CHAPTER SEVEN

Mind the Gap

Compensatory education has been tried, and it apparently has failed.

—Arthur Jensen (1969)

There is no evidence that school reform can substantially reduce the extent of cognitive inequality as measured by ability] tests.

—Christopher Jencks and others (1972)

There is no reason to believe that raising intelligence significantly and permanently is a current policy option, no matter how much money we are willing to spend.

—Charles Murray (2007)

IN 2002 THE U.S. CONGRESS passed the No Child Left Behind Act, which mandated that American schools eliminate the gap between the social classes and between minority groups and whites by 2014. I don’t know if most members of Congress actually believed that such accomplishments are possible. But if so, they are deeply ignorant of the forces that operate to produce high academic achievement.

Intellectual capital is the result of stimulation and support for exploration and achievement in the home, the neighborhood, and the schools. To think that this can be changed by mandate—operating only through the schools—is preposterous. Moreover, the schools attended by minorities and the poor are wanting in ways that cannot be drastically improved overnight. The problems include quality of teachers willing to work in these less rewarding schools, the caliber of school management, the disruptiveness produced by high levels of student turnover, and the nature of the schools’ clientele, whose homes and neighborhoods make it unlikely that they will be encouraged toward high academic achievement.

It should be clear from the previous chapter that there is no theoretical limit on the degree to which the achievement gap between blacks and whites can ultimately be closed. Though there is far less evidence on the native intellectual ability of the extremely broad and diverse group of cultures labeled as “Hispanic,” I see no reason why the gap cannot ultimately be bridged there as well.

On the other hand, it should be clear that unlike the black/white and Hispanic/white gaps in achievement and IQ, the social-class gap is never going to be closed. This is true, if for no other reason, because the well-off are always going to find ways to get a better education for their children and are always going to find ways to be ahead in terms of parenting skills and are always going to be able to provide superior neighborhood environments. In addition, there is always going to be at least some difference in the gene pools of the lower class and the middle class. Recall from Chapter 1 that within a given family the sibling with a substantially higher IQ achieves much higher socioeconomic status (SES) than less favored brothers and sisters. And since the higher IQ is attained in part by virtue of a better luck of the draw from the gene pool of the parents, higher SES is always going to be in part a result of better genes for intelligence. So higher-SES people are going to pass along better prospects for intelligence to their offspring by virtue of having, on average, better genes and by offering better environmental advantages to their offspring.

But these considerations should not be cause for pessimism about the degree to which the intellectual lot of lower-SES people can be improved. Recall from Chapter 2 (on heredity) that the effect of an upper-middle class upbringing on children born to lower-SES parents is to raise the IQ by 12 to 18 points. The theoretical ceiling for improvement of lower-SES intellectual capital is very high indeed.

But how much improvement can we realistically hope to produce for lower-SES individuals and for currently disadvantaged minorities?

Early Childhood Education

When I tell people that I am writing a book on the modifiability of intelligence, they sometimes tell me, so as to encourage me to avoid wasting my time, that Head Start does not work. For many people their belief about one particular program settles the matter of whether intelligence can be manipulated.

Head Start is a compensatory program initially aimed primarily at improving the health and welfare of poor children three and four years old. Some of the founders hoped that it would also lead to improvements in the children’s intelligence, school performance, and subsequent success in life. Head Start sessions run for a half-day five days a week, typically for thirty-four weeks, but only a small portion of each session specifically targets cognitive concerns.

Is Head Start a failure? It depends on your perspective. With respect to physical health, Head Start has been a resounding success. It results in mortality rates for children that are 33 to 75 percent lower than for comparable children not in the program. It has driven mortality rates down, in fact, to a level not substantially different from those for children in general.

In earlier days, Head Start was associated with a gain of about .35 SD on cognitive tests, or about 5 IQ points, when the child finished the program at age five, and more recent studies showed that there is still a .10 to .20 SD effect on some IQ and achievement variables at age six or seven. These fade into nothingness by late elementary school. Recent reports show lower effect sizes at age five—on the order of .25 SD. But it is difficult these days to find true control groups for pre-kindergarten interventions because the majority of even poor minority children receive some kind of pre-kindergarten care. As a result, effects are compared, not to no-treatment controls, but to control groups in which typically half of the children have some day care.

There are shockingly few evaluations on the long-term academic effects of Head Start. What little there is indicates a slight effect of Head Start on completion of high school, which is about 2 to 5 percent greater than for controls, and a small effect as well on likelihood of attendance in college, which is about 3 to 6 percent greater than for controls. The cost of Head Start is about $7,000 per child, so whether the program is worth the gain from an intellectual and academic standpoint is an open question.

Early Head Start, which begins at birth and continues to age three, has proved to be no more successful than Head Start in improving educational outcomes. Services include child development, child care, home visiting, parenting education, and family support services. Individual programs have been given substantial latitude about which services to emphasize. Effect sizes on a range of variables from the purely cognitive to the emotional and social are in the range of .10 to .30 SD—slightly higher for minority children than for white children. Even the best versions of the program produced IQ gains of less than 4 points in the short term (though vocabulary scores increased by .40 SD). The program is expensive, and it is not clear whether long-term gains (which apparently will not be measured) would justify the cost.

But there are more ambitious programs than Head Start, and some of them have much bigger and more lasting results. A review of about a dozen of the better small-scale preschool and kindergarten programs focusing on black children showed that they lead to big gains in IQ—of as much as .70 SD or even more at age five. They also result in significant achievement gains in the first few grades of elementary school, but these gains generally fade, often completely.

Fading is to be expected if high-quality environments are not maintained. Only if children’s brains are like clay would we expect them to remain in good shape years after they were formed. If children’s brains are like muscles, however, then we would expect exercise, in the form of stimulating environments and activities, to be necessary to maintain good performance. I favor the muscle view, and so do the data.

As it turns out, several early childhood education programs actually do produce large immediate gains in IQ, as well as long-term gains in IQ or academic achievement, or both. Let’s look at three of the more effective programs that randomly assigned children to treated versus untreated conditions and that followed the children until late adolescence or adulthood.

The Perry Preschool Program was carried out in Ypsilanti, Michigan, between 1962 and 1967 by Lawrence Schweinhart and David Weikart. It was administered to fifty-eight black children living in poverty whose mothers had IQs between 75 and 85 on the Stanford-Binet IQ test. Children entered the program at age four in the first year of operation and at age three in the remaining four years. With the exception of the first group, children spent two years in the program.

