[The Emperor Frederick II] wanted to find out what kind of speech and what manner of speech children would have when they grew up, if they spoke to no one beforehand. So he bade foster mothers and nurses to suckle the children, to bathe and wash them., hut in no way to prattle with them, or to speak to them, for he wanted to learn whether they would speak the Hebrew language, which was the oldest, or Greek, or Latin, or Arabic, or perhaps the language of their parents, of whom, they had been born. But he labored in vain, because the children all died. For they could not live without the petting and the joyful faces and loving words of their foster mothers.

—Salimbene (13th century) in the
Portable Medieval Reader

_____________

 

The ambition of American social science is to arrive at general laws of society and human behavior: laws that shall be as universal, as precise, and as useful as Newton’s. Indeed, it is the conviction that it can accomplish this aim that represents much of the attraction of that vast field which now includes thousands of practitioners, tens of thousands of students, and which every year spends millions of dollars gathered from business, government, scholarly foundations, and universities.

But its every effort towards the formulation and testing of such laws has to date been frustrated. Social science has been able to present its results only in the form of specific historical statements, and not as general social and psychological laws. It can only say, “at this time and this place”—in short, in this sample—“the educated are less prejudiced,” and has found no way of saying with any great confidence what the relation between education and prejudice will be at any other time or any other place—in short, in the next sample.

But the call—both from within and without—to discover general laws is so great, and the resulting frustration so intense, that it sometimes leads to what can only be called hallucinations: some social scientists begin to believe that these laws have already been discovered. For example, Stuart C. Dodd, now a professor of sociology at the University of Washington, fills his big book on The Dimensions of Society with formulas, and makes it look very much like a text in physics or engineering: except that the symbols in his formulas can by no stretch of the imagination be replaced by numbers, and can be applied to actual human behavior by no one, not even himself. Professor Dodd’s book lies on library shelves, virtually undisturbed, but, still a respected figure, he continues to present his point of view in the various professional journals. In the Winter 1948-9 issue of the International Journal of Opinion and Attitude Research, we hear him saying: “Each of the three factors [leading to the mispredicting of elections] is measurable, [and] its share assessable. . . .” The appropriate formula is “S = ^tT^-1:₁L²😛_p:I:²₁,” which means that a situation is dependent on such factors as time, location, and the number of people. But this formula rests unused in a footnote, and Professor Dodd must satisfy himself with only the symbols of science.

_____________

 

Our interest here, however, is not in the more exotic blooms along the borders of the social sciences, but in the central thickets themselves; not the social sciences where they leave themselves most vulnerable but where they present their strongest effort. The American Soldier is unquestionably such a strong effort. It has been written by Samuel A. Stouffer and many associates, and is being published by Princeton University Press. Of the projected four volumes, the first two—Adjustment During Army Life, by Samuel A. Stouffer, Edward A. Suchman, Leland C. DeVinney, Shirley A. Star, and Robin M. Williams, Jr., and Combat and Its Aftermath, by Stouffer, Arthur A. Lumsdaine, Marion Harper Lumsdaine, Robin M. Williams Jr., M. Brewster Smith, Irving L. Janis, Shirley A. Star, and Leonard S. Cottrell, Jr. ($13.50 for the two volumes)—are already published, form a single unit, and constitute the heart of the whole enterprise.

These volumes have been hailed as standing at the summit of American social science. “Never before had such modem methods of social science been employed on so large a scale, by such competent technicians . . .,” says Major General Osborn, formerly director of the Information and Education division of the Army, in the introduction. And further: “In no other country could there have been found so many men with such a high level of scientific training in investigating and interpreting the behavior of human beings.” The authors themselves write of their book that it contains “a mine of data, perhaps unparalleled in the history of any single research enterprise in social psychology or sociology.”

The mine of data referred to was garnered through some two hundred questionnaires that the Research Branch of the Information and Education division drew up and administered to perhaps half a million soldiers all told (on the average, it would seem, 2500 men were given a single questionnaire). Not all these questionnaires were used in the writing of The American Soldier: only those form part of this report which deal with problems that are generally of interest to the social sciences, and those which have behind them enough data to “at least illustrate if not demonstrate the behavior thought to be involved.”

