Assumptions Versus History in Ethnic Education
by Thomas Sowell - 1981
In an attempt to investigate factors surrounding educational performance among different ethnic and racial groups, several areas are examined: (1) segregation; (2) performance levels and intelligence quotient variables among various ethnic groups; (3) intergroup differences among segregated and nonsegregated groups; (4) socioeconomic variables; and (5) the effect of upward mobility. (Source: ERIC)
Collection of the data presented here was financed by grants from (in alphabetical order) the Ford Foundation, National Affairs, Inc., the Rockefeller Foundation, and the U.S. Office of Economic Opportunity. A special debt of gratitude is due Janella Moore for her role in collecting and supervising the collection of the raw data from the coded schools. Raw data for the two named schools from the study in The Public Interest were collected by Yolande Brower and Margo Logan. Responsibility for all conclusions or errors is mine.
Even more important than the assumptions and beliefs that guide educational policy is the extent to which these assumptions and beliefs are tested against facts, rather than judged by their individual plausibility or by their consonance with some general vision of education or society. In the area of race and ethnicity, many key assumptions behind current policies remain untested against either current or historical facts.
Perhaps the best-publicized assumptions guiding judicial educational policy today is that segregated schools are inherently inferior (Brown v. Board of Education), and that the equalization of per-pupil expenditures is essential to an equalization of education (Serrano v. Priest). A more general assumption, encompassing these and other policies, is that large disparities in school performances among racial or ethnic groups (1) are unusual and suspicious, and (2) reflect differences in the way those groups are treated by the schools and/or the society. These are not unreasonable assumptions, but reasonableness is no substitute for empirical verification, especially when so much is at stake. There is no a priori reason why statistics collected at a given institution must be solely the result of the policies of that institution, rather than the characteristics of the population in question.
Other common educational doctrines seldom seriously tested include the belief that school performance is greatly influenced by family socioeconomic status, class size, teacher-student differences in ethnicity, and the cultural bias of tests. Statistical correlations are abundantly available in support of some of these doctrines, but the principle that "correlation is not causation" cannot be simply a pious disclaimer uttered in passing, while proceeding post haste to equate the two in cognitive conclusion or policy application.
History is important because it allows a given principle to be tested under a far wider variety of conditions than are likely to be found contemporaneously. For example, we need not confound the effect of segregation with the effect of a particular cultural background or set of values and beliefs by or about blacks. Despite the unplanned nature of historical "experiments," they can sometimes provide a richer set of data. At the very least, history provides an additional set of evidence.
There is no serious question that the segregated black schools long traditional in the South generally had educational results inferior to those in the white schools in the same communities. The Supreme Court in Brown v. Board of Education attributed causation, thereby making segregation the reason for educational and psychological problems in the black schools, and in turn this state-enforced educational inferiority constituted a denial of the "equal protection" required under the Fourteenth Amendment to the Constitution. But the mere contemporaneous existence of two striking social phenomena-rigid racial segregation and large differences in academic performance, in this case-does not automatically establish one as the cause of the other. At the very least, such a conclusion should await consideration of alternative hypotheses and the derivation of evidence that would distinguish one hypothesis from the other.
Among the best-known competing theories are that deficiencies in educational performance are the result of either (1) a unique black heredity or (2) a unique black environment or history. Tempting as it is to plunge into the Jensen controversy (as I have done elsewhere1), we must recognize how limited these hotly disputed theories are for the issue at hand. Before resorting to either hereditary or environmental theories, which-even if true-would be applicable only to the special case of blacks, we must first determine whether the performance disparities between blacks and whites are themselves unique. In other words, we need to frame some general hypotheses, going beyond black-white differences, and at least see whether these larger patterns apply to racial as well as ethnic or other socioeconomic group differences. If such an attempt fails, there will then be time enough to formulate theories applying solely to blacks and whites. But we should not begin by presuming that such an attempt must fail, before even trying.