The Perry treatment consisted of daily morning sessions in a classroom lasting two and a half hours for thirty weeks each year and focused on activities that would foster cognitive growth and social development. The average child-teacher ratio was 6 to 1, which is extremely low, and staff were very well trained in early childhood development and education. Once a week, the teacher would also visit each child’s home for ninety minutes to encourage the mother to become involved in the educational process. The fifty-eight children comprised the treatment group, and sixty-five were in a control group. When children completed the program, they entered school in the disadvantaged neighborhoods where they lived.

The mean IQ of the children in the control group at the end of the program, at age five, was 83. The mean IQ of the treatment group was 95. The IQ of the treatment children dropped progressively over the years of grade school until it was the same at age ten as the control mean, 85.

The IQ results are disappointing, though scarcely surprising given the circumstances of home, neighborhood, and school after termination of the program. What is surprising is that the academic gains and subsequent economic and social gains were enormous. These are summarized in Figure 7.1A Chapter Seven. About a third of the control-group children were assigned to special education classes at some point, versus 13 percent of the treatment group. At age fourteen, only 14 percent of the control group tested above the 10th percentile on the California Achievement Test, compared with almost half of the treatment group. Effect sizes for reading, math, and language achievement scores ranged from 0.50 to 0.75 SD. Forty-three percent of the control group managed to graduate from high school, compared with 65 percent of the treatment group. High school grades were .57 SD higher for the treatment group than for the control group. At age twenty-seven, only about 6 percent of the control group earned as much as $2,000 per month versus 28 percent of the treatment group. Eleven percent of the control group owned their own home, whereas 33 percent of the treatment group owned their own home. About 20 percent of the control group had managed never to be on welfare, compared with 40 percent of the treatment group. Eight percent of the control-group women and 40 percent of the treatment-group women were married. By the age of forty, 55 percent of the controls had five or more arrests; 36 percent of the treatment group had five or more arrests. These results have tremendous social and economic implications.

The Perry program is by no means unique in showing fading IQ gains for intervention groups combined with big achievement advantages later in life. These advantages include lower percentage retained in grade, lower percentage in special education, and higher percentage graduating from high school. The fact that gains in academic and life achievement can be so great even though IQ gains had completely dissipated has led many to suspect that the achievement gains are attained not through increased intelligence per se but rather through temperamental or motivational changes that resulted from the intervention and that persisted even when the IQ gains were no longer supported by the environment.

An intervention even more ambitious than the Perry program was started by researchers in Milwaukee. Rick Heber, the program’s initiator, discovered that one particular area of the city, which had 3 percent of the population, accounted for 33 percent of the mentally retarded children in the district. He decided to concentrate his resources on that section of the city. All of the children recruited for the study were African Americans at high risk for mental retardation because their mothers were poor and had IQs of 75 or less. The children were randomly assigned either to a control group (eighteen children) or to an intervention group (seventeen children), which was an intensive day-care program lasting from the time the children were less than six months old until they enrolled in first grade.

The intention of the Milwaukee Project was to give children the equivalent of a middle-class environment. The program focused on developing their language skills and cognitive capacities. Paraprofessionals frequently interacted with the children in an enjoyable way, using the best developmental programs and educational toys known at the time. Sessions lasted seven hours each day, for five days each week. The program provided the children with good food and high-quality medical and dental care and offered training in homemaking and child care to the mothers. The treatment group was compared not only to the control group but also to a low-risk comparison group of children born to mothers of average to above-average intelligence (108 on the Wechsler Adult Intelligence Scale). At the age of thirty months, the control-group children had an average IQ of 94 on the Stanford-Binet test, compared to 124 for the treatment-group children. The treatment group actually scored higher than children born to mothers with average or above-average IQs. Those children scored 113. By age five, the control group scored 83 on the Wechsler Preschool and Primary Scale of Intelligence, compared to 110 for the treatment group, which was still higher than the IQ score of 101 for the low-risk comparison group.

At termination of the program at age seven, the average treatment-group IQ was still 22 points higher than the average for the control group. (Note that this gives an even higher upper bound for the difference between typical lower-SES rearing strategies and superior strategies.) For grade school the Milwaukee children, unlike the Perry children, were put into reasonably high-quality schools—all rated at or above the national average for scores on achievement tests. This enabled children in the treated group to maintain their IQ gains. Nine years after the program was over, when the children were adolescents, the control-group children scored 91 on the Wechsler Intelligence Scale for Children and the treatment-group children scored 101—on par with the 97 IQ for the comparison group born to mothers with average or higher IQs.

Achievement scores for grades 1 through 4 for the treatment group were higher than those for the control group, and the difference was great in standard deviation terms—around .75. The number of children was low, however, and there is a 10 percent probability that a difference that big could have been obtained by chance (though there is only one chance in twenty for a reliable difference specifically in favor of the intervention group, as opposed to a difference in favor of either group).

A yet more intensive intervention was attempted by Craig Ramey, Frances Campbell, and their colleagues in a project called the Abecedarian Program. Almost all of the 111 children involved in the program, who were born between 1972 and 1977, were African American. The children were considered as being at high risk of retardation in light of their mother’s IQ, which averaged 85, and mother’s education, which averaged ten years, in addition to other risk factors such as low family income, absence of father, poor social or family support for the mother, poor academic performance of siblings, employment of parents in unskilled labor, and reliance on public agencies for support. Abecedarian was a full-day intervention that began before children were six months old and lasted all year-round. The infant-teacher ratio was 3 to 1 at first and increased to a child-teacher ratio of 6 to 1 as the program progressed. The intervention program continued up to kindergarten age. Data were collected on the participants regularly up until age twenty-one.

The program included four groups of about twenty-five children each. One group was assigned to the infant-kindergarten intervention and also to a school-age intervention. The latter intervention provided, for the first three elementary school grades, a home-school teacher who met with the parents and showed them how to supplement the educational activities at home. Parents were encouraged to work with the child for at least fifteen minutes a day. This home-school teacher was also a link between the school teachers and the family. She met with teachers and families once every two weeks. She also helped the family with finding jobs, dealing with social service agencies, and taking the children to doctor appointments. One group of children was assigned only to the preschool intervention, and one group only to the school-age intervention. The fourth group was assigned to neither intervention.

