Stanford News


CONTACT: Kathleen O'Toole, News Service (650) 723-2558;

American SAT, English A-level exam trends reflect changes in sibling environment, psychologist says

"We see encouraging signs among these students, many of whom will be college graduates in the year 2000, the start of the new millennium. . . . These students clearly show that the harder you work, the better you do in school and on the SAT."
­ Donald Stewart, president of the College Board last August, announcing a slight rise in SAT scores
If the test scores go down again, "we are going to think that it's all due to our schools, that government and Congress should do more, that we are not competing well with the Chinese or the Germans, when actually it could be our birth order driving the indicators."
­ Claude Steele, Stanford professor of psychology, commenting on the meaning of a study and theory by his colleague, Robert Zajonc

Slight changes in the academic performance of college-bound students can be expected over the next decade and a half in both the United States and the United Kingdom because of the number of the students' brothers and sisters, Stanford psychologist Robert Zajonc predicts.

When the American and English test scores are announced in a few days, Zajonc says, we will probably see a slight rise in the SAT scores and a decline in the English A-level exam scores. Like SATs, which are taken by American 18-year-olds planning to go to college, A-level exams, given to 17-year-old English students, are important criteria in college admissions.

Zajonc bases his predictions on changes in the average birth order of the test-takers, which he analyzed with research associate Patricia Mullally and published in the July issue of American Psychologist. Children born earlier in their families tend to do slightly better on academic tests than younger siblings, according to several past studies comparing individuals' test scores to their birth rank within their families.

"The differences are apparent when looking over a period of years at the average scores and the average birth order of a nation's population," Zajonc says. (See graphs) "They are much less apparent when comparing the test scores of any two individual siblings."

Graphic courtesy of Robert Zajonc and Patricia Mullally

These graphs show the trends in average birth order of babies born between 1950 and 1983 and trends in the mean test score for the subset of each birth cohort that later took the exam in question. In the birth order scale, at right, a rank of 3.0 would mean that, on average, the babies born in that year were the third child in their families. Average birth order is calculated by averaging the number of first, second, third and so on, births for a given year. The three test scores shown are the math portion of the SAT test, which is taken by about 40 percent of American 18-year-olds in recent years; the British A-levels, which are taken by 17-year-olds who plan to attend college in the United Kingdom; and the Iowa Test of Basic Skills, which is taken by all children in Iowa schools in certain grades. The test score trends track birth order trends most closely in the Iowa test case, the researchers say, because nearly all of the original birth cohort later took that test.

Zajonc (pronounced Zy-unce) first predicted future SAT scores in 1976 in the journal Science. He wrote that the SAT scores, which were then in a free fall, would bottom out in 1980, and a look back now shows that they did.

Now, he and Mullally say that the average score on the SAT exam, taken by most American 18-year-olds planning to go to college, will hover pretty much where it is over the next 15 years ­ at levels substantially below their peak 34 years ago, but not nearly as low as the scores were in the late 1970s and early 1980s. The A-levels will decline slightly until the year 2000. Their predictions would not hold, they caution, if education officials make changes in how they score the tests.

The trends and changes already seen in scores for age cohorts born and tested since World War II are "a clear reflection of family patterns, confirming the important role of environmental factors and especially, family and peer influences" that cannot be explained by heredity, they wrote.

Seeming paradox

There are two reasons for the seeming paradox that birth order makes little difference for individuals but makes a significant difference for a society as a whole, Zajonc says. The first is based on mathematical effects of aggregation and the second is that a phenomenon that applies to individuals can have larger collective effects ­ where the sum is greater than its parts.

Aggregation of birth order effects is similar to what happens in epidemiology, Zajonc says. "We can't predict whether a particular individual will get lung cancer, given that he smokes or does not, but when the data are aggregated, we know that smoking raises the amount of cancers in a society."

The second reason birth order can be more important for a society, Zajonc believes, is that the intellectual environment is raised for an entire age group of children when there are more early-borns in their peer group.

"Because the quality of our education depends to a great extent on our peers, a given year's offspring with a higher proportion of early-borns has a 'bonus.' They will benefit from higher standards of excellence throughout their schooling and advance more rapidly in the more challenging intellectual environment. Not only students would be affected by a greater proportion of firstborns. Teachers, too, would advance their students more rapidly because they would be confronted with a more favorable distribution of grades than would be the case in a generation of students with few firstborns."

Zajonc and Mullally argue that this phenomenon, which they call "collective potentiation," can have significant regional, national and international ramifications. In the case of technology adoption, for example, historians of science have noted that new technology tends to cluster. A particular place becomes known for producing the best software or the best violins or even the best hockey coaches. "In these locations and in these times, a few highly talented individuals set the standards for others, invent new techniques and propagate new approaches," Zajonc and Mullally wrote. "They inspire others who build on examples set for them, and in turn, improve on previous products that challenge the original contributors to achieve even higher standards."