_____________

 

We Find no hallucinations in The American Soldier. Professor Stouffer himself is a very able man, a clear and (for the social sciences) even good writer, and he is, one is assured and stands ready to believe, a brilliant statistician (the statistical presentation throughout these volumes is always clear and often ingenious). He was assisted in the work of the research branch by the best men social science has, men like Robert K. Merton and Paul F. Lazarsfeld of Columbia and John Dollard of Yale, and in writing this book he also had the help of some of the most promising of the younger men in the field. Professor Stouffer and his associates, we learn very early, do not believe, as Professor Dodd does, that they can today measure and weigh—that is, give meaningful numerical weights to—the interacting factors in any given situation, but they believe that some day they will. For example, we may read such a sentence as: “The problem of developing appropriate psychological tools for measuring the extent to which an individual’s behavior toward a particular rule has been ‘internalized,’ remains largely for the future.” But there is no breath of doubt that such tools will eventually be developed.

The overpowering obsession with the physical sciences and their great achievement makes its mark on every page, and defines the general aim. “There are several streams of influence which are converging to develop social psychology and sociology into sciences with conceptual schemes from which, it is hoped, empirically verifiable inferences and predictions can eventually be made.” This is a modest and sober evaluation of present achievement. The extravagant and silly claims for the social sciences are not made by men like Stouffer, Merton, Lazarsfeld, and Dollard: they come from enthusiastic journalists like Stuart Chase, and the methodologically naive anthropologists and “social psychiatrists.” The type of work that is represented in The American Soldier—that in which a highly technical, carefully tested type of public opinion research is used within the framework of ideas created by social thinkers like Pareto and Durkheim so as to test them and build a science—is practiced by sophisticated and learned men who know their accomplishments do not warrant brash enthusiasm. Nevertheless, however much they admit their present limitations, their aim is clearly stated: to create sciences; and even the present volumes, we may read, make some (“severely limited”) “contributions to the verification of hypotheses of some generality.”

The last two quotations leave no doubt that they have taken as their model physical science, the most advanced and successful of all. Physical science does indeed create sets of universal propositions (conceptual models), and from them deduces almost everything in the world with which it deals. Given the law of gravitation, plus some historical conditions, practically nothing is left unexplained in the movement of the planets. Other sciences are not as strikingly successful in ordering their worlds. If the writers of The American Soldier, and the branch of social science they represent, had taken, let us say, biology as a model, their presumption would have been less and their success in approximating it possibly greater.

_____________

 

The claim must guide the critic. This book, we are clearly told, is not history, though it will be useful to historians; it is not the casual or organized impressions of scholars, however gifted or intelligent; it is science.

But what is science? Science, we have recently been reminded in a brilliant article by Eliseo Vivas (“Two Notes on the New Naturalism,” Sewanee Review, July 1948) is not intelligence and not wisdom, and it is not even reason. Aristotle and Spinoza, Francis Bacon and Giordano Bruno, John Dewey and Rudolf Carnap—whatever may be their varied relations to science and reason, none of them is a scientist. Rational thinking and writing, the awareness of one’s own biases, the consideration of all aspects of a question, and the use of mathematical symbols—all these were present in men before the rise of modem science, and some perfectly sound examples of scientific investigation, such as the premature one reported at the head of this article, involve hardly any of these modes of thought.

In its most ideal form—and this is the form the writers of The American Soldier have chosen to emulate—science is a technical enterprise which involves, among other things, the observation of phenomena, often indirectly through technical devices, the discovery of uniformities—of patterns—in these phenomena, the formulation of hypotheses about these uniformities, and the determination, by experiment, prediction, and further observations, which of two or more hypotheses best describes the phenomena and is worthy of being called a law. Science is also an enterprise which, in contradistinction to all other efforts of the human mind, is, as Professor Stouffer himself points out, cumulative. Every novelist, for example, must begin from scratch: because someone else has described a love affair does not mean that no more love affairs need be described. The work of earlier philosophers is like the advice of a parent: perhaps sound and even usable; but life must still be lived again. John Dewey, like Plato, must again take up the old questions of what is just, good, and worthwhile, and how do we know what we know, and so on. But scientists do not begin from scratch: they build on each other’s work, and do not have to go over the ground others have secured unless they have some special reason to think there was an error. Each scientific achievement forms part of a chain closely joined, one link immediately dependent on the other, in a way that the achievements of art and philosophy do not.