At age three, children who had no preschool had an average IQ of 84; children who received the preschool intervention had an average IQ of 101. At termination of the intervention, control children had an average IQ of 94; preschool intervention children, an average of 101. Then, instead of attending poor-quality inner-city schools, all children went to schools where most of the children were reasonably well-off white kids. At age twelve, only 13 percent of children exposed to the intervention had IQs of less than 85, compared to 44 percent of the control children. Even at age twenty-one, those who had the preschool intervention had an average IQ 4.5 points higher than the average for those who had been in the control group. The children whose mothers had the lowest IQs (less than 70) benefited the most from the program. There is no evidence that the school-age treatment added anything in the way of IQ to the preschool treatment, nor did the school-age treatment by itself accomplish much. This study is one of many prompting pessimism about the effects of home visits—unless they are very ambitious in their scope.

By the time the subjects in the study reached the age of twenty-one, it was clear that the Abecedarian preschool intervention had had a major influence on many educational outcomes, summarized in Figure 7.1.B. Almost half of those in the control group had been assigned to special education classes at some point in their educational career, versus fewer than a fourth of those in the intervention group. Over half of the control group had repeated a grade, whereas 30 percent of the intervention group repeated a grade. At age fifteen, the reading scores of the intervention group were 1.40 SDs higher than those of the control group, and the math scores were .86 SD higher than those of the control group. By age twenty-one, the intervention group was two grade-years ahead of the control group in reading scores and more than a year ahead in mathematics scores. Half of the control group had graduated from high school, versus two-thirds of the intervention group, and 12 percent of the control group had attended a four-year college, versus a third of the intervention group. At age twenty-one, fewer than 40 percent of the control group were either in a skilled job or in higher education, compared with two-thirds of the intervention group.

Figure 7.1. Academic, economic, and social outcomes for the Perry Preschool and Abecedarian programs. (A) Data from the Perry program collected when the individuals were twenty-seven years old. > 10th percentile achievement, children who scored above the lowest 10 percent on the California Achievement Test (1970 at age fourteen; HS Grad., percentage of children who graduated from high school on time. (B) Data from the Abecedarian Program collected when the individuals were twenty-one years old (Carolina Abecedarian Project and the Carolina Approach to Responsive Education, 1972–92. Light bars, intervention group; dark bars, control group. From Knudsen, Heckman, Cameron, and Shonkoff (2006).

The Abecedarian effects reported are probably underestimates of the actual effects because the control-group children received pre-kindergarten care of some kind.

There have been several replications of at least part of the Abecedarian intervention. One is important because it refutes the contention by Herrnstein and Murray that the results of the Abecedarian study are suspect since substantial IQ differences were found at age one. Herrnstein and Murray believed such differences could not be due to the program at such an early age and therefore were indicative not of program effectiveness but of a failure of random assignment to produce equivalence of control and intervention groups at the outset. Project Care, using methods essentially like those of Abecedarian, started with children assigned to intervention versus control groups who had identical scores on the Bayley Mental Development Index at six months. By age one the scores for the two groups differed by 11 points. So Abecedarian has effects on the IQ of even very young children.

Another replication of Abecedarian is particularly important because it shows that the program can be used to boost the IQ of children at risk of mental retardation owing to prematurity and low birth weight—2,001 to 2,500 grams. Two-thirds of the infants in the study were black or Hispanic. About 40 percent of the mothers had not graduated from high school, and only about 13 percent had completed college. The program was different from Abecedarian in that rather than starting shortly after birth, it began at age one, and continued only until age three. At the end of the intervention at age three, IQs of the intervention group averaged 9.2 to 12.5 points higher than control IQs, depending on the test used.

Two years after the end of the intervention, IQs of the intervention group were 2.5 to 5.4 points higher, depending on the test used (the lower estimate was not statistically significant). At age eight, there were still detectable effects of the intervention, with intervention-group IQs ranging from 3.6 to 5.4 points higher, depending on the test used. Even at age eighteen, there was a detectable difference in IQ of 3.8 to 5.3 points between youth who had been in the intervention program and those who had not. Achievement differences were not impressive at any point.

IQ gains for the intervention group were much larger for the children who stayed in the program the longest. But this kind of analysis—showing that participants who stayed longest benefited most—invites a self-selection bias, which was not completely compensated for by control procedures used by the investigators. And it should be noted that treatment effects were not significant beyond age three for infants who weighed extremely little at birth (less than 2,001 grams).

A particularly important fact is that early childhood intervention programs benefit black and Hispanic children more than white children (who already have more of the advantages conferred by the interventions, on average), and benefit poor children more than middle-class children.

In short, early childhood intervention for disadvantaged and minority children works—when it is strenuous and well conducted. Many different programs get high gains in IQ by the time they end. These gains generally fade over the course of elementary school, but there is some evidence that this is less true if children are placed in high-quality elementary schools. Much more important are the achievement gains that are possible: lower percentage of children assigned to special education, less grade repetition, higher achievement on standardized tests, better rates of high school completion and college attendance, less delinquency, higher incomes, and less dependence on welfare. And these changes can be very large.

There has not been much research on the effectiveness of home-visitation programs having the intent of improving parenting practices. Some reasonably ambitious parenting interventions led to improvements in mothers’ behaviors as well as children’s emotional and cognitive behaviors. The key to success appears to be coaching parents on specific behaviors.

One particularly effective intervention indicates that such programs might have great value. This was conducted by Susan Landry and her coworkers. They went into the homes of mostly lower-income, black and Hispanic mothers of one-to two-year-old infants and provided ten to twenty 1½-hour sessions showing the mothers beneficial patterns of response to their infants. They worked on teaching mothers how to interpret the intent of their child’s positive and negative signals, how to respond to the child’s behavior in warm and sensitive ways even when the child denies the mother’s requests, how to attend to the child’s focus of attention and maintain and build the child’s interest, when to introduce interactions and games, and how to use vocabulary-rich language, labeling objects and actions. The effect on mothers’ behavior was marked for many dimensions, including warm and sensitive behavior, responding appropriately to child’s intents, maintaining the child’s interest in activities, and verbally encouraging the child. Effect sizes for such outcomes ranged as high as 1 SD.

The effect on the children’s behavior was also substantial. Children became more cooperative, were more engaged when interacting with their mothers, used words more frequently, were better able to coordinate their words with what they were attending to, and scored higher on a picture vocabulary test. These effects ranged as high as .70 SD.

We do not know yet what the long-term effects of parental interventions are, but there seems to be good reason to be hopeful. The benefit-cost ratio of such programs could be very high.

School-Age Interventions

How about interventions for school-age children? What can be done for children who have not had powerful preschool interventions, or could be done to sustain the gains experienced by children who have been in center-based day-care programs?

