Keep the Faucet Flowing

Summer Learning and Home Environment

The findings in the article that follows are based on the authors’ Baltimore School Study, which began in 1982 and is still in progress. Entwisle, Alexander, and Olson chose twenty schools on the basis of their racial composition (six were predominantly African American, six predominantly white, and eight were integrated) and their socioeconomic status (fourteen were inner-city working class and six were middle class) and then randomly selected 790 first graders from those schools. Although the authors went on to follow these students throughout their school years and beyond, the findings discussed here pertain to the children’s elementary school experiences. In addition to standardized test scores, the authors gathered data from interviews with the students and their parents, questionnaires completed by their teachers, and school records.


By Doris R. Entwisle, Karl L. Alexander, and Linda Steffel Olson

In seeking to explain why poor children do worse academically than children from middle-class and wealthy families, analysts have focused on two major topics: differences in schools and differences in home environments.

Because government has more of a handle on schooling than on home environment, public policy has emphasized the former. This has led to a widespread impression that poor children are routinely shortchanged by their schools. In fact, poor and middle-class children make comparable achievement gains during the school year. But while the middle-class children make gains during the summer when they are out of school, poor or disadvantaged children often lose ground academically. So far, one appealing remedy for reducing the achievement gap—summer school—has been disappointing. How to explain this paradox? If summers are the time when differences are established, why does summer school do so little to close the gap?

Seasonal learning

Given all that has been said about the strong correlation between parents’ resources and school performance, it is astonishing that resources of Baltimore parents—both financial and psychological—did not predict how much children learned in winters when school was open. Their resources mattered only for predicting their children’s gains in summer. Many studies1 besides the Baltimore study indicate that when schools are closed for summer vacation, the achievement scores of children from disadvantaged families either stay the same or slip back a little. To see how these seasonal patterns contribute to children’s achievement, we calculated the children’s gains on standardized tests in summer, when schools were closed, separately from school-year scores in winter, when schools were open.

The achievement levels of all children, regardless of their socioeconomic status,2 moved up substantially during the winter of first grade. Between the fall and spring of that first year, poor children in the Baltimore sample gained fifty-seven points in reading and forty-nine points in math, and their more affluent counterparts gained almost exactly the same number of points—sixty-one points in reading and forty-five points in math.

In the summer after first grade, however, more affluent students gained fifteen points in reading and nine points in math, while the less affluent children lost ground. For example, in the summer after first grade, they lost four points in reading and five points in math.

And this pattern continued. In the course of the first five summers in elementary school, the low-SES students gained less than one point total in reading, and they lost eight points in math. At the same time, the higher SES children gained forty-seven points in reading and twenty-five points in math. However, during the winters, when children were in school, both groups gained virtually the same amount (one hundred ninety-three points in reading for low-SES children versus one hundred ninety-one points for higher SES and one hundred eighty-six points for both in math). So the increasing gap in test scores between the two groups of children over the first five years in elementary school accrued entirely from the fact that relatively affluent children continued to gain when school was closed whereas poor children stopped gaining or even lost ground.

A faucet theory

We think a "faucet theory" makes sense of these seasonal patterns. That is, when school was in session, the resource faucet was turned on for all children, and all gained equally; when school was not in session, the school resource faucet was turned off. In summers, poor families could not make up for the resources the school had been providing, and so their children’s achievement reached a plateau or even fell back. Middle-class families could make up for the school’s resources to a considerable extent so their children’s growth continued, though at a slower pace.

This seasonal pattern is not obvious because most schools give standardized tests once a year, and spring-to-spring comparisons convey the distinct, but wrong, impression that middle-class children learn more over the entire year than poorer children. Thus, it looks as though home resources help year round—and as though schools are failing poor children. The seasonal scores, however, show that home resources matter mainly—or only—in summer.

The seasonal data from Baltimore are not a statistical fluke—they agree with seasonal data for children in Atlanta, New Haven, and several other localities. In addition, as the scores demonstrate, poorer children in Baltimore derive just as much benefit from school as their better-off classmates do when school is open in winter. And because they have lower scores when they begin school and gains are usually proportional to starting scores, the progress they make when school is open is quite extraordinary. If we could get poorer children up to speed before they start school, perhaps schools could do even more to close the achievement gap.