The American Soldier forms no part of a cumulative record of science in this sense. It rests on no hypotheses or laws established by previous social science, nor does it pass on any. No one will ever in the future begin a scientific article with the words, “As Stouffer et al. have established. . . .” They will have to use instead words like “suggested” or “illustrated.”

Now it is true that this work does form part of a cumulative development in respect to techniques. It uses those developed by other social scientists, and its own new techniques will be used by future social scientists: there is, indeed, no question but that these techniques are scientific, that is, systematic and empirical. But as to the scientific conclusions established in a general form by their use, the authors are necessarily silent.

The Pirate outfielder Ralph Kiner is called in one press report “scientific”: he keeps a card file in which he indexes the characteristics of every pitcher he faces. The technique seems an excellent one, breathing the spirit of modern scientific enterprise, and aspiring batters in the future will perhaps use it. But no one believes that the art of hitting home runs can thereby be transformed into a science. In the same way, The American Soldier uses and creates techniques for the study of human behavior. Unquestionably, we have scientific method in The American Soldier, as we may have it in baseball. The question however remains, do we have an approach to the organized body of conclusions that form a science?

_____________

 

To Demonstrate—as I believe—that social science has not succeeded in establishing the kind of uniformities or laws concerning society and human behavior that might be first material for the building up of a science; that all that is cumulative in social science is its techniques; and that, for various reasons having to do with its subject-matter, and not with its techniques, no true social science can be established—to demonstrate these assertions, would require a close analysis of The American Soldier and other works that social scientists claim are stepping stones toward the building of a science. It would involve, if we were to begin the task with the first two volumes of The American Soldier, deciding what, in scientific terms, is the status of each of the statements in its 1250 pages of text—where we deal with demonstrated fact, where with uniformity, hypothesis, law, prediction, and what-not.

This is properly a job for the philosophers of science: those gray eminences who, themselves usually not concerned with the work of science, devote themselves to the study of its method and its essence.¹ But the philosophers of science, many of whom hold to the position that there can be a social science, have done very little to show—in detail, and using the materials of the social sciences—what its structure might be, nor have they subjected social science, in its monographic, workaday form, to the close and detailed analysis it needs.

Failing their intervention, the less qualified must do what they can. I would like to take from The American Soldier three examples of what would be generally considered scientific work as good as any the social sciences have to offer and see just what the scientific status of each is. The first is a set of a questionnaires forming what is called an “index”—the chief technical tool that has been developed for observation in the social sciences; the second is a “concept,” which in this case corresponds most closely to the ordinary scientific notion of a hypothesis; and the third a prediction.

_____________

 

Observation: The American Soldier’s data, as we have said, are taken almost entirely from questionnaires. The questionnaire is a relatively recent development. Formerly, only those social scientists who studied opinions and attitudes resorted to questionnaires. Now, every problem, no matter how remote from matters of opinion on its surface, is considered approachable by questionnaire methods; and one can be sure that if someone set out today to study anomie (a word used by the French sociologist Emile Durkheim to refer to the feeling of “normlessness,” rootlessness, and loneliness that he felt was endemic in industrialized and urbanized societies), he would not, like Durkheim, use suicide statistics, but set to work creating his own data with questionnaires.

Without the questionnaires, it would hardly be possible to make such extensive use of The Machines, and it is perhaps not unfair to suggest that this is one of the reasons we have had the increasing emphasis on questionnaires in recent years. The questionnaires supply far more fodder for The Machines than the relatively sparse and unmalleable data of the census, or any other source of “real” figures, i.e., figures which are independent of the particular way in which we try to estimate them or of the specific agents who compute them.