To present an alternative hypothesis: What would we expect to see if segregation were not a significant cause of educational differences? Unless we presume a genetic basis for unique black intellectual or educational results, we might expect to find at least three major phenomena:
1. Black intellectual or educational performance would not be unique in level or pattern, but would be closely approximated by some other group(s).
2. Some groups who live in the same neighborhoods and attend the same schools together would be expected to exhibit intellectual or educational differences of a magnitude comparable to black-white differences in the segregated South.
3. Performance differences within the set of segregated black schools should be of a magnitude comparable to those between black and white schools in the segregated South.
A case could be made that the unique historical background of blacks might take the place of a unique genetic background in preventing these phenomena from emerging. However, if these phenomena do emerge, despite some unique features of black history, then the argument that segregation has had the devastating educational effect attributed to it is undermined all the more. That is, the uniqueness of racial segregation plus all the other unique features of black history together would have been unable to present the emergence of a pattern found among other American ethnic groups not subject to these unique influences.
PERFORMANCE LEVELS AMONG ETHNIC GROUPS
The habit of comparing black IQs, reading scores, or other indices of aptitude or performances with the "national average" glosses over the question of whether that "national average" is itself only an amalgam of results as disparate as the black-white differences under discussion. This is also true of comparison of economic or other indices of any given group with the so-called national average.
Despite a long and bitter controversy over how best to explain a supposedly unique black IQ level, neither the hereditary nor the environmental advocates have established that uniqueness in the first place. History presents an entirely different picture from that from which both sets of controversialists begin. There has been nothing unique about the black IQ level. The average IQ of blacks in the United States has been consistently around 85, compared with the national norm of 100. Group IQ averages at or below 85 have been common in history and currently. Back in the 1920s, for example, numerous studies showed similar IQ averages for such American ethnic groups as the Italians, Greeks, Poles, Hispanics, Slovaks, and Portugese.2 A more recent study shows Mexican Americans with lower average IQs than blacks in the 1940s, 1950s, and 1960s, and Puerto Ricans with lower average IQs than blacks in the 1970s.3 Similar group averages have been found in white mountaineer communities in the United States, and among culturally isolated people in the Hebrides Islands off Scotland and among children raised in canal boat communities in England.4
A similar picture emerges in comparisons of educational performances of black schools in Harlem in the 1940s and 1950s compared with (1) the citywide average, and (2) performances on the same tests given at the same time in the ethnic neighborhoods on New York's lower east side. Table 1 shows that schools in Harlem and on the lower east side were typically below the citywide averages, but without any consistent or decisive advantage over each other in either verbal or mathematical areas.
Sometimes it is not the performance level but the performance pattern that is considered unique to blacks. The so-called Moynihan Report in the 1960s demonstrated that black female performance on tests and grades significantly exceeded black male performance, explaining this by a supposedly "matriarchial" black culture going back to slavery.5 However, higher female performances have been common among low-IQ groups, now and in the past, in the United States and abroad.6 Sex differences are especially pronounced among high IQ members of low-IQ groups.7 A plausible speculation is that this is an example of a more general phenomenon of greater female insulation from either the positive or the negative features of the environment.8 But whatever the explanation, the pattern extends well beyond blacks. There was, for example, a time when Jews scored below the national average on mental tests,9 and in that era Jewish girls scored higher than Jewish boys.10 Today, Mexican American girls score higher than boys on IQ tests and are overrepresented among high-IQ Mexican American students.11
INTERGROUP DIFFERENCES AMONG NONSEGREGATED ETHNICS
There are serious practical difficulties in testing the proposition that intergroup differences among nonsegregated ethnics would be comparable to black-white differences, in the absence of the segregation effect postulated in Brown v. Board of Education. Interethnic comparisons of educational performance among European-origin groups have become rare since the 1920s, though there was a substantial literature before then, during the immigration controversy preceding the restrictive laws that went into force in 1924. These early studies did, however, show IQ differences, for example, as great as (or greater than) those between blacks and whites. While Polish and Italian youngsters, for example, scored in the low to middle 80s in IQ, scores at or above 100 were common among youngsters who were German or Irish or from a number of other groups that had immigrated far enough in the past to be assimilated to American norms in general.12 Diane Ravitch's history of the New York City school system showed German and Jewish school children in the early twentieth century completing high school at a rate more than one hundred times greater than that of Irish or Italian school children.13
A recent study by the present writer has attempted to trace the IQ records of a number of ethnic groups from that early period to the present. The research has involved the collection of more than 70,000 IQ records for students attending fifty-eight schools in communities across the country. The general results of that survey have already been published,14 but what is relevant here is the performance of different groups in the same schools at the same time.
