Making the Visible Invisible

Scientific Management, Measurement, and Positivism in the New Science of Education Research

To anchor our narrative, we begin with the Cleveland School Survey, a massive study of education in Cleveland conducted by a team of 30 researchers led by Leonard Ayres, the head of the Departments of Statistics and Education at the Russell Sage Foundation (Lagemann, 2000). An exemplar of the new emphasis on measurement and scientific management, the Cleveland School Survey provides a window into key cultural and institutional shifts in the American educational system and in the research community that was developing around it. Rooted in progressive-era surveys of communities conducted by settlement workers, the Cleveland School Survey was among the most famous of the hundreds of school surveys that were conducted in U.S. public schools between 1911 and 1930, and represented a new phase of education research oriented toward increasing the social efficiency of schools (Lagemann, 2000; Sears, 1922; Tyack & Hansot, 1982). Yet even while the Cleveland Survey provided voluminous information about the city’s public schools, it made little mention of social class, apart from a brief discussion of the occupations of the inhabitants of the city at the beginning of the report. Most of the report’s recommendations were carried out, including a proposal to hire a superintendent of schools charged with overseeing “a system of scientific general supervision” informed by the continual analysis of self-surveys (Ayres, 1917, p. 55; see also Tyack & Hansot, 1982).

The school surveys of this period—large-scale compendiums of data on educational agencies, conducted by education experts using the techniques of social science—reflected the new emphasis on professional models of research and the disciplinary specialization occurring in higher education in the early 20th century. These education researchers were also a key constituency of the “administrative progressives”—the university professors, urban school superintendents, state education leaders, businessmen, lawyers, and elite men and women engaged in reforming elementary and secondary education (Tyack, 1974). Deeply influenced by ideas associated with scientific management, the administrative progressives’ goal was to empower experts who would make public schools more efficient and productive by using the data and analyses generated by researchers (Fass, 1989; Mehta, 2013; Tyack & Hansot, 1982).

Poverty Knowledge in the New Science of Education

One major focus of this “movement for the use of scientific methods in education” was to develop and validate tools for measuring intelligence (Courtis & Packer, 1920, p. 5). Fass (1980) observed that “intelligence testing sharpened and accelerated the cultural awareness of individual and group differences” (p. 439). Given the emphasis on measurement, it is perhaps not surprising that researchers also turned their efforts toward creating objective and reliable measures of social class, using indicators such as home conditions, social status, and economic group (e.g., J. C. Chapman & Sims, 1925; Holley, 1916; Williams, 1918). Researchers quickly linked those goals and used both measures to confirm their assumptions about intelligence and social class, which gave the assumptions the imprimatur of science. For example, Terman et al. (1917) initially drew on Taussig’s (1911) division of occupations into five noncompeting and hierarchical groups that ranged from day laborers to the professional classes in order to assess the relationship between students’ social status and intelligence. ¹⁶

Subsequent measures allowed Terman and other researchers to quantify and assess the relationship between class status and intelligence. Researchers replaced the more subjective assessments they had been relying on with new, seemingly precise measures of students’ class backgrounds. ¹⁷ Yet some of the measures were not as rigorous as they appeared. Barr (1918), one of Terman’s students, developed a numerical ranking of occupations for use in vocational guidance for his master’s thesis project at Stanford University, which Terman later popularized (see Terman, 1926). Barr (1918) selected a list of 100 representative occupations and asked 20 judges—all students in Terman’s intelligence testing class at Stanford—to rate the occupations based on their assessment of the “mental ability required for success in the occupation” (p. 28). A statistical analysis of the rankings yielded numeric ratings that ranged between 0.00 (hobo) to 19.62 (surgeon). ¹⁸ While Barr viewed his scale as a tool for vocational counselors for determining appropriate occupational choices for advising purposes, education researchers began to use his scale and other indicators to examine the relationship between intelligence and social class. The use of tests for social purposes and the connection between education research and policymaking are discussed by Kett (2013), who noted that researchers involved in intelligence testing for the army during World War I began to analyze the association between intelligence and occupational prestige. The economists of the era engaged in similar analyses.

Most of these analyses did not address poverty per se but legitimized existing patterns of social stratification by documenting and interpreting the association between measures of social class and IQ, which was viewed as an objective and merit-based criterion. In his influential Genetic Studies of Genius (1926), a study of 643 children with IQ scores of 130 or higher, Terman used these and other measures including (a) Taussig’s five-category ranking of occupations; (b) the Barr scale, which elaborated and quantified Taussig’s typology; (c) ratings of students’ home and neighborhood conditions; (d) parental education; and (e) the size of students’ home libraries as an indicator of the “cultural status” of the home (p. 81). Terman assessed his sample on these measures against comparison groups and concluded that “the heredity of our gifted subjects is much superior to that of the average individual” (p. 83). ¹⁹ Because Terman was studying gifted children rather than adults, he inferred that the children’s social origins were attributable to “original endowment rather than . . . environmental influences” (p. 66; see also Terman, 1916).

