group. They don't bother competing for their mother's attention because it wouldn't work -- almost all of her attention is given to the latest baby. Siblings look after each other in these societies. Brothers are allies, not rivals.

The same is true of chimpanzee brothers (Goodall, 1986). Among nonhuman animals, the kind of sibling rivalry that results in physical aggression is found almost exclusively in species that rear their young in litters, rather than one at a time (Mock & Parker, 1998).

What About Twins?

In all human societies, a distinction is made between relatives and nonrelatives; as a rule, relatives are treated better. Humans are nice to their kin even if they don't think their favors will ever be reciprocated. Parents feed their children and older siblings defend their younger brothers and sisters against bullies without expecting to be repaid.

Among the psychological mechanisms that permit mammals to distinguish relatives from nonrelatives and close relatives from distant ones, an important one for humans is the recognition of similarity. Because multiple marriages and sex outside of marriage are common in most human societies, siblings often have different biological fathers and thus share only 25 percent of their genes. Similarly, a man is not necessarily the biological father of all his wife's children; he may share no genes at all with her latest baby. Consistent with the predictions of evolutionary psychology, people form closer emotional ties with others who resemble them physically or psychologically. For instance, parents feel more grief at the death of a child who resembles them than at the death of one who does not (Littlefield & Rushton, 1986; Segal, 1993).

Because twinning is relatively uncommon in the human species, it is unlikely that humans evolved special psychological mechanisms to deal with the twin relationship. But the same mechanism that causes people to feel closer to those who are similar to themselves makes identical twins -- who share 100 percent of their genes -- feel the closest of all. The loss of an identical twin is more devastating than the loss of a child or of a husband or wife. Compared to fraternal twins (who, like ordinary siblings, share about 50 percent of their genes), identical twins are less competitive and more cooperative with each other (Segal, 1993, 1999).

And yet the nongenetic differences in personality between identical twins, reared together from birth, are about as large as the nongenetic differences in personality between ordinary siblings (click here for an explanation of this statement). Reared-together identical twins aren't nearly as alike as most people think; in fact, they are no more alike in personality than identical twins separated in infancy and reared in different homes. Reared together or apart, the correlation between their scores on personality tests is only about .50 (Bouchard et al., 1990). This correlation is due entirely to genetic influences on personality; their environments have made them neither more alike nor less alike.

It is a point that bears repeating: their environments have made them neither more alike nor less alike. Some people who read about the behavioral genetic results and see the question "Why are siblings so different?" get the idea that something in the home environment -- perhaps the rivalry between them, perhaps the way their parents encourage them to follow separate paths -- actually creates or widens differences between siblings. That is not the case: the home environment simply fails to have the expected effect of making them more alike. Adoptive siblings reared in the same family are not less alike in personality than adoptees picked at random from the population -- the correlation between them is approximately zero. If their shared home environment were creating or widening differences between them, the correlation would be a negative number. The net effect of the home environment is approximately zero. It was this baffling finding that caused me to conclude that prevailing theories of development must be wrong and that a new theory was needed (see The Nurture Assumption).

Any theory that purports to account for the differences between siblings -- the unexplained nongenetic variance in personality -- has to account for the fact that the same results are found for identical twins, fraternal twins, ordinary biological siblings, and adoptive siblings. The amount of variance that is left unexplained by shared genes and a shared home environment is about the same -- roughly 50 percent -- for all these sibling pairs. A theory that attributes the sibling differences to birth order effects or competition between siblings cannot account for the fact that the same differences in personality are found between identical twins, who do not differ in birth order and who compete less and cooperate more than ordinary siblings.

Recent Attempts to Account for the Unexplained Variance

Ever since the publication of Plomin and Daniel's 1987 article, cutting-edge researchers have been searching for the source of the unexplained variance -- measuring differences in the experiences of pairs of siblings who live in the same home and trying to link the direction and magnitude of these differences to the differences in the siblings' personalities. The best of this research is summed up in two recent reports.

The first, by Turkheimer and Waldron (2000), is a meta-analysis. (A meta-analysis is a method for combining the results of a collection of individual studies, taking into account the number of subjects who participated in each study and the size of the obtained effect.) The studies included in this meta-analysis investigated a variety of factors -- including birth order -- that might (in the opinion of the researchers who did the work) account for the differences between siblings. None of the studies managed to account for much of the variance, but the least successful studies were those that looked at "family constellation variables" such as birth order and age differences between siblings. Family constellation variables managed to account, on average, for only 1 percent of the variance -- a negligible fraction of the 50 percent of the variance that needs to be accounted for.