I’ll start with some bad news—and unfortunately there is a lot of it. In Chapter 4 (on improving schools) I discussed some of the efforts made to improve the academic performance of children in general. Sheer amount of money spent doesn’t do a lot of good for students in general, so we would not expect money by itself to narrow the achievement gap between lower-and upper-SES students or between minority and white students. Vouchers for attending private schools have been given out to poor and minority children, but there is not much evidence that they are the answer. The same is largely true for charter schools. Some might do a better job than the public schools—in fact, you will read about one later—but charter schools in general seem to be little better than the public schools even after they have been in operation for a while, and they may actually be worse in their early years.

Are there some regular public schools that do a particularly good job with disadvantaged minority populations? Two different reports maintain that at least some schools do a superb job with their underprivileged, minority clientele. One report is by a conservative institution, and one is by a liberal institution.

The Heritage Foundation, a conservative institution, claims to have found twenty-one high-poverty schools whose students have achievement levels above the national norms. These schools are held up as mavericks that escape from low performance by dint of having visionary principals who are willing to buck the “cult of public education,” fire bad teachers, and teach the basics instead of progressive nonsense.

Richard Rothstein, the former national education columnist for the New York Times, debunked these claims. According to him, only six of the twenty-one schools were fully nonselective neighborhood schools. The rest were (a) magnet schools; (b) schools where many of the parents were “impoverished” graduate students; (c) schools that did indeed produce high scores early on by concentrating on the basics such as phonics, but scores declined precipitously in later grades precisely because the emphasis on the basics precluded learning reasoning and interpretation skills necessary for success in the upper grades; or (d) schools where parents had to apply to get their children accepted, thereby introducing a potentially heavy self-selection bias.

The Education Trust, a liberal organization, claims to have uncovered 1,320 schools in which at least half the students were both poor and minority and whose test scores were in the top third for their states. The claims about these schools do not hold up either, according to Rothstein. The 1,320 schools did in fact have high scores, but in only one grade, in only one subject, and for only one year. These accomplishments were mostly statistical flukes.

Rothstein gives an even more bracing back of the hand to yet another claim, this one by Douglas Reeves, who says he has identified a group of “90/90/90” schools in Milwaukee where 90 percent of the students are poor, 90 percent are minority, and 90 percent meet “high academic standards.” Those standards turn out to be only basic, nonproficiency level scores as defined by the state of Wisconsin.

How about claims that if you put poor minority kids into a school with children whose families are well off, their academic achievement can soar? The New York Times announced on its front page in 2005 (and then repeated in another front-page story in 2007, and yet again in a magazine story in 2008) that such effects had been achieved in Wake County in Raleigh, North Carolina. Under the headline “As Test Scores Jump, Raleigh Credits Integration by Income,” the Times said this:

Over the last decade, black and Hispanic students here in Wake County have made such dramatic strides in standardized reading and math tests that it has caught the attention of education experts around the country.

The main reason for the students’ dramatic improvement, say officials and parents in the county, which includes Raleigh and its sprawling suburbs, is that the district has made a concerted effort to integrate the schools economically.

The article goes on to state that since 2000, officials used income to guide assignment to schools, with the intention of holding the proportion of low-income students to less than 40 percent. The result?

In Wake County, only 40 percent of black students in grades three through eight scored at grade level on state tests a decade ago. Last spring, 80 percent did. Hispanic students have made similar strides. Overall, 91 percent of students in those grades scored at grade level in the spring, up from 79 percent 10 years ago.

Unfortunately, there is no evidence that schools in Raleigh do a better job for minority students than schools statewide. The black/ white gap was actually slightly less for students in the state as a whole in 2004/05 than it was in Wake County. Moreover, the apparent gains in Wake County by minority students were almost surely due to an easing of statewide standards for proficiency. By the new standards 95 percent of whites were proficient, which indicates that the standard was so lenient it is impossible to get a read on the actual size of the gap.

The data comparing Wake County with the state as a whole are encouraging in one respect, though. They indicate that integration of low-income children with higher-income children may hurt the academic achievement of neither—and may result in social gains for both. But we would have to know much more—for example, the SES level of blacks and whites both statewide and for Wake County—before we could reach that conclusion. There is good evidence, incidentally, that black and Hispanic children learn more in integrated classrooms than in majority black or Hispanic classrooms.

How about efforts to improve schools by instituting top-to-bottom changes in administration and curriculum? We know from Chapter 4 that there are a lot of such whole-school interventions, in which entire curricula and educational strategies are set in place in a school. But those interventions tend not to produce very impressive results for students in general, so we have to be skeptical as to whether they would be very successful in reducing the achievement gap.

Teacher certification and higher academic degrees do not improve the achievement of anyone very much either. Experience in teaching counts, though, and there is a possibility that experience matters more for lower-class than middle-class kids or more for minority kids than for whites. Teacher quality also makes a big difference for students’ achievement scores. Again, there is the possibility that teacher quality matters more for poor and minority kids. We certainly know that Miss A, heralded in Chapter 4, made a huge difference for the poor kids in her first-grade class. And we know that the performance of poor kids and of kids who had been disruptive in kindergarten is greatly affected by the quality of instructional support and emotional care that they get in first grade.

We also know that smaller class sizes result in better performance on achievement tests, and that the effects are bigger for black kids than for white kids—.33 SD versus .25 SD—and bigger for poor kids than for middle-class kids.

There are some big success stories in K–12 education though. Two educational programs for poor and minority children—one for math and one for reading—have been shown to be quite effective.

The math training program is called Project SEED. It is an enrichment program that hires and trains mathematicians, scientists, and engineers to teach poor minority students. The program teacher uses Socratic questioning to introduce abstract mathematical concepts and students become active participants in the lessons—through the use of dialogue, debates, and choral response. The regular teacher sits in on all sessions. Project SEED is not a replacement for the regular mathematics curriculum but an add-on. One study of Project SEED’s effectiveness, carried out in Dallas, compared the California Achievement Test (CAT) scores of 244 fourth-grade students in SEED classrooms with 244 students in the same schools who did not have SEED instruction and 244 students who were not in SEED schools but were deemed comparable to them. The SEED students outscored the comparison-group students by .37 SD—a very significant benefit. But SEED students outperformed the non-SEED students in their own school by only .19 SD. Arguments as to which is the more appropriate control can be made both ways, and it makes a good deal of difference which side one comes down on. A gain of .37 SD is surely worth the cost, which was not great; a gain of .19 SD may not be.