Because it was refracted through scientific racism—one of the dominant racial ideologies of the late 19th and early 20th centuries (Powers & Patton, 2008)—poverty knowledge was also racialized. In the first decades of the 20th century, the mass immigration of southern and eastern Europeans to the eastern United States and of Asians to the western United States complicated the racial order built on slavery, White privilege, and Black subordination (Smedley, 1993). Within this context, intelligence testers also used the new technologies of testing and measurement to rank racial and national groups (e.g., Terman, 1916; see also Stern, 2005; Valencia & Suzuki, 2001). Early researchers found that non-Whites had low IQs and also tended to be poor, and attributed these differences to hereditary factors. While European immigrants did not experience institutionalized boundaries based on race (Fox & Guglielmo, 2012), intelligence testing provided scientific proof of their apparent inferiority to native-born Whites and justified less formalized discrimination and immigration restrictions (Fass, 1989; Ngai, 2007). ²⁰

The ideas associated with scientific management and intelligence testing were also wedded with eugenics, the science of racial improvement (Baker, 1998; Stern, 2005). Indeed, the intelligence test was translated into English and popularized by a prominent eugenicist, Henry Herbert Goddard, who was one of Terman’s early colleagues (Lagemann, 2000). Terman, Thorndike, and other education researchers were active in eugenicist organizations (Barkan, 1992; Stern, 2005), and their psychometric research was also used to rationalize school segregation and the institutionalization and sterilization of juvenile delinquents. ²¹ While other education researchers may have been more neutral about the relative roles of hereditary and environmental factors in shaping intelligence, Fass (1980) observed that “the clear direction of American interpretation and the construction of experiments with tests were toward the view that intelligence tests were measuring something that was pure and inborn” (p. 441). This legacy left a heavy framing effect on the profession, which in turn shaped how the tests were incorporated into the work and organization of schools in the decades that followed (P. D. Chapman, 1988; Fass, 1980, 1989; Walters, 2009).

Competing Perspectives on the Tools of the Science of Education

During the first four decades of the 20th century, few education researchers questioned how the ideologies and techniques associated with the new science of education facilitated the sorting of students for differentiated school experiences. Because the field was dominated by psychology, most ignored structural issues or how such policies largely reproduced existing class and race inequalities in the guise of objectivity and meritocracy (e.g., Collins, 1928; Goodenough, 1928; Haggerty & Nash, 1924; see also Cohen & Barnes, 1999; Fass, 1989; Kett, 2013; Mehta, 2013; Rury, 2005; Tyack & Hansot, 1982; Valencia & Suzuki, 2001). Only a small cadre of scholars critiqued the relationship between social class and educational outcomes. While their perspective was a minority view among researchers until the Great Society programs of the 1960s, policymakers of the same period were even less concerned about the connection between poverty and educational outcomes (G. Davies, 1996; Katz, 2013; O’Connor, 2001). ²²

One of the critics was Walter Lippman. In a series of articles published in the New Republic in 1922, Lippman directly engaged three core tenets of the new science of education: that hereditary intelligence was measurable, that intelligence tests provided a valid measure of intelligence, and that intelligence tests should be used to determine students’ access to educational opportunities. To support his claims, Lippman reproduced figures from Terman et al. (1917) that illustrated the association between social class and IQ and argued the results could be interpreted as evidence of the “considerable connection between education and environment” rather than the heritability of IQ (Lippman, 1922, p. 329). Lippman also observed that Terman’s finding that the correlation between social class and IQ declined with age could be viewed as “a rather strong argument . . . for the traditional American theory that the public school is an agency for equalizing the opportunities of the privileged and the unprivileged” (p. 329).

John Dewey (1922) also joined the fray in the pages of the New Republic. While endorsing Lippman’s (1922) argument, Dewey expressed concern that the methodology underlying intelligence tests, which measured students against group averages, obscured the individuality of students. Instead of differentiating students’ educational experiences based on a crude measure, Dewey argued for an “inquiring and creative education” as early as elementary school (p. 37). He advocated for an education that expanded the capacities of all students. Until that time, Dewey concluded, “we shall never have any light upon what are the limits of intelligence set by innate qualities” (see also Hlebowitsh & Wraga, 1995). ²³

Among the dissenting voices whose criticism never gained much traction in the field were Stephen S. Colvin and George Counts (Valencia, 1997). ²⁴ Colvin (1922) distinguished between general intelligence, or the “inborn” ability to learn, and acquired intelligence or learning. General intelligence cannot be directly measured; rather, it is “infer[red] from differences in acquired intelligence” (p. 19). In Colvin’s view, the appropriate benchmark for assessing the results of individual or group intelligence tests is the results from another group of students who share similar knowledge and experiences. If students are tested on material with which they are not familiar, then the test will not provide an accurate measure of their ability to learn. “Hence children of different social and economic status may score differently in such tests not because of any real difference in native intelligence but because of such differences in home surroundings that some are favored while others are handicapped” (p. 43). Unlike Terman, Colvin also advocated interpreting test results in the context of other information about students, including teachers’ assessments of students.

