Abstract
Much research has tried to parse the school’s contribution to children’s learning apart from the family’s and the family’s contribution apart from the school’s as though they were discrete and separable. The 1966 Equality of Educational Opportunity report helped launch this agenda, finding in favor of family. In this essay I argue that the framing of the issue as “family versus school” is fundamentally flawed. Rather, family and school (and neighborhood) together shape children’s academic development. I argue that the strong effect associated with school socioeconomic composition in the original report, and stronger still in more recent studies, is in fact an expression of family influence: family determines where children live and the schools they attend. But it is a school influence as well. When properties of family, neighborhood, and school overlap, as they do under conditions of extreme neighborhood and school segregation, poor children’s profile has them triply disadvantaged. The same ecological perspective on children’s learning implies that by reducing the degree of overlap across these “overlapping spheres of influence,” school socioeconomic context can function instead to offset family disadvantage. Relevant literatures are reviewed and the concluding section considers the potential of socioeconomic integration at the school level as a policy lever for improving poor children’s educational prospects.
What are the social forces that govern children’s academic development? This question arguably has been, and remains, the core problematic for the sociology of education as a field of inquiry, with the achievement gap across social lines a particular focus. In pursuit of answers, studies typically attempt to parse the school’s contribution to children’s learning apart from the family’s and the family’s contribution apart from the school’s. Which of the two has the greater weight? James S. Coleman’s Equality of Educational Opportunity report (Coleman et al. 1966; hereafter Coleman Report), which helped launch this agenda, found decidedly in favor of family.
In this essay, I argue that framing the issue as “family versus school” asks the wrong question. Rather, family and school, along with neighborhood, together shape children’s academic development, and in ways that may not always be separable. When we think of family resources in support of children’s schooling, it is natural to look to the interior of family life—for example, the family’s material well-being, the structural integrity of the family, and the parents’ engagement with their children’s learning and their children’s schools. But the family’s reach extends beyond the confines of the household. Where children live and the schools they attend also are parental decisions, and owing to the deeply entrenched residential and school segregation of life in the United States today, the imperatives of family, neighborhood, and school tend to be mutually reinforcing—privileging those already privileged and disadvantaging those already disadvantaged.
THE COLEMAN REPORT AS BACKDROP
Daniel Patrick Moynihan’s New York Times obituary (1995) for his good friend James Coleman began by recounting an incident at the Harvard Faculty Club on the occasion of the 1966 release of the Coleman Report. Moynihan tells us that he was approached by Seymour Martin Lipset, another eminent member of the faculty, who excitedly pronounced: “You know what Coleman is finding don’t you? … All family.” A few years later, Godfrey Hodgson (1975) introduced his expansive essay on the Coleman Report with the same story: “Hello Pat,” Lipset began, “guess what Coleman found? … schools make no difference, families make the difference.” Though shaded differently, both renderings convey the large takeaway point at the time of the report’s rollout: in the tug of war between family and school in shaping children’s academic development, family wins. And it is a decisive victory.
The conclusion that “schools make no difference” was a disheartening revelation for those who believed that poor and minority children suffer under the weight of woefully deficient schools. The Coleman Report was expected to provide scientific justification for school improvement as the remedy for generations of racial injustice. Instead, it implicated the private sphere of family life, seeming to leave little room for school reform as a solution.
Some who were distressed by the report’s message dismissed it on technical grounds. Others took what they could from it—for example, the “news” that school segregation remained widespread throughout the United States a decade after the Supreme Court struck down the South’s dual school system, that differences in children’s educational experiences within individual schools counted for more than did average differences across schools (the latter being the report’s perspective), and that teacher quality seemed to matter some, as did the socioeconomic makeup of a school’s enrollment, such that poor and minority children perform better academically in schools with a diverse student body. But against the weight of family advantage and disadvantage, these were mere crumbs, and the decades that followed offered little relief from the deep malaise that set in:
In 1969, Arthur Jensen began his famous paper in the Harvard Educational Review (Jensen 1969, 2) with the assertion that “Compensatory Education has been tried and apparently has failed,” following with: “Why has there been such uniform failure of compensatory education programs wherever they have been tried?” (3).
In 1972, Frederick Mosteller, a Harvard statistician, and Daniel Patrick Moynihan published the results of a faculty seminar in which distinguished academics from several disciplines revisited the Coleman Report data, analytic procedures, and conclusions. The result? According to the book’s dust jacket: “This study turned understanding of a major area of social policy upside down, as had no comparable event in the history of social science” (Mosteller and Moynihan 1972).
In 1989, Robert Slavin’s literature review on the educational effectiveness of small classes concluded that there was little benefit short of one-on-one tutoring.
Also in 1989, and then updated a decade later, Eric Hanushek’s (1989, 1997) review of the evidence on school funding concluded that “variations in school expenditures are not systematically related to variations in student performance.”
In light of such pronouncements, Barbara Heyns (1978, 186) was moved to elevate education research above economics as the “dismal science,” a legacy of the Coleman Report that remains with us still. According to Debra Viadero (2006, 23), writing on the occasion of the fortieth anniversary of the report: “What most people took away from the report … was the notion that ‘schools don’t matter.’” She then followed with a telling quote from David Armor, one of the Harvard seminar participants: “No one has found a way, on a large-scale basis, to overcome the influence of family.”