The machines in question—and we must understand them before we can understand The American Soldier—are IBM or Hollerith machines, which take punched cards, each with about seventy columns, in each of which there are about a dozen positions that can be punched. The punches, the full record of the questionnaire, indicate one of two things: either something about the respondent’s social position—his sex, race, age, marital status, schooling, and so on, the sort of “real” figures the census collects; or an answer given to a question about attitudes. Most of the card is taken up with punches representing these latter items, the special creation of the social scientist. There will be one or more cards for each respondent, depending on how many questions were asked (it is not hard to record a hundred items on a single card). The original questionnaire is transferred by the operator of a punch machine to the cards, according to a system devised for this questionnaire by the social scientist, who tries to arrange the punches so as to permit the maximum amount of manipulation of the data in search of significant correlations. The whole process of thinking up questions, “testing” them (that is, asking them of a few people to see if they are understood), training interviewers and administrators, drawing up a code and training coders (the code is the system by which the questionnaires are translated into punching arrangements)—this whole process reaches its apotheosis when the cards are “run”—that is, fed into the machine, with whatever punch one wishes counted at some superhuman speed by electrical impulses.

Without such a technical process. The American Soldier would not have been possible. With it, we can discover in no time at all—to take a hypothetical example—how many men prefer oatmeal to farina, how many of these have been in the Army one, two, or three years, how many of each of these categories are married, and of these how many have completed college; and thus we may know how many unmarried college men who have been in the Army one year prefer farina. In short, questions that would otherwise never have come to trouble the human mind can now be asked and answered by the machines. Any reader of The American Soldier, seeing the same information recur again and again, broken down in a variety of different ways, will soon come to feel that the possibilities and limitations of the machine dictate a large part of the contents of such a volume.

_____________

 

The utilization of the reckoning devices of modem technology requires the production of numbers; and questionnaires, along with the way in which they are analyzed, can supply us with a vast number of numbers. The question, however, now arises whether they are important numbers, whether they are numbers which tell us about the real world of social behavior, and permit us safely to go on to higher steps of the scientific process.

Now it becomes quickly apparent they do not immediately reflect the objects of our concern, since in most cases we are interested in the study of behavior, and questionnaires only give us the expression of attitudes, which may be related to behavior in a variety of ways: as conscious lie, unconscious rationalization, rational but inaccurate prediction, and so on. But the fundamental assumption of the use of questionnaires in a volume such as The American Soldier is that there will be a regular connection between the sets of numbers in which we sum up questionnaires, and some social behavior—that is, that by means of the questionnaire, we will measure a reality of social behavior. For this reason, sets of questions are sometimes called “scales” (if they fulfil certain technical qualifications) and can also be called “indexes.” (Anything which is believed to vary regularly with something else may be considered an “index” to it; e.g., we may consider suicide an index to anomie.) The critical question now becomes: do questionnaires give us good indexes? Do they really reflect the social reality they are used to study?

Let us see. A good part of the first volume of The American Soldier is devoted to the presentation of the data accumulated by the use of an index—that is, a set of questionnaires—that was established to study morale. Morale, we are told at the outset, is a group phenomenon; questionnaires, of course, report only on individuals, so there seemed no simple way of studying the morale of a group by way of questionnaires. “The problem of measuring the cooperative effort of a unit [that is, morale] was never solved satisfactorily. . . . Faced with the necessity of giving the Army command, quickly and reliably, information which would be useful in policy making, the Research Branch concentrated primarily not on the evaluation of the cooperative zeal of groups toward Army goals, but rather on study of personal adjustment.” For, it was reasoned: “If most of the soldiers in one outfit show evidence of adjustment . . . and if most of the soldiers in another outfit show little evidence of adjustment, it is not unlikely that the former outfit would be found to have, if it could be measured, higher morale than the latter.”

In this way, individual “personal adjustment” was substituted for group morale. But personal adjustment, too, cannot be directly observed: “ . . . it is [therefore] assumed that, on the average, men who said they were in good spirits, that they were more useful in the Army than as civilians, that they were satisfied with their Army jobs and status, and that in general they liked the Army, were better adjusted to the Army than men who were negative.”