Counts was one of the few education researchers who suggested that access to educational opportunity was a function of social class rather than intelligence. In The Selective Character of American Secondary Education, Counts (1922a) analyzed the factors that shaped high school attendance in four large cities. Counts observed that while enrollment in American public high schools had expanded considerably in the 40 years prior to his study, the students attending high schools were a highly select group, and the most important factor determining high school attendance was parental occupation. Children whose parents had lower status occupations were less likely than their more privileged peers to enter and persist in high school. Students were also tracked by parental occupation within high schools: The children of the “laboring classes” were overrepresented in the vocational tracks (p. 142). While there was some degree of selection based on intelligence, Counts observed that there was a considerable overlap in the distribution of ability, such that “there was much excellence out of, as well as much mediocrity in, the high school” (p. 147). While he stopped short of arguing that schools were reproducing class inequalities, Counts viewed the unequal distribution of the opportunity to attend high school as profoundly undemocratic. ²⁵

During the 1930s, the work of George Stoddard and his colleagues at the Iowa Child Welfare Research Station was also influential in producing a body of research that questioned the assumption that intelligence was fixed and determined primarily by heredity (Beatty, 2012a; Lagemann, 2000). By the mid-1930s, the consensus that the relationship between intelligence and socioeconomic status was rooted in genetics was also beginning to break down, and an increasing number of researchers were attributing the association to environmental factors (e.g., Byrns & Henmon, 1936; Jordan, 1933; Neff, 1938). However, the primary focus of many of these analyses was to document the relationship between socioeconomic status and intelligence. In assessing the significance of their findings, few researchers fundamentally questioned the use of intelligence tests for distributing educational opportunities or the role of schools in sorting, apart from raising relatively mild concerns in passing (e.g., Saltzman, 1940). ²⁶

Led by W. Lloyd Warner, the Committee on Human Development at the University of Chicago was one of the few research groups that raised more substantive questions about the use of intelligence tests to sort students within schools (Foley, 2010; Lagemann, 2000). For example, in Who Shall Be Educated (1944), Warner, Havighurst, and Loeb provided an analysis of the role of schools in perpetuating class-based status hierarchies in U.S. society. Warner et al. observed that although the expansion in access to schooling had promoted a widely held belief that education would provide most children with a path to social mobility, in practice the differentiated educational system sorted and selected students based on their class backgrounds.

Citing a study that examined educational outcomes for a group of students that had IQ scores of 110 or higher, Warner et al. (1944) noted that few students of high ability but below-average socioeconomic status completed high school or college. ²⁷ The authors attributed this phenomenon to the financial and opportunity costs of attending high school and college rather than to a lack of desire on the part of the children and families of the “lower socio-economic levels” (p. 53). When students did attend high school, they were sorted into different curricula, informally or formally, by class background. More specifically, while many people believed that differentiation within schools occurred based on ability, two of the principle methods for determining ability—teachers’ assessments and IQ tests—tended to favor middle-class children. ²⁸

Anticipating the turn toward educational programs in the War on Poverty, Warner et al. (1944) argued for a more fully realized meritocracy that was less dependent on class as a sorting mechanism and, rather, oriented toward identifying and supporting talented students from the lower classes at a young age. While stratification could not be avoided, Warner et al. proposed that the negative effects of stratification could be ameliorated by ensuring that “those from the bottom . . . be given more than a fighting chance to compete with those above them” (p. 146). Warner et al.’s ideas were embraced and promoted by prominent education reformers during this period: James B. Conant, president of Harvard University and later the author of three influential Carnegie Foundation–sponsored reports on American education between 1955 and 1964, and John W. Gardner, president of the Carnegie Foundation (Lagemann, 1989). The dominant view of reformers was that schools needed to more effectively facilitate the mobility of the gifted and talented.

The Zeitgeist of the Coleman Report: The “Culture of Poverty” and Human Capital Theory

Although poverty and other social inequalities were not a primary focus of education researchers’ concerns, the work of Warner et al. (1944) exemplifies a broader effort in the social sciences to identify and describe class cultures (O’Connor, 2001). In the 1950s, poverty was “rediscovered” by social scientists and liberal politicians in the United States (Brauer, 1982, p. 99; see also Katz, 2013). While the economy was growing and the middle class was perceived as expanding, John Kenneth Galbraith and others called attention to segments of the country and the population that were experiencing economic dislocation (Galbraith, 1969). As the notion of class cultures was taken up and elaborated by scholars, it offered an appealing explanation for persistent inequalities in “the affluent society.” ²⁹

