Of all the tools utilized by the social sciences, public opinion polls have made the greatest impact on the American mind. Their amazing accuracy in the forecasting of election results, and their proved value to business and advertising in gauging consumer tastes and wants, suggested their use as an instrument for the study of prejudice, with special reference to anti-Semitism. Scientific polling is based on the simple principle that generalizations, within insignificant margins of error, can be made about a total population from information obtained about a representative sample of that population.
But there are two “ifs.” The most obvious is that the sample must be a true cross-section, a “random” sample. That there are no easy short-cuts in the painstaking procedure of drawing such a sample was decisively demonstrated by the debacle of the Literary Digest straw-poll in the 1936 election; the telephone subscribers who constituted the sample were too upper-bracket an income group to be representative of the American voting public.
To safeguard reliability, nation-wide polls in the United States now stratify their samples so that they are characteristic of the total population with respect to such factors as geographical and rural-urban distribution, color and economic status. Age and sex are often taken into account, but other variables such as religious affiliation, nationality origin, amount of education and occupation are usually automatically taken care of—it is hoped—by the process of random selection in obtaining interviews. The polling process itself consists usually in the asking of questions in personal, door-to-door interviews by trained interviewers. And, using these techniques upon samples of no more than 3,500 to 5,000 interviews, the outcome of the last three presidential elections has been predicted by major polling organizations within a 3 per cent margin of error.
_____________
Results of the Polls
We May assume that the major polling groups have, for all practical purposes, solved the problem of getting representative samples. Before going on to the second “if,” which is by no means as simple as the first, let us sum up what polls that have been taken to date claim to have found out about anti-Semitism.
Altogether, about twenty to thirty nationwide polls, conducted by such agencies as Elmo Roper’s Fortune Survey, Gallup’s American Institute of Public Opinion and the University of Colorado’s National Opinion Research Center, as well as private organizations who take polls to guide their own work, have used questions designed to reveal anti-Semitism.
Polls try to estimate the extent of anti-Semitism according to the proportions of respondents who make a presumed anti-Semitic response to a wide variety of questions. Some of these questions ask explicitly for agreement or disagreement with a specific statement about Jews, two perennial favorites being: Do you think that the Jews have too much power in the United States? and Do you believe that in this country hostility toward the Jewish people is growing or not? Other questions may require the respondent to name groups who in some way are a threat to the security of the country or the respondents: In your opinion, what religious, nationality or racial groups are a threat (menace) to America? or Are there any groups of people you think are trying to get ahead at the expense of people like you?
The “too much power” question has been used time and again by opinion-research organizations with some variations, “influence” being substituted for the word “power,” and with the kinds of influence, whether in business, government, finance, etc., specified. Whichever way this question is phrased, it always yields a considerable proportion—usually more than 50 per cent—of “yes” answers.
The “growing hostility” question is usually answered in the affirmative by about one-third of respondents, with Jews acknowledging to a somewhat greater extent than any other religious group their awareness of this trend.
The “what group” type of question—which leaves it to the respondent to identify the threatening groups—yields yet another quantitative picture of how the American people feel toward Jews. In all published poll results, the proportion of those spontaneously naming Jews has always been less than 10 per cent. In the latest Fortune Survey, published February 1946, 5.1 per cent of the respondents named Jews as a “group harmful to the country unless they are curbed” (note here the addition of “unless they are curbed”: it may reduce the percentage somewhat); 6.5 per cent designated Jews as people trying to “get ahead” at the expense of people “like themselves.” And 8.8 per cent of the respondents spontaneously gave “Jews” in answer to one or the other or both of these questions.
Obviously, a conclusive general statement about the extent of anti-Semitism cannot easily be derived from answers to different kinds of questions. Gordon W. Allport in Commonweal, October 6, I944, arrives at the conclusion that “five to ten per cent [of the population] are violently anti-Semitic, while perhaps 45 per cent more are mildly bigoted in the same direction.” Elmo Roper, in Fortune, ventures the opinion that “competent authorities agree that in the United States anti-Jewish feeling reached new heights as war approached. Now [February 1946], however, anti-Semitism’s growth in the United States appears to have been halted.”