Several reading programs geared to minority and poor children have been developed, of which one of the most promising is Reading Recovery (Descubriendo La Lectura in its Spanish version). Reading Recovery is a tutoring program for low-performing first-graders that was developed by Ohio State University researchers. Tutors provide a daily one-on-one 30-minute lesson for twelve to twenty weeks. Children read stories they already know verbally, read a story they read the day before, and write stories. The Ohio State group conducted randomized studies to evaluate the program. They found effect sizes ranging from .57 to .72 SD for most indicators of reading proficiency. The effect sizes faded over time, though they were still detectable at about .20 SD in the third grade. One independent study evaluating the Spanish-language program demonstrated effect sizes ranging from about 1.00 to 1.70 SDs. This finding would need to be duplicated before we could place much confidence in it.

There is at least one extremely important exception to the rule that whole-school interventions, and charter schools, have only modest effects on student achievement: the Knowledge Is Power Program, or KIPP. Michael Feinberg and David Levin, two Houston-area elementary-school teachers with four years of teaching experience between them, founded this extraordinarily ambitious educational project in 1994. They designed it around the needs of poor children, especially minority children. They freely admit that they invented the program as they went along. But they had help from a much more experienced mentor—Harriett Ball—a teacher who had grown up in the racially segregated neighborhoods of Houston and who was known for having classes with students who were well behaved and high achieving.

Feinberg and Levin developed a type of school, initially mostly for middle-school children, based on 7:30 a.m. to 5:00 p.m. school days (yes, 9½ hours), mandatory attendance for three additional weeks in the summer, a half-day of Saturday classes biweekly, visits to students’ homes, an insistence on kindness and good behavior, the principal having the power to hire and fire teachers, cooperation among teachers, and a system of rewards and penalties for behavior and academic accomplishment. The huge amount of extra contact time allows for the KIPP students to get exposure to what upper-middle-class students get through their homes and expensive public or private schools—sports, museums, dance, art, musical instruments, theater, and photography. The first two schools—in Houston and the Bronx—scored highest on achievement tests of any schools in their areas. Since 2001 KIPP has begun to expand with financial support from Doris and Don Fisher, the founders of Gap stores. Most KIPP schools are charter schools that are essentially franchises of the KIPP Foundation. Two other similar programs, called Achievement First and North Star, have been less well researched.

KIPP’s students are economically disadvantaged as a group. More than 80 percent are eligible for federal free or subsidized lunches. Most are African American or Hispanic. KIPP maintains that “while the average fifth-grader enters KIPP in the bottom third of test-takers nationwide (28th percentile), the average KIPP eighth-grader outperforms nearly three out of four test-takers nationwide (74th percentile) on norm-referenced reading and math assessments.”

But some KIPP schools have posted losses for their students (to KIPP’s credit, by its own admission), and KIPP’s claims for successes have been based mostly on its own teachers’ tests and not on independent research. However, SRI International conducted an independent study of KIPP schools in the San Francisco Bay Area, in the early years of this century, and I report on it at length here.

In the five Bay Area KIPP schools, 72 percent of the children were economically disadvantaged and 75 percent were African American or Latino. Each school in the study was compared to two comparison schools that were similar socioeconomically and with respect to minority composition. The KIPP schools began in 2002/03 with an entering class of fifth-graders and added a class each year.

Principals study the KIPP model for a year before they get their own schools. But the KIPP model does not prescribe a particular instructional approach or curriculum, so teachers do not have to learn a new system required by the model, in contrast to most whole-school reforms. At the time of the study, about half the Bay Area KIPP teachers came from the Teach for America program, and their median teaching experience before joining KIPP was two years.

It is possible for children to fail in KIPP schools. Principals believe that the option to hold students back is essential because the school “couldn’t let other kids see that kids who did nothing could move ahead.” It is impressed on staff, students, and parents that certain behaviors are demanded of them. The KIPP credo is “If there is a problem, we look for a solution. If there is a better way, we find it. If we need help, we ask. If a teammate needs help, we give it.” Some sample KIPP slogans are “Work hard, be nice,” “All of us will learn,” and “KIPPsters do the right thing when no one is watching.”

Students earn “paychecks” each week, based on a point system that adds on to their value or subtracts from it depending on behavior and academic performance. The paycheck can be used to buy items from the KIPP store, including snacks and school supplies, and to buy attendance on field trips. Students get publicly “benched” for bad behavior and failure to do work. At one school, students have to go three days in a row without getting deductions from their paychecks in order to get off the bench. Teachers insisted to SRI researchers that all student actions have consequences. But teachers also maintained that good behavior was not due to fear.

“We’ve never had a kid talk back to a teacher, and we’ve never had kids fight. I don’t attribute this to the discipline system. It’s from setting expectations from the start. The smallest detail was called out…It’s because kids believe that this is an extraordinary place, and we’ve taught them that. I don’t think they don’t tease because they are afraid of the [bench]. It’s just something that they would not do at KIPP. This is the one school they’ve been to where there’s no teasing. They feel safe, and they are learning more.”

Another teacher said, “At this school, it’s okay to be smart, and that’s something that is lacking at most inner-city public schools…In [my former district], I worked with traditionally underserved kids and found the schools to be an inadequate solution to kids’ needs…I visited a KIPP school [nearby], and the school was like an oasis.”

Comments of students show that they are well aware of the difference between the KIPP school and others they have attended. “Everyone is committed to learning.” “The other school was not challenging enough, and I’ve found it has been here.” “Now I realize you have to work to get to college.” In all the schools, students said there were fewer fights, if in fact there were any at all, than in previous schools they had attended: “At this school, something holds me back from fighting.”

Of course the demands on teachers are enormous. They must be in school from 7:15 a.m. to 5:15 p.m. and are usually there longer. In addition to planning curriculum, they provide instruction, monitor study halls, lead enrichment activities like trips to zoos and museums, tutor, and call parents at night to discuss their children’s performance. They also work several weeks during the summer and every other Saturday. Not surprisingly, they fight emotional exhaustion. And in general, most KIPP teachers say they will be able to stay for only a few years.