The two European immigrant groups with the most pronounced cultural differences in their approach to education were the Jews and the Italians. The centuries-old tradition of reverence for learning in the Jewish culture is well known. Among the people of southern Italyfrom whom most Italian Americans are descendedan opposite tradition of hostility to formal schooling existed. For example, the introduction of compulsory school attendance laws in Italy in 1877 provoked riots in southern Italy, in the course of which schoolhouses were burned to the ground.15 Much of the literature on the Italian immigrants to the United States mentions various indicators of their low esteem for formal schooling.16 Moreover, the peak of the Italian and the Jewish immigrations coincided in time (late nineteenth and early twentieth centuries), so that they would have been present in the immigrant ghettos at the same time and their assimilation processes would not have been out of phase, as would those of the Irish, for example. The Italians and the / Jews would then clearly be among the candidates for testing whether nonsegregated ethnics have had as substantial differences in their performances in the same school as segregated races have had in different schools. Another pair who would have been in the same schools at the same time and at a similar phase of their assimilation would be the Japanese Americans and Mexican Americans in the West. They also have many cultural contrasts, well documented in the literature. For example, over half of all Mexican American women get married in their teens, while only 10 percent of Japanese American women get married that young.17 The implications of this for higher education and subsequent careers is obviously important. Here we are concerned about the differences in cultural values implied.
The nationwide sample of fifty-eight schools was searched to find schools in which students from either of these pairs (Italian-Jewish or Japanese-Mexican) were present, with at least ten students per year from each group for at least five years. Only two sets of data in the nationwide sample proved to meet these specificationsone for each ethnic pair. One was a school (coded 0918) in a northeastern metropolis with a population of more than one million. An unexpected bonus was that the same school also turned out to contain a significant Puerto Rican population during the same years as the Italian and Jewish populations, and their data are included in Table 2 to extend the intergroup comparisons. The other data (coded 0610) are for a unified school district in a western community with fewer than 20,000 inhabitants. The data for the nationwide survey were collected under a pledge of confidentiality, so no institutional identification is possible.
How do these data bear on our hypothesis concerning the effect of segregation on education? Table 2 shows substantial and persistent IQ differences between nonsegregated ethnic groupsdifferences comparable in magnitude to national black-white differences. As already noted, the national black-white IQ difference is 15 points. In the segregated southern schools, black-white IQ differences have been slightly greateralmost 20 points.18 Over a period of more than twenty years, Jewish youngsters averaged 13 points higher IQs than Italian youngsters attending the same school. Indeed, the highest Italian IQ average for any of these years is lower than the lowest Jewish IQ average for any of these years. Puerto Rican IQs averaged 26 points lower than the Jewish IQ. In short, the Jewish-Italian IQ difference was almost the same as the national black-white IQ differences, and the Jewish-Puerto Rican IQ difference was even greater than black-white IQ difference in segregated southern schools. These huge disparities existed among children living in the same neighborhoods and sitting side-by-side in the same classrooms.
The data coded 0610 show a similar pattern of difference between Japanese Americans and Mexican Americans (see Table 3). Although these were school district data, rather than individual school data, children of Japanese and Mexican ancestry were not separated from each other in different schools, either de jure or de facto. These data include parental occupation, which was unavailable in school 0918, and these occupations indicate that the large IQ differences between Japanese and Mexican children were not due to "middle class" occupations or incomes of the Japanese parents during the 1950s, when data were available for both groups. The parents of the high-IQ Japanese Americans were generally "unskilled and semiskilled" workers to an even greater extent than the parents of low-IQ Mexican American youngsters. Only 2 percent of the parents of either group of children in this school were white-collar workers, and another 2 percent had skilled or supervisory occupations. Nevertheless, the average IQ differences between these Japanese American and Mexican American school children ranged from 16 to 30 points for the six years for which data are available for both-averaging 20 points difference for the whole period. Again, this is larger than national black-white IQ differences, and about the same as the IQ difference between southern blacks and whites in racially segregated schools.