Explaining summer gains

What is it that better-off parents and neighborhoods do in summers that poorer parents and neighborhoods do not? To answer this question, we need to step back and consider parental attitudes toward school and learning. First of all, middle-class parents see themselves as partners in the learning process while blue-collar or poorer parents see education as the school’s job.3 Because middle-class parents take an active role, they know more about their children’s school programs than poorer parents do. They understand how schools work, what determines success, how to get along in a complex bureaucracy, and how present actions relate to future interests. Middle-class parents themselves have been successful in school (e.g., they are more highly educated) and in the workplace (e.g., they have higher income and job status), so they are in a position to encourage activities at home that will lead to success in school.

For many poor parents, schools are intimidating—the rules and conventions are foreign and the middle-class professionalism of school personnel, threatening. Poor parents tend to defer to school personnel, they advise their children to "follow the rules," and they rely on "professional authority" to decide what needs to be done for their children rather than deciding themselves.

Higher family income allows expenditures for books, games, computers, and other resources that could promote learning in summer, but more income is far from being the whole story. Parents’ financial capital overlaps their human capital, their social capital, and especially their psychological capital. In the Baltimore study, parents’ psychological capital, as measured by parents’ expectations for children’s school performance even before the children started school, was of about the same importance as family socioeconomic status in predicting cognitive growth.

These expectations continue to be of great importance when children enter school, and for poor families they may be unfairly undercut by the grades poor children get. Researchers do not gauge children’s progress in school by looking at their marks because marking standards vary so much from one school to another and from one teacher to another. Instead, they use standardized test scores. However, most families and children pay much more attention to marks than to test scores, and although many do not understand the significance of test scores, the youngest children and the poorest parents know a low mark when they see one. Further, children come home every day with marks on papers and homework, and they receive report cards several times a year. Parents who see low marks react by believing that their children are not learning very much, but they are often at a loss as to what to do. Children themselves are disappointed and also confused. All of this turns marks into a key dynamic in the link between poverty and learning deficits.

If poor children are progressing as well as their more affluent counterparts, why does this fact not show up in their marks? We found that children’s marks corresponded to the socioeconomic status of their neighborhoods. In schools where 30 percent or fewer children were in poverty, over one-third received a first reading mark of A or B, while in schools with more than 30 percent of children in poverty, only 5 percent received an A or a B. It strains credulity, but in one school where 88 percent of children received meal subsidies, all the students in our study failed reading in the first quarter of first grade.

In other words, poor children were not being marked in terms of how much they advanced during the school year but in terms of where they started—even though, judged by the gains they made on standardized tests, they improved as much as the youngsters from more affluent families. Tragically, the message sent home on report cards was thus that many of these children were already academic failures. This negative picture helps to shape poor parents’ reactions to their children, thus further eroding parents’ valuable psychic capital, which is essential to undergirding children’s long-term academic prospects.

The level of marks, generally, in a school also seemed to affect how teachers viewed their students. At the end of first grade, when asked to predict how their students would perform in grade two, teachers’ predictions shadowed the marking patterns of the school. Teachers in the top ten schools, judged by their economic status, expected their pupils to get more A’s and B’s in reading than C’s or below, while teachers in the bottom ten schools expected nearly all their students to get C’s or below.

And teachers’ ratings of children’s classroom behavior corresponded to these marks. In a school with only 11 percent of children on meal subsidy, teachers rated their pupils significantly higher in general interest and classroom participation than did teachers in a school where 90 percent of children were on subsidy. Actually, the correlation between the meal subsidy level of the school and teachers’ average ratings of their students’ class participation is almost perfect. In schools with high percentages of children on subsidy, some children were rated so low in terms of class participation that they were at the bottom of the scale. In the more affluent schools no student was rated at the bottom of the scale on these qualities.

The overall picture is one of poor children assigned poor marks, expected by teachers to get poor marks, and perceived as deficient in classroom behaviors known to foster learning. The great inequity is, of course, that during the school year, the poor children learned at a rate equivalent to that of the better-off children in the same school system. Parents and the community as a whole accepted the evaluations with which they were furnished without being aware that the level of these evaluations reflected children’s home addresses rather than children’s progress in reading or math on standardized tests.

The school’s role

School systems and society in general are misinformed about the origins of social difference in children’s school performance. The strong impression that home resources help all year long is mistaken. Instead, family resources make a difference mainly when school is closed. One implication is that schools are doing a far better job than they have been credited with. Another is that middle-class parents’ aspirations, attitudes, activities, and psychic investments in their children are major reasons for the social class differences in children’s cognitive growth when schools are not open.