What educational results does all this produce? Children in KIPP schools in the Bay Area succeed at levels well beyond what would be expected considering their demographic composition. The children were given the reading, language arts, and math Stanford Achievement Test (SAT 10), which is normed nationally, in the fall and spring. For fifth-graders, fall scores indicate where students are at the outset of the KIPP experience. The improvement from fall to spring was calculated in terms of the percentage of students scoring at or above the 50th percentile for national norms at each point. In the fall, for fifth-grade language arts, the proportion of students scoring at that level was 25 percent (on average for all classes at the four schools that gave the test), slightly more than we would expect given the demographic characteristics of the schools. In the spring, it was 44 percent, substantially higher than we would expect given the demographics. In the fall, for fifth-grade math, the proportion was 37 percent; in the spring it was 65 percent. Gains for sixth-graders were also marked. These increases are extremely large. They indicate that after just one year at a KIPP school, disadvantaged and largely minority children were scoring close to or above the national average on standardized tests. Since scores were not very impressive at the beginning of the year for those same children, we can rule out the possibility that self-selection is entirely responsible for the outcome in the spring.

In the spring, all children (at all five schools) were also given the California Achievement Test, which is required by law. For English Language Arts, 43 percent of KIPP fifth-graders scored at the level of proficient or above, versus 19 percent of the fifth-graders at comparison schools chosen for their demographic similarity to the KIPP schools. For Mathematics, 55 percent of KIPP fifth-grade students scored at the level of proficient or above, versus 20 percent of comparison-school students. Results on the CAT were entirely comparable for the sixth grade. Again, these results are extremely striking. (Baseline, beginning-of-the-year CAT scores were comparable for KIPP and the comparison schools.)

I have violated my own implicit research standards in reporting the SRI study. Students were not randomly assigned to be in KIPP schools versus control schools. And it could be done: KIPP schools sometimes have waiting lists and use lotteries to pick students. Unchosen children could be assigned to a randomized control group and given the same tests as the KIPP students. KIPP itself is worried about picking the cream of the crop. To their credit, some schools have taken steps to ensure that they get as many underprivileged minority students as possible, the population they regard as their target.

But there is no question that there is a severe self-selection problem, because parents, not researchers, decide whether their kids will get into a KIPP school. This automatically means that some of the performance advantages of the KIPP students might come not from the school, but from having parents, and perhaps other advantages, that work to their benefit. The question is, could self-selection have produced the results that were obtained from the students?

As a start to answering this question, we can at least be reasonably sure that attrition is not a major contributor to the favorable results. Only about 9 percent of students at schools studied by SRI International exit each year, some to avoid being held back. The schools themselves discourage some students from continuing, though no one is ever actually expelled. But relatively few leave, so self-selection out of the school seems not to be a serious contributor to the high scores of KIPP children.

In short, though I do not doubt that students whose parents choose KIPP schools are more promising than their demographics would indicate, I cannot imagine that the results obtained by KIPP could be primarily due to self-selection. However, before society invests a huge amount of money in KIPP schools, some fully randomized studies should be conducted.

I have little doubt that such randomized studies will show that poor minority children—at any rate, those whose parents care enough to get them into KIPP schools—can perform academically at levels as high as those that approximate those of middle-class whites. The next step will be to find out whether children whose parents are not so concerned about their children’s education can also benefit from KIPP-type programs.

High School Math for Poor Hispanics

You may have seen Stand and Deliver, the movie about math teacher Jaime Escalante’s achievement in getting East Los Angeles barrio students—who typically did not graduate from high school—to pass AP calculus at higher rates than students at rich Beverly Hills High, and for that matter at most elite high schools in the country. But is the story told by the movie true?

There’s good news and bad news about Escalante’s feat. Most importantly, that it happened is perfectly true.

But unfortunately it did not happen in the way the movie implies: In real life Escalante did not announce to unsuspecting seniors that he was going to make them into math whizzes that year. He built up math programs at junior high feeder schools that brought highly prepared students into his three-year high school. And he made sure his students had excellent courses in high school math before they ever came into his class. But all that was accomplished with substantial opposition from the first principal he worked under. Things began to move smoothly only when a sympathetic principal came on board (and banned students from being on athletic teams if they maintained less than a C average). And then Escalante had to fight a teachers’ union when his classes grew to be much larger than union rules allowed. He couldn’t find enough good teachers to increase the number of classes taught, so classes grew too large.

Then the sympathetic principal was replaced by one less sympathetic. Escalante left his school over the class-size issue and other problems. The program became progressively less successful after his departure, though the high school continues to do far better with AP math than most schools of its type.

The importance of Escalante’s achievement is very great. It serves as an existence proof for the contention that disadvantaged minority kids can function in math at a level far higher than the national average.

Inexpensive Interventions by Social Psychologists

Some of my fellow social psychologists have recently come on the education scene, bringing interventions that are very simple to carry out and extraordinarily cost-effective.

Many Americans believe that abilities are essentially fixed at birth: either you have math ability or you don’t. Others believe abilities are highly susceptible to manipulation: if you work hard, you will be better at a given skill than if you don’t. Carol Dweck and her coworkers have measured attitudes about ability in a group of mostly minority junior high school students, asking for beliefs about such questions as “You have a certain amount of intelligence, and you really can’t do much to change it” and “You can always greatly change how intelligent you are.” They showed, not surprisingly, that students who believe that ability is a matter of hard work get higher grades than students who believe ability comes from the genes.

Dweck and her colleagues then tried to convince a group of poor minority junior high school students that intelligence is highly malleable and can be developed by hard work. The thrust of the intervention was that learning changes the brain by forming new neurological connections and that students are in charge of this change process. Dweck reported that some of her tough junior high school boys were reduced to tears by the news that their intelligence was substantially under their control. Students exposed to the intervention worked harder, according to their teachers, and got higher grades than students in a control condition. The intervention was more effective for children who initially believed that intelligence was a matter of genes than it was for children who already were inclined to believe that it was a matter of hard work.

Joshua Aronson and his colleagues have performed similar experiments, with dramatic results. One study was conducted with poor minority students in Texas who were just beginning junior high school. Their intervention was intensive and the results were dramatic.

Each student in the Texas study was assigned a college-student mentor for their first year in junior high. The mentors discussed a variety of issues related to school adjustment. The mentors for the control participants gave information about drugs and encouraged their students to avoid taking them. Experimental-group mentors told their students about the expandable nature of intelligence and taught them how the brain can make new connections throughout life. Every student was exposed to a Web page that reinforced the mentor’s message. For students in the experimental group, this Web site showed animated pictures of the brain, including images of neurons and dendrites, and provided narratives explaining how the brain forms new connections when new problems are being solved. The mentors also helped the students design a Web page in which the students presented, through words and pictures of their own making, the message that the mentor had been presenting.