Around the same time, Adam Gamoran and Daniel Long (2007, 23) credited the report with “the seminal finding in U.S. sociology of education.”1 Such acclaim after some forty years is quite remarkable, and not much has changed over the ensuing decade. The Coleman Report and the idea that “schools make no difference” continue to be invoked as authoritative, despite the following four facts that bear on the family-school tug-of-war:
The Coleman Report did not conclude that “schools make no difference.”
Studies since, using more rigorous methods applied to both the same data and to new data, buttress—indeed strengthen—the report’s actual conclusions.
Conclusions from the other studies cited as reinforcing the view that schools make no difference all have been refuted by scholars of comparable standing.
Pitting family against school as a contest between “this” and “that” is a flawed framing of the social forces that drive student learning and has led to much wrongheaded thinking.
I address these several points in the sections that follow, starting with the fourth, as it is fundamental.
THE FRAMING: IS IT “FAMILY OR SCHOOL?” OR “FAMILY AND SCHOOL?”
Richard Jessor’s 1993 American Psychologist essay sketches an ecological perspective on children’s development that is at once familiar and profoundly insightful. It follows in the social ecological tradition of Uri Bronfenbrenner (1979), but absent the obscure language. For Jessor, children’s development is governed by experiences in the three institutional settings they encounter daily, up close and personal: family, neighborhood, and school. The resources available to children in these three settings, and how they are deployed, combine to channel youth along different developmental paths, ones that often overlap lines of race, gender, and family background.
At a farther remove, but still relevant, are the broader sociostructural and sociocultural contexts within which family, neighborhood, and school themselves are embedded. In my research (for example, Alexander, Entwisle, and Olson 2014), the background context is deindustrialized Baltimore over the last two decades of the twentieth century into the first decade of the twenty-first. The Annie Casey Foundation (2010, 2) has characterized this span of years as a time ofcrippling trends and tragic events—the dramatic loss of manufacturing jobs and tax base, the ruinous riots of 1967 and 1968; the exodus of first white then African-American, middle class families; the sequential epidemics of heroin, crack cocaine, and HIV; the intensified crime and gang activity that fed and feasted off the drug trade; and the activities of slumlords, property flippers and predatory lenders. The end result has been an ever-deepening cycle of disinvestment and decline.
It is fundamental to the life-course perspective on human development that the life paths children follow and the characteristic differences across social lines in those life paths are historically situated. It is self-evident that the conditions just described pose challenges for the healthy development of many of Baltimore’s children. Those challenges play out at the interior of family life, in Baltimore’s high-poverty neighborhoods, and in Baltimore’s public schools, where enrollment systemwide in 2014 was 84 percent low-income and 83 percent African American.2
There is nothing exceptional in Jessor’s rendering to this point. That comes with his insight that family, neighborhood, and school are overlapping spheres of influence in children’s development. That is to say, children’s experience of family life, their neighborhood exposure, and the character of the schools they attend are not isolated silos. Rather, they are experienced holistically, as a mosaic. Children who grow up in poor families, live in high-poverty neighborhoods, and attend schools with high-poverty enrollments are triply disadvantaged. This profile, sadly, is all too common, and children who are burdened by it find themselves disadvantaged at every turn. It does not auger well.
Now to my point: poor children’s experiences across these settings are an expression of family disadvantage. In deciding where to live, parents, poor and nonpoor alike, determine the character of their children’s neighborhood exposure and of the schools they attend. Family, neighborhood, and school as overlapping spheres of influence in children’s development characterize both family advantage (a middle-class experience profile) and family disadvantage (a low-income experience profile).
To speak of school influence as something apart from family influence is a false divide. We might try to force their separation heuristically for analytical purposes, but the counterfactual reality so imagined is just that: imaginary. In children’s lived lives, school influence is an expression of family influence. That realization is critical to a proper understanding of the Coleman Report’s results.
WHAT THE COLEMAN REPORT REALLY CONCLUDED
To see clearly that school influence is an expression of family influence requires going beyond thumbnail accounts, most of which misconstrue the report’s import. Here, then, is the original source material (Coleman et al. 1966, 258): Of the many implications of the study of school effects on achievement, one appears to be of overriding importance.
This is the implication that follows from the following results taken together:
The great importance of family background for achievement;
The fact that the relation of family background to achievement does not appear to diminish over the years of school;
The relatively small amount of school-to-school variation that is not accounted for by differences in family background …
The small amount of variation in achievement explicitly accounted for by variations in facilities and curriculum;
Given the fact that no school factors account for much variation in achievement, teacher’s characteristics account for more than any other …
The fact that the social composition of the student body is more highly related to achievement, independently of the student’s own social background, than is any school factor …
Taking all these results together, one implication stands out above all: The schools bring little influence to bear on a child’s achievement that is independent of his background and general social context.
The report continues with specific reference to home, neighborhood, and peer environments: “For equality of educational opportunity through the schools must imply a strong effect of schools that is independent of the child’s immediate social environment; and that strong independent effect is not present in American schools.”
Points 1, 3 and 6 are key, as is the language of “independent effects.” Family background is of overriding importance—no one would dispute that.3 But the fact that school-to-school variation in school achievement closely tracks family background precisely exemplifies Jessor’s “overlapping spheres of influence” thesis.