However, at present, misperceptions about the process of schooling needlessly depress poorer parents’ psychological resources. In addition, this mismatch between children’s actual progress and how that progress is viewed is highly inequitable. When Baltimore children started school, their pre-reading and pre-math skills reflected their uneven family situations, but despite this, children in our study, regardless of socioeconomic level, progressed at the same rate over first grade. In June of first grade, though, the unevenness in test scores present at the start was still there. Poor children started from a lower point than better-off children did, so when school let out for the summer they ended up at a lower point even though both groups made equal gains during the year. In addition, in summer the poor children’s growth just about stopped while better-off children’s continued to rise.

Summer school as a solution

If economically disadvantaged children fall behind their better-off classmates in summer, it seems obvious that attending summer school could, or should, bring poorer children up to speed. Sad to say, this course of action so far does not work. The few careful evaluations that have focused on attending summer school for the purpose of closing the gap between social groups (racial, economic, or both) find just the opposite. The evidence on summer school for this purpose is clearly negative. Summer school increases the gap.

On average, the summer school gain for students of all socioeconomic levels is quite small: about one-seventh of a standard deviation—roughly one month on average or a few test points (out of three or four hundred) on standardized tests like those used in the Baltimore study. This small gain is for rich and poor combined, so the first question is whether disadvantaged students attending summer school make any gains. The literature suggests they do not.

However, the failure of summer school to narrow the learning gap is not really surprising. Many other programs undertaken in the past have also had disappointing results. A major aim of "Sesame Street," for example, was to reduce the knowledge gap between minority (or poor) preschool children and their majority (or better-off) counterparts. But though, on average, it was clearly of benefit to preschoolers, it backfired in terms of decreasing the learning gap, which grew larger rather than smaller.4

The counter-intuitive outcome of summer school, like that of "Sesame Street," is an example of the "Matthew Effect," an often-observed phenomenon that takes its name from the gospel of Matthew 25:29—and that can be roughly paraphrased by the old saw, "the rich get richer and the poor get poorer." Providing add-on services across the board benefits advantaged students more than poor, bright rather than not-so-bright, majority more than minority, and so on. Why? One reason is that higher scoring and better-off children’s parents find out about special programs and see that their children attend them more often than do poorer children’s parents. In Atlanta, where summer school was voluntary, the Atlanta children who chose to attend summer school had higher achievement levels during the school year than those who chose not to attend.5 Similarly, the children least likely to attend preschool programs these days are children from low-income and single-parent families and those whose parents have the least education. For summer school and other programs to close the learning gap, they will have to be designed especially for poor children and provided specifically for them, as has been done with some compensatory education preschool programs.

Proposing a strategy

To determine the content of a summer program that will boost the summer achievement of poorer children, we need to know the kinds of learning experiences they lack and when to offer such programs. An important consideration for timing of programs, which we have not mentioned so far, is that children’s cognitive growth slows down precipitously as they progress upward through the grades. The reading achievement of Baltimore children improved twice as fast in grade one as grade three, for example, and the gap in summer gains between better-off and poorer children shrank over time.

Considering (1) the relatively small gap in children’s test scores associated with family income at the point when they start first grade, (2) the seasonal profiles of achievement growth with better-off children gaining more in summers, and (3) the marked deceleration in the rates of cognitive growth over the early school years, we suggest the following course of action: Provide poor children with high-quality preschools, and then follow up with summer school just for poor children in the summers before and after first grade.

Where to start? A good place would be to provide programs to help bring poorer children up to speed before they start first grade. We already know this can be done, because scientifically impeccable data show that good preschools can improve the early school success of disadvantaged children. A major effect is to reduce the retention rate in first grade,6 which is higher than in any subsequent grade,7 and holding poor children back in first grade mortgages their futures. By age 23, Baltimore students who had been held back in grade one were three times more likely to have dropped out of school, even when family economic status, minority status, and actual school performance were taken into account. Attending a good preschool could be enough to protect economically disadvantaged youngsters against low placements in first-grade reading groups or early retention.8 However, such programs must be specifically targeted at those children. If preschools reach more of the wealthier than the poorer children, or if wealthier children find their way to higher-quality programs, the gap will get bigger rather than smaller.

More disadvantaged children also need to attend kindergarten, which is a kind of preschool. Because kindergarten is not compulsory, a surprisingly large number of children still attend half-day programs, and back when our study children were in the primary grades, many skipped kindergarten altogether. In Baltimore City, for example, which is one of the poorest school districts in Maryland, 10 percent of first-graders in our study had not attended kindergarten, compared to about 1 percent nationwide today. In addition, the Baltimore study children who came from the poorest families were more often enrolled in half-day than full-day sessions. (Of children who attended half-day, 77 percent were on meal subsidy compared to 32 percent of those who attended full days.)