The effects of the intervention were very powerful. On the math portion of the Texas Assessment of Academic Skills (TAAS), performance by male students exposed to the intervention was .64 SD higher than for males not exposed to the intervention. For females, who tend to have worries about whether their gender makes them less talented in math, the difference was 1.13 SDs. For reading, students exposed to the intervention did .52 SD better than students in the control group.

Daphna Oyserman and her coworkers set up an elaborate intervention with poor minority junior high school students. They ran several sessions designed to make the students think about what kind of future they wanted to have, what difficulties they would likely have along the way, how they could deal with those difficulties, and which of their friends would be most helpful in dealing with them. These were supplemented with sessions during which students worked in small groups on how to deal with everyday problems, social difficulties, academic issues, and the process of working toward high school graduation. The intervention had a modest effect on grade point average of .23 SD, a moderately large effect of .36 SD on standardized tests, and a very big effect on likelihood of retention in grade of .60 SD.

Small interventions can also make a difference in college. Most students worry about social acceptance and fitting in on campus, but for minority students these concerns can be particularly worrisome. If they fail to make friends, because there are not that many minority students on campus and because they may feel ill at ease with majority students, they may begin to wonder if they belong on campus. It is common for minority students’ motivation to decline and for their grade point average to suffer as they go through school.

Social psychologists Gregory Walton and Geoffrey Cohen reasoned that lagging performance could be nipped in the bud if minority students knew that worries about social acceptance were common for all students, regardless of ethnicity, and that their situation would likely improve in the future. The researchers performed a modest intervention with black students at a prestigious private university. They invited black and white freshmen to participate in a psychology study at the end of their freshman year. The intent of the experimenters was to convince an intervention group that worries about social acceptance were common but tended to vanish as they developed more friendships. The experimenters expected that this would help black students to realize that the best way to understand their social difficulties was not in terms of their race (“I guess my kind of people don’t really belong at this kind of place”) and to replace that with the belief that their experiences were shared (“I guess everybody has these kinds of problems”). The researchers believed that recognition of their common problem—and its probable solution—would likely keep the students from worrying about belonging and help them focus on academic achievement.

To drive the point home, Walton and Cohen had students in the intervention group write an essay about the likelihood of improvement in their social situation in the future and deliver a speech in front of a video camera, which they were told would be shown to new students at the school “so that they know what college will be like.” The investigators then measured academic achievement behavior over the next week, as well as the students’ grade point average the subsequent semester.

The intervention had a big positive effect on blacks but not on whites. In the period after the intervention, blacks reported studying more, contacting professors more, and attending more review sessions and study group meetings. The subsequent term, grades of the blacks in the intervention group reflected these behaviors: their grades were a full standard deviation higher than those of blacks in the control group.

College as a Gap Reducer

It turns out that college itself has a huge effect on the intellectual abilities of blacks, substantially more than its effect on whites.

The black/white IQ gap grows in high school. Some hereditarians interpret this fact to indicate that the genes assert themselves more and more over the course of development. At each higher level of education, therefore, blacks could be expected to be farther and farther behind. Herrnstein and Murray, for example, maintained that it was unlikely that education beyond high school would serve to reduce the racial gap in IQ.

Data showing that the black/white gap grows during high school come from the National Longitudinal Survey of Youth, which administered the Armed Forces Qualification Test (AFQT). That test is a portion of the entire ability battery that the armed forces gives to potential soldiers. The AFQT correlates so highly with intelligence tests that it is appropriate to regard it as a measure of IQ. The test was given to participants in the survey at different points in their educational career from the age of fourteen to the age of twenty-one.

There is no question that, as Herrnstein and Murray showed, black ability increases less in high school than does white ability. The increase in the size of the gap is great enough to be quite disturbing. Blacks start out high school with scores less than three-fifths of a standard deviation lower on the AFQT than white scores, but end high school with scores almost a whole standard deviation lower than white scores.

Psychologist Joel Myerson and his coworkers set out to determine whether the same dismal failure of growth occurs for blacks in college. Based on the theory that ability differences manifest themselves more and more over time, we would expect less value added to IQ for blacks than for whites in college, resulting in a gap larger than that found in high school.

What Myerson and his colleagues found is the opposite. At the end of high school, black students who ultimately graduated from college performed more than 1 SD worse than white students who ultimately finished college. But the white students gained very little in IQ over the course of college, whereas the black students gained IQ at a remarkable clip, ending up with an average IQ only a little more than .40 SD below the average for whites. This difference in value added by a college education is huge.

Why do blacks gain so much in college? It may make more sense to ask why they gained so little in high school. The most obvious answer to that question is that blacks go to worse high schools than whites. A second answer is that pressure not to act white is harder to resist in high school than in college (if the pressure even exists at college).

A third possible answer lies in the research on stereotype threat, which demonstrates the remarkable variation in both test performance and motivation among black students, depending on the nature of the social circumstances they face. For example, Steele and Aronson’s experimental research documented remarkably better test performance among black students when the test was presented in a nonthreatening way—that is, when test-takers weren’t confronted with the explicit scrutiny of their intellectual abilities, presumably because doing so makes them anxious that their performance will mark them as fitting the stereotype of intellectual inferiority. In addition to impairment on tests, black high school students are more likely than white students to adapt to this evaluative discomfort in unhelpful ways, such as avoiding challenge and disengaging from academic pursuits, which are characteristic responses to stereotype threat seen among middle-and high-school-age students. One study that followed black students over their high school careers found a particularly sharp decline in black males’ engagement with academics as they progressed through high school, so that by the time they were in twelfth grade there was no connection between their feelings of self-worth and their academic achievement. These responses seem particularly likely among students who buy into the stereotypes about their group. Another long-term study found a direct link between students’ worries that the negative stereotypes about their group might be true and their later reduction in effort. So the circumstances that maximize stereotype threat may be especially prevalent in high school.

But the truth is, we do not know the reasons for the black gains in college. What we do know is that college produces a huge reduction in the ability gap. We also know that we have yet another important piece of evidence contradicting the idea that the IQ gap increases with age because of a genetic deficiency that manifests itself more over time. The IQ gap actually narrows very substantially over the college years.

Summing Up

So what do we know about intervention with minority children and the poor? Several very striking things. Perhaps the main lesson is that what works and what doesn’t is an empirical question.

Some early childhood programs that seem very reasonable do not have very big—or very lasting—effects. Head Start is a very sensible program that ought to make a big difference—and it does at first. But when students are left in poor families, in poor neighborhoods, and in poor schools, the Head Start gains in IQ fade and the academic achievement effects are not very large. But there are pre-kindergarten programs that have huge initial effects on IQ, along with significant residual effects on IQ if its graduates attend high-quality public schools. Even when children are not in good schools, the best programs nevertheless have very large academic achievement effects and enormous social benefits in terms of reduced crime and welfare dependence. Pre-kindergarten programs in general benefit poor kids and minority kids more than better-off white kids.

A similar pattern exists for elementary and junior high school interventions. I’ve described many programs that do not have much effect on student achievement and therefore are unlikely to be able to close the racial and SES gaps. The factors that do make a difference are the quality of the individual teacher, teacher incentives (probably, though much more work needs to be done to find out just what kinds of incentive programs work best and are most practical), and class size (which seems to have a bigger effect on black kids than on white kids).

Some teaching techniques discussed in Chapter 4 that are not terribly expensive are known to have fairly sizable effects and might help to reduce the achievement gap. These include computer programs to teach math and writing (which might actually be less expensive than standard instructional methods) and “cooperative learning” situations (which would add little if anything to cost). Even if these techniques were not found to reduce the gap, the fact that they improve the achievement of children in general indicates that they should be used for educating all kinds of students. A rising tide is welcome even if it lifts all boats equally.

We know that one math program—SEEDS—and one reading program—Reading Recovery—have big effects on minority kids. Reading Recovery is particularly inexpensive and provides a big bang for the buck.

A host of whole-school interventions have been tried, and the results are mostly disappointing. However, there is one hugely important exception—the KIPP program. Like Jaime Escalante, KIPP teachers can get poor minority pupils to perform at or above the level typical of white middle-class pupils. And that’s true even when they start the program late: surprisingly, the effects are big for students who start it in fifth grade. It remains to be seen what KIPP could accomplish if it were to start with much younger kids. Let’s hope we find out soon.

Paying for Gap Reduction

But can we afford the effective programs? A better question to ask is, can we afford not to have them? Many economists have evaluated the benefit-cost ratio of the most successful pre-kindergarten interventions. The Nobel Prize–winning economist James Heckman estimates the payback of the Perry Preschool Program—in terms of special education classes avoided, extra years of schooling avoided, crime and welfare costs avoided, and higher incomes for its participants—to be eight to one. This is equivalent to a return on investment of 17 percent per year. And that is just the cold monetary calculation. The gains to participants’ quality of life and those of their families and neighbors are not even included in those calculations. The initial cost of the Perry program is high—estimates range between $12,000 and $16,000 per student in 2007 dollars. It is much more expensive than Head Start (though the costs of future Perry-type programs should be lower because there would be no research component, which was a significant part of the Perry program costs) but it would be vastly more effective. The same is true of the Abecedarian intervention, which has been calculated as having a benefit-cost ratio of $3.78 to the dollar. Even when benefits are calculated strictly to the taxpayer, in terms of education, welfare, and criminal justice charges saved, the costs of the most successful pre-kindergarten programs are repaid in time.

So how much would it cost to enroll, say, the poorest third of children in a Perry or Abecedarian program from birth to the time of entry into kindergarten? There are about 7 million such children in the United States, and either the Perry or the Abecedarian program might cost as much as $15,000 per child per year. That works out to about $105 billion. But to soften the sticker shock, let me quickly point out that about $20 billion in public money is currently spent on pre-kindergarten programs. An unknown amount of privately spent money should also be subtracted. So should the additional earnings by the mother in the years during and after the program. The ultimate benefits to the child and to society should also be subtracted. Moreover, not nearly enough studies have been done on the most expensive programs—it is possible that very substantial improvements in cognitive and social functioning could be achieved by spending notably less than $15,000 per child. Finally, recall that gains in pre-kindergarten programs are proportional to need. The lower the mother’s IQ and SES, the greater the gains. So even if we provide intensive pre-kindergarten to only the neediest one-sixth or one-twelfth of the under-five population, the payoff would be very great.

As a yardstick for measuring the gross initial cost of early childhood education, note that the projected post-2001 tax cuts for the richest 1 percent of U.S. families will cost the treasury $94 billion in 2009 alone.

How about the apparently extremely effective KIPP programs? KIPP schools are actually not much more expensive than the regular public schools. (Some KIPP schools cost less than the public schools in their districts.) But KIPP gets its results by dint of the willing slave labor of its young, idealistic teachers who are paid only slightly more than public school teachers of comparable experience. And the KIPP teachers cannot keep up the pace for many years. Not surprisingly, unions have begun to oppose KIPP schools for their rate-busting.

How much would it cost to run KIPP schools if teachers were paid the same rate per hour that public school teachers earn—which seems not only fair but necessary if sufficient teachers are to be found? KIPP students get about 60 percent more contact time with their teachers. Most of the cost of education is in the form of physical plant, administrative and maintenance costs, and interest on debt, which are not increased by KIPP techniques. The cost per pupil in the United States for the average public school was about $8,000 in 2005, and about a third of this is for teachers’ pay. If we assume that teachers’ pay would be 60 percent more, and if we assume that a KIPP-type program would be made available to one-third of the 40 million children five to fourteen years old, it would cost an extra $35 billion. But again, these costs would be offset to a significant degree by child-care costs saved plus additional earnings by the mother. The economic gains over the lifetime for children in such schools cannot be calculated at this point. At a minimum they could be expected to defray a significant part of their extra cost.

To be clear, I am not advocating instituting particular programs at this time. A huge amount of research needs to be done to establish whether something like the Perry or Milwaukee or Abecedarian program would be effective and feasible if scaled up to national proportions, and the same thing is true of KIPP-type programs. In the case of KIPP, we would have to see how much gain could be expected for children of parents who did not exert effort to get their children into the program and keep them there.

We have existence proofs, however, that a marked reduction in the IQ and achievement gaps is possible, and we know that the costs of the effective interventions are at least conceivable. It would be irresponsible to fail to do the necessary research to find out what kinds of intensive programs do the most good.

Finally, if we want to make the poor smarter, a good way to do it might be to make them richer. The Scandinavian countries are much more egalitarian in their income distribution than the United States is, and the achievement gap between their richest and poorest children reflects that relative equality. Honest employment in a job having social value should pay enough to support a family. This could be achieved in part by increasing the minimum wage (which even with the new increases will be only 73 percent of what it was forty years ago), the Earned Income Tax Credit, and child tax credits.

The economic cost of at least some of this—and probably even more than the cost—would be recouped by increasing the productivity of the poor and reducing crime and welfare rates. We would likely do well by doing good.