But the following general trends show up rather consistently regardless of the type of question asked: Assuming that answers to these questions can be regarded as true indices of anti-Semitism then anti-Semitism is strongest in the Northeast and Middle West where Jews are found in greatest concentration, and weakest in the South and West; stronger in urban populations, particularly large cities, than in rural populations; stronger in upper-income brackets than in lower-income brackets; stronger among whites than among Negroes; and stronger among men than among women. And on the same assumption, anti-Semitism rose regularly up to Pearl Harbor, when there was a sharp drop; it has recovered somewhat, but has not returned to its pre-war peak.
_____________
Experiment and Life
This brings us to the second “if.” In any experiment, such as public opinion polls, the basic problem of prediction rests upon the extent to which performance under the testing procedure constitutes a valid estimate of performance in the real-life situation. When we use polls to predict election behavior—where we ask “Whom will you vote for?”—the experimental situation of the polling interview is very similar to the real-life situation of voting. But how closely does the attitude expressed in the experimental situation approximate the anti-Semitism of the real-life situation?
The polling interview (the experimental situation) usually takes place in a rational atmosphere, in stranger-to-stranger contact. In the real-life context, attitudes of prejudice are often manifested in an emotionally charged atmosphere, frequently in group relations where the responding individual has the opportunity to perceive the cues that mean group approval or group rejection of his behavior. In the real-life situation, anti-Semitic behavior is elicited by a constellation of stimuli rather than by a unique, isolated stimulus (the interviewer’s questions). Different individuals ex-hibit anti-Semitism under quite different sets of conditions, and not all individuals will respond to the same stimuli. One person may blame the Jew for everything that goes awry in the routine of his daily life; another will blame the Jew when aroused by a rabble-rouser. (This is the kind of behavior called “scapegoating,” because the individual who blames the Jew does so out of a psychological necessity to displace the aggression that arises from his own frustrations in life.)
We are not suggesting that, in order to increase the validity of polls for studying anti-Semitism, experimental situations should be deliberately constructed so as to approximate closely the emotionally charged and frustrating life experiences that evoke anti-Semitic responses. What does seem to be called for, however, is the formulation of polling questions that approach the stimuli of real-life situations.
An example may be cited to illustrate this problem: An non-Semitic farmer in an isolated rural farm district might be classified as non-anti-Semitic if he did not answer “Jews” in response to the question: Are there any groups of people you think are trying to get ahead at the expense of people like you? While he may hate Jews intensely he may not name them in answer to this question, because he may not have had to compete with Jews in his work. On the other hand, the person who agrees in the poll situation that everybody, even the Jew, ought to be treated alike, might well be the one who in an emotionally loaded real-life situation takes to violence.
It is here that the essential dilemma in using the results of polls to estimate the amount, the extent and the intensity of anti-Semitism, begins to be uncovered. Even if we formulated more valid verbal stimuli, we would still have to know—as we do not now—what anti-Semitism is. We have still to isolate and identify the major psychological factors that constitute anti-Semitism. For that, more intensive and different kinds of study are needed. We will have more to say about this later in our discussion.
_____________
Truth and Falsehood
An addditional possibility of serious error—also stemming from the disparity between experiment and real-life—arises when the respondent is asked to give some indication of his own probable behavior in a life situation that may never arise. (This, too, constitutes an important difference between testing prejudice and testing voting behavior.) A respondent may be asked: Would you actively support a campaign against Jews? or Would you vote for a candidate who has an anti-Semitic program?, or Would you be willing to deport all Jews? These examples of anticipated action are fortunately only rarely put to the test, and the respondent is usually well aware of this; and even should such a chance occur the respondent who answered “Yes” may be the last to go into action.
This is different from the often-raised question as to the truthfulness of respondents. Conscious lying has been found to be a negligible factor in polling pre-election behavior. This is probably because it is eminently respectable to belong to one of the two major political parties and to state openly one’s choice of a political candidate; these are quite acceptable forms of behavior in our culture. But in investigating prejudice one cannot discount the possibility of defensiveness, which is a “scientific” way of suggesting that respondents may lie to hide their real attitudes, especially when these attitudes run contrary to the American creed. (It is reasonable to suppose that at least some respondents are aware of the fact that anti-Semitism is supposed to be “un-American.”) There is some evidence which throws light on how important this .factor may be. W. Turnbull (in Cantril’s Gauging Public Opinion) asked two equivalent samples of about 300 each whether in their opinion the Jews have “too much power and influence in this country.” He used two methods: the interview, and the secret ballot. “Yes” answers were given by 56 per cent of those interviewed and by 66 per cent of those using the secret ballot.