Highlighting class-related cultural practices and beliefs allowed researchers to explain poverty as a function of deeply ingrained behavioral and psychological traits rather than structural inequalities (O’Connor, 2001; see also Katz, 2013). Most prominently, anthropologist Oscar Lewis argued that the poor adapted to structural dislocation and economic marginalization by developing a culture of poverty that was transmitted to their children through socialization. While originally conceptualized to analyze poverty in underdeveloped countries, the culture of poverty was taken up by social scientists to explain persistent poverty in the United States and introduced to a popular audience by Harrington’s (1962) The Other America. Riessman’s (1962) The Culturally Disadvantaged Child provided a detailed analysis of the social psychology of the “culturally deprived child” and a set of strategies aimed at helping schools better address the needs of such children (Beatty, 2012a). In economics, human capital theorists applied market principles to the supply-side of the labor market (O’Connor, 2001) and argued that, much like companies that invest in physical capital, individuals invest in education and training to maximize their returns in the labor market. While the empirical support for human capital theory is not robust (Carnoy, 2009; Karabel & Halsey, 1977; Klees, 2012), it cast the commonsense assumptions of the early school reformers in econometric terms (Labaree, 2010). As we explain below, culture-of-poverty arguments and human capital theory were the conceptual underpinnings for domestic policy initiatives focused on poverty.

This attention to poverty occurred during a period when the Supreme Court’s decision in Brown v. Board of Education (1954) made segregation and the resistance to desegregation that occurred in its wake a pressing public problem. ³⁰ Yet many prominent educators and researchers, including Conant (a high-profile member of that community), viewed segregation as largely a Southern problem and believed that the major issue facing large metropolitan areas outside the South was the de facto segregation of minority students in urban schools (Urban, 2009; see also Conant, 1961). ³¹

The lack of awareness among these researchers of their own politics of expertise and of the sociocultural processes involved in constructing “expert research knowledge” (Walters, 2009) was consistent with a culture of education research that continued to bracket inequality related to poverty and racism. These processes of willful ignorance allowed many education researchers in the late 1950s and early 1960s to pivot from the issue of segregation and its attendant focus on race-based inequalities in educational opportunity, to the more racially neutral problem of the cultural disadvantages of urban youth. For example, in the introduction to The Educationally Retarded and Disadvantaged, the 66th yearbook of the National Society for the Study of Education, the yearbook committee noted that “although large numbers of Negro pupils are among the ‘disadvantaged,’ there are also very large numbers of retarded and ‘disadvantaged’ white pupils who similarly need ‘compensatory’ education” (Whitty, 1967, p. 4; see also Artiles, 2011; Martinez & Rury, 2012; Smiley, 1967).

These ideas from the social sciences shaped the perspectives of policymakers, and in particular those of the economists associated with President John F. Kennedy’s Council of Economic Advisors, who were also influenced by pragmatic political considerations (O’Connor, 2001). In the early 1960s, the Kennedy administration had begun working on proposals to address poverty (Brauer, 1982). While Kennedy had advocated for civil rights legislation, he also recognized that Southern members of Congress would oppose civil rights but might be more willing to support antipoverty programs because of the high concentration of poverty in the South. Both culture-of-poverty arguments and human capital theory pointed to the benefits of investments in education targeted at youth. Policies aimed at redistributing income to the poor had little political support, so the Kennedy administration focused on targeting a modest amount of funds toward small-scale programs in local communities (Brauer, 1982; Kantor & Lowe, 2006; H. M. Levin, 1989; O’Connor, 2001).

After Kennedy was assassinated, antipoverty proposals, which had not been publicly announced, became key elements of the War on Poverty, the signature initiative of President Lyndon B. Johnson’s administration. Johnson also promoted other key policies of the Kennedy administration, including the Civil Rights Act of 1964. Brown v. Board of Education and the conflicts it engendered around desegregation also helped focus the Johnson administration’s efforts on education policy (Kantor & Lowe, 2006). The Elementary and Secondary Education Act (ESEA) of 1965 funded compensatory education for poor children, although Congress distributed the funds widely and its budgetary appropriations fell short of meeting all students’ needs (Kantor & Lowe, 2006). While Johnson’s education officials initially saw ESEA as a way to force Southern states to comply with the Civil Rights Act, which prohibited discrimination in schools that received federal dollars, the backlash against efforts to enforce desegregation also threatened the ESEA’s political viability. By the late 1960s, government officials began to decouple federal desegregation efforts from Title I of ESEA, which targeted federal funds to low-income schools. Urban (2012) observed that the compensatory education programs of the 1960s, and the researchers that created and championed them, tended to take segregated schools for granted and addressed achievement gaps within that structure rather than engaging in efforts to desegregate schools. While it was unlikely that this approach was motivated by racism, it “reinforce[d] the racism of white educational actors who sought to distance white students from black and other minority students in all settings at all costs” (Urban, 2012, p. 3; see also Beatty, 2012a).