The other report is a single large, exceptionally well-done study that directly measured sibling differences, as well as many of the environmental factors that the researchers thought might be responsible for the differences. David Reiss and his colleagues (2000) studied 720 adolescent sibling pairs living in stable two-parent families. The siblings were the same sex and included identical and fraternal twins, ordinary siblings, half-siblings, and stepsiblings. Each pair was examined twice over a three-year period; the researchers took hundreds of measurements of their home environments and of the siblings themselves. The measures of the home environment included judgments of parent-offspring closeness, rapport, monitoring, disagreement, and control (made separately for each sibling by both parents and by the sibling in question); judgments of parental rejection, conflict, closeness, warmth, support, and communication (made separately for each sibling by the researchers); judgments of sibling conflict, negativity, warmth, and support (made by both parents and both siblings); and judgments of parental conflict about the adolescents (made by both parents). The measures of the adolescents included judgments of their antisocial behavior, depressive symptoms, industriousness, autonomy, school performance, sociability, social success, social responsibility, and self-esteem. These judgments were again made for each sibling separately, by the siblings themselves, their parents, and the researchers. Because the measures obtained in this study were based on the combined judgments of several different people, they were exceptionally accurate and reliable.

The study took twelve years to complete. When it was done the researchers were more than disappointed: Reiss has confessed to a journalist that he was "shocked" by the results (Paul, 1998, p. 46). I, on the other hand, was neither surprised nor disappointed: the results were exactly what my theory (Harris, 1995, 1998) predicted.

But Reiss has my sympathy. Though he made a valiant effort to find something to say that would justify his unshakable faith in the importance of the family environment, the data speak louder than the words (Reiss, 2000). What he and his colleagues got out of all their hard work was pretty close to zilch. Yes, they confirmed that there are sizable differences between siblings that can't be attributed either to genes or to aspects of the home environment that the siblings share; and yes, they confirmed that parents behave differently toward different offspring. But the researchers' attempt to find the sources of the differences between the siblings was a total flop. The differences in parental behavior were almost entirely a response to innate differences in the offspring themselves, rather than a cause of the differences between them. Nor was it possible to pin the differences between siblings on disagreements between the parents on how to deal with them. Nor was it possible to pin the differences on the sibling relationship itself -- the hostility or camaraderie between them, the domination of one by the other. Reiss summed up his negative results this way:

We can say with confidence that, on the basis of the data we collected, the following family characteristics do not reflect nongenetic, nonshared influences on the adolescent: differential marital conflict about the adolescent versus the sib, differential parenting toward siblings, and asymmetrical relationships the sibs construct with each other. . . . Given that our very large twelve-year study was designed to identify nongenetic, nonshared factors, this dearth of findings is not only disappointing but galvanizing. (Reiss, 2000, pp. 406-407)

One reason why "asymmetrical relationships the sibs construct with each other" could not explain the differences between them was that usually the sibling relationships were not asymmetrical. Whether they were twins, full siblings, half-siblings, or stepsiblings, most relationships between siblings were found to be remarkably symmetrical -- what evolutionary psychologists call "tit for tat." If Sibling A was nice to Sibling B, then B was nice to A. If A said that B was a jerk, B was likely to say the same about A. Relationships that work the same in both directions cannot explain the differences between siblings.

Occasionally asymmetries were found in the sibling relationship. In some cases, Sibling A showed more kindness or more hostility to Sibling B than B did to A. In such cases, other differences were usually found between A and B -- differences, for example, in their autonomy, sociability, and antisocial behavior. But the data indicated that the asymmetries in the sibling relationship and the other measured differences between the siblings could both be attributed to genetic factors. A kid who was hard to get along with from birth was likely to be nasty to his sibling and to be troublesome in other areas of his life as well. Like parent-child relationships, sibling relationships reflect the genetically influenced characteristics of both parties. And like differential parenting toward children, asymmetric sibling relationships are a result, rather than a cause, of differences between the siblings.

