PSYCHOLOGICAL TESTS IN BUSINESS K0RN1 JAUSER and KINGSBURY MATERIALS FOR THE STUDY OF BUSINESS Materials for the Study of Business Industrial Society. By Leon C. Marshall. 1,082 pages, cloth, $4.50, postpaid $4.65. Financial Organization of Society. By H. G. Moulton. 790 pages, cloth, $4.00, postpaid $4.12. Principles of Accounting. By Albert C. Hodge and J. O. McKinsey. 390 pages, cloth, $3.00, postpaid $3.10. Law and Business. By William H. Spencer. 3 vols., cloth, $4.50, postpaid $4.62 each. Vol. I. Introduction. 612 pages. Vol. II. Law and the Market. Law and Finance. 670 pages. Vol. III. Law and Labor. Law and Risk-Bearing. Law and the Form of the Business Unit. 654 pages. Business Administration. By Leon C. Marshall. 920 pages, cloth, $4.00, postpaid $4.12. Education for Business. By Leverett S. Lyon. 618 pages, cloth, $3.50, postpaid $3.60. Social Studies in Secondary Schools. By a Commission of the Association of Collegiate Schools of Business. 114 pages, boards, $1.00, postpaid $1.10. Forms, Records, and Reports in Personnel Administration. Edited by C. N. Hitchcock. 128 pages, paper, $1.75, postpaid $1.79. Recent British Economics. By D. H. MacGregor, R. Lennard, and J. A. Hobson. 134 pages, boards, $1.50, postpaid $1.60. The Worker in Modern Economic Society. By Paul H. Douglas, Curtice N. Hitchcock, and Willard E. Atkins. 962 pages, cloth, $4.50, postpaid $4.65. Dumping: A Problem in International Trade. By Jacob Viner, 344 pages, cloth, $3.00, postpaid $3.12. The Economics of Overhead Costs. By J. Maurice Clark, xiv+ 502 pages, cloth, $4.00, postpaid $4.15. The Packing Industry. By the Institute of American Meat Packers. 358 pages, cloth, $3.00, postpaid $3.10. Risk and Risk-Bearing. By Charles O. Hardy. 400 pages, cloth, $3.50, postpaid $3.60. Readings in Risk and Risk-Bearing. By Charles O. Hardy. 368 pages, cloth, $3.50, postpaid $3.60. Psychological Tests in Business. By Arthur W. Kornhauser and Forrest A. Kingsbury. 194 pages, cloth, $1.75, postpaid$ 1.85. IN PREPARATION The Technique of Business Commu- nication. Managerial Accounting. Commercial Cost-Accounting. The Manager's Administration of Finance. The Place of the Market in our Eco- nomic Society. The Manager's Administration of Labor. The Physical Environment of Busi- ness. Social Control. THE UNIVERSITY OF CHICAGO PRESS CHICAGO. ILLINOIS PSYCHOLOGICAL TESTS IN BUSINESS THE UNIVERSITY OF CHICAGO PRESS CHICAGO, ILLINOIS THE BAKER & TAYLOR COMPANY NEW YORK THE CAMBRIDGE UNIVERSITY PRESS LONDON THE MARUZEN-KABUSHIKI-KAISHA TOKYO, OSAKA, KYOTO, FUKUOKA, SENDAI THE MISSION BOOK COMPANY SHANGHAI PSYCHOLOGICAL TESTS IN BUSINESS BY ARTHUR W. KORNHAUSER AND FORREST A. KINGSBURY THE UNIVERSITY OF CHICAGO PRESS CHICAGO, ILLINOIS Copyright 1924 By The University of Chicago All Rights Reserved Published June 1924 Composed and Printed By The University of Chicago Prejs Chicago, Illinois, U.S.A, ' PREFACE The chapters in this book attempt to state the main facts about tests, test methods, and test accomplishments, in the field of business. The aim throughout is to present a simple and impartial picture which will enable the reader to arrive at sound conclusions regarding the actual usefulness of tests in the business world, and the possibilities of future development. It is hoped that the material included will prove useful to anyone interested in the problem of more adequately adjusting men to jobs. The book is written for students of personnel practice and of applied psychology, and for all business executives who have been hearing the clamor about tests and who want to form a critical opinion on the basis of facts. Many of the statements of the book will seem overcautious to people whose hopes have run high concerning tests. But there is need of caution. A too enthusiastic attitude toward tests is bound, in the long run, to bring an unfavorable reaction. In some quarters it has already done so. At present it is fully as important to emphasize the limitations and shortcomings of tests as to speak in glowing terms of their possibilities. Some readers, too, may feel disappointed at the absence of concrete guidance that would enable them immediately to procure some tests and begin testing. The reason should, however, be obvious; tests are scientific instruments. Test work should be supervised by someone with thorough training in applied psychol- ogy. Otherwise it should be let alone. That is why the present book contains no set of tests "ready for use," or lists of tests for particular purposes. As patent medicines, tests are dangerous. Arthur W. Kornhauser Forrest A. Kingsbury VII TABLE OF CONTENTS CHAPTER PAGE I. The Nature of Psychological Tests x II. Scientific Method in Constructing Psychological Tests for Business 43 III. Psychological Tests for Office Occupations 77 IV. Psychological Tests for Non-Office Occupations . . . 112 V. The Place of Tests in the Personnel Program . . . . 141 VI. The Outlook for Tests in Business 164 Suggestions for Further Reading 185 Index 189 IX CHAPTER I THE NATURE OF PSYCHOLOGICAL TESTS Ever since men first began to have dealings with one another, they have endeavored to find successful ways of predicting one another's conduct. Earlier generations laid much stress on the effort to detect, through signs and portents, those supernatural forces and mysterious influences which were regarded