In fact, school influence on children’s achievement is so deeply embedded in children’s family life that they hardly are separable.4 These tight linkages across institutional contexts additionally imply that the social organization of schooling, as constituted back then and still today, functions mainly to maintain or reproduce children’s place in the social order. To illustrate, in 2005 nationally, poor students were in the majority in 84 percent of schools with minority enrollments of 90 to 100 percent; in schools with minority enrollments of 10 percent or less, just 18 percent of schools had majority low-poverty enrollments (Orfield and Lee 2007).
The insight that the social composition of the student body is the strongest school-based correlate of student achievement, independent of the child’s family background, pinpoints the particular mechanism that channels family influence through the school: neighborhood residential segregation.
Owing to such segregation, most schools enroll mainly children of like background. High-poverty neighborhoods and high-poverty schools are population aggregates. Their properties do not inhere in any single family, and they have consequences beyond those located at the interior of family life. In that sense, they are independently consequential. But their force derives from the residential decisions of individual families, which are historically grounded and contemporaneously maintained. Change the demographics of neighborhood and school and the entire formula potentially changes: with less overlap across spheres, the evidence suggests that the social composition of the school would reverse function and serve to weaken the link between family background and school achievement. But such a circumstance today is the exception, not the rule.
As the national commitment to school desegregation has waned, segregation at the school level has increased: “Nearly 40 years after the assassination of Dr. Martin Luther King, Jr., we have now lost almost all the progress made in the decades after his death in desegregating our schools” (Orfield and Lee 2007, 11). That reversal has been most dramatic in the South, which for a brief period could claim the lowest levels of racial segregation in the country: in 1988 the percentage of black students attending majority-white schools in the South stood at 43.5 percent (up from essentially zero in 1954 and 2.3 percent at the passage of the 1964 Civil Rights Act); by 2005 it had dropped to 27 percent, just a bit above the figure for 1968 (Orfield and Lee 2007, 23).5
In an earlier era, the force of law, commercial practices, and, not infrequently, violence maintained segregated schooling and housing and kept neighborhoods separate along lines of race, ethnicity, and income. Today there is formal commitment to “equality of opportunity,” but the largely informal practices that maintain such separation are no less powerful than in the past (Bonilla-Silva 1996).
Neighborhood segregation begets school segregation. Consider the parallels in Baltimore, as reported by John Logan (2002): during the 1989–1990 school year, neighborhood and school segregation indices were 79.1 and 76.3, respectively; a decade later, at 74.3 and 79.0, they had barely changed.6 That the paired figures are similar and high is hardly coincidental.
This degree of alignment reveals why attempting to separate the influence of family from the influence of school as though they are discrete entities serves little purpose. It is not that such a separation cannot be done—the Coleman Report achieved an approximation of it using the tools then available, and an even better job of it can be done today. The problem, rather, is conceptual: focusing on the independent contributions of family and school misses their overlapping contributions. It is not a matter of family or school, but of family and school.
AMPLIFYING THE “KEY FINDING”
In this section, I review studies from the contemporary literature that add weight to the Coleman Report’s conclusions regarding poor and minority children’s learning in relation to their school’s enrollment mix. These studies instruct us that the report very likely understates the importance of school context, an insight that is startlingly important. I also review studies that compare school-year learning with summer learning, as these comparisons afford a rather different perspective on the contributions of family and school.
Borman and Dowling (2010)
Geoffrey Borman and Maritza Dowling (2010) reanalyze the original Coleman Report data for ninth-graders, acquired through the University of Michigan’s Inter-University Consortium for Political and Social Research, a secondary analysis data archive. Some of the original ninth-grade data were missing, but otherwise Borman and Dowling use the same data, predictor set, and variable operationalizations as in the original Coleman Report. Their dependent variable is a test of verbal achievement, a short vocabulary test administered in the original project. The Coleman Report centered on the same test, which it characterized as a test of “verbal ability.”
With these parallels to the Coleman analysis, Borman and Dowling’s study also suffers the original project’s limitations, a large one being its cross-sectional research design. The project did not monitor children’s learning over time; rather, it assessed achievement levels at a single point in time and how those covaried in relation to school and family characteristics. Static comparisons across grade levels for different students might look like learning trajectories, but they are not. Still, if the original analysis is accorded credibility, as clearly it has been, then so too should Borman and Dowling’s.
In addition to these many parallels, there is one large difference that sets the Borman-Dowling study apart from the original: its mode of analysis. The original report relied on simple correlational and regression analyses. Borman and Dowling instead use an approach that adjusts for the nested data structure that is characteristic of classroom- and school-based research. The methods used in the Coleman Report assume independence of observations, meaning, in this instance, that what happens to one child has nothing to do with what happens to another child. But children who attend the same school or are enrolled in the same classroom share many experiences, including some that no doubt have bearing on their learning.
Owing to this shared experience, the forces that impinge on children’s learning vary less across persons than they would if, for example, each child attended a different school. As a result, there is a strain toward homogeneity in the data collected from them and about them. This nesting can be multilayered—for example, children within classrooms, within schools, within school districts, or within states—and to ignore that possibility risks getting biased results. The distortion can be large or small, but since the extent of bias usually is unknown, it looms large as a concern.
