In the first two essays on this website, I summarized the results of more than half a century of research on birth order. My conclusion was that birth order does not have noticeable effects on adult personality. The question I must deal with now is: Why do most people -- including most psychologists -- continue to believe that birth order does have important effects on adult personality?

In this essay I will discuss several sources of the belief in birth order, including people's subjective impressions based on their own personal experiences, flawed or misleading research, the tendency for research to be published and publicized only if it supports the belief in birth order, the impressions psychotherapists get from listening to their patients, and biological factors.

It isn't enough, however, to explain why people came to hold a particular belief. I also have to explain why they cling to it so tenaciously in the face of disconfirming evidence.

Impressions Based on Personal Experiences

As I explained in the first two essays on this website, birth order effects unquestionably do exist within the context of the family of origin. When they are with their parents and siblings, firstborns behave differently from laterborns -- even in adulthood. But most people don't spend their adult lives in their childhood home; they see their parents and siblings only occasionally. In my two previous essays (and in The Nurture Assumption), I presented evidence showing that the patterns of behavior acquired in the childhood home don't affect the way people behave outside that home -- even in childhood.

Now ask yourself this: Whose birth order do you know? Certainly you know your own, those of your sisters and brothers, and those of your children. In all likelihood you also know the birth orders of other relatives, your husband or wife, your close friends, and the children of your friends and neighbors.

Of the people in your life, these are the ones you are most likely to have seen in the presence of their parents and siblings. To put it another way, you are unlikely to know the birth orders of people whom you've never seen interacting with their parents and siblings.

You can see birth order effects in action when you observe people (including your own brother or sister) interacting with their parents or siblings (including you). So the question now becomes: Why do you assume that people behave in a similar way when they're not with their parents or siblings?

The answer is that humans have a strong tendency to underestimate the effects of context on other people's behavior. This tendency is well-known to social psychologists, who call it the fundamental attribution error. In accounting for people's behavior, we tend to give too much weight to what we assume are enduring aspects of their personality and too little to the exigencies of the situation we see them in. It's easiest to make this error about someone we are meeting for the first time: if the newcomer acts grouchy, we tend to peg her as a grouchy person, even if we know that she got tied up in traffic, couldn't find a parking space, and had her jacket splattered with mud. In a classic study (Napolitan & Goethals, 1979), subjects were informed in advance that, for the purposes of the experiment, a research assistant had been instructed to act friendly (or unfriendly). Nevertheless, they attributed her behavior to genuine friendliness (or unfriendliness).

But the fundamental attribution error can also affect the way we characterize people we see again and again, if we often see them in the same setting. David Myers, a professor of social psychology at Hope College, Michigan, is well aware that small differences in the situation can have big effects on people's behavior, and that students are likely to be more alert in the evening than early in the morning. And yet, he reports, "Even knowing the effect of the time of day on classroom conversation, I find it terribly tempting to assume that the people in the 7:00 p.m. class are more extraverted than the 'silent types' who come at 8:30 a.m." (Myers, 1996, p. 80).

Even knowing the effect of family get-togethers on people, you might find it terribly tempting to conclude that Jeff -- who acts bossy and superior whenever he's around his younger brother -- is a domineering person with an inflated ego; and that Sarah -- who acts clueless and irresponsible whenever she's with her parents and older siblings -- is basically an airhead. In fact, even if you were aware that Jeff acts bossy and superior only when he's with his younger brother, or that Sarah acts clueless and irresponsible only when she's with her parents and older siblings, you might still find it terribly tempting to peg Jeff as an egotist and Sarah as an airhead. The reason is the nurture assumption -- the assumption that, as birth order psychologist Kevin Leman put it, "For a young child growing up, there is no greater influence than his or her family" (1998, p. 24). The nurture assumption causes people to believe that, when they see Jeff interacting with his brother, or Sarah interacting with her parents and siblings, they are getting a glimpse of the real Jeff or the real Sarah. But if Jeff and Sarah spend only a small fraction of their time interacting with their parents and siblings, and if they behave differently (in all probability, better!) when they're not interacting with their parents or siblings, isn't that behavior a more accurate indication of the real Jeff and the real Sarah?