So, faced with the difficult problem of finding out what makes for high morale and what for low morale—ultimately, to find out what makes armies win or lose, as the problem is put in the very first page of this book in a quotation from War and Peace—we see that the social scientist carefully picks his way back along a chain of shaky reasoning until he comes to the point where he can ask a few questions, and record what soldiers say. This permits him to draw up charts and reports for the guidance of policy-makers showing how personal adjustment varies with education, with age, and so on. Meanwhile, what has happened to morale?

“Mit gier’ger Hand nach Schätzen gräbt, Und froh ist, wenn er Regenwürmer findet!” (“With greedy hand he digs for treasures, And is happy when he finds earthworms!”)

—this was Faust’s conclusion on man’s efforts to achieve knowledge. The social scientists, free from any such scepticism, offer their earthworms with delight.

_____________

 

How can the social scientist “prove” to the policy-maker that his questionnaires really get at “personal adjustment”? Well, it appears that those men who come out highest on his questionnaires also are more likely than not to be NCO’s and officers, and those who come out lowest are more likely than not to be AWOL’s and psychoneurotics.² The point of this demonstration seems to be that while we cannot show directly that our questions measure personal adjustment, everyone agrees that officers and NCO’s are better adjusted than AWOL’s and psychoneurotics, and the questions do select them—that is, good answers correlate positively with one’s position in the Army hierarchy, negatively with the degree of psychoneurosis and AWOL.

But then the whole point of the index is thrown into question. For it is as much as admitted that a soldier’s position in one of the four categories—officers, NCO’s, AWOL’s, and psychoneurotics—is a more certain measure of his personal adjustment than his score on the set of questions forming the index: for we have used the former to validate the latter. Then why not study the number of psychoneurotics and AWOL’s, and the way they are distributed in different theaters, at different times, and by different background characteristics, directly, rather than the scores on a questionnaire index? Why resort at all to the shifting sands of subjective report?³

Perhaps because the contemporary social scientist, instead of studying a social actuality to the limited and tentative extent he can, too often prefers to devote himself to the multiplication of his special measuring rods, all suffering the same congenital inaccuracy. (“At least sixteen indices of intensity of opinion are known,” reports Stuart C. Dodd in the article previously referred to, “and more seem inventable.”) Professor Stouffer does indeed consider the possibility of what is called an “objective index” for morale, that is, something which measures what people do, not what they say. “The chief difficulty with such indexes . . .,” we are told, “turned out to stem from the fact that so few men in a given unit would be involved. If only one or two per cent of the men would be arrested by MP’s in a given time interval, offenses are not a very useful index of relative adjustment. . . .” I cannot understand the argument. One can draw the same conclusion by comparing .5 with I as by comparing 50 with 100, if the samples are large enough. And if one is using a simple measure, as the number of men arrested, it is easy to work with large samples. And, practically, the point also seems ill-taken. If I were a general, I would be far more interested in the proportions of men who had, say, gone AWOL—even if the figure were very small—than in the proportion who scored above a passing grade in an index of adjustment, based on their oral reports. It might be shown to my complete satisfaction that the latter had a high correlation with the former. But why use a shaky “instrument” when the reality it is supposed to measure is immediately at hand—and can be read off more easily than the instrument?

But perhaps that is just the point. The relatively few objective indexes one can use mean technological unemployment for the machines, while the instrument for measuring personal adjustment involves no less than twenty-three questions, divided into four areas. And we can run the machines far into the night correlating scores on the total index with scores in each area, even on each question, and then correlating these with our background information on age, marital status, rank, and so on. A good part of this book is filled with just such information. And the clicking of these intricate machines so impressively resembles the cold, hard rhythm of real science!

_____________

 

Hypothesis: We may speak of hypotheses in history. Louis XIV, one may hypothesize, did not expect that England would go to war if his grandson ascended the Spanish throne. One may then find documents to prove or disprove the hypothesis.

Similar hypotheses can be found scattered through The American Soldier, and one can conceive of thousands more that can be verified by data in it: for example, American soldiers of World War II who were married were unhappier about being in the Army than soldiers who were unmarried. And, indeed, for the first time it becomes a relatively simple matter to verify hypotheses in history referring to the attitudes of large numbers of people. This is a real achievement of public opinion research.