Borman and Dowling use hierarchical linear modeling (HLM) to adjust for the fact that observations are nested and not independent, and this does indeed affect the results. We know this because they also report results using the report’s original mode of analysis. Here is what they conclude:Going to a high-poverty school or a highly segregated African American school has a profound effect on a student’s achievement outcomes, above and beyond the effect of his or her individual poverty or minority status. Specifically, both the racial/ethnic and social class composition of a student’s school are more than 1¾ times more important than a student’s individual race/ethnicity or social class for understanding educational outcomes. In dramatic contrast to previous analyses of the Coleman data, these findings reveal that school contexts dwarf the effects of family background. (Borman and Dowling 2010, 1239)
All family? Schools don’t matter? That hardly seems the case.
Rumberger and Palardy (2005)
Russell Rumberger and Gregory Palardy (2005) pose similar questions using a different data source: the National Educational Longitudinal Survey of 1988, which affords national coverage. This project began in 1988 with a large, representative sample of eighth-grade students and their schools and tracked their educational progress through twelfth grade. Because it monitored the learning of the same students over time, Rumberger and Palardy are able to evaluate changes in test scores in relation to characteristics of children’s families and schools and in relation to experiences at school. They analyze learning in several achievement domains: math, science, reading, social science, and a composite of all four via HLM, the same mode of analysis used by Borman and Dowling.
Their results are complicated and nuanced, with comparisons across testing domains, different kinds of students (classified by race-ethnicity and family socioeconomic status), and different kinds of schools (for example, low-, mid-, and high-SES). Still, the results of most immediate interest are straightforward:The results of this study confirm a widely held belief of many parents: that whom you go to school with matters. But what appears to matter most is the socioeconomic, not the racial composition of schools. … While students’ own social backgrounds were related to their achievement, so too were the average social class backgrounds of all the students in their school. In fact, the effects of school SES were almost as large, and sometimes much larger, than the effects of student SES. (Rumberger and Palardy 2005, 2020)
Rumberger and Palardy continue by noting that their results confirm the original conclusions of the Coleman Report, although with much closer balance between the influence of school and the influence of family.
Here too, then, the school’s enrollment mix—its socioeconomic makeup in particular—emerges as highly consequential for children’s learning. For poor and minority children, attending a school with a diverse student body boosts achievement and attending a school with mainly others of like background depresses achievement; thus, when poor children live in high-poverty neighborhoods and attend high-poverty schools, their learning suffers. The original Coleman Report suggested this; these studies strengthen the point.
Schwartz (2010)
Heather Schwartz (2010) focuses on a single community, Montgomery County, Maryland, comparing the school performance of poor children who live in predominantly middle-class neighborhoods against other poor children who live in less affluent communities. This uncommon circumstance is afforded by the county’s inclusionary housing policy, which requires real estate developers to set aside a portion of homes to be rented or sold at below market rates. There were 12,000 such units in wealthy Montgomery County at the time of Schwartz’s project, one-third of which had been acquired for public housing.
Public housing applicants were randomly selected for placement. Schwartz (2010, 5) examines the experience of 850 children so placed who, as she puts it, “attended elementary schools and lived in neighborhoods that fell along a spectrum of very-low-poverty to moderate-poverty rates.” With 40 percent to 85 percent the range of low-income enrollments, Montgomery County has few schools with extreme levels of concentrated poverty. This contrasts sharply with the experience in Baltimore, where most black children attend hypersegregated schools with African American enrollments of 90 percent or more. Still, these differences in Montgomery County are large, and potentially meaningful. Additionally, most of the higher-poverty schools there received supplemental resources not provided to schools with low-poverty enrollments: funding for full-day as opposed to half-day kindergarten, staffing to reduce class size (from twenty-five to seventeen), extra staff professional development, and an enhanced literacy curriculum. Schwartz takes advantage of these resource differences to compare the benefits of income integration to resource enhancement in higher-poverty settings, two very different approaches to addressing needy children’s educational challenges.
Assigning families to neighborhoods and children to schools by a random draw effectively eliminates any self-selection of more highly motivated parents into stronger schools. At the time of the Coleman Report, with near-universal segregation by race, one has to wonder whether the small number of black children in schools with diverse enrollments would have done equally as well if they had attended the high-poverty, segregated schools typically attended by other black children. If family is all that matters, then perhaps so, and with residential and school segregation still high, the same uncertainty prevails today. But not in Montgomery County. Taking parental choice out of the picture shows that school differences are likely to signify school effects. And what are those schools effects? According to Schwartz:School-based economic integration effects accrued over time. After five to seven years, students in public housing who were randomly assigned to low-poverty elementary schools significantly outperformed their peers in public housing who attended moderate-poverty schools in both math and reading. Further, by the end of elementary school, the initial large achievement gap between children in public housing who attended the district’s most advantaged schools and their non-poor students in the district was cut in half for math and one-third for reading. (Schwartz 2010, 6)
Imagine that—improved performance by poor students compared to other poor students attending resource-enhanced schools and a greatly diminished achievement gap compared to nonpoor students. These gains took time to materialize, with economic integration in children’s neighborhoods and schools both implicated (the latter in larger measure). Schwartz adds that these impressive results are probably a lower bound to the benefits that poor children could realize by attending low-poverty schools: she notes that fewer than 1 percent of Montgomery County’s elementary schools have high-poverty enrollments, compared to 40 percent of urban elementary schools nationwide.