The characteristics often associated with the various birth-order positions -- the belief that firstborns are bossy, aggressive, and achievement-oriented; that middleborns are diplomats, able to get along with anyone; and that lastborns are irresponsible and rebellious -- are descriptions of the way these people behave within their family of origin. These behaviors are not enduring aspects of their personality but are reactions to the social context in which they find themselves; in other contexts, they behave differently. As I explained in my first essay (see "Why Are Birth Order Effects Dependent on Context?" on this website), the evidence shows that when children go to school or to the playground, they leave behind the strategies they learned at home for dealing with their siblings. The firstborn who is bossy or aggressive with his younger siblings behaves differently with his peers: outside the home, firstborns are not bossier or more aggressive than laterborns, even in childhood (Abramovitch et al., 1986; Deater-Deckard & Plomin, 1999). Firstborns are often seen as high achievers by their parents and younger siblings, but in school they perform no better, on the average, than laterborns (Blake, 1989; Ernst & Angst, 1983).

At home, children must deal with siblings who differ from them in age; outside the home, children in developed societies spend their day in the company of their agemates. At home there is only one firstborn but outside there are many; the fact that a child is older than his siblings, and can dominate them or serve as their teacher, doesn't matter when he's in school or on the playground. All that matters is how he compares -- in size, strength, intelligence, knowledge, and so on -- with other children his own age. Even if he goes out into the world assuming that it will just be a larger version of his home (and the evidence suggests that he makes no such assumption), he will soon find out that he is wrong. Behaviors that worked at home will not work in the wider world outside.

Various personality traits have been ascribed to firstborns and to laterborns, but there is one theme that crops up again and again: the description of firstborns can be summed up as "more mature." They're more responsible, more adult-like in their behavior, take life more seriously. Well, firstborns are more mature than their younger siblings, for the simple reason that they're older! Although the age difference between siblings becomes less important as they get older, I suspect that popular ideas about birth order are based largely on comparisons between siblings (such as the offspring of friends and neighbors) who are still living at home, which means that an older child is being compared to a younger one. Popular ideas about birth order are also influenced by people's thoughts and feelings about their own siblings. But people's thoughts and feelings about their own siblings date back to the years when they lived together under the same roof and saw each other every day -- a time when an age difference of two or three years meant a big difference in maturity.

Resistance to Disconfirming Evidence

When people believe something, their belief influences their perception of new information. Confirmation bias is the tendency to seek, notice, and remember evidence that confirms one's belief, and to ignore, forget, or explain away contrary evidence. It's a universal human failing, demonstrated over and over again in laboratory studies and in real life (Myers, 1996). In one experiment, for example, subjects were asked to evaluate two (fictitious) research studies on the deterrence effect of the death penalty: one supporting, the second disconfirming. The subjects who started out believing in the death penalty readily accepted the first study but were highly critical of the second. Those who were opposed to the death penalty accepted the second study but found flaws in the first. Regardless of their prior belief, all the subjects ended up even more convinced that they were right -- despite the fact that they had all been exposed to exactly the same evidence (Lord, Ross, & Lepper, 1979).

When faced with mixed evidence, people focus on the parts that confirm their beliefs and ignore the parts that don't. A believer in birth order who didn't happen to know the birth order of Bill Gates, the founder of Microsoft, might think he was a firstborn, because of his leadership ability and drive to succeed. A believer in birth order who happens to know that Bill Gates is the second of three siblings will focus on some other aspect of his personality, such as his openness to new ideas or his loyalty to old friends. Because a wide variety of personality characteristics have been attributed to the different birth orders (laterborns, for example, have been described as rebellious, easy-going, cooperative, competitive, irresponsible, shy, sympathetic, open-minded, adventurous, affectionate, and lacking in self-confidence), most people can find something in the list that fits them -- just as most people can find something pertinent in their astrological forecast in the newspaper. Of the three or four vague and cryptic statements in an astrological forecast, we pay attention only to the one or two that seem to apply to us.

If the disconfirmatory evidence can't be ignored, it can be explained away. Kevin Leman, who accurately describes himself as "North America's 'pop' birth order psychologist" (1998, p. 13), believes that firstborns are meticulous, well-organized perfectionists, but was undismayed when he encountered a firstborn with a messy desk. "Your desk is sloppy," he observed, "but can you find what you need on it?" "Of course," was the firstborn's reply. Satisfied, Leman went on to speculate: "My guess is you are something of a perfectionist and perfectionists are known for having sloppy desks as a means of covering their discouragement for not always having life go just the way they want it" (1998, p. 93). It's a win-win setup for Leman: If a firstborn has a neat desk, it's because he's a perfectionist, and that confirms Leman's theory. If a firstborn doesn't have a neat desk, it's because he's covering up his discouragement, and Leman's theory is again confirmed.