But history is not science. The social scientist is after bigger game than this. He wants to infer, from general conceptual schemes, hypotheses of such a nature that eventually all the conditions that might prevent them from holding true are known and no more have to be added to explain future deviations. He wants to be able to say, for example, “persons in subordinate positions tend to obstruct the activities of those superior to them,” and eventually discover all the limiting cases, all the conditions tending to operate against this law, so that when one brings up a case where the law did not hold, one of the previously listed limiting conditions will be found present.

There are not too many good hypotheses in The American Soldier, and many of these seemed to be operating under pseudonyms. I believe that what the authors speak of as the “concept” of “relative deprivation” is such as hypothesis.

The concept is brought in to explain some of the data collected by the index of personal adjustment just discussed. If we take the questions that refer to “personal esprit” and “personal commitment” alone, we discover that the less educated, the older men, and the married men come out with more negative attitudes than the better educated, the younger men, and the unmarried men. An explanation for this is offered by the concept of relative deprivation, which suggests that one feels better or worse on the basis of whether one has been treated better or worse than other people who are around. The older men compare themselves with the younger, feel it is unfair they should have to give up careers, jobs, etc., and have less personal esprit and commitment. The married men compare themselves with the unmarried, and go through the same line of reasoning.

But the less educated men—how explain this apparent deviation from the hypothesis of relative deprivation? In what way are they relatively deprived? Professor Stouffer here suggests that perhaps, on the average, they were not in as good health as the better educated (since education is, in turn, correlated with income!), and, consequently, felt more deprived by being drafted. Then he refers to jobs. It’s true they were giving up worse jobs, and shouldn’t have felt deprived on that score, but their friends—assumed for this purpose to be the less educated too—were being deferred as skilled workers or farmers, which must have made those less educated who were drafted feel relatively more deprived.

Now, obviously, if the better educated had turned out feeling more deprived than the less educated, we could have found rather easier ways of fitting that under the concept of relative deprivation. Thus a little imagination will permit us to cover any conceivable outcome of the questionnaires. When we read that Air Corps men were more dissatisfied than non-Air Corps men about promotions—even though the tables of organization in the Air Corps called for far more NCO’s and officers than anywhere else in the army—the versatile concept of relative deprivation is again called upon, and we are asked to believe that just because there were so many opportunities for advancement in the Air Corps, men felt dissatisfied with whatever grade they had achieved.

It would seem that a concept in the social sciences is something unique. It is an explanation, or a hypothesis, which need not be tested, but which it is legitimate to use wherever, by twisting and turning, it may be made applicable. It cannot be refuted by facts, and it will be found to hold true whatever the outcome of a given set of data. If the non-Air Corps men are dissatisfied with their promotions—why clearly, they are deprived relative to the Air Corps men. If they are not—there is recourse to the mental gymnastics we have just described.

It is a simple rule of scientific method that if a hypothesis is such that it can neither be proved nor disproved, if it is held to be correct no matter what the facts, it does not belong to the realm of science—it is a matter for philosophy or faith. The concept of relative deprivation would seem to fall into this category. And every other hypothesis on this level of generality offered in The American Soldier would, I believe, be found similarly wanting if subjected to analysis.

_____________

 

Prediction: Perhaps that part of The American Soldier of which its authors are proudest is an example of prediction—the crown and glory of science. Before D-Day, a large number of companies were given a questionnaire and rated for “willingness for combat,” “confidence in combat stamina,” and “confidence in combat skill.” The average score of each company was then compared with the nonbattle casualty rate of that company for the two months after D-Day. What we have here are two indexes: the questions (eight in all) that are used to rate soldiers for “willingness for combat,” etc., form an index to the attitudes which determine how effectively a unit will fight; the rate of nonbattle casualties (which includes cases of exhaustion, psychoneurotic cases, and so on) is then an index to combat effectiveness. One might argue—as I did earlier in the discussion of the index of personal adjustment—against the validity of both indexes, and the second one in particular seems questionable, but for the purposes of the discussion, let us assume both are sound. It is now asserted that performance on the first (“willingness for combat,” etc.) permits us to predict performance on the second (nonbattle casualties).