These results also accord with Coleman’s conclusions. Indeed, to this point, we have encountered no contradictions to the Coleman Report, although the school context effects documented in these more recent studies are vastly larger than those in the report. These studies thus add force to the report’s conclusions, help establish their contemporary relevance, and suggest, in a way that the report analysis could not, a path to school improvement for poor children: reduce the degree of overlap in the “overlapping spheres of influence” that impedes their academic development.
THE SEASONALITY OF LEARNING
Interest in summer learning as distinct from school-year learning extends back to the first decade of the twentieth century (see Cooper et al. 1996). The modern era of research on the topic was launched by Barbara Heyns in her remarkable book Summer Learning and the Effects of Schooling (1978). Her great insight was that comparing children’s summer learning to their school-year learning affords leverage for disentangling the influence of school from the influence of family.
It is clear that the overlap of school characteristics with family characteristics poses challenges for poor children, but that same overlap also poses challenges for researchers. Observational (that is, non-experimental) data oblige the separation of family influence from school influence by statistical means, yet the social forces combined by nature do not yield easily to artificial devices. When the overlap is severe the separation is uncertain, and confounds such as parental self-selection into schools with diverse enrollments cloud interpretation. Random assignment, as in Montgomery County, is uncommon, but the school-year calendar affords the same kind of contrast in a natural experiment that is near-universal.
Children are in their families and neighborhoods throughout the year but are in school intermittently. The long summer break, typically three months in the United States, isolates the school’s contribution to children’s learning: if achievement gains track differently over the summer months, when children are not in school, than during the school year, then time out from school must be implicated. This also eliminates the parental self-selection confound, as the school calendar applies to everyone.7
Heyns analyzes achievement gains by family income level and race-ethnicity (African American or white) for a large sample (nearly 3,000) of Atlanta middle school students over an eighteen-month period bracketing two school years and the summer between. Her findings are both expected and unexpected. Among the former is that children, regardless of their background, learn more and learn more efficiently when they are in school—thank goodness for that! Among the latter findings is that poor children and African American children come close to keeping up academically during the school year. That is to say, their learning from fall to spring is nearly at parity with that registered by whites and children from higher-income families. But these children are not performing at the same level at year’s end, and the reason why is revealing: they fall behind during the summer months owing to a dearth of learning resources in their families and neighborhoods.
Heyns’s study thus reveals that the achievement gap across social lines traces substantially to differences in out-of-school learning opportunities over the long summer break.8 On this basis, she concludes that school—regular school—is compensatory education for poor and disadvantaged minority children (Heyns 1978, 188).
We find much the same in our Baltimore research (Alexander, Entwisle, and Olson 2001; Entwisle, Alexander, and Olson 1997; Entwisle, Alexander, and Olson 2001). Our summer learning studies track test score gains for a representative sample of children who began first grade in twenty Baltimore City public schools in the fall of 1982, with seasonal comparisons through the end of elementary school (five school years and the four summers between).9
In reading comprehension (assessed via the California Achievement Test battery), low-income and middle-class children’s scores differed by a half-grade equivalent in the fall of first grade; by the end of fifth grade the gap had increased to three grade equivalents.10 To be reading at a third- or fourth-grade level when poised to transition to middle school is hardly what we would want for our children, but in Baltimore, just as in Atlanta, most of the gap increase over this span of years happened during the summer months. During the summer, middle-class children’s reading skills continued to improve, but low-income children’s did not, a pattern known as “summer slide” or “summer setback.” The school-year pattern was altogether different, as then low-income children and middle-class children registered similar progress.
In the late 1990s, the Early Childhood Longitudinal Study-Kindergarten cohort (ECLS-K), national in scope, yielded broadly similar results for learning over kindergarten, first grade, and the summer in between (Downey, von Hippel, and Broh 2004; Burkam et al. 2004). This was clearest for disparities along lines of family socioeconomic level, which paralleled the patterns seen in Atlanta and Baltimore:Past seasonal researchers have argued that inequality in cognitive skills emerges primarily when school is not in session, and that it likely is a function of different family and neighborhood experiences. … With substantially better data than previous researchers, we provide the strongest support to date for this position. … With respect to socioeconomic status, the primary source of inequality lies in children’s disparate non-school environments. (Downey, von Hippel, and Broh 2004, 632)
“Better data” here refers to the ECLS-K’s national coverage and large sample size. Also, with testing dates and school-year starting and ending dates known, ECLS-K analyses more accurately bracket the relevant seasonal intervals (in-school or summer).11
The patterning of learning disparities along lines of family socioeconomic background is consistent across these several studies, despite their many differences.12 They span three decades, cover different student populations in different localities at different grade levels, and monitor different domains of achievement using different instruments. Overriding these differences of detail are two profound insights: (1) poor children are capable learners in that they come close to keeping up when they are in school; (2) it is mainly family disadvantage, not school disadvantage, that holds children back.