While the War on Poverty was largely framed in color-blind terms, Johnson’s June 1965 speech at Howard University signaled a shift in the administration’s understanding of Black poverty as a unique problem and in need of targeted policy interventions (G. Davies, 1996; see also Katz, 2013). Drawing on the Moynihan Report, which had not yet been released, Johnson merged a critique of White racism with culture-of-poverty arguments. According to Johnson, the dynamics of Black poverty were fundamentally “the consequence of ancient brutality, past injustice, and present prejudice,” which isolated Blacks in ghettos and contributed to the breakdown of the Black family and communities (Johnson, 1965, para. 36). After the Moynihan Report was released and its key arguments and claims were distorted in the national media (Katz, 2013), and as the urban riots of the 1960s unfolded, urban poverty was increasingly understood in racial terms (G. Davies, 1996; Martinez & Rury, 2012).

The Genesis of the Coleman Report and Its Findings

Released in 1966, the Coleman Report had a less auspicious reception than its current standing in the field would suggest (G. Grant, 1973). The Civil Rights Act of 1964 required the Commissioner of Education to assess the “lack of availability of equal educational opportunities for individuals by reason of race, color, religion, or national origin in public educational institutions at all levels in the United States” (Section 402). The contract for the project was awarded to sociologist James Coleman, who reframed the goals of the survey required by the legislation from a focus on educational inequalities (i.e., school-based resources) to an analysis of the relationship between educational inputs and outcomes. ³² Coleman and his team of researchers executed the ambitious, state-of-the-art Equality of Educational Opportunity (EEO) survey, which surveyed and tested 645,000 elementary and secondary students at five grade levels attending 4,000 schools, and also surveyed their teachers and principals (Coleman et al., 1966).

Using then cutting-edge statistical techniques, Coleman and his team analyzed the relationships between student achievement and school resources, teacher characteristics, and students’ family backgrounds. ³³ The latter measures were constructed from students’ responses on a set of questions that were aligned with the long-established tradition of measuring socioeconomic status in educational psychology (e.g., J. C. Chapman & Sims, 1925): urbanism, mother’s and father’s education, family size, items in the home (e.g., telephone, refrigerator, car), and reading material in the home (e.g., books, magazines, encyclopedias). ³⁴ The analysis indicated that while schools were highly segregated, the differences in the characteristics of schools attended by White and Black students were not as large as many of the people involved in the study expected, including Coleman himself (G. Grant, 1973; see also Gamoran & Long, 2007; Mosteller & Moynihan, 1972).

After documenting the substantial achievement gaps between White and Black students, Coleman determined that most of the variation in students’ achievement scores occurred within schools rather than between schools and that school characteristics accounted for a small proportion of this within-school variation. Rather, Coleman and his research team found that much of the variation in student achievement was attributable to students’ family backgrounds and the backgrounds of the other students attending their schools: ³⁵

Schools bring little influence to bear on a child’s achievement that is independent of his background and general social context; and . . . this very lack of an independent effect means that the inequalities imposed on children by their home, neighborhood, and peer environment are carried along to become the inequalities with which they confront adult life at the end of school. 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. (Coleman et al., 1966, p. 325)

While the Coleman Report’s focus on the outcomes of schooling has been widely viewed as shifting scholarly and policy attention from the inputs of schooling to outputs (e.g., Carnoy, 2009; Mosteller & Moynihan, 1972; Walters & Lareau, 2009), we might also see it as both the culmination and extension of long-standing blind spots in education research and the willful ignorance that framed the nexus between education research and policymaking in the first half of the 20th century. These blind spots include positivism and the emphasis on measurement, human capital theory, culture-of-poverty perspectives, and, perhaps more constraining, the systematic willful ignorance that resulted from using the White middle-class male student as the implicit norm and the othering of students who did not conform to “normalized” class, race/ethnicity, gender, and ability categories (Artiles, 2011; Valencia & Suzuki, 2001; Varenne & McDermott, 1999).

The school survey, measures of socioeconomic status, and achievement tests were joined in multivariate analyses of massive nationwide samples of students and schools. ³⁶ To a certain degree, the initial legislative charge to focus the analysis on race-based inequalities in inputs might be viewed as a departure from prior studies, which did not systematically document inequalities in access to educational resources (e.g., Warner et al., 1944). However, Coleman’s decision to focus on outputs rather than inputs was a departure only in the context of the heightened awareness of race-based educational and social inequalities fostered by Brown and the civil rights movement. His decision can also be read as rooted in early research efforts such as the school survey, even while the report’s findings challenged the widely held assumption of early school surveyors that increasing school resources would result in increased efficiency and productivity (Lagemann, 2000; see also Cubberley, 1916).

However, as Mosteller and Moynihan (1972) observed, Coleman and his colleagues did not address social class directly in their analysis. The “presence of social class was implicit in the stated findings that family background, measured in social class terms . . . is apparently a major determinant of educational achievement” (p. 22). Drawing on another analysis of the EEO data that assessed class differences in achievement within and across racial groups, Mosteller and Moynihan highlighted how some of the racial achievement gaps documented by the Coleman Report were attributable to social class (see Okada, Cohen, & Mayeske, 1969).