The benefits of full-day as compared to half-day kindergartens for the Baltimore children were striking. With family background and many other variables allowed for, first graders who attended full-day kindergarten were absent fewer days in first grade, were less often retained, and earned higher marks and test scores than the half-day attendees. So in addition to preschools (age four and younger), having poorer children attend high-quality full-day kindergarten (age five) could help close the gap.

Summer school programs

The next logical step after increasing poor children’s attendance in full-day preschool is to develop summer school programs for poor children that add on to preschool. Preschools can reduce the achievement gap when children start first grade, but then we need to keep the faucet open during the summer to give poor children the extra resources that middle-class parents provide for their children.

What should these summer programs consist of? Summer activities related to reading top the list. Low-income children involved in Atlanta’s summer schools tended to read more on their own than did students not attending.9 Likewise, in Baltimore, first- and second-graders who went to the library more often in summer and who took out more books did better than other children. Both math and reading growth benefited from library activities.

Better-off children also did things in summer different from what they did during the school year—they attended day camps, took swimming lessons, went on trips, visited local parks and zoos, and played organized sports, to name a few. These activities provided children with experiences unlike their experiences in school. Probably summer programs for disadvantaged children should feature activities that include a substantial amount of physical activity for both boys and girls, especially games like soccer, field hockey, or softball that require very little equipment but have complicated rule systems and require children to take multiple roles. Adult leaders need to be cast in the role of "coach" rather than teacher.10

Program content is not the only concern, however. Higher-income parents have psychological capital of a kind that summer school coaches could emulate: using positive rather than negative reinforcement, teaching productive problem-solving strategies, encouraging children to be self-directed, having high expectations, and seeing that the means are there for children to meet high expectations. Perhaps most important, coaches need to encourage children to enjoy themselves: Engagement is key to learning, and engagement can be difficult to achieve if summer programs are perceived as punitive.

The logistics of summer programs need careful planning, especially in terms of teachers who can establish strong attachments to students and parents. The programs need to be located near pupils’ homes, so children can get to them easily and so parents can become involved. Changing the summer environment of children in low-income families may require community intervention.

No single approach is likely to close the academic gap between low- and high-income children, but summer programs bracketing first grade could help. It is absolutely essential to be aware that special programs, including summer school, given to children of all income levels would probably enlarge the gap between rich and poor. High-quality preschools and kindergartens can definitely improve the school performance of low-income children. But as they go through the first three grades, these children—especially the most disadvantaged—need additional resources to stay even. Programs mounted in summers before the first and second grades that emphasize voluntary activities—recreational reading, organized sports, and a variety of summer activities that middle-class families often pursue—hold promise. The programs should not be scheduled as "make-up" or billed as being for children who have "failed." The success of these programs, we believe, hinges on their non-school flavor and on providing them specifically for disadvantaged children.

At age six, when children’s cognitive development is proceeding at probably twice the rate it does two or three years later, the trajectory of children’s long-term educational careers is being established. For this reason, it is imperative to concentrate on the pre- and primary schooling of disadvantaged youth.

The larger picture

People who think and talk about inequality often ask why it is perpetuated and how we can get those at the very bottom of the ladder to move up. When social theorists and policymakers propose schools as a solution, they often seem to go along with the notion that education is a sorting device. Thus, students leaving school are channeled into job slots that correspond with how long they have stayed in school and how successful they have been while there. Those who win out in school will win out in the labor market as well.

To us this image of sorting students at the end of their schooling misrepresents the nature of inequality. Families sort themselves by income into neighborhoods. Then schools, which reflect the social strata of their neighborhoods, tend to eliminate any real contest between students from different income levels. Because the unequal distribution of resources across families is the engine that drives the system, tinkering with schools has not, thus far, eliminated this inequity. And, judging from the available research, we believe inequity would be exacerbated by noncompulsory summer schools open to all children.

The good news is that despite poverty and family disruption, young children’s ability to learn during the school year seems little impaired by scarce family resources. In seeking to address the achievement gap between rich and poor, we should begin by recognizing the efficacy of elementary schools in leveling the playing field. Most press coverage of American education today emphasizes the system’s failures, especially its failures vis-a-vis the most disadvantaged students. These negative perceptions often undercut popular support for elementary schools and public education in general, and they miss the extent to which schools make up for the deficits in poor children’s backgrounds. The real tragedy of current educational practice is that schools are organized—and children perceived—as though the more-advantaged groups are better able to benefit from the schooling process. Poor children are assumed to be "slower" learners—less capable of absorbing the curriculum—and these lower expectations color poor parents’ own perceptions about their children’s academic futures. This is especially unfortunate in the early grades when students’ achievement trajectories are being set and their cognitive growth is most rapid.