Two further factors may militate against truthfulness. An individual may honestly believe that he is telling the truth yet be unaware of the existence and depth of a prejudice he has, since it may be like an “iceberg,” with its greatest part submerged in his unconscious. This was revealed in one of the studies conducted by the Department of Scientific Research of the American Jewish Committee in cooperation with psychotherapists, where cases were encountered of “liberal” individuals who, in the psychoanalytical situation, showed symptoms of deep-seated anti-Semitic prejudice of which they had been completely unaware.
The other factor limiting accuracy of responses is the inability of some individuals to understand questions and to verbalize answers on a general, abstract level. This may partially explain why polls show greater anti-Semitism for the better-educated higher-income levels.
_____________
The Social Effects of Polling
One argument often raised against polls on anti-Semitism asserts that the taking of polls on anti-Semitism helps increase it. This seems to us to be completely unwarranted. There have been altogether about twenty to thirty nation-wide polls directly or indirectly concerned with the problem of anti-Semitism. Assuming that each poll sample consisted of 5,000 respondents (some have been smaller), this would mean that at the most 100,000 to 50,000 people have been questioned, assuming also that there has been no overlapping. Thus less than one-tenth of one per cent of the population has participated in these polls. Even if the interviews were not innocuous and did create anti-Semitic feeling in a few cases where none had existed before, which we doubt, the total number of those affected must be insignificant.
A more serious charge concerns the publication of poll results, particularly in such media as Fortune and the daily newspapers. The readers of an exclusive publication like Fortune, and of a “class” newspaper like the New York Herald Tribune, are probably those who may be regarded as opinion-molders. Since the average man in the street does not read these publications it can have no direct effect on him. But what is the effect on the opinion-molder? If he is anti-Semitic to begin with, does reading the results of the polls, particularly when they indicate an increase in anti-Semitism, reinforce his prejudice?
This is hardly likely. Despite our inadequate knowledge of anti-Semitism, we do know that it is so irrational and deep-seated that it can scarcely be seriously affected by the reading of poll results. However, it is alleged that the anti-Semitic opinion-molder is given additional ammunition to use with Gentiles against Jews, for he can capitalize on the “band wagon” effect. Unfortunately, the confirmed anti-Semite needs no facts to feed his prejudice; if new facts are required he readily, invents them. Certainly it is hard to see how keeping poll results secret—or publishing only results favorable to the Jews, if such were available—would in any real measure diminish anti-Semitism.
While harm done by publishing of poll results is, in our opinion, negligible, the publication of poll results may serve a useful purpose as a clarion call to danger, as a means of awakening the non-anti-Semitic opinionmolder to a peril to democracy.
_____________
New Roads For Research
We have hinted above at what is the major and basic difficulty in using polls to measure anti-Semitism: We do not know just what the polls measure when they attempt to find the incidence of anti-Semitism. Polls do tell us the extent to which certain statements which we call anti-Semitic are accepted by the American public—within the limitations we have discussed above. But it is much more crucial to find out the implications of the various responses uncovered by polls. Are we dealing with a harmless parroting of American folklore or something deeply grounded in the individual’s experiences and an integral part of his personality? Will acceptance of anti-Semitic statements lead one to engage in or support pogroms? Is “slight” anti-Semitism a fore-runner of “acute” anti-Semitism? Is there a “social” and a “political” anti-Semitism?
If we knew what anti—Semitism or anti-Semitisms—there may be more than one kind—meant, we could construct better polling questions and we could better interpret the answers to these questions. The question arises whether the conventional polls can be improved to give us this insight into the nature of anti-Semitism or whether completely new techniques must be devised.