Is there any disconfirming evidence that Leman can't explain away? On one occasion he went out on a limb and guessed that Bryant Gumbel, co-host of the Today Show at the time, was a firstborn. When he was told that Gumbel actually has a brother three years older, he explained it this way: "Clearly, there had been a 'role reversal' in the Gumbel family. Bryant had taken over as the 'firstborn' while his brother, who today is a broadcaster and certainly no slouch in his own right, became 'number two' " (Leman, 1998, p. 30).

The idea of sibling role reversals is a useful escape clause for psychotherapists who are followers of birth order theorist Alfred Adler (1870-1937). According to Adlerian theory, it's psychological birth order, not actual birth order, that matters: "A child's birth order position may be seized by another child if circumstances permit." This explains the firstborns who act like secondborns and vice versa. But what about the people who don't fit any of the stereotypes? The Adlerians have an answer for that, too: "Birth order is sometimes not a major influence on personality development" (Stein, 2001).

There is no question that siblings differ from one another in personality. But, as I explained in my previous essay (see "Why Can't Birth Order Account for the Differences Between Siblings?" on this website), two children reared in the same home don't differ more than two reared in separate homes. The evidence does not show that growing up together causes children to develop distinctive characteristics that make them less alike. Birth order theories, including those that allow  "role reversals," are based on the idea that sharing a home creates or widens differences between siblings. This idea is erroneous.

Misleading Research

Most people not only believe in birth order: they also believe that there is plenty of research to support their belief in birth order. Indeed, many studies have found statistically significant differences in personality between firstborns and laterborns. On the other hand, many other studies -- especially the largest and most rigorous ones -- have failed to find birth order differences. Why the discrepancy? Sometimes it's just a matter of chance: so many birth order studies are carried out that some of them are bound to pay off. But sometimes it's the result of the research method. Certain research methods can produce misleading results.

Making use of context effects.   Because birth order effects do exist within the family of origin, the easiest way to demonstrate them is to put people back, psychologically, into that context. Researchers can do this quite easily: for example, they can ask their subjects to compare themselves to their siblings, or they can assess their subjects' personalities by asking their parents to describe them. Many of the studies that reported significant effects of birth order on personality have used such techniques. The results of these studies are not inaccurate; the differences they report are real. But they are misleading, because the researchers would like us to believe that the results tell us something about how the subjects behave in contexts outside the family -- in their careers, for example -- and they do not.

Context effects are discussed in detail in my first birth order essay (see "Why Are Birth Order Effects Dependent on Context?" on this website), so I won't belabor this point here.

Failure to control for family size and socioeconomic status.   People from large families and people from small families differ, on average, in at least three measurable ways: socioeconomic status (SES), education, and IQ. Large families are more common at lower SES levels. On the average, mothers who have many children have less education and lower IQs than mothers of the same age with fewer children. On the average, children who have many siblings have lower IQs than children with few or no siblings, and they are less likely to graduate from high school (Blake, 1989; Ernst & Angst, 1983, Rodgers et al., 2000).

Of a sibship of two (sibship size is the number of offspring in a family), half are firstborns. Of a sibship of four, only a quarter are firstborns. Small families produce a relatively larger proportion of firstborns; large families produce more laterborns. A firstborn plucked at random from the population is more likely to be from a small family, a laterborn more likely to be from a large one.

Taken together, the facts in the previous two paragraphs imply the following: a firstborn is more likely than a laterborn to come from a higher SES family, to be born to and reared by intelligent parents, to have a higher IQ, and to be better educated.

Any study that fails to take family size and/or SES into account, and simply lumps together subjects from large and small sibships, is likely to find differences between firstborns and laterborns. But these differences are misleading: they are due to the fact that, on average, the firstborns come from higher SES homes and have higher IQs than the laterborns.