Coleman et al. (1966) addressed poverty explicitly only once in the summary of the report, in a discussion of the consistent and widening gaps in achievement between minority and White students across the grade span:

Whatever may be the combination of non-school factors—poverty, community attitudes, low educational level of parents—which put minority students at a disadvantage in verbal and nonverbal skills when they enter the first grade, the fact is that schools have not overcome it. (p. 21)

Likewise, in an expanded discussion of achievement gaps in the body of the report, Coleman et al. (1966) invoked culture-of-poverty arguments when he noted that the achievement differences suggested that “the ecology of educational disadvantages experienced by particular minority groups in the United States” were largely attributable to “the background cultures from which these groups came” (pp. 273, 275). Coleman noted that if schools provided different opportunities for students, or if they did not address students’ cultural disadvantages either by default or by design, these differences would persist and restrict students’ opportunities into adulthood.

The Political Impact of the Coleman Report

Coleman et al.’s (1966) central finding that family background was the key factor shaping student outcomes contradicted the common wisdom among officials in the Johnson administration, as well as one of the central myths of American society: that schools are a mechanism for social mobility (Lippman, 1922; Warner et al., 1944). Not incidentally, it also suggested that the billion dollars targeted toward compensatory education programs by the ESEA, which Congress had approved the year before, were being misdirected (G. Grant, 1973). A short summary of the Coleman Report was released on a Friday afternoon prior to a holiday weekend, without a press release. Initial newspaper reports mentioned the findings related to family background but tended to place greater emphasis on the report’s findings about segregation and highlighted the areas where there was evidence of school resource gaps between majority-Black and majority-White schools (e.g., Herbers, 1966; see also G. Grant, 1973). The full report was released a month later. Although Congress passed an extension of the ESEA in 1966, few Congressional staff members were aware of the Coleman Report’s findings (G. Grant, 1973).

The Coleman Report had relatively little political impact until Daniel Patrick Moynihan, the editor of The Public Interest, asked Coleman to write an article for the journal summarizing the findings and their policy implications (G. Grant, 1973). Moynihan also promoted the report’s finding in his own speeches and articles. When Moynihan assumed a position in President Richard Nixon’s administration, the findings from the Coleman Report were used to justify the administration’s unsuccessful attempt to cut funding for education. Four years later, Coleman’s results on peer effects and, more specifically, his finding that the family backgrounds of the students attending a school had a strong effect on the achievement of minority students were used to justify federal funding for desegregation by the Nixon administration. Coleman’s own advocacy of his findings on Capitol Hill as well as Moynihan’s ongoing support were important factors in ensuring that the report gained relevance and momentum as a policy document. While the initial policy impact of the report was to support the Nixon’s administration’s approach to desegregation, the most enduring findings were that (a) most of the variation in student achievement occurs within rather than between schools, and (b) there is strong relationship between family background and student achievement relative to school factors (see, e.g., Vinovskis, 2009; Walters & Lareau, 2009).

Replications and Extensions: The Coleman Report and Its Influence on Research

Another indicator of the Coleman Report’s impact, in addition to the many studies that have engaged its findings, ³⁷ is the repeated analysis of the EEO data, including studies published more than three decades after the report was released (Borman & Dowling, 2010; Konstantopoulos & Borman, 2011). This is striking because there are few replication studies in education research (Makel & Plucker, 2014; Schneider, 2004). Some of the most prominent and early appraisals of Coleman’s findings, including the reanalyses of the EEO data, were begun almost immediately after the report was issued in the form of a faculty seminar conducted at Harvard University in 1966–1967, led by Moynihan (Mosteller & Moynihan, 1972). Funded by the Carnegie Corporation, more than 70 faculty members participated in the seminar, including Coleman and some of his coauthors. The seminar generated an edited book addressing the report’s findings and their implications for policy and future research (Mosteller & Moynihan, 1972). Most of the reanalyses have tended to confirm the Coleman Report’s main conclusions. Here, we highlight those that have addressed social class more directly.

In the Mosteller and Moyhihan (1972) volume, Jencks (1972) and Smith (1972) conducted reanalyses of EEO data that addressed the relationships among social class, school resources, and student achievement more directly than the Coleman Report had. Jencks (1972) focused on Northern elementary schools and found that, on average, poor students (students who had seven or fewer home items) attended schools with resources similar to those of the schools attended by their middle-class peers. Jencks and his colleagues incorporated the analyses of EEO data into a book length treatment of the topic of economic inequality in American society (Jencks et al., 1972). In Inequality, Jencks et al. (1972) reached a conclusion that he had hinted at in the Mosteller and Moynihan (1972) volume—that it would be impossible to alter existing patterns of poverty and social inequality through school reform. Rather, Jencks et al. (1972) argued that a more effective way of ending inequality and, in particular, of ameliorating poverty—the goal of the War on Poverty—was to redistribute income. This argument has since been elaborated by education researchers but has had little influence on policymaking (e.g., Anyon, 1997; Apple, 2012; Berliner, 2006; Darling-Hammond, 2010; Kozol, 1992).