Because Reiss was trying to explain the differences between twins, as well as those between other kinds of siblings, he didn't look for birth order effects. But he did look at the two factors that are believed to be the basis of birth order effects: differences in how parents treat their children, and asymmetric relationships between the children themselves. Reiss's findings are consistent with those of Turkheimer and Waldron (2000). Both studies demonstrated that birth order cannot account for the unexplained variance in personality.

Getting Along at Home and Elsewhere

It's easy to misunderstand the results of the behavioral genetic research. Many people assume that because the home doesn't make siblings more alike, it must make them more different. Not so. Siblings are not more different than a bunch of children picked at random from the population. The unexplained variance represents both differences between siblings and differences between children who are not siblings. Researchers expected that the environment shared by children who grow up in the same home would account for some of that variance, but it didn't. They thought that knowing that Justin is aggressive would give them a clue about what to expect from his brother Jared, but (apart from the similarities due to their shared genes) it didn't. The variance left to account for is the same variance they had to begin with, before they entered into their calculations the fact that Justin and Jared were being reared in the same home by the same parents. The question "Why are siblings so different?" really means "Why aren't siblings more alike?" Prevailing theories of development led people to expect that siblings would be more alike. "Why are they so different?" is the mournful cry of unfulfilled expectations.

Those unfulfilled expectations have produced a lot of useless research. Looking for things that make siblings different -- instead of looking for things that produce differences among all children, whether or not they sleep under the same roof -- implies that we need a special explanation for sibling differences. It implies that shared experiences in the home must have similar effects on the children who live there, and that therefore we must look for other factors in the home that undo the similarity-producing effects of the shared experiences.

But maybe the puzzle of the unexplained variance has a simpler solution: maybe what happens to children at home doesn't contribute anything at all to the variance, one way or the other. Maybe what happens to children at home has -- as Trivers predicted -- no long-term effects on their personalities. Maybe the strategies that children work out for getting along with their parents and siblings affect their behavior only when they're with their parents and siblings; they don't use these strategies in other social contexts because they are useless, or worse than useless, for dealing with people outside the family. (See "Why Are Birth Order Effects Dependent on Context?" on this website.)

Birth order is important at home because, unless they're twins, siblings differ in age. Differences between them in size, strength, and know-how persist for years, and children have to accommodate themselves to these differences. But outside the home, birth order loses its importance. This is true both in modern societies, where children's companions outside the home are mainly others of the same age, and in traditional societies, where children play in mixed-age groups. In the same-age groups of modern societies, children who are a bit bigger, stronger, or more knowledgeable than their agemates have higher status; it doesn't matter whether they are the biggest or the smallest at home. In the mixed-age play groups of our ancestors, children's status increased as they got older. Each child -- whether he was his mother's firstborn child or her fifth -- began as the smallest child in the play group and worked his way up, eventually becoming one of the largest, as those ahead of him graduated out of the group.


Birth order cannot explain the personality differences between siblings, though it can shed light on their relationships with each other and with their parents. Because family relationships are important to us, birth order seems important too (see "Why Do People Believe in Birth Order Effects?" forthcoming on this website). But, as important as they are in other ways, family relationships do not shape personality.

For the answer to the mystery of environmental influences on personality, we must look at the experiences children have in their other environment -- the world outside their home. This is the world in which they will spend their adult lives.


I thank William Dickens, Charles Harris, Steven Pinker, Robert Plomin, and David Rowe for their comments on earlier versions of this essay.


Abramovitch, R., Corter, C., Pepler, D. J., & Stanhope, L. (1986). Sibling and peer interaction: A final follow-up and a comparison. Child Development, 57, 217-229.

Bouchard, T. J., Jr. (1994, June 17). Genes, environment, and personality. Science, 264, 1700-1701.

Bouchard, T. J., Jr., Lykken, D. T., McGue, M., Segal, N. L., & Tellegen, A. (1990, October 12). Sources of human psychological differences: The Minnesota study of twins reared apart. Science, 250, 223-228.

Dawkins, R. (1976). The selfish gene. New York: Oxford University Press.

Deater-Deckard, K., & Plomin, R. (1999). An adoption study of the etiology of teacher and parent reports of externalizing behavior problems in middle childhood. Child Development, 70, 144-154.

Dunn, J., & Plomin, R. (1990). Separate lives: Why siblings are so different. New York: Basic Books.

Eibl-Eibesfeldt, I. (1989). Human ethology. Hawthorne, NY: Aldine de Gruyter.