The best way to control for the effects of family size is to do the birth order comparisons separately for each sibship size: compare firstborns from sibships of two with laterborns from sibships of two, firstborns from sibships of three with laterborns from sibships of three, and so on. To do this properly requires collecting data from a large number of subjects -- hundreds or, better still, thousands. Small samples don't contain an adequate number of subjects from each sibship size.

Nonetheless, small studies often make it into the news. Here, for example, is a description of a study that got a lot of publicity in June, 2001:

Dr. Leong's conclusions were based on two small studies: one with 159 subjects and one with 119 (Leong et al., 2001). That's not nearly enough to control for family size. The results he reported on career choice were exactly what you would expect if the firstborns and laterborns in his samples differed, on average, in socioeconomic class and intelligence.

There is likely to be a surplus of firstborns in any group of eminent people -- famous scientists, political leaders, and so on. This surplus is explained by the correlations between birth order, family size, SES, IQ, and education. Even strong believers in birth order effects admit that  "Once eminence is controlled for family size and social class, differences by birth order generally disappear (Sulloway, 1996, p. 109). The same is true of IQ: once the research method controls for family size and parental intelligence, firstborns are no more intelligent than their younger siblings (Rodgers et al., 2000).

In previous generations, another factor may have contributed to the surplus of eminent firstborns. Before it became common for parents to see that all their children received an education, the offspring who was most likely to be educated was the firstborn son. This educational edge helps to explain, for example, the slight surplus of firstborns and only children among U.S. presidents. (Fourteen of the 42 U.S. presidents -- 33% -- have been firstborns or onlies. The expected percentage, based on historical family sizes, is 23%, according to Somit et al., 1996.)

Another complication is that the proportion of firstborns in the population -- and therefore the decision about whether the proportion in a given sample is higher or lower than expected -- varies over time, due to changes in average family size. In U.S. and Canadian populations, the proportion of firstborns was 27% in 1960, 39% in 1970, and 44% in 1984 (Somit et al., 1996).

Average SES and educational levels have also risen steadily over time. These social changes may affect the outcome of a birth order study even if the subjects are all the same age, because the parents of a laterborn are likely to be older than the parents of a firstborn (to put it another way, there's likely to be a bigger age gap between a laterborn and his parents than between a firstborn and his parents). Older parents grew up at a time when average SES and educational levels were lower. Thus, the laterborns in a sample may have come from lower SES homes than the firstborns, even if the researchers controlled for family size.

That age gap between parents and their offspring -- the gap that is larger for laterborns than for firstborns -- can also contribute to the erroneous impression that laterborns are more rebellious. During times of rapid cultural change, firstborns' ideas are likely to be more similar to those of their parents, because they are closer to their parents in age and thus were reared in more similar cultural settings.

"Birth order research seems very simple," the Swiss researchers Ernst and Angst wryly observed (1983, p. xi). Its apparent simplicity makes it a perennial favorite among students in search of a science fair or thesis topic. In reality, this kind of research requires sophisticated knowledge of statistics and sociology, as well as a sizable pool of subjects.

Massaging the data.   Very few researchers go to the trouble of carrying out a study unless they have a idea -- a belief or theory -- that they hope will be supported by the results of the study. They are disappointed if the results don't support their idea. They may do another study to see if a change in the procedure will demonstrate the effect they are looking for, or they may look more closely at the data to see if they missed something. Some give up. A very small minority change their minds.

When lots of studies get done, some of them are bound to produce statistically significant effects, just by chance. If significant is defined in the usual way, as "only a 5% probability that these results were due to chance," there is a 5% probability that the results were due to chance. An honest pair of dice can, on occasion, come up snake eyes four times in a row, due to chance. Chance is probably the reason why a larger-than-expected proportion of a sample of stripteasers, and a larger-than-expected proportion of the scientists who founded SETI (the Search for Extraterrestrial Intelligence), were firstborns (Skipper & McCaghy, 1970; Shermer, 2001). Taking off one's clothes in public and searching for extraterrestials are not the sort of dull but prestigious occupations we usually think of in connection with the "firstborn personality"!

Chance can't happen in the absence of opportunity. The more times you throw the dice, the more opportunities you give yourself to produce a run of four snake eyes. Many studies of birth order measure several different variables at once, thus increasing the chances that significant birth order effects will be found in at least one of them. Even if the researchers are interested in only one outcome variable -- say, achievement motivation -- they may have two or three different ways of measuring it. This doubles or triples their chances of finding a significant effect.