Certainly, it is possible to learn more by improving poll techniques proper. Most wellknown polling organizations constantly engage in experimental studies to improve their techniques and methodology, and though these studies are not explicitly concerned with anti-Semitism, their findings do and will in the future benefit polls on anti—Semitism. To indicate briefly lines along which exploration is possible: The stimuli difference between questions containing the word “Jew” and questions in answer to which the word “Jew” may be spontaneously offered, should be determined more accurately. As we have said above, not more than 10 per cent of respondents are “spontaneously” anti-Semitic-that is, give anti—Semitic responses to questions which do not refer to Jews—while many more respondents, from 30 to 60 per cent, are “non-spontaneously” anti-Semitic—will give anti-Semitic responses to questions referring specifically to Jews. Are these differences due to such factors as suggestibility, respectability, articulateness, and does the discrepancy have some meaning in terms of a typology or basic pattern of anti-Semitism?
Another important investigation within the realm of poll techniques would be the study of individuals regarded as anti-Semitic because of their responses to one question, and non-anti-Semitic because of their responses to another. For example, some individuals answer “yes” to the question about Jews being a “problem” in their community, but “no” to the question about Jews being a “menace” to the country. Perhaps this may be interpreted as indicating a lesser intensity of anti-Semitism. Yet it is possible that a confirmed anti-Semite may be so vain that he cannot bear to admit that the objects of his scorn are sufficiently menacing as to present anything stronger than a “problem” to a “superman” like himself.
Many polls have attempted to correlate anti-Semitism with other “anti”-attitudes of a reactionary nature (anti-labor, anti-government spending, anti-Russia and Britain, and so on). Uncertainty as to what the responses to given questions mean makes it risky to draw conclusions from such correlations. Yet some of the correlations obtained by polling groups jibe with clinical investigations of the anti-Semite: he tends to hate labor, Great Britain, Russia, Negroes, foreigners and a long list of other groups and people. Qualitatively, we can accept the suggestion of the polls that the individual whose responses are deemed anti-Semitic is also anti-other groups. But quantitatively we can make no inferences about these relationships as drawn from the polls.
This leads us to consider relatively new approaches for studying anti-Semitism. For, no matter what the improvement of traditional poll techniques, it is essential that new ways be utilized so that we can know just what we are measuring and hence can develop the kind of poll questions—if any—that will give us a true “index” (measure) of the amount of anti-Semitism. These new approaches involve “depth” interviewing and various “projective” techniques.
In “depth” interviewing, we go beyond the simple “yes,” “no” or “don’t know” of the poll interview, and by intensive, time-consuming interviewing, try to grasp the subject as a dynamic whole, in whom anti-Semitism arises from and fulfils certain drives or needs. “Projective” techniques utilize the reactions of the individual to some “unconventional” situation (for example, what stories a set of pictures suggest to him) to get a deeper insight into his character and personality, again so that we can really discover what this “anti-Semitism” in the person is. And of course, it is essential that these techniques be used to study non anti-Semitic as well as anti-Semitic individuals. It is possible to use these techniques to “test” the public opinion polls. Groups classified as anti-Semitic or non-anti-Semitic by the polls could be followed up. By studying the differences between these groups more deeply through these new techniques we may discover what the poll is measuring, and what it is possible to learn from it.
These new techniques are now being utilized in a “battery” of related studies, planned and directed by the Department of Scientific Research of the American Jewish Committee. They are designed to answer such questions as: What is the personality structure of the anti-Semite and what is the function of anti-Semitism in his emotional economy? What is the relationship between anti-Semitism and other “anti-” ideologies? How is anti-Semitism expressed in a living community? To what socio-psychological conditions in childhood has the prejudiced adult been exposed? At what age does prejudice develop in the child, what role does it play, and in what social and psychological context does it appear? Under what circumstances will latent anti-Semitism break out into the open? What educational means can be used to transform attitudes of prejudice into socially desirable and personally useful attitudes?
If one could evolve an inventory of anti-Semitic manifestations, a hierarchy of their intensity, an insight into their causes and links with other “anti-” attitudes, then polling organizations could establish a rationale for their questions on anti-Semitism and the social scientist could accept the findings of such polling studies with a greater degree of confidence.
Until more is known about the phenomenon of anti-Semitism and the indicators which reveal its presence, results based upon present polling techniques should be used with caution—for scientific reasons, however, rather than for their allegedly dangerous impact on intergroup relations in the United States. In the meantime, the polls give us some clues and hunches in an area about which there has been infinitely more speculation than facts.