The book uses a number of ways to represent the order of success achieved. Perhaps the most impressive and simplest is when it takes in each regiment the three best and three worst companies, according to the score achieved on the attitude test (in this example, we use only the “willingness for combat” part of the test, which gets the best results), and compares the average nonbattle casualty rates for the two groups. In one regiment, the worst companies had 132 per cent more nonbattle casualties than the best; in another regiment, the worst companies had fewer nonbattle casualties than the best: only 92 per cent as much. On the average, the companies with negative attitudes before D-Day had 62 per cent more nonbattle casualties after D-Day than did the best companies.

In order to evaluate this achievement—by no means insignificant—correctly, let us consider two alternate means of getting at the same results.

We might ask each regimental commander to list the companies in his command in the order of their estimated combat effectiveness. The results might be better, worse, or about the same as the results arrived at by using attitude tests. There is no question, however, that the method is less scientific, in so far as it is less systematic.

We might also have doctors and psychiatrists examine each man, and estimate the health and physical fitness of each unit. The results might be better, worse, or about the same as those achieved by attitude tests. The method, however, is rather more scientific than attitude tests, in so far as our procedures for measuring health are better than our procedures for measuring attitudes.

The point is a simple one. In science, the role of prediction is to test hypotheses and its accuracy is the most compelling proof of the correctness of a hypothesis (as when we predict, from formulas, that this structure will break when so and so much weight is put on it). A prediction is based on a clear conception, stated in hypotheses, of the connection between two sets of things: in this book, a prediction would involve the tracing of a line of connection between the way soldiers feel, the way they report their feelings, and the way they fight. The connection, it would have to be asserted if we were to have a real prediction, is thus and so, and consequently I predict combat effectiveness will be thus and so. Now this “prediction” of combat effectiveness, and the two possible alternatives I suggested, test no hypotheses: they are merely elaborate ways of making estimates.

An estimate is based on no rigorous explication of connections, and plays no role in testing either special hypotheses or the structure of a science. Drawing on general experience and whatever specialized knowledge may be available, one may try to forecast a certain outcome; one may be even so fortunate as to find a regularity between two sets of data. But as long as there is no clear and explicit line of necessary connection between the two sets of data, we have an estimate, and the structure of science is in no way advanced. Perhaps next time we will not observe the uniformity; and perhaps there will be a higher correlation if we use some completely different means to get one or both sets of statistics.

Prediction, as the term is used in scientific procedure, would have to be based on some theory of the connections between expressed attitudes and nonbattle casualties, and would eventually have to involve some such statement, made before the battle, as, “on the basis of these hypotheses, suggested by such and such preliminary work, we predict nonbattle casualties will be thus and so.”

In The American Soldier, the numbers are allowed to speak for themselves, and the authors, impressed with the evidence of the numbers and hoping the readers will be equally impressed, make no effort to show that their “prediction” forms part of a chain of scientific procedure: whether it tests hypotheses, applies hypotheses, proves or illustrates a logical connection. As a matter of fact, it does none of these things, and we have no way of knowing whether the means I suggest above, or ten others, more or less scientific, might not be better.

_____________

 

The use of scientific method, of abstract conceptual schemes and fabulously complex calculations, and the adherence to canons of scientific objectivity, all these do not necessarily create a science. Why this should be so—why such heartbreaking effort in the social sciences should have led only to The American Soldier, when, I am sure, much less effort in astronomy had already led to such an achievement as the prediction of eclipses—such a question can be answered, according to each man’s preference, by listing all the conceivable differences between nature and history, even between “mind” and “matter.” But whatever the reason why human affairs are so peculiarly resistant to the type of organization that has been successful in so many other spheres, it is so.