Another way of increasing the chances of finding a significant effect is to look for statistical "interactions": effects that turn up in one subgroup of subjects but not in another. Many of the significant findings reported in Ernst and Angst's (1983) survey of the birth order literature were interactions, rather than "main effects." Significant birth order effects might be found in male subjects but not in females (or vice versa), or in high-SES subjects but not in low-SES ones (or vice versa), or in subjects from large families but not from small ones (or vice versa). Researchers who are unsophisticated about statistics might see nothing wrong with dividing up their data in as many ways as they can think of and reporting only the one that pays off.

Researcher bias.   Earlier I said that confirmation bias -- the tendency to notice and accept evidence that supports our beliefs and to ignore or explain away contradictory evidence -- is a universal human failing. Scientists are not exempt from this failing. Some sciences are more advanced than others in finding ways to guard against researcher bias. Medicine is way ahead of psychology in this respect.

Medical researchers use a variety of procedures to reduce the chances that researchers' biases will affect the outcome of their studies. In testing the effects of a new drug, for example, the condition of the patients at the end of the trial must be assessed by physicians who don't know which patients received the drug and which received the placebo. Otherwise it would be too easy for the physicians who make the assessment to err in the direction of their prior bias. Trusting physicians to do their best to be objective turned out not to be good enough; nowadays medical researchers use double-blinding whenever possible.

Researcher bias is a psychological phenomenon; when it does affect the results of a study, it usually happens without the researcher's conscious awareness. One would think that psychologists, of all people, would be alert to its dangers. Yet procedures such as double-blinding are virtually never used in psychology. Unlike medical researchers, psychologists simply trust each other to do their best to be objective.

Some methods used to study birth order are particularly susceptible to the effects of researcher bias. For example, there's the question of whether biological birth order or  "functional" birth order affects personality. Someone who was biologically his parents' secondborn child, for instance, would be a functional firstborn if his older sibling died in infancy and he grew up as the oldest child in the family. In this case it makes sense that functional birth order is what matters. The problem is that researchers may be more motivated to ask questions about an individual's functional birth order if his biological birth order doesn't support their views. The problem becomes more serious if the individual in question has been dead for years: it takes a good deal of effort to ferret out the kind of personal information needed to assess the functional birth order of a historical figure. Jeremy Freese and his colleagues (1999) have explained how researcher bias could have influenced the results reported in Born to Rebel (Sulloway, 1996), a book that is mainly about historical figures:

When a researcher re-examines results that go against his theory, and accepts without question results that support it, the researcher's biases can determine the outcome of a study. The same thing can happen in an entire field, if a bias is widely shared. As medical epidemiologist Alvan Feinstein observed,

An example is the debunking of the work of the late British psychologist Cyril Burt by an American psychologist, Leon Kamin. According to Kamin (1974), Burt faked some of the data he used to support his view that intelligence is largely inherited. Whether Burt was or was not a phony is not the issue here: the point is that Kamin looked for anomalies in Burt's data only because he strongly disagreed with Burt's conclusions: Kamin doesn't believe that genes have any effect on intelligence. Researchers who fake their data (or who use other means to nudge their results in the desired direction) are much more likely to go undetected if the results happen to agree with widely held beliefs.

Publication bias.   More studies get done than get published. If the results don't come out the way the researchers expected or hoped, they may never see the light of day. Publication bias is the tendency for statistically significant findings to get published and nonsignificant findings -- or findings contrary to expectations -- to be tossed, with a shrug of the shoulders, into a file drawer or the recycling bin.

The failure to publish no-difference results is a problem in all fields of science, including medical research, where such results can be vitally important. If a new drug, or a new surgical procedure, is no better than the standard treatment, physicians and their patients need to know that. (Would you want to undergo surgery if it didn't increase your chances of getting better?) And yet, even in medicine, studies that fail to find a significant difference are less likely to be published (Ioannidis, 1998).

When positive findings are published and no-difference or negative findings are not, it can lead physicians to the wrong conclusions. To guard against this, medical researchers have devised a statistical procedure called a funnel plot to test for publication bias. They gather together all the published studies they can find on a given topic and check to see if the smaller studies -- those involving fewer patients -- tended to produce larger effects (that is, more substantial differences between the tested drug or procedure and the placebo). If larger effects are found for smaller studies, it's an indication of publication bias (Egger et al., 1997). The reasoning is that a big study is likely to be published regardless of its outcome, whereas a small study that doesn't produce a statistically significant effect may just be tossed out. Thus, publication bias results in a dearth of small studies producing no-difference or nonsignificant results.