This recognition of the power of schools to make a difference in the lives of poor students needs to be coupled with efforts to involve parents and communities in the schooling process so that all parents, not just middle-class parents, are active collaborators in the education of their children. Preschool and summer programs, properly organized, can help develop economically disadvantaged parents and their neighborhoods into active supporters of children’s academic endeavors. These parents need to know, for example, that such simple activities as reading aloud to their children can have big academic pay-offs. Neighborhoods need playgrounds and coaches to encourage organized sports and craft activities in summer. Workshops and other outreach efforts could help disadvantaged adults develop some of the psychological and social capital that is so important to undergirding their children’s learning.

Doris R. Entwisle is emeritus professor, Karl L. Alexander is professor, and Linda Steffel Olson is senior research assistant in the Department of Sociology, Johns Hopkins University, Baltimore. This article is adapted from "Summer Learning and Home Environment" in A Notion At Risk, edited by Richard D. Kahlenberg, and it appears with the permission of The Century Foundation Press. A full account of the study and what Entwisle, Alexander, and Olson found appears in Children, Schools, and Inequality (Boulder, Colo: Westview Press, 1997).


1. As examples, see Barbara Heyns (1978), Summer Learning and the Effects of Schooling (New York: Academic Press), Richard Murnane (1975), The Impact of School Resources on the Learning of Inner-City Children, (Cambridge, Mass.: Ballinger). Much of this literature is reviewed in Harris Cooper, Barbara Nye, Kelly Charlton, James Lindsay, and Scott Greathouse (Fall 1996), "The Effects of Summer Vacation on Achievement Test Scores: A Narrative and Meta-Analytic Review," Review of Educational Research 66: 227-269.

2. Family socioeconomic status is indexed by information on student participation in the school meal subsidy program, a rough not-low/low-income measure, mother’s and father’s educational levels, and occupational status.

3. Annette Lareau (1987), "Social Class Differences in Family-School Relationships: The Importance of Cultural Capital," Sociology of Education 60: 73-85.

4. See Thomas D. Cook, Hilary Appleton, Ross F. Conner, Ann Shaffer, Gary Tamkin, and Stephen J. Weber (1975), "Sesame Street" Revisited (New York: Russell Sage), especially Chapter 1. For another example, see Launor F. Carter (1983), "Sustaining Effects Study" (prepared for the Systems Development Corporation). Students who entered Title I programs at near-average achievement levels profited most, whereas those entering at a low level profited only little or not at all. More recently in 1997, the New York State Legislature enacted a universal kindergarten for four-year-olds, but there has been underenrollment in the communities with limited access to quality preschools. See Foundation for Child Development, March 1999 Update (New York: Foundation for Child Development).

5. See Heyns, Summer Learning, p. 128.

6. See Irving Lazar and Richard Darlington (1982), "Lasting Effects of Early Education: A Report from the Consortium for Longitudinal Studies," Monographs of the Society for Research in Child Development 47, No. 2-3: ix-139; Consortium for Longitudinal Studies (1983), As the Twig Is Bent: Lasting Effect of Preschool Programs (Hilldale, N.J.: Erlbaum). For good general review, see W. Steven Barnett (Winter 1995), "Long-term Effects of Early Childhood Programs on Cognitive and School Outcomes," The Future of Children, Long-term Outcomes of Early Childhood Programs, 5 (3), 25-50.

7. Karl L. Alexander, Doris R. Entwisle, and Susan L. Dauber (1994), On the Success of Failure: A Reassessment of the Effects of Retention in the Primary Grades (Cambridge, Mass.: Cambridge University).

8. See Doris R. Entwisle (Winter 1995), "The Role of Schools in Sustaining Benefits of Early Childhood Programs," The Future of Children 5, No. 3: 133-44.

9. Heyns, Summer Learning, p. 191.

10. For a discussion of the link between organized sports and academic progress, see Doris R. Entwisle, Karl L. Alexander, and Linda Steffel Olson (1994), "The Gender Gap in Math: Its Possible Origins in Neighborhood Effects," American Sociological Review 59: 822-38.

American Educator, Fall 2001