From the literature on summer learning we see that when the influences of family and school are convincingly separated, both emerge as powerful forces in children’s learning. And what of David Armor’s assertion that “no one has found a way, on a large-scale basis, to overcome the influence of family”? Research into the seasonality of learning gives the lie to any such claim. School—and better still, a high-quality school experience—is the key to counteracting family disadvantage.
REGARDING THE NAYSAYERS
And what of those “dismal science” voices proclaiming the irrelevance of educational resources? After examining the relevant literatures, expanded to include trends in test performance over time and international comparisons, the economist Alan Krueger (1998, 30) had this to say: “The … widely held belief that American schools have failed … is not supported by the evidence. The evidence suggests that the perceived crisis in education has been greatly exaggerated, if indeed there is a crisis at all.”
And Krueger was hardly alone. Other voices from around the same time also pushed back against the assertion that America’s schools are in decline and ineffectual (see, for example, Bracey 2004; Berliner and Biddle 1995; Rothstein 1995), including mine (Alexander 1997). And what of the naysayers’ specific assertions? In each area, there have been countervailing voices, also highly regarded and with supporting evidence:
Against Jensen’s (1969) assertion that compensatory education has been tried and failed is evidence of impressive benefits of the Perry Preschool, ABECEDARIAN, and other early education programs that extend beyond achievement effects to include reduced risk of grade retention and high school dropout and, in young adulthood, lower levels of criminality and unemployment; see Barnett (2008, 2011), Heckman (2008), and Schweinhart and Weikart (1997) for small-scale studies that extend over many years. Large-scale studies, but of shorter duration, have demonstrated effectiveness for academic outcomes specifically—for example, in Oklahoma (Gormley 2013) and New Jersey (Barnett, Jung, and Youn 2013).
Against Slavin’s (1989) conclusion that smaller class sizes are no more effective than larger enrollments is evidence from the Tennessee class size experiment—a true experiment in which poor and minority children were found to derive substantial and long-lasting benefits from smaller classes in the early elementary grades (for example, Krueger and Whitmore 2002), extending, it seems, even to earnings in young adulthood (for example, Chetty et al. 2010).
And against Hanushek’s (1989, 1997) assertion that expenditures and the things that expenditures are used to purchase have little bearing on achievement, Larry Hedges, Richard Laine, and Rob Greenwald (1994, 11) conclude, after reviewing the same body of evidence, that effects for per pupil expenditures are “substantially positive” and effects for teacher salary, administrative inputs, and facilities are “typically positive.”
There have always been conflicting evidence, conclusions, and claims about the efficacy of schools and schooling in promoting student success, and it seems safe to predict that this will continue. But the extent to which the negative voices have dominated public perceptions—and I would say professional perceptions as well—really is quite striking. Why that is the case is hard to say, although I suspect that the foundation laid by the Coleman Report has played a role.
My pushback against these negative voices should not be construed as a belief that we have arrived at some blissful state of educational nirvana. To the contrary, the challenges we confront, especially for poor children in high-poverty school systems, are daunting and deeply entrenched. They will not yield easily, and the resources that we know can help these children often are in short supply in the schools they attend. One of those resources, the one I take up in my concluding comments, is the opportunity to attend schools not burdened by the drag of concentrated poverty.
CONCLUDING THOUGHTS
My conclusion from the review in this essay of five decades of research is that there indeed are enduring truths to be found in the Coleman Report, but they are not the ones typically touted. Moreover, I can imagine a future in which they no longer maintain; that is to say, though the report’s insights remain relevant, they are not immutable. At issue are consequences for children’s learning that flow from the overlapping spheres of influence emanating from family, neighborhood, and school. Alter the degree of overlap and the entire formula potentially changes. The Coleman Report instructs us that the surest route to helping poor and minority children keep up academically is to enable them to attend schools that are not just desegregated but authentically integrated—by which I mean that the experience of diversity infuses children’s daily experience: As Coleman notes, “School integration is vital … because it is the most consistent mechanism for improving the quality of education of disadvantaged children.” He continues: “So long as middle-class students remain a majority in a given school they establish the achievement tone … and by attending such a school disadvantaged students make more consistent educational gains than by any other mechanism.”
This was Coleman in 1970, sharing lessons he learned from the report that bears his name.13 His conclusion implies that weakening the link between family background and the character of the schools that children attend could well be transformative. School choice has that potential, although at present it is largely unrealized.14 So too would a renewed commitment to neighborhood and school desegregation along lines of family income (see, for example, Kahlenberg 2010; Semuels 2015); certainly that is what the Montgomery County research indicates. Consider this, from recent commentary: “Desegregation is the best way to improve our schools. Racial achievement gaps were narrowest during the height of school integration.”15 And as Nikole Hannah-Jones told Ira Glass, “Integration works. … We have this thing that we know works, that the data shows works, that we know is best for kids, and we will not talk about it. … It’s not even on the table.”16
Hannah-Jones overstates the case, but only somewhat. The Century Foundation has identified ninety-one school districts and individual charter schools throughout the United States that have a purposeful commitment to diversity (Potter, Quick, and Davies 2016). Those include some large school systems, including Hartford, Connecticut (Eaton 2013) and Louisville, Kentucky (Semuels 2015), both of which are pursuing a metropolitan regional approach. And Louisville holds the distinction of being a Southern school system that maintained its commitment to integration even after the court order that forced desegregation was lifted. Hartford and Louisville are exceptions, however, and ninety-one districts and schools is too few altogether—many too few.
If we have known for a half-century that school desegregation works—and not just for disadvantaged minority children17—why isn’t that the conventional wisdom, rather than the rhetoric of “schools make no difference”? And why the retreat from one of the few demonstrable interventions that is known to work? No doubt there are many considerations, but I suspect that the Coleman Report’s “family versus school” framing has played a role, together with the misconstruals of its results and their implications.
As noted by George Theoharis, when the Supreme Court’s school desegregation mandate did eventually begin to be enforced vigorously, levels of school segregation nationally declined dramatically.18 In parallel, and I think not coincidentally, the achievement gap across lines of race-ethnicity also declined dramatically. In a 1997 essay, I wrote: “Would it surprise you to learn that the ‘IQ’ gap separating black and white youth declined by almost a third between 1970 and 1990?” (Alexander 1997, 1). A year later, Krueger (1998, 31) pointed out that the black-white gap in math National Assessment of Educational Progress (NAEP) scores among seventeen-year-olds declined by nearly half between 1970 and 1990.19
Nothing is fixed. Attempts to parse the “whether” of school versus family seek a definitive answer, but this false dichotomy fundamentally misconstrues the backdrop to children’s learning. Family matters, to be sure, but school also matters, and it is how the two intersect that sets children on their developmental paths. The consequences that follow when many poor children live in high-poverty communities and attend schools with high-poverty enrollments are easy to anticipate, but are those consequences really removed from family? That the school’s enrollment mix emerges consistently in research as the school factor most strongly implicated in children’s learning makes it a difficult to case to argue.
In generating opportunity, family and school are indeed in tension, but it is a tension not captured in the “school versus family” framing. Here too Coleman understood the sense of it better than most: what counts is the balance between private family resources and public resources in support of children’s learning. At present, the private and public resources invested in children’s schooling are highly unequal, and they favor families of means. On that basis, Coleman concluded that complete parity across social lines in schooling probably is not feasible. The goal, rather, should be to move toward greater parity or gap reduction, which he believed was both feasible and desirable (Coleman 1975).
The resources at issue are not just material, as important as those are. In our Baltimore research, the typical “urban disadvantaged” parent, white or African American, had not finished high school; many were single parents weighed down by the so-called feminization of poverty. These parents themselves struggled at school, and many suffered a low literacy level and a weak command of formal English (see, for example, Farkas and Beron 2004; Hart and Risley 1995).
Given that it is hard for these parents to model and support the kinds of learning that are valued in school and to provide the enriching experiences so critical to children’s healthy development, is it realistic to expect that their children will arrive at school as well prepared as the children of middle-class and professional parents? The consequences of family resource inequality for children’s schooling are apparent at kindergarten entry (Lee and Burkam 2002). And because those inequalities are ever-present in children’s lives, so too is their drag (Alexander, Entwisle, and Olson 2007; Entwisle, Alexander, and Olson 1997)—absent, that is, an effective counterbalance to them.
The public resources invested in children’s schooling could be that counterbalance, but these resources are less abundant than the private resources commanded by some families. Moreover, public investments in children do not always, or even usually, favor the disadvantaged. Parents of means understand that a good school matters, and they see to it that their children attend one, either by paying a premium to purchase a home in a community that maintains good schools or by enrolling their children in a private school. Schools in wealthier communities serve the interests of their residents, and when those communities are isolated by race and family income, as is typical, the interests they serve are those of the well-to-do.
But it also needs to be said that schools do not simply reinforce patterns of family advantage and disadvantage. Rather, poor children fall behind when their learning depends on the sparse resources available to them at home and in their communities. Their schools, even those burdened by concentrated poverty, help them to keep up academically. From research on summer learning loss we learn that the portion of school influence that is separable from family serves to lift up poor children, not hold them back.
The key insight here is that the “overlap” of these overlapping spheres of influence is not perfect. Schools are at the very same time agents of social reproduction, favoring those already favored, and agents of social mobility. There is no logical contradiction in this duality, but to achieve greater public resource balance and so tilt the scales more in favor of social mobility, poor children need to have access to the same quality schooling as do the nonpoor.
And what exactly might that look like? Decades of research anchored in the Coleman Report instruct us that the most valuable school resource for poor and minority children inheres in the school’s demographic makeup. Poor children do best academically when they attend schools not burdened by concentrated poverty. Resource enhancement in high-poverty schools is a second-best option, as results under that approach are less impressive (Schwartz 2010) and there is little precedent for achieving sustained excellence in such settings on a broad scale.20
Richard Kahlenberg (2001, 2012), a leading advocate for school socioeconomic integration, has cataloged the many advantages that accrue to poor children when they attend schools that enroll a solid middle-class core: beneficial peer influences, such as stronger academic motivation and preparation; a more orderly classroom environment with less student mobility and fewer absences; improvement in parental resources through more effective advocacy and involvement; higher-quality teachers with higher expectations for their students; and a more demanding curriculum. To this list Rumberger and Palardy (2005) add the amount of homework students do and their feelings of safety at school.21 Meanwhile, Roslyn Mickelson (2005) and Karolyn Tyson (2011), among others, have revealed the harm done to poor and minority children in nominally desegregated schools when they are isolated in low-level remedial programs—internal resegregation through tracking. The goal must be authentic integration; simply having children of different backgrounds in the same building is not enough.
Not every wealthy suburban school provides an optimal learning environment, and not all high-poverty urban schools are distressed. These are tendencies, but as tendencies they are quite real and they matter. They matter for the poor children who are resource-deprived in their homes, their communities, and their schools, and they ought to matter for those of us who wish a brighter future for them.
To advance the cause of socioeconomic integration will not be easy, and certainly the demographic profile districtwide in places like Baltimore poses daunting challenges. But from the experience in those communities and individual schools throughout the country that are committed to this goal (Kahlenberg 2012; Kahlenberg and Potter 2014), we know that it is not impossible. If the Coleman Report helped trigger what I have called a deep malaise, perhaps the time has come to reverse course. From that document, and Coleman’s own good counsel, we know “what works.” What is needed is the will to follow through.
FOOTNOTES
↵1. For Gamoran and Long (2007, 23), the seminal finding is that “variation between schools in their resources mattered little for variation among individual students.”
↵2. Baltimore City Public Schools, “By the Numbers,” available at: http://www.baltimorecityschools.org/about/by_the_numbers (updated July 17, 2015).
↵3. It also should be noted, however, that the report’s assessment of family background is quite limited; it includes just a handful of surface indicators as reported by children and absent a meaningful theory or conceptualization of the processes by which parents and a family’s resources are thought to guide children’s academic development.
↵4. Research on summer learning loss, reviewed later in this essay, complicates that conclusion, but for now I let the point stand.
↵5. The situation for residential segregation is more complicated, as over time segregation by race has decreased while segregation by family income has increased (see, for example, Reardon and Bischoff 2011; Reardon et al. 2009). However, poor blacks specifically remain as isolated as ever (Massey and Brodmann 2014, 7).
↵6. Segregation indices represent the percentages of children who would have to change their school or neighborhood for every school and neighborhood in the city to have the same percentage African American as the citywide percentage.
↵7. Heyns also monitors summer school attendance, another facet of the “in school” versus “not in school” comparison.
↵8. This is not apparent under the more common accountability regime that tracks testing from spring to spring, blurring together school year and summer learning.
↵9. The city schools at the time were doing twice-annual achievement testing, in the fall and spring. It is this schedule that allows for the separation of school-year learning (fall to spring) from summer learning (spring to fall across successive school years).
↵10. A similar pattern maintained for math concepts and applications.
↵11. Having these dates provided an important technical check. Such adjustments might have altered estimates of school-year and summer learning gains, as well as of differences in the seasonal patterning of gains along social lines, but those details instead proved to be robust.
↵12. The same does not hold for the achievement gap by race-ethnicity: in the ECLS-K data, that gap expanded during the school year and there was little difference in learning rates summer versus school year.
↵13. Jack Rosenthal, “School Expert Calls Integration Vital Aid to Educating the Disadvantaged,” New York Times, March 9, 1970.
↵14. See, for example, Diane Ravitch’s (2010) critique of the pro–charter school documentary Waiting for “Superman.”
↵15. George Theoharis, “‘Forced Busing’ Didn’t Fail. Desegregation Is the Best Way to Improve Our Schools,” Washington Post, October 23, 2015.
↵16. Nikole Hannah-Jones, “The Problem We All Live With,” interview with Ira Glass, This American Life, July 31, 2015, available at: http://thisamericanlife.org/radio-archives/episode/562/transcript (accessed June 28, 2016).
↵17. Anya Kamenetz, “The Evidence That White Children Benefit from Integrated Schools,” nprED, October 19, 2015, available at: http://www.npr.org/sections/ed/2015/10/19/446085513/the-evidence-that-white-children-benefit-from-integrated-schools (accessed June 28, 2016).
↵18. Theoharis, “‘Forced Busing’ Didn’t Fail”; see also Orfield and Lee 2005.
↵19. Recent NAEP data indicate an upward track in scores since 2000 among whites, Latinos, and African Americans, accompanied by a modest reduction in the gap (Education Trust 2015). This trend, unlike the earlier one, is probably not driven by school desegregation.
↵20. Samuel Casey Carter (2000) identifies a number of individual schools that appear to meet this standard, but his report has received much critical commentary; see, for example, Billing and Bracey (2000) and Schmidt (2001). For a counterargument, see Duncan and Murnane (2013).
↵21. Many high-poverty schools are located in high-crime neighborhoods. Trauma surrounding exposure to violence affects children’s school performance (see Burdick-Will et al. 2011; Sharkey 2010), yet another way in which outside conditions filter into the school.
- Copyright © 2016 by Russell Sage Foundation. All rights reserved. Printed in the United States of America. No part of this publication may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, electronic, mechanical, photocopying, recording, or otherwise, without the prior written permission of the publisher. Reproduction by the United States Government in whole or in part is permitted for any purpose. Direct correspondence to: Karl Alexander at karl{at}jhu.edu, School of Education, 2800 N. Charles St., Johns Hopkins University, Baltimore, MD 21218.
Open Access Policy: RSF: The Russell Sage Foundation Journal of the Social Sciences is an open access journal. This article is published under a Creative Commons Attribution-NonCommercial-NoDerivs 3.0 Unported License.