Smith (1972) observed that the argument that most of the variation in achievement between schools is attributable to students’ family backgrounds rests on the assumption that family background and school characteristics are relatively independent. However, if more advantaged students attend better resourced schools, then the causal claim could be reversed. That is, differences in student achievement would be largely attributable to school resources rather than to family background. Smith suggested that there was little evidence for the latter claim and extended Coleman’s conclusions to highlight the sorting function of schools, much like Warner et al. (1944) three decades earlier.

A somewhat different conclusion can be drawn from a school-level reanalysis of the EEO data conducted by Department of Education staff. Mayeske et al. (1969) examined the relationship between indexes constructed from clusters of variables and found that schools’ characteristics were highly correlated with the socioeconomic status of their students, and that socioeconomic status and school characteristics had a substantial joint effect on school achievement. Notably, the index of socioeconomic status used in the Mayeske et al. analysis included a survey item that asked students to report their fathers’ occupations, whereas Coleman’s analyses did not. ³⁸

While overall the findings from the Mayeske et al. (1969) reanalysis tended to confirm Coleman’s initial conclusions, they also highlighted how the “influence of the school is bound up with the social background of the students that they get initially. Very little influence of the schools can be separated from the social background of their students and very little of the influence of social background can be separated from the influence of the schools” (p. 327). In a subsequent analysis of student achievement using the same techniques, Mayeske, Okada, Cohen, Beaton, and Wisler (1973) reached similar conclusions. These findings indicate that students’ socioeconomic status was undermeasured in the Coleman report and in reanalyses that used a more limited set of variables to assess it (see also Rury & Saatcioglu, 2015). The undermeasurement of socioeconomic status may explain why the Coleman Report has been less influential in raising public awareness about poverty.

Another influential critique of the Coleman Report came from scholars who argued that schools were not effectively educating the children of the poor (Edmonds, 1979). Studies of “effective schools,” including a reanalysis of EEO data (Edmonds & Frederiksen, 1979), focused on analyzing the academic efficacy of schools serving poor children and identifying common practices within the group of higher achieving schools. ³⁹ In the decades that followed, researchers, including Coleman himself, continued to engage the questions raised by the Coleman Report about the relative influence of family background and school factors on student achievement (e.g., Coleman, 1975; ⁴⁰ Hanushek, 1996; Hedges & Greenwald, 1996; see also Gamoran & Long, 2007, for a review of the school effects literature).

More recent reanalyses of the EEO data using newer and more sophisticated analytical techniques than those available to Coleman and his team have confirmed the Coleman Report’s main findings related to family background. They have also indicated that school effects are more substantial than Coleman’s analysis suggested (e.g., Konstantopoulos, 2006; Konstantopoulos & Borman, 2011; see also Borman & Dowling, 2010, and the review in Gamoran & Long, 2007). Much like Mayeske et al. (1969) and Mayeske et al. (1973), these newer analyses suggest that it is difficult to disentangle the relationship between students’ backgrounds and school resources, at least in the ways that they are conventionally measured (e.g., Konstantopoulos & Borman, 2011). In other words, the effects of school resources are complex and are likely contingent on the social dynamics within schools, or on how resources and opportunities are distributed and engaged by teachers and students (e.g., Bidwell & Kasarda, 1980).

Is the Coleman Report Still Relevant?

If we view the Coleman Report through the lens of the research-policy nexus, it represents a powerful prototypical case of a research study that was legislatively mandated; recognized as rigorous, comprehensive, and connected to what was viewed as the real needs of school reform; and ultimately acknowledged as relevant by the research community, policymakers, and the media. Particularly notable is that a study of this type helped catalyze specific research-based policy recommendations. However, despite all these positive attributes and its enormous potential, the report was constrained by the long-standing blind spots and willful ignorance at the center of the research-policy nexus that we have described.

The six research and policy recommendations outlined by Mosteller and Moynihan (1972) are illustrative in this regard. First, Mosteller and Moynihan argued that “equality of educational achievement for the several racial/ethnic groups, [should be adopted] as a national goal” (p. 52). This is what we would describe in contemporary terms as closing the achievement gap. Second, they pointed to the need for long-term and experimental studies aimed at understanding the effects of educational programs (see also Dyer, 1972; Smith, 1972). Third, they advocated for regular and national-level assessments of student achievement, such as the National Assessment of Educational Progress, which had been initiated recently. Fourth, echoing Jencks (1972), they highlighted the need for employment and income programs aimed at improving children’s home environments. They also emphasized the need to evaluate new social programs. Fifth, they recommended supporting the creation of new schools as laboratories for educational innovations. Finally, they exhorted the public and policymakers to embrace “optimism” (Mosteller & Moynihan, 1972, p. 57). They urged the public to pressure elected officials to set educational goals and assess the nation’s progress toward meeting those goals. Yet they also noted that gains would be a “rarity” and should be celebrated when they occurred (p. 57). They highlighted the considerable advances that had been made in the 20th century in expanding access to education and reducing inequalities, and pointed to the progress that had been made in desegregating Southern schools since the release of the Coleman Report.