This conclusion still leaves us with important questions. For example: even if scientific method when applied to human affairs does not produce a science, is it still not the best way of studying anything? If we take scientific method—as John Dewey and Sidney Hook do—to include all the modes of rational and systematic procedure, of course. It is the best way to proceed in any matter, whether we are studying beliefs in ghosts or the movements of ocean currents. But we cannot argue that because scientific method has achieved certain unique results in some spheres, it will achieve the same results in all spheres. We cannot argue—as certain social scientists do—that because scientific method produces agreement on what is true for nature, it will produce agreement on what is true for man. I do not believe that the problem of extending science over the sphere of human behavior is primarily one of overcoming our human biases and prejudices; that our success in physics is owing to the fact that we are not emotionally involved with masses and energies, as we are with human wills and motivations—though this difference is certainly very important. The fundamental problem seems to me to arise from the nature of the materials which first attract our attention. There are aspects of human behavior—like language and simple learning—that seem to be, by nature of the facts about them that interest us, susceptible to organization into scientific systems. As we get further away from human behavior rooted in physiology, however, the application of scientific method becomes less and less fruitful, and it becomes harder and harder to rise above the immediately given facts through organized scientific procedures.

If our aim in applying scientific method to human affairs is more than the relatively humble one of assisting reason (which, unlike science produces neither near-universal agreement nor relative certainty)—if it is to find “conceptual schemes from which . . . empirically verifiable inferences and predictions can . . . be made”—then, from the evidence of The American Soldier and the many years work that have preceded it, we must either fool ourselves into thinking we are social scientists when we are not, or remain silent.

_____________

 

If The American Soldier is not, strictly speaking, science, is it at least useful? To the Army, I think hardly at all. We might take every statistic in this book, halve or double it: the Army could have done very little about it, or would have wanted to do very little about it. The generals seem to have been motivated by a kind of pure curiosity in supporting the Research Branch studies (though, as General Osborn says, “with research, we . . . were able to defend our course before Congress and the press”). At one point Professor Stouffer writes: “The general picture in this volume of men preoccupied with minimizing their discomforts, acquiring higher rank or pay, securing safe jobs . . . does not suggest a particularly inspired work performance in the American Army. But Americans fought, and fought brilliantly and tenaciously when they had to. . . .” This seems to come pretty close to saying that all the attitudes “measured” did not matter when it came to fighting, which was the prime function of the Army, and the ultimate purpose for which the studies were undertaken.

However, The American Soldier, if not useful to the generals, is useful to the historian and the scholar. In the course of the efforts to create a science, a great deal of interesting and valuable information about the American Army and the American soldier in World War II has been turned up. (I have not reported on it here, since it was not my aim; just as I have not reported on oddities and gaucheries of interpretation, and some incredible examples of moral and social blindness which wise men would not be capable of, but which scientists cannot especially be blamed for—for it is primarily as science that I consider The American Soldier.)

But actually this information was produced only as a by-product of the scientific undertaking. If a group of scholars of the same capacities as those who produced The American Soldier had set to work writing an account of the American soldier in World War II, without the illusion they were creating a science, and using whatever means came to hand—including, of course, all the statistical procedures that have been developed since the rise of capitalism—we would have had far more information on the subject. Indeed, one might say that rarely was so little useful information about so large a question spread over so many pages. All because the aim was science, not understanding; the mechanical and formal confining of knowledge, not the increase of it.

_____________

 

¹ Strangely enough, not only do those who today know most about scientific method and science as a system of logically connected assertions—the philosophers of science—have little to do with science, but the physical scientists, whose scientific accomplishments have been most amazing, and which form the subject matter for speculation and analysis by the philosophers of science, concern themselves very little with the general understanding of science. The unfortunate social scientist is in the middle; he knows more about scientific method and “methodology” than the practicing physical scientist, but tries to make up with this formal knowledge for the poverty of his concrete scientific accomplishments.

² More impressive is the demonstration that it is possible to “predict” personal adjustment from the questionnaire index, since privates with positive attitudes are more likely to become NCO’s in the future. But I will take up the whole question of this type of prediction later.

³ We leave out of account, in this analysis, the enormous problems involved in getting people to give truthful answers, interviewers to use uniform techniques, coders to feel a sense of responsibility—and so on through the chain of bureaucratic organization necessary before the social scientist is presented with a neat pile of cards.