Though I didn't realize it at the time, I demonstrated a publication bias for birth order studies in Appendix 1 of The Nurture Assumption (p. 372), when I showed a clear trend among the birth order studies reviewed by Ernst and Angst (1983): no-difference or nonsignificant results were reported far less frequently for the smaller studies than for the larger ones. There is no question that publication bias exists in birth order research. It probably exists in all areas of psychology.

Psychologists sometimes use a simple calculation called the file-drawer test to check for the effects of publication bias. This test isn't used by medical researchers. I will have more to say about the file-drawer test in my fourth essay ( "The Mystery of Born to Rebel: Why Did Sulloway's Results Differ from Those of Ernst & Angst?", forthcoming on this website).

As I said at the beginning, most psychologists do believe in birth order. Their beliefs influence their choice of a research topic, the methods they use in collecting and analyzing data, and whether they decide to write up their results and submit them to a journal. Journal editors and people who review manuscripts for the journals also have their biases. Albert Somit and his colleagues (1996), who looked for birth effects on political behavior and concluded that there weren't any, discovered that it is very hard to change people's minds:

Fortunately, not completely impossible. But the extensive work by Somit and his colleagues has received little or no publicity, whereas the small study by Frederick Leong got a fair amount of attention. Getting research mentioned in a newspaper or magazine article is far more difficult than getting it accepted by a journal. I'll bet you've never seen a newspaper article with the headline STUDY FINDS NO DIFFERENCE BETWEEN FIRSTBORNS AND LATERBORNS. And yet there have no doubt been thousands of such studies.

Observations by Psychotherapists

Some of the most dedicated believers in the importance of birth order are psychotherapists -- psychiatrists and clinical psychologists who help people with their personal problems. Therapists believe in birth order not only because it fits in well with an idea that Freud made popular -- the idea that the psychological problems of adults have their origins in things that happened to them in early childhood -- but also because it fits in well with what their patients tell them.

Why do people who seek psychotherapy talk so much about their parents and siblings? Why does what they say tend to reinforce their therapist's belief in the importance of these family relationships?

The first question is easy to answer. Thanks to Freud and his followers, almost everyone who decides to go to a psychotherapist takes it for granted that he or she will be expected to talk about Mom and Dad. Since one of the ways Mom and Dad cause problems is by apportioning love, attention, responsibility, and blame unequally among their children, memories involving brothers and sisters are also assumed to be relevant.

The answer to the second question is that talking about their parents and siblings puts the patients back into the context of the family they grew up in -- the one context in which birth order is truly important. By encouraging their patients to relive their childhood experiences with their parents and siblings, therapists are evoking the feelings associated with being the firstborn, or the lastborn, or a middle child, or an only child. What patients say under these conditions is likely to reinforce the therapist's belief in the power of family relationships to shape, and perhaps to damage, a child's personality.

Whatever type of therapy they administer, most psychotherapists feel that they're helping their patients. If their patients talk about their childhood experiences with their parents and siblings and end up feeling better, therapists interpret the improvement as confirmation of their belief that the childhood experiences were at the root of their patients' troubles. But most troubled people feel better after a while whether or not they go to a psychotherapist. Carefully controlled studies have failed to support the view that talking about childhood experiences has therapeutic value. In fact, the forms of psychotherapy that these studies have shown to be the most effective are those that focus on people's current problems, rather than their ancient history (DeRubeis & Crits-Christoph, 1998). Alan Stone, a professor of psychiatry at Harvard and a former psychoanalyst, has admitted in print that he now questions the premises he started out with: that every psychological disorder has its roots in the experiences of infancy and childhood, and that reconstructing these experiences is an essential part of psychotherapy.

Stone no longer puts his patients on a couch and asks them to free associate; he has changed the way he does psychotherapy. "My focus," he explains, "is almost entirely on the here and now, on problem-solving, and on helping patients find new strategies and new ways of interacting with the important people in their lives" (p. 39). He means the important people in their current lives.