Epstein, D. H. (2001, February 1). Re: Cheesy Debate. TIPS  (Teaching in the Psychological Sciences) online discussion group ( ).

Ernst, C., & Angst, J. (1983). Birth order: Its influence on personality. Berlin, Germany: Springer-Verlag.

Goodall, J. (1986). The chimpanzees of Gombe: Patterns of behavior. Cambridge: Harvard University Press.

Hamilton, W. D. (1964). The genetical evolution of social behaviour (I and II). Journal of Theoretical Biology, 7, 1-52.

Harris, J. R. (1995). Where is the child's environment? A group socialization theory of development. Psychological Review, 102, 458-489.

Harris, J. R. (1998). The nurture assumption: Why children turn out the way they do. New York: Free Press.

Hoffman, L. W. (1991). The influence of the family environment on personality: Accounting for sibling differences. Psychological Bulletin, 110, 187-203.

Jefferson, T., Jr., Herbst, J. H., & McCrae, R. R. (1998). Associations between birth order and personality traits: Evidence from self-report and observer ratings. Journal of Research in Personality, 32, 498-500.

Kagan, J. (1976). The role of the family during the first half decade. In V. C. Vaughn, III, & T. B. Brazelton (Eds.), The family--Can it be saved? Chicago, IL: Yearbook Medical Publishers.

Kagan, J. (1998, September 13). A parent's influence is peerless: "Nurture Assumption" ignores more theory than it proposes. Boston Globe, p. E3.

Littlefield, C. H., & Rushton, J. P. (1986). When a child dies: The sociobiology of bereavement. Journal of Personality & Social Psychology, 51, 797-802.

McHale, S. M., Crouter, A. C., McGuire, S. A., & Updegraff, K. A. (1995). Congruence between mothers' and fathers' differential treatment of siblings: Links with family relations and children's well-being. Child Development, 66, 116-128.

Mock, D. W., & Parker, G. A. (1998). The evolution of sibling rivalry. Oxford, UK: Oxford University Press.

Pfennig, D. W., & Sherman, P. W. (1995, June). Kin recognition. Scientific American, 272, 98-103.

Pinker, S. (1997). How the mind works. New York: Norton.

Plomin, R., & Daniels, D. (1987). Why are children in the same family so different from one another? Behavioral and Brain Sciences, 10, 1-16.

Reiss, D., with Neiderhiser, J. M., Hetherington, E. M., & Plomin, R. (2000). The relationship code: Deciphering genetic and social influences on adolescent development. Cambridge, MA: Harvard University Press.

Rowe, D. C. (1994). The limits of family influence: Genes, experience, and behavior. New York: Guilford Press.

Segal, N. L. (1993). Twin, sibling, and adoption methods: Tests of evolutionary hypotheses. American Psychologist, 48, 943-956.

Segal, N. L. (1999). Entwined lives: Twins and what they tell us about human behavior. New York: Penguin.

Stocker, C., & Dunn, J. (1990). Sibling relationships in childhood: Links with friendships and peer relationships. British Journal of Developmental Psychology, 8, 227-244.

Sulloway, F. J. (1996). Born to rebel: Birth order, family dynamics, and creative lives. New York: Pantheon.

Sulloway, F. J. (1998, November). Birth order and the nurture misassumption: A reply to Judith Harris. Edge (

Sulloway, F. J. (1999). Birth order. In M. A. Runco & S. Pritzker (Eds.), Encyclopedia of creativity (vol. 1, pp. 189-202). San Diego, CA: Academic Press.

Trivers, R. (1985). Social evolution. Menlo Park, CA: Benjamin/Cummings.

Turkheimer, E., & Waldron, M. (2000). Nonshared environment: A theoretical, methodological, and quantitative review. Psychological Bulletin, 126, 78-108.

Version 1.0
September 3, 2001

Citation (American Psychological Association format):
Harris, J. R. (2001, September 3). Why can't birth order account for the differences between siblings? Retrieved [insert date] from the World Wide Web: http://judithrichharris/tna/birth-order/index.htm

Copyright Notice
Copyright 2001 by Judith Rich Harris.
Permission is granted to link to this essay and to quote from it briefly. All other rights reserved.
For permission to reprint, contact Charles S. Harris, .

To the birth order page
To The Nurture Assumption home page
Back to top Visits to this page: Visits to this page: