Morton Clurman here reports on a recent large-scale study of prejudice which was undertaken by the Roper organization to determine whether colleges discriminate against Jewish applicants. Mr. Clurman tries to sum up the meaning of the Roper findings, at the same time taking into account criticisms by other authorities.
_____________
Among the problems of our age, that of prejudice is not the least complex. In its action and reactions, a web is woven in which we are all caught. Like it or not, where a set of attitudes toward any group exists, the very awareness of this, regardless of how we ourselves react, makes it impossible to be unprejudiced in the literal sense of the word—that is, to think without pre-judging. The awkwardness of the Gentile telling an acquaintance whom he has just discovered to be a Jew that some of his best friends are Jews—and really meaning it—or the white who is painfully aware of the Negro’s suspicion explaining that he really has no special “feeling” about Negroes—and believing he means it—merely indicates we are all subject, one way or another, to prejudice.
The psychological complications of prejudice are virtually infinite: ranging from the self-deception and rationalizations of the “superior” group to the super-sensitivity (one recalls the joke about the stutterer telling a friend that he was turned down for a job as a radio announcer because of anti-Semitism) and self-hatred of the “inferior” group. These complications make accurate measurements of prejudice extremely difficult. Yet we keep trying to measure it because we must.
The measurement of discrimination, the effect, as distinct from prejudice, the cause, is not quite so hopeless. Prejudice is an attitude, as intangible as most attitudes. But discrimination, after all, is an act that can be observed, reported, even legislated against.
Yet since discrimination means an act motivated by prejudice, and since acts that appear to be very similar can have different causes, the problem of locating discrimination ultimately comes down to that of locating motive. And that, as Dr. Watson knew, is not always easy.
_____________
Even the case mentioned of the Jewish stutterer who wanted to be a radio announcer is not quite so simple as it seems. It is possible, after all, that he was turned down because he was Jewish. For example, he may not have had an interview, or having had one, may not have stuttered during it. The obvious point of the joke, however, is the rather slim chance that discrimination was involved.
There are several ways in which this hypothetical case of the stuttering applicant could be cleared up. A purely “scientific” technique might be to proceed from the assumption that nothing was known about anti-Semitism, stutterers, or radio announcers, and to do a survey of present employees of the radio station. If a large proportion of non-stuttering Jewish announcers were found, the probability of discrimination would be low. If none, or almost none of the announcers was Jewish, the probability of discrimination would be higher. If many of them were non-Jewish stutterers, it would be higher still. But if the station already employed many Jewish stutterers as radio announcers . . . But enough—that way madness lies. Anyhow, all these data could be run through IBM machines on punch cards and the probabilities of discrimination in this particular case computed to any number of decimal places.
A less elaborate method might be to assume that something was known about stutterers, anti-Semitism, and radio announcers, and to ask the interviewer why the applicant was not hired. If he said “stuttering,” the case against discrimination could be considered proven.
To the stupefied bystander, the second method would seem to have obvious advantages. But social scientists do not always see things that way. Instead, a minimum number of reasonable assumptions, and a maximum use of IBM machines, involved statistical procedures, and lengthy questionnaires seem to be prescribed, even where less devious methods would appear to be adequate.
_____________
A cash in point is a recent monumental study of discrimination in college admissions sponsored by the American Council on Education and conducted by the Elmo Roper organization under the direction of Julian Woodward. To determine the incidence of discrimination, a nationwide sample of more than 10,000 white high school seniors were personally interviewed in May 1947. Negroes were omitted from interviews because there were too few Negro seniors in Northern high schools to furnish a statistically reliable sample by this method, and the segregated Southern schools clearly presented a special problem.
In addition to this nationwide sampling of 10,000 seniors—of whom 3,500 were applicants for college admission—a second “big city” sample of almost 5,500 seniors in cities of half a million population or over were interviewed, of whom 2,338 were college applicants.
Obviously, there were bound to be difficulties in this approach, and the Roper organization ran into more than their share of them. The assumption, undoubtedly correct, was that many factors influenced the admission of students to colleges. But the larger the number of factors assumed to have some causal connection with a single event—in this case admission to college—the more difficult it becomes to determine the relative weight of each factor. If our hypothetical would-be radio announcer also had a chronic smoker’s cough, were illiterate, and couldn’t speak English, it might be very difficult indeed to discover the exact weight of each disability in his rejection.
In the Roper study matters are further complicated by the interdependence of many of the factors. For example, it was found that many colleges preferred the sons and daughters of alumni. Yet most parents of Jewish high school seniors were foreign-born and had had little formal education when compared to the Protestant parents. How evaluate this element in assaying “discrimination” against Jews?
Or—to take another example—it was found that many colleges showed a definite preference for applicants from their own state. A much higher proportion of Protestants than of either Jews or Catholics applied to out-of-state colleges in the Northeastern states. How important was this factor in reducing apparent discrimination against Jews and Catholics?
Theoretically, of course, there is a way around these difficulties. You isolate one factor after another, continually whittling down your sample, until finally you come up with the pure distillation of, say, Jewish seniors whose fathers and mothers were native-born college graduates, who applied to out-of-state colleges, who were ex-GI’s in the highest fifth of their respective senior classes, and who were also leaders in extracurricular activities. Then, from this pure residue, you should be able to determine the incidence of discrimination against Jewish applicants. The practical difficulty, however, is that of your original sample of 10,000 seniors, by the time you complete your distillation you may be left with something like 26 cases of this type, a number that is, naturally, “statistically unreliable.”
_____________
The Roper people were not unaware of these problems. Indeed, ample verbal obeisance is made to them. The Report is filled with warnings against drawing unwarranted conclusions from “statistically unreliable” samples, against assuming that all the possible factors had been accounted for and correctly weighted, and against closing the door to the possibility of other interpretations of the data.
Yet these caveats, far from helping matters, actually confront the reader with a real dilemma. If he takes them seriously, he must then decide how to weight them, while knowing that whatever substance he does attach subtracts in equal measure from the weight of the original conclusions. But if he ignores these warnings, he opens himself to the charge of accepting part of the package in place of the whole.
A case in point is a conclusion, reached in the Report, on discrimination against Jewish seniors in the “big city” sample. It was found that there was a “statistically significant” difference between the admission rate of Jewish and non-Jewish seniors only with regard to top-fifth Jewish students with a high record of extracurricular activity who were native-born and whose parents were college graduates and members of the executive and professional classes. To account for this surprising fact the Roper Report suggests that “it is possible that they are the ones who have the temerity to apply to the colleges who do discriminate and thus acquire a low batting average, while the Jews with less qualifications content themselves with applying to colleges where discrimination does not take place.” Perhaps—and perhaps not. In any case, this hypothesis has a tremendous bearing on the whole study. Yet the reader must fend for himself.
_____________
Perhaps the most weighty criticism of this method of investigating discrimination has been provided by Professor Frank Shuttleworth of City College of New York. Professor Shuttleworth, who advises the predominantly Jewish students of City College on their future careers, was concerned lest they begin to think that their chances of getting into good colleges were worse than they actually were—and they already thought them bad enough. For almost three years he has been conducting what amounts to a running battle with the Roper study—a battle recorded in a hot and protracted correspondence between himself and the American Council on Education and, finally, two magazine articles from his pen.
The main point of Professor Shuttleworth’s criticisms is that the Report has not proved what it says it has: namely, the existence of discrimination on the part of colleges against Jewish seniors in the Northeastern United States and in American big cities. Instead, he maintains that a correct analysis of the Roper data would show that Jews, far from being discriminated against, are given equal and perhaps even preferential treatment. Actually, Professor Shuttleworth does not believe that Jews are given preferential treatment. He believes that the Roper data must be incorrect.
In Shuttleworth’s opinion, two major errors, plus a host of minor ones, are responsible for the erroneous Roper conclusions. The first major one is the failure to make allowance for the factor of multiple applications. Jewish seniors, it seems, are much more eager to get into college than their non-Jewish classmates, and the Jewish applicants average 2.2 applications each as contrasted to 1.3 applications for Protestants and 1.4 for Catholics. Put another way, only 22 per cent of Protestants applied to more than one college, as against 62 per cent of Jewish applicants.
As a result of this disparity between applicants and applications, it turns out that 87 per cent of Jewish applicants are accepted in some college compared to 88 per cent of Protestant and 81 per cent of Catholic, but that only 56 per cent of Jewish applications are accepted compared to 77 per cent of Protestant and 67 per cent of Catholic.
Professor Shuttleworth’s point is that the Roper study wrongly stresses the relative unsuccess of Jewish applications and ignores the great success of Jewish applicants. To this, the Roper partisans reply that the colleges must necessarily treat each application as an individual, and that the action they take on applications is the best index of their attitude toward individuals. Professor Shuttleworth’s rejoinder is that colleges know of the other applications that multiple applicants make, and of their order of preference among colleges, from the reports of the College Entrance Examination Board; and that almost invariably, colleges with enough well-qualified first-choice applications tend to give little consideration to second-, third-, or lower-choice applications. Thus any group that makes multiple applications would tend to show less success per application irrespective of discrimination.
Social scientists are not a tongue-tied group and the Roper people are not left at a loss by this last point. They wave their data sheets showing that 82 per cent of the first-choice applications of Protestants were accepted compared to 71 per cent of Catholic and 63 per cent of Jewish first-choice applications. Whereupon Shuttleworth answers with a table he has prepared based on the Roper data which shows that when Jewish and non-Jewish groups filing equal numbers of applications are matched in their respective academic quintiles, the Jews do as well or better in most quintiles than the non-Jews in getting into college. To this, the Roper partisans reply that comparing Jews and non-Jews by number of applications within quintiles is invalid because it neglects the possibility that Jews making only one application may do so because their academic standing within the top quintile virtually assures them of acceptance anywhere. And Shuttleworth comes back with an analysis of the Roper data showing that such is not the case.
This gives in barest outline the differences between Professor Shuttleworth and the Roper people on just one point in their Report. Clearly, these things can get complicated.
_____________
Shuttleworth’s second major point is that the Roper figures are distorted by a failure to take into account the different admission rates of the colleges to which Jews and non-Jews apply. Thus the Roper data on the national sample show that in the upper quintile 66 per cent of Protestant applications to privately controlled colleges are accepted compared to 52 per cent of Jewish applications. However, Professor Shuttleworth points out, almost half the Protestant applications to privately controlled colleges are made to denominational schools which are known to have a very high acceptance rate, while over 90 per cent of the Jewish applications go to non-sectarian schools with a much lower rate of acceptance. He then proceeds to show that if the acceptance rate for denominational schools is assumed to be about 82 per cent and that for non-sectarian schools about 50 per cent (reasonable assumptions in the light of known admissions policies), the disparity between Protestant and Jewish acceptances is accounted for without involving discrimination at all.
This would appear to be a very telling point, but it is weakened somewhat by the fact that the Roper Report stresses that discrimination is most evident in colleges outside the home community of the applicant. And if applications to home-city schools are excluded, it turns out that the disparity between Protestant and Jewish success is not 66 per cent to 52 per cent but about 64 per cent to 30 per cent. That gap cannot be as easily explained away by Professor Shuttleworth’s analysis.
There is no doubt that many of his criticisms are valid. But the more important question is not the validity of one or another of his points—verdicts over which statisticians can debate without end—but the fact that the Roper Report seems to lend itself so easily to this type of negative analysis. The reasons are not hard to find. Instead of going as directly as possible to the heart of the problem—discrimination—and trying to find out where it exists and how much of it there is, an attempt is made to deduce and distill it from a huge mass of data that is assumed to be relevant. And of course, it always turns out that some conceivably relevant data have been overlooked or wrongly weighted.
Ironically, the summary report and analysis of the Roper study issued by the American Council of Education itself makes the most melancholy admission of all when it states “we have seen that the samples used in this study have not been quite large enough to handle the complexity of the situation disclosed. The national picture is seriously obscured in the tabulations, by reason of the attempt to accommodate too many details.”
In other words, what is required is a bigger, more complex study to account for absolutely all the details anybody can think of—and then we shall see what we have.
_____________
Yet as Professor Shuttleworth himself points out, without however seeming to grasp the implications of his observation, far simpler and more effective studies of discrimination in college admissions have been made.
In 1946, for example, a study was done by David Berkowitz for a Legislative Commission on Discrimination in Higher Education in New York State. Mr. Berkowitz simply examined all the applications to fourteen liberal arts colleges in upstate New York. He found that five did not discriminate against Jewish applicants at all. Five others discriminated somewhat, accepting two qualified non-Jews to every Jew. The remaining four institutions discriminated sharply on a four-to-one ratio.
Although most Jewish applicants came from New York City, the possibility that what seemed like discrimination might simply be the normal workings of geographical preferences was excluded by an analysis which showed that upstate and out-of-state Jewish applicants were even less favored than those from New York City. A further analysis showed that academic standing could not be a factor since Jewish and non-Jewish applicants had virtually identical qualifications, on an average, in this respect.
Although Shuttleworth praises the Berkowitz study and contrasts it with the Roper Report, he criticizes it for failing to take two factors into account: multiple applications and preferences for children of alumni. These omissions, he believes, increased the amount of discrimination found by Berkowitz, although he agrees that even after making generous allowance for these factors the major conclusions of the Berkowitz study would hold.
Aside from Professor Shuttleworth’s criticisms, which would appear to be valid, the great virtue of the Berkowitz study is that it tells us something we didn’t know before about discrimination, namely how much there was in 1946 in upstate New York and where it existed. The great defect of the Roper study is that it obscures these facts instead of pin-pointing them. The only conclusion one seems justified in drawing from the latter is that in 1947, while there was undoubtedly discrimination in college admissions throughout the country, there was much less than most people had imagined—a conclusion inherently probable but not very helpful.
Every trade in every age has its special delusions, and a major affliction of social science today might be called the IBM fallacy. This delusion reflects the endemic conviction of 20th-century man that machines can do everything for him—including thinking. In the case of the social scientist it takes the form of a certainty that if you feed enough data through enough electric circuits what you are looking for is bound to come out. The corollary of this hypothesis is the conviction that only a minimum of human cerebration need be combined with a maximum of electronics to produce miraculous results.
But little fallacies survive and multiply in the shelter of big ones. The belief in the omnipotent machine is supported by the idea that the social sciences can attain the same exactitude as the natural sciences, with help of the statistician and his machine. This idea, in turn, depends on a mistaken notion of how data are used in the natural sciences. Like the movie cowboy who is almost never seen herding cattle, the physicist or chemist is too often pictured collecting or processing data amid a jungle of laboratory gadgets, and seldom in that crucial act of scientific creation—thinking. The laboratory experiment, or natural observation, which are analogous to the collection and processing of data in the social sciences, are simply ways of verifying the scientist’s hypothesis. They cannot create a hypothesis, only confirm one. Where that hypothesis comes from, God may know, but certainly no one else does. Where it doesn’t come from, however, is a machine or any specific body of data. If it did, scientific creation would be possible for almost any high school boy.
_____________
Mr. Berkowitz was lucky. His hypothesis, the existence of discrimination, was given to him. All he had to do was to prove it in the simplest way possible. The Roper people, although primarily interested in discrimination, were grimly determined to study all “factors affecting the admission of high school seniors to college.” The result: 300 pages of tables in which the fact and the nature of discrimination are virtually buried. Instead of giving the undoubtedly true picture of many white spots interspersed with a few black ones concentrated in those states where the Jewish population is large, a kind of pale gray emerges in which the significant details are muted.
Yet there is a strange paradox about the Roper Report. While it does not contribute too much to its objective, which is reliable knowledge about discrimination, it does, by the sheer magic of accumulated information, permit various pictures to emerge of American high school seniors and their parents that are a valuable contribution to contemporary sociology.
For example, the Roper data strikingly confirms the widely held picture of a native white Protestant upper class, a Jewish middle class and a Catholic lower class. The Protestant seniors came from the best-educated and best—established families economically, with 68 per cent of their fathers in the business and executive classes. Almost half of these were in the very top occupational and income scale. Thirty-eight per cent of the Protestant fathers had been to college as had almost one-third of the Protestant mothers. And 78 per cent of the Protestant parents were native-born.
In comparison with the Protestants, the families of the Catholic seniors were grossly underprivileged. Only 17 per cent of the Catholic fathers and 9 per cent of their wives had been to college. One-fourth of the fathers were factory workers, while 18 per cent were in the service trades and 22 per cent had white-collar jobs. The Catholics also showed a surprisingly high percentage of foreign-born parents. In only 58 per cent of the Catholic cases were both parents native-born; and more than half the foreign-born Catholics came from Ireland or Italy.
In education and income, Jews held an intermediary position. While 69 per cent of the Jewish fathers were in business or the professions, compared with 68 per cent of the Protestants, more than half of the Jewish businesses were small. And the Jewish parents were only slightly better educated than the Catholics. Only 22 per cent of the fathers and 12 per cent of the mothers had gone beyond high school. Here, however, one of those complicating factors that make life miserable for social scientists intervenes. A very large proportion of Jewish seniors (more than twice as many as Catholics and Protestants) were unable to answer the question about their parents’ education. This uncertainty, as Roper points out, was probably due to their unfamiliarity with European education and their inability to equate it with American standards. For the single most striking characteristic of Jewish seniors is the fact that only one-quarter of them had two native-born parents. And three-quarters of the Jewish foreign-born parents came from Russia, Poland, or one of the Baltic states, none of which was particularly celebrated for the educational opportunities offered Jews.
The data also confirm the powerful influence of family background on educational incentive and opportunity. Of the seniors, almost three-quarters of the children of professional men and executives applied to college, compared with less than one-quarter of those whose parents were factory workers. Family had a similar influence on quintile rating. Thirty-five per cent of the children from professional and executive families made the top quintile of their senior class, while only 16 per cent of the factory workers’ children did.
_____________
How do the attitudes, abilities, and ambitions of Jewish seniors compare with those of their classmates’? In some respects the differences are striking, in others less than might be expected.
For example, if desire to go to college is any index, Jewish seniors are far more ambitious than their classmates. More than two-thirds of the Jewish seniors apply to college compared with slightly more than one-third of the Protestants and one-fourth of the Catholics. And, as we have seen above, once having decided to go to college, Jews file almost twice as many applications as do their non-Jewish classmates.
Jewish seniors also think more about the future, apparently. Thus, 93 per cent of the Jewish males named a specific career they had in mind compared with 90 per cent of the Catholics and 87 per cent of the Protestants. In three particular careers the difference in choice is significant. Engineering is chosen by more than 22 per cent of the non-Jews, less than 9 per cent of the Jews. Business, on the other hand, is chosen by almost 26 per cent of the Jewish seniors, compared to 14 per cent of the non-Jews. Medicine is chosen by 19 per cent of the Jews in contrast with less than 9 per cent of the non-Jews.
Some interesting differences appear also in extracurricular activities. Jews are more interested in student government, with 47 per cent having held school office compared to 32 per cent of non-Jews. But they are much poorer athletes, less than 35 per cent having been on a school team, compared with 60 per cent of the Protestant and 55 per cent of the Catholic seniors.
As to scholarship—there’s the big surprise. Jews are not as smart as they—and perhaps others—think they are. We have seen that the Berkowitz study of discrimination in New York State disclosed little difference in quintile grouping between Jews and non-Jews in New York. Roper’s national study does show a greater spread but it is not very great: 28:6 per cent of the Jewish seniors were in the top quintile of their class compared with 21.8 per cent of the Protestants and 18.5 per cent of the Catholics. It is possible, of course, that if the high schools were weighted by some yardstick such as College Entrance Examination records, the spread might be greater, for the high schools in middle-class urban neighborhoods, where most of the Jewish seniors were concentrated, tend to have somewhat higher standards.
One point above all emerges from the Berkowitz study. Every time an intelligent reporter like Samuel Lubell or a skilled investigator like Roper pokes into the myth of American uniformity, either in its radical, disenchanted version that shows us as robots in a mass-production civilization, or in its liberal, patriotic variant according to which we are as alike as peas in a pod despite differences of race, religion, etc., what is revealed is the most amazing diversity of attitudes and cultures, inhabiting as it were separate worlds and indicating an astonishing variegation in American life.
This variegation imposes an automatic caveat on die social scientist. The simpler and more direct the techniques and the surer the initial hypothesis, the greater the chances of success. Perhaps the miracle of the Roper Report is that, ignoring these precautions, and using an intricate mechanical approach to the problem of discrimination, it managed to uncover so much useful material about American life.
_____________