INTERGROUP DIFFERENCES AMONG SEGREGATED BLACKS
If racial segregation is not the crucial determinant of disparate educational performances that it has been assumed to be, we should expect to find performance disparities among all-black schools comparable in magnitude to black-white disparities. The same survey of fifty-eight schools shows the two highest IQ and two lowest IQ all-black schools to differ by more than 20 pointsthat is, by more than the IQ differences found in racially segregated schools in the South. Nor are the two high-IQ black schools unique. Similar IQ levels have been found in other all-black schools surveyed elsewhere.19 The low-IQ schools are likewise not unique. Just among the all-black schools in this study, there were five more with IQs at least 20 points below the IQ level of school 0508/0598 (see Table 4).20
The school coded 0508/0598 is a public high school in a northeastern city of between half a million and one million population. (Because it was handpicked for special study, its data are not included in the published national data for blacks in the IQ study already mentioned.) The school coded 3169 is a private elementary school in the South, in a city of between 100,000 and 500,000 population. The schools coded 2756 and 2757 are both public schools in the same southern town of less than 20,000 population. These various schools are not alleged to be comparable in any way other than being all-black schools. No doubt there are many reasons for their large IQ differences. Segregation is not among them.
SEGREGATED NONBLACK ETHNIC GROUPS
Blacks have not been the only group segregated even de jure, much less de facto. Other groups have attended schools whose student bodies have consisted exclusively, or almost exclusively, of members of their own ethnic group. In our nationwide IQ sample, there were two school codes that were more than 95 percent Chinese, one that was 95 percent Puerto Rican, one that was 100 percent American Indian, and one that was more than 99 percent Mexican American, Did these segregated school children of various ethnicities have test performances inferior to their respective compatriots in nonsegregated schools? Let us consider these groups one by one.
The above-average IQ of the more than 1,500 segregated Chinese school children in our sample (Table 5) is in complete contradiction to the Brown v. Board of Education doctrine that separate schools are inherently inferior. The IQs of the segregated Chinese are certainly no lower than the IQs of Chinese Americans in general nor Americans in general.
The segregated Chinese children in our sample come from four schools in the same city (population between half a million and one million) in the West-three public elementary schools coded together as 0115 and one private school coded 0106 (see Table 5). Like most other private schools in our sample, it is not located in an affluent area. Such schools, in this study at least, are much more likely to be Catholic parochial schools than the academically selective and socially exclusive schools conjured up as the image of "private" schools. The median family income of the census tract in which school 0106 was located was about half the national average in the 1950 census, and just under a third of the national average in the 1960 and 1970 censuses. Such income disparities have been common in Chinatown neighborhoods, even after the Chinese rose above the national average in income, for the more affluent Chinese tend to live away from Chinatowns.
Our Puerto Rican school sample was only one school (0921) and for only one year (1952). The mean IQ there was 81-compared with 79 for Puerto Ricans nationally in the same sample for the decade of the 1950s.21 Although the IQs here are below the national average, the all-Puerto Rican school did not have results inferior to those of Puerto Ricans scattered through other schools, which is the point at issue. As might be expected, this all-Puerto Rican public school was in a low-income urban neighborhood. The median family income in its census tract in 1950 was just under half the national average. The city was in the Northeast, and had a population of more than one million.
Our American Indian sample was also for only one school (and only one tribe), but included data for a number of years (see Table 6). For this tribe of Indians, at least, an all-Indian school did not mean an inferior performance. Their IQs were consistently just above the national norms. They were also just above the national IQ level found among American Indians in this survey (106)-not significantly above, but not below, which is what is relevant to the Brown v. Board of Education doctrine. School 1016 was a public school located in the Northeast in a community of fewer than 20,000 inhabitants. No income data were available, and 90 percent of the parents' occupations were unknown. This one school provided 86 percent of all the American Indian IQs found in our nationwide survey, so a comparison among Indians would mean somewhat less than otherwise, and no conclusions about the national IQ level of Indians in general are drawn from them here, nor were any of these data presented in the published IQ study referred to,22 because of the narrowness of the sample. But for our more limited purpose here, this sample is one more piece of evidence at odds with the doctrine that separate is inferior.
With Mexican Americans as well, the evidence goes directly counter to the doctrine that segregation reduces the group's performance. The Mexican American mean IQ of 90 in school 3167 is below the national average, but above the average of Mexican Americans nationally in the same study (see Table 7). No income data are available for school 3167, a public school in the West, in a community of between 20,000 and 50,000 inhabitants. Parental occupations were slightly higher than for Mexican Americans nationally. While only 2 percent of the parents were professionals and 4 percent small businessmen, about a quarter were skilled or supervisory workers. However, the mean IQ of the children whose parents fell into the "small business" or the "skilled or supervisory" category was the same as the school average, and the children of the "unskilled and semiskilled" were only one point below. The children whose parents were professionals had mean IQs of 102, but because they were only 2 percent of the student body, this could hardly explain the school's IQ level. For this 99-percent Mexican school at least, economic status seems to have had as little overall effect as segregation.
II. SOCIOECONOMIC VARIABLES
We are well familiar with the fact that the social class and economic position of the parents affects the school performance of the child, and that the quantity and quality of schooling affects the eventual socioeconomic position of the child. Perhaps we are too well acquainted with these relationships, and therefore too inclined to be deterministic in our thinking about education.
One of the problems in trying to disentangle the effects of family background on school performance is that so many of the relevant variables vary together. High-income parents tend to live in high-income neighborhoods with higher quality schools and have homes where books, magazines, conversation, and child-rearing patterns all enhance the development of the child's intellectual potential. Sorting out which of these factors is most responsible is like trying to unscramble an egg.
A more manageable analysis may be possible when dealing with historical data. At particular times in history, there have been groups with the values and aspirations that go with good educational performance, but that had not yet acquired the parental education, incomes, or occupations considered "middle class." Have their children's school performances matched their parents' socioeconomic realities or their parents aspirations and pressures?
Two of the classic cases of groups whose history in America began in poverty and ended in affluence are the Jews on the East Coast and the Japanese on the West Coast.23 We have already seen, from the record of Japanese American children in one school district (0610), that parental occupations overwhelmingly in the unskilled and semiskilled category did not prevent their offspring from meeting or exceeding national test norms. For our nationwide sample as well (see Appendix), Japanese American youngsters whose parents' occupations were low skilled still had IQs of 102 for the one decade (1950-1959) for which we have a national sample size of 50 or more. The national sample of Jewish school children likewise shows that those whose parents were unskilled and semiskilled had mean IQs of 104 through 106 for the three decades for which sufficient data are available.
While our historical data (see Appendix) for ten ethnic groups generally shows a higher IQ for the children of professionals than for the children of low-skilled workers, it is difficult to explain the substantial intergroup differences by parental occupations. For example, the children of Irish, German, or Chinese low-skilled workers scored consistently higher than the children of Mexican white-collar workers, small businessmen, or skilled and supervisory personnel. This is not to say that parental background meant nothing and the school everything. Clearly the school is not everything, for we have already seen (Tables 2 and 3) different ethnic groups with very different test performances in the same schools. The point here is simply that the kinds of parental backgrounds we traditionally measureoccupations or incomeseem to explain little. No doubt the values and attitudes of the parents meant much. That, however, does not mean that good schooling requires parental "participation" in school decisions. Neither the immigrant Jewish parents nor the immigrant Japanese parents participated in school decisionsexcept to back up whatever the teacher said.24 The same was true of the role of the parents in high-quality black schools.25
What of the effectiveness of schools? Do they make little difference, as many studies suggest? Before determining the potential effectiveness of school differences, we must first have some idea of how different the schools are in the first place. For example, if most ghetto public schools differ little among themselves in the first place, we should expect little difference in their pupils performancesbut this in no way indicates that it is futile to expect them to make a real difference, if they were to be themselves very different from what they are, or even to vary more than they do. A growing literature has shown numerous ghetto schools with good-to-high academic performance,26 often among children either wholly unselected, or differing in no demonstrable socioeconomic way from other ghetto children. The local example I studied was P.S. 91 in Brooklyn, where over half the children were eligible for the free lunch program, and where many came from families on welfare. Yet whole grades were reading above the national norms, and the school's performance was far above the performances of the other schools in the same school district (see Table 8).
Parental occupational data from a number of high-performance black schools (see Table 9) reinforce the point that socioeconomic background is no more of an insurmountable handicap to good school performance among blacks than among other groups. These are children whose parents' occupations fall in the "unskilled and semiskilled" category. The children's IQs around the national norm are all the more significant because three of the schools are in the South as commonly understood, and all four are in the South as defined by the census. As noted in Section I, the average IQ of southern blacks has usually been around 80. These four schools could all be considered "selective, " in that the individual student had to choose to attend them, unlike P.S. 91, which is a neighborhood school to which students were automatically assigned. However, they are by no means selective in the sense in which Andover or Exeter is selective. None had cut-off scores that students had to meet to gain admission, all admitted substantial numbers of students with IQs in the 80s or below, and one (0508/0598) enrolled at least one-third of all the black students in its city during most of the period covered by our data.
"Selective," like "private," is a label that must be used cautiously. Parental occupational data for P.S. 91a nonselective schoolwould have been more appropriate, but was unavailable. However, by all available indicators of socioeconomic status, those occupations are very unlikely to be higher than those for the schools shown in Table 9.
The point here is not to claim that schools can achieve success in spite of parents, or even to apportion the blame for low-performance ghetto schools between the parents and the schools. There is plenty of blame to go around. Instead, the point is to determine which of the many factors that differ among groups make no major demonstrable difference in results, so that we can concentrate attention on those that do. Clearly schools cannot do it alone, as shown by the historical examples of groups with vast performance difference in the same schools. But clearly, too, poverty and low parental status are not fatal either.
Nothing is easier than to put together a list of desired things and call them the "prerequisites" for quality education, and to blame the absence of these things for all present shortcomings. But, historically, many of these prerequisites have been missing in high-performance schools, as well as low performance schools. Adequate physical plant, small class size, parental participation in the school, teacher "role models" of the student's own ethnic background, bilingual educationthese are among the many prerequisites almost universally lacking in the schools from which the children of Jewish or Japanese immigrants emerged, and in the high-quality black schools I have studied.
My study of high-performance black schools repeatedly took me into ancient school buildingsincluding P.S. 91 in Brooklyn, where there were still gas jets in the halls, from the era before there was electric lighting. Others have found good academic performances in "storefront" ghetto schools. At least one of the schools in my study had lacked central heating for most of its history, and relied on pot-bellied stoves in each classroom. This was neither desirable nor morally right. But it did not prevent good education from taking place.
Historically, the most outstanding of all black schools-whether measured by IQ or by the later achievements of its alumniwas Dunbar High School in Washington, in the period from 1870 to 1955. In the nineteenth century, the average class size at Dunbar was over 40, and it continued to have the largest class size of any high school in Washington on into the twentieth century. In New York at the turn of the century, it was not unusual for Jewish school children to be in classes of 60 or more pupils.27 And there were no teacher's aides.
Teacher role models of the student's own ethnic background have been equally rare in history. When the Irish Catholic immigrant children went to public school in the middle of the nineteenth century, they were likely to be taught by Protestant Anglo-Saxon teachers. By the time the Jewish immigrant children were flooding into the public schools of New York at the turn of the century, they were far more likely to be taught by Irish Catholic teachers than by Jewish teachers. By the time I went to school in Harlem in the early 1940s, Jewish teachers outnumbered black teachers many times over. Similar patterns of group succession existed in labor unions and other organizations. All that is different about today is our naive insistence that all statistical disparities are unusual and/or evidence of sinister designs. Among the high performance black schools in my study for The Public Interest, some had all black teaching staffs, some all-white, some predominantly clerical, some exclusively laymen. Some of the principals were warm and friendly, others tough and blunt. The notion that there is one formula that can be applied simply does not fit the facts, either as regards schools or homes. Much has been made of the home well stocked with books and magazines, and enriched with a continuous stream of conversation between parents and children. This model fits the history of Jewish immigrants on the East Coast, but not the history of Japanese immigrants on the West Coast. Books were rare in their homes and conversation between the generations infrequent and usually unilateral.28 But the Japanese children were the delight of their teachers,29 even in an era of anti-Japanese sentiment in the society at large, and their educational performances demonstrated the payoff to diligent application, even in the absence of other prerequisites of good education.
Among the many fashionable prerequisites for good minority education (or excuses for its absence) is bilingual schooling. As in so many other areas of social policy discussion, this expression defines a process not by its own characteristics, but by its hopes. Described as a process, so-called bilingual education in practice means that a certain portion of the school work (sometimes a majority or virtually all) is taught in the student's native language, rather than in English. Whether or not that leads to bilingual students is a question, not a foregone conclusion. History does show that preceding generations of studentsfrom the Jews and Italians on the East Coast to the Chinese and Japanese on the West Coastbecame bilingual through opposite policies: teaching exclusively in English in school and speaking their respective native tongues at home.
The argument here is not over whether decent buildings, smaller class sizes. and so forth, are desirable but over whether they are prerequisites-or (more to the point) whether their absence provides blanket excuses for educational failure.
The effect of the school on socioeconomic mobility has become as deeply embedded in the folklore as the effect of socioeconomic variables on the schools. The strength of this belief is demonstrated, not by the evidence marshaled to support it, but by the lack of any felt necessity to produce evidence. Most prosperous groups are of course well-educatedand usually well-housed, well-clothed, and well-entertained. Yet no one regards that as proof that housing, clothing, or entertainment produces economic advancement. History again permits us to see various ethnic groups before they became prosperous, to find out which came first, the chicken or the egg.
It is by now a familiar story how the immigrant Jews brought to New York a long tradition of reverence for learning, how they crowded into the public libraries, the lecture halls, and the free colleges and universities of the city.30 What is not so familiar is that the occupations through which they rose out of poverty were not primarily occupations requiring or utilizing formal schooling. In 1880 and in 1905, over half the Russian Jews in New York worked in manual occupationsand that is not counting the great number who were pushcart peddlers, who were classified as "white-collar" workers. Only 2 percent of the Russian Jews in 1880 were clerks or semiprofessionals31the kinds of occupations normally thought of as "white collar." After a quarter of a century of upward mobility, there was a smaller proportion of Russian Jews in lower white-collar jobs, because pushcart peddling had declined.32 But although Jewish upward mobility was well under way by the early twentieth century, Jewish children at the turn of the century seldom went far in school, because they were working-at home or outside. In 1908, 38 percent of all clothing workers in New York City were teenage Jews.33 Even among Jewish children under ten years of age, between 7 and 10 percent were out of school and working at home34in the "sweatshops."
The upward mobility of Jews in the last two decades of the nineteenth century could hardly have been due to education, and certainly not to higher education. No Jewish youngster graduated from a New York City public high school in the nineteenth century, for the first graduating class for any New York City public high school was the class of 1902.35 A 1951 survey of City College students (mostly Jewish) showed that only 17 percent of their fathers born before 1911 had completed the eighth grade.36 The massive Eastern European influx into City College and Hunter College occurred later, in the 1920s and 1930s.37 By then their parents could afford to support them through high school.