Biological Factors

According to Ray Blanchard, a researcher at the Centre for Addiction and Mental Health in Toronto, Canada, birth order is related to sexual orientation: laterborn males who have older brothers are more likely to be gay. "The best available data," Blanchard reports, "indicate that each additional older brother increases the odds of homosexuality by 33%" (2001, p. 106). Older sisters don't have this effect; nor do older siblings of either sex increase the odds of homosexuality in a female.

The explanation of the connection between sexual orientation and "fraternal birth order" is almost certainly biological. Blanchard has proposed that it results from a reaction by the mother's immune system to a male fetus, and that this reaction becomes stronger with each pregnancy that results in a baby boy. Male fetuses produce substances called male-specific Y-linked minor histocompatibility antigens. These antigens, which are not normally present in a female's body, get into the mother's bloodstream and are regarded as foreign invaders by her immune system, which responds by producing antibodies to them. "When that happens," Blanchard conjectures, "these antibodies partly prevent the fetal brain from developing in the male-typical pattern" (2001, p. 109). The result might be a child who is less masculine in his interests than most boys and who, as an adult, will be attracted to men rather than women.

I was intrigued when I encountered Blanchard's theory, because it can explain a finding reported years earlier by the Swiss researchers Ernst and Angst (1983). In addition to reviewing the work of earlier birth order researchers, Ernst and Angst did a study of their own -- to my knowledge, the largest study of personality and birth order ever carried out: 7,582 subjects took part. Among subjects from sibships of two, there were no differences between firstborns and secondborns in any of the twelve dimensions of personality that were measured. But for subjects from sibships of three or more, Ernst and Angst found one statistically significant difference: the youngest sibling was a little lower in masculinity than the older ones.

Other investigators have described laterborns as less dominant, less aggressive, and less conventional than firstborns (Sulloway, 1996). If laterborn males are less masculine than other men, that could explain why they are less dominant and less aggressive. If laterborn males are more likely to adopt a gay lifestyle, that could explain why they are seen as unconventional. Thus, as researchers Jeremy Beer and Joseph Horn (2001) have pointed out, some of the characteristics attributed to laterborns may be due, not to their experiences with parents and siblings, but to prenatal influences on brain development.

It's a good explanation but there may be nothing for it to explain. The evidence for birth order differences in dominance, aggressiveness, and conventionality is very weak -- perhaps nonexistent, if you eliminate judgments by parents and siblings (see my discussion of the study by Deater-Deckard and Plomin in "Why Are Birth Order Effects Dependent on Context?" on this website). Not all cases of male homosexuality can be attributed to the prenatal effects of the mother's prior pregnancies: some gay men are firstborns. In a typical birth order study of 1000 subjects or less, the number of laterborn gay men is probably too small to affect the outcome.

When Deeply Held Beliefs Come Head to Head with Science

The knowledge that there is a lot of conflicting or disconfirming evidence on birth order doesn't stop people from believing in it. Here is a statement from psychologist Ginger Moore, of Duke University:

(Notice that Moore offers only two choices: either birth order is incredibly important or it's one factor among many. She doesn't even consider the possibility that it might have no effect at all.)

Moore is sure that birth order influences personality; she's just not sure in what direction!

Somit and his colleagues (1996, p. vi) referred to "the inherent non-rational nature of deeply held beliefs" in their attempt to explain the persistence of the faith in birth order. Deeply held beliefs are often part of a culture and are extremely resistant to disproof.

The Yanomamö, natives of the Amazon rain forest, believe that when someone in their village falls ill, it's due to witchcraft -- an evil spell perpetrated by one of their enemies. From our standpoint, this is clearly a non-rational belief. But the Yanomamö support their beliefs with evidence, just as we do. A stray gust of wind or the sudden cry of a bird is accepted as proof of the presence of the enemy's malevolent spirit (Valero, 1970).

Our ancestors, too, believed in witchcraft. It was medical science that persuaded them to give up the idea that illnesses are caused by evil spells. Scientific evidence and reasoning can, in the long run, win out over cultural myths. Call me an optimist, but I believe that psychology will someday catch up with medical science.

I thank Ray Blanchard, Allen Esterson, Jeremy Freese, Charles S. Harris, Marilyn Heins, David G. Myers, Albert Somit, and Frederic Townsend for their helpful comments on earlier versions of this essay.