It is a leap from showing socioeconomic-status influences on IQ to arguing that school programs have the same kinds of effects.
A recent study of IQ concludes that environment is more important than heredity for the IQs of children from low-socioeconomic-status families. (“IQ Study Weighs Genes, Environment,” Sept. 10, 2003). This is a very important result, particularly because the study applied a sophisticated methodology based on identical-twin data—one used by other psychologists to argue that IQ is determined mostly by genetic characteristics (for everyone).
While I agree that socioeconomic status has an important influence on IQ, as would many social scientists, the study by University of Virginia psychologist Eric Turkheimer and colleagues is limited for the purpose of policy because it offers no data on the specific types of environmental conditions that influence IQ.
The authors say that socioeconomic status is measured by a combined index of parent income, education, and occupational status, all of which are clearly parent characteristics. No information is provided about other family characteristics or other environmental conditions. Are these the only environmental factors that influence IQ, or do other family or parenting characteristics also count? Does the larger environment matter, such as school programs, neighborhoods, or peer groups? The children in the study were only 7 years old, so most would have no more than three years of schooling plus, possibly, a preschool experience.
The authors of the study are quite candid about this limitation, and the paper itself does not comment about what environmental conditions matter except their socioeconomic status, or SES, index. But that does not prevent speculation—by the researchers and by education experts—about other environmental conditions that might be important. Mr. Turkheimer, for example, was quoted in Education Week as saying that these environmental factors could include “Head Start and other high-quality programs for children of low socioeconomic status.”
It is quite a leap from showing socioeconomic- status influences on IQ to arguing that school programs have the same kinds of effects. There is a substantial body of research that identifies the most important “risk factors” for a child’s IQ (and for academic achievement in general), and most of them reside in the family. In contrast, the evidence on the effect of compensatory school and preschool programs is decidedly mixed. Moreover, the effects of special school programs are generally small when compared with the influences of family conditions.
My own research, based partly on this literature and partly on data from the National Longitudinal Study of Youth, identifies 10 of the most important risk factors for a child’s IQ and academic achievement. Since most of these risk factors influence a child’s IQ by the age of 5, they clearly occur within the family and, equally important for policy, they operate during the early years of a child’s intellectual development.
The major risk factors for a child’s IQ include the well-known socioeconomic characteristics identified in the Turkheimer study, but they also include other family and parenting characteristics, some of which have stronger effects than parent income and education. Ranked by strength of influence, these risk factors are: parent IQ; parent cognitive stimulation (instructional activities) for that child; parental nurturance of that child; family income; parent education; family structure (two parents vs. never married mother); nutrition (for example, breast feeding); number of children in the family; mother’s age at first child; and child’s birth weight.
There is growing evidence that the likelihood of changing IQ diminishes with age, and that it is very hard to change after the 1st or 2nd grade.
While parent IQ probably includes genetic effects, and low birth weight can reflect physical illness, the remaining eight risk factors in this list are clearly environmental conditions provided by the child’s parents.
Most social scientists and education researchers would agree with the basic premise that family environment and parenting behaviors have strong influences on cognitive skills; indeed, this is why low- and high-SES children start school with a large cognitive-skill gap. But there is disagreement on whether school programs or other types of special interventions can overcome the cognitive disadvantages that low-socioeconomic-status children bring to school. There is growing evidence that the likelihood of changing IQ diminishes with age, and that it is very hard to change after the 1st or 2nd grade.
This does not mean that children stop learning after the early elementary years; all children learn considerable amounts of new knowledge and skills every year they stay in school. But children with lower IQ scores have to learn just as much as children with higher scores, simply to keep the gap constant. And to reduce the IQ gap, they have to learn much more. This situation arises because IQ tests and most standardized achievement tests are normed to specific ages, and therefore maintaining the same IQ or achievement score from one year to the next assumes an “average” amount of learning. It is very difficult for a low-IQ child to learn more than a high-IQ child, and it rarely happens for large groups of children such as an entire school.
Research on the effects of preschool, compensatory school programs, and school resources generally supports this argument. Although regular Head Start programs have been shown to improve various motivational and behavioral outcomes (in studies of the Perry Preschool Project, for example), results are disappointing for sustained gains in IQ or basic cognitive skills. The experimental Early Head Start program, which begins during infancy, has been somewhat more promising for improving the IQ of at- risk children, but the reason may be that this program replicates the parenting behaviors of cognitive stimulation during the very early developmental years.
Compensatory programs during the regular school years, such as the federal Title I program, have also been disappointing. The best evidence was produced by the well-designed longitudinal evaluation called Prospects, which followed a large national sample of children from the fall semester of 1st grade until the spring semester of 3rd grade. The Prospects study found no significant gains in reading or math scores for children who received Title I services when compared with children who did not.
Finally, there is the ongoing debate about the effects of increasing or enhancing regular school resources and teacher characteristics, such as overall school expenditures, class sizes, teacher salary, teacher education, teacher certification, teacher subject-matter mastery, and so forth. The strongest evidence for achievement benefits may be from smaller class sizes and improved teacher subject-matter mastery, but even here there are conflicting results. For example, California’s attempt to implement a class- size reduction comparable to that in the Tennessee STAR project on class size has so far failed to produce any important gains in achievement.
For policy purposes, the variability in results is more important than whether a particular study demonstrated a large effect for a particular group of students. Studies of family characteristics invariably show large effects on IQ or achievement, regardless of location, sample sizes, and so forth. In contrast, studies of school interventions yield mixed results at best, and the effects are usually small when compared with family characteristics.
If we are really serious about improving the IQs of low- socioeconomic-status children, we must look hard at programs that aim to change family behaviors or supplement parenting behaviors for infants.
For example, my recent study Maximizing Intelligence compares five family socioeconomic characteristics with six commonly cited school resources, using the 1996 National Assessment of Educational Progress math scores for 13- year-olds. All five of the socioeconomic characteristics had very strong and statistically significant effects on math achievement; all but one of the school and teacher characteristics had effects near zero and were not significant. The one exception was a teacher characteristic: having a college major or minor in mathematics (subject-matter mastery).
This is not to say that a particular school intervention cannot have a beneficial effect on the academic achievement on some children in some schools at some point in time. The problem is being able to reproduce and maintain that positive effect in a consistent fashion, so that it can be replicated for at-risk children in any school. The family risk factors, operating during early childhood, do not have such weak and transient effects; their influence appears to be substantial and ubiquitous.
It would seem to me, then, that the immediate reaction of policymakers to the Turkheimer IQ study should be to support policies and programs aimed at improving family risk factors, rather than trying yet another special school program with uncertain outcomes. Of course, family policies and programs lie outside the reach of conventional education policies, and trying to change family characteristics or parenting behaviors is not something that most educators would easily embrace.
But if we are really serious about improving the IQs of low-socioeconomic-status children, we must look hard at programs that aim to change family behaviors or, at least, to supplement parenting behaviors for infants. The best example of the latter approach is Early Head Start, which does the best job so far of emulating some of the critical parenting behaviors. If we consider this to be a form of education (supplementing parent instruction), perhaps educators should embrace this type of intervention as a legitimate form of education policy.
David J. Armor is a professor of public policy at George Mason University, in Fairfax, Va., and the director of the Ph.D. program in the university’s school of public policy. His book Maximizing Intelligence was published in July by Transaction Publishers.