In the same way, they suggest, "many group differences, especially those in test scores, might well derive from collective potentiation effects."

SAT trends

A precipitous drop in SAT scores in the 1970s prompted a great deal of concern in the United States, but no consensus has yet developed among testing experts and social scientists on what caused it. In the 12-year period from 1968 to 1980, the average score dropped by 25 points.

Families were larger after World War II and more siblings were born closer together, but the SAT scores should have dropped only 5 or 6 points, Zajonc says, according to the relationship previously reported by other researchers who compared individual test scores and birth order. He believes it is possible that aggregate effects were supplemented by collective effects: As the baby boom proceeded after World War II, the preponderance of younger siblings in larger families set less rigorous standards for each other, leading to collective effects that resulted in test scores dropping at a faster rate than family fertility rose.

When SAT scores went up slightly last year, College Board President Donald Stewart said it was because the test-takers had worked harder than their predecessors. "They have taken more honors courses and more pre-calculus, calculus, chemistry, physics and other academic courses, and are more computer literate. They are also more ambitious with over half planning to go beyond the bachelor's degree."

Zajonc's work agrees with Stewart's evaluation. He points out, however, that the students may work harder because of a more challenging intellectual environment created by their generation, not because schools consciously adopted stricter standards, as some education officials and politicians suggested in last year's media reports on SAT trends.

Aggregate birth order effects could account for recent changes in British A-level exam scores as well as for SATs, Zajonc says. When the average scores rose on the A-level last year, a debate developed in Parliament, with some politicians suggesting the exam graders must have let the standards slip. But the size of the gain in the aggregate A-level scores, Zajonc says, is comparable to what could be expected from aggregate changes in birth order alone.

Confluence model

In 1976, Zajonc, who was then at the University of Michigan, developed a mathematical model that could explain several other patterns in the data on birth order and test scores. It can account for higher scores for older siblings, firstborns with siblings outperforming only children, firstborns not scoring better than later-borns when tested at young ages, and twins performing below average.

In his "confluence" model Zajonc assigns the arbitrary "mental age" of 30 to each parent and uses the children's real ages. He adds the ages up and divides them by the number of people in the family to get a number quantifying the intellectual climate of a family in a given year. The model characterizes the family's climate changes over time and shows the eventual advantage for the firstborn.

To explain the model's logic, Zajonc asks people to compare "the linguistic experiences of a firstborn son before the birth of his younger sibling with those of a girl who already has several younger brothers and sisters. During his pre-sibling period, this firstborn interacts mainly with adults. He is exposed to a fairly sophisticated language, a rich vocabulary and a variety of life's domains. He lives in an adult world. Just being surrounded by adults gives him auditory access to a pool of words, many of which he will be asked about in his SATs."

In contrast, the girl with younger siblings "hears the language of toddlers because she lives in a toddler's world. The pool of words in which she is immersed is restricted, more primitive, contains very few low-frequency items and does not afford the formation of complex sentences." The parents may even change their own vocabulary as toddlers dominate.

Zajonc suggests that older children may benefit by acting as tutors to their younger siblings. By tutoring a younger brother or sister on how to hold a baseball bat or on the meaning of a word, the older child eventually may benefit more than the sibling and give him or her an advantage over children with no siblings. Some learning research suggests that people learn by teaching others. In the current era, where there are many only children in classes, teachers might be wise to instruct half the class in decimals and the other half in the lowest common denominator, letting them tutor each other, he suggests.

Richard Snow, a standardized testing expert on the faculty of Stanford's School of Education, calls Zajonc's theory a "plausible idea based on data collected around the world over 30 or 40 years." Snow says he has been persuaded that birth order is a factor in achievement or intelligence test scores and that a change in birth patterns over decades could explain at least part of the rise and fall in average tests scores for nations or states. Snow and others say it is important for policy makers to take into account such effects, lest they misinterpret aggregate scores and attribute changes in generational performance to other factors.

Zajonc and Mullally are careful to point out that their comparisons aren't very meaningful to couples planning their own family's size or spacing. "Many factors such as parental support, enthusiasm for one's teachers, effort, or on the negative side, a neighborhood environment that derogates academic values all contribute to a particular student's SATs," they wrote. Those factors don't affect the aggregated data because the researchers are comparing age cohorts. By including every test-taker's score within a given cohort, differences that have a larger impact on individual scores than birth order have been reduced to a single average for the cohort.

Moreover, they write, "test scores are not everything. Birth order and family size may contribute quite differently to psychological well-being and economic attainment than to test scores." For example, it could be that later-born children from larger families might be more affectionate, better leaders or less prone to depression, Zajonc and Mullally say. Research recently reported by Frank Sulloway, a visiting scholar at MIT, indicates that later-borns are less likely to defend the status quo than firstborns and that the later-borns' "openness to experience" has made them more numerous among leaders of scientific and political revolutions in history.


By Kathleen O'Toole