The experience of the Japanese immigrants on the West Coast demonstrates even more dramatically that upward mobility occurred first, and the sending of the next generation to college was the consequencenot the causeof that socioeconomic rise. As late as 1940, a majority of Japanese American males were in farmingand they produced about a third of all the commercial truck crops grown in California.38 They were also successful as small businessmen. As early as 1919, Japanese Americans owned almost half the hotels and one-fourth of the grocery stores in Seattle.39 First-generation Japanese also owned hundreds of produce markets in Los Angeles,40 and more than a thousand were contract gardeners in southern California.41 None of this required education, and most of the first-generation Japanese immigrants spoke little English.42 Education was one of the things the next generation acquired with its affluencenot the cause of it.
The dire state of American public education has been well documented, as is the fact that it has generally been getting worse, rather than better, over the past decade or more. As public trustand public moneythreaten to ebb away, many panicky "explanations" of failure or "prerequisites" for success have emerged. The quarter-of-a-century-old crusade for racial integration overlaps and complicates the educational picture, and generates its own explanations and prerequisites.
What is "the" answer? It is not clear that there is any single answer. The constant grasping for one answer or some magic formula or "innovation" may itself be part of the problem. What should be clear, however, is that self-serving mythology is not the answer. We are never going to solve the problem unless we can first face the problem as it is-not with a long list of excuses, pious hopes, or open-ended demands for "more."
More generally and more fundamentally, how have we all managed for decades to repeat dogmas and mount crusades without first testing our facts? Much of the discussion of the educational effects of segregation, or of socioeconomic status, has proceeded as if these were irrefutable facts rather than fashionable assumptions and moralistic pronouncements. They were never facts but only what the great sociologist Robert Merton has called "pseudo facts." The problem with pseudo facts, as Merton warns, is that they "have a way of inducing pseudoproblems which cannot be solved because matters are not as they purport to be."43 The problems of education, and particularly of minority education, are all too real. But what we are trying to solve are the pseudoproblemshow to extract more government money for more projects, studies, and "innovative" gimmicks. The taxpayers' revolt, symbolized by spending limitations initiatives (beginning with California's celebrated Proposition 13) and by the defeat of school bond issues that used to pass routinely, suggests that the pseudoproblem may not be solved for many years, if at all. The real problemteaching youngsters to read and thinkhas already been solved by many institutions in both rich and poor neighborhoods, by people with varying ethnicities, personalities, and approaches. The real problem can be solved, but it must first be addressed.
A strong word of caution is necessary before presenting the nationwide IQ data that follow. Unlike the coded school data tabulated from the same study, the data in the following tables do not represent ethnic performance differences in the same schools on the same tests, or even in the same neighborhoods or geographic regions of the country. Intergroup comparisons from the national tables are therefore very hazardous, for many relevant variablessocioeconomic status, test type, geographic mixare not held constant from one group to another though they could be, if someone wished to retabulate the data from the computer tape. However, the tables may be useful in a variety of other ways, including showing patterns of internal IQ differences (by parental occupation) for a given group, and following the changes that have occurred in that group's scores over time. Other uses and more detailed breakdowns of these data (by sex, test types, etc.) may be found in Thomas Sowell, "Race and I.Q. Reconsidered," in Essays and Data on American Ethnic Groups, ed. Thomas Sowell (Washington, D.C.: The Urban Institute, 1978), pp. 213-238; and Leon J. Kamin, "Sibling IQ Correlations among Ethnic Groups," ibid., pp. 239-49. The raw data itself, including information not included in any of the published studies (foreign-born parents, for example) and a more detailed description of the data categories and collection methods are all available from National Technical Information Service, U.S. Department of Commerce, Springfield, VA 22161 (Accession Number PB265 8.13).
In the tables that follow, a dash is entered in any cell for which the national IQ sample for the whole group is less than 50, or in which the sample is less than 10 for any given parental occupation. An asterisk indicates that, while the sample in the occupational cell was 10 or more, this was less than one-half of 1 percent of the parental occupations for that decade.