Of Mosteller and Moynihan’s (1972) six policy recommendations, some are easily recognizable in contemporary policies such as No Child Left Behind, Race to the Top, and the enshrining of studies using experimental research designs as the “gold standard” (Walters, 2009). Mosteller and Moynihan’s first and third recommendations are echoed in the ways that No Child Left Behind institutionalized accountability policies aimed at documenting and closing achievement gaps, annual testing, and the assessment of schools against benchmarks that continued to reverberate in Race to the Top (RTT). ⁴¹ The second recommendation, regarding more rigorous and experimental studies, is clearly manifested in the Education Sciences Reform Act of 2002 and the What Works Clearinghouse; the implications of these policies for the research community have been discussed extensively elsewhere (Berliner, 2002; Erickson, 2014; Moss et al., 2009; Pellegrino & Goldman, 2002; Shavelson & Towne, 2002; St. Pierre, 2006). ⁴² The fifth recommendation is mirrored in the embrace and promotion of charter schools in federal and state policies.

Two of Mosteller and Moynihan’s (1972) recommendations have had less resonance for contemporary policymakers. Their sixth recommendation, that the public and policymakers should embrace measured optimism in assessing educational progress, is striking in light of the decades-long perception of crisis in education engendered by A Nation at Risk (National Commission on Excellence in Education, 1983; see also Berliner & Biddle, 1995; Mehta, 2013; “Testing Doesn’t Measure up,” 2015). If their sixth recommendation was rendered irrelevant by the displacement of optimism by crisis-driven pessimism, the fourth (employment, income, and social programs aimed at improving children’s home environments) was plainly and willfully ignored. In this regard, it is instructive to reflect on how Mosteller and Moynihan introduced the six recommendations:

No single program can be expected to close the gap in educational achievement between the disadvantaged minorities and the white group. Furthermore, we do not know what school programs might offer the largest improvement for the cost involved. We must also recognize that strengthened educational achievement may not be the most important social reform needed. Indeed, higher income and better occupational changes probably are more immediate targets of reform groups, with educational achievement regarded as part of the means toward such change, as well as having value in itself. (p. 52) ⁴³

Even with its blind spots, Coleman’s analysis gave the relationship between poverty and school outcomes heightened visibility among researchers. As we indicated earlier, a robust “school effects” literature engaged and debated Coleman’s findings, using increasingly sophisticated analytical techniques such as multilevel modeling (e.g., Konstantopoulos, 2006). Likewise, as in the founding years of the field, some researchers have consistently addressed issues of poverty, its effects on schools, and access to educational opportunities (e.g., Anyon, 1997; Berliner, 2006; Carter & Welner, 2013; Darling-Hammond, 2010; Gamoran, 2001, 2008; Kozol, 1992; Oakes, Hunter Quartz, Ryan & Lipton, 2000; and many others), but their arguments have not had the traction they deserve among policymakers, given their empirical robustness and conceptual contributions. Instead, Coleman’s analytical approach has become common sense in educational policymaking, in the form of a narrowly conceived and punitive accountability movement (Mehta, 2013). That is, policymakers have tended to embrace the Coleman Report’s measurement technologies while ignoring the implications of the findings. ⁴⁴ It has been more politically expedient for policymakers to implement relatively limited policies that focus on schools rather than redistributive policies aimed at ameliorating poverty. When the more limited policies invariably fail to address poverty or class-based achievement gaps, policymakers tend to blame schools or the poor (Beatty, 2012b). While education research has come full circle to the emphasis on measurement embodied in the early decades of the field, the public seems to have lost faith in schooling as a democratic institution, another hallmark of the administrative progressives’ belief system in the early 20th century (Fischman & Haas, 2012).

We are convinced that although many education researchers were instrumental in the development and legitimation of new regimes of research expertise that willfully ignored poverty and other educational inequalities, and their intersectionalities, researchers have also begun to challenge those regimes as the field enters its second century. We are not naive; we understand that in highly polarized and politicized contexts, the biggest challenge in developing a more effective research-policy nexus is not to produce more or better data—the field is already doing that—but to overcome the lack of trust among potential allies and to intervene in the political arena to confront those who manipulate research for political gain. ⁴⁵ The scenario is complex, and as our own review shows, education researchers do not have a stellar record. Yet we remain cautiously optimistic. There are significant manifestations of communities, teachers, parents, social movements, universities, researchers, and allied groups resisting, denouncing, and demanding changes to the punishing regimes affecting so many children and their schools. A number of foundations, including the William T. Grant and Spencer Foundations, have targeted resources and attention to addressing poverty and improving the connections between education research, policy, and practice.

Even with all the challenges and limitations confronting a professional research organization, AERA has been and remains a key institution in challenging structures and actors that benefit from the visible invisibility of the oppressive effects of unequal educational opportunities. The road to excellent education in democratic societies needs education researchers and allied policymakers who abandon the willful ignorance of poverty and other social inequalities, and instead work together to address the complex relationships between them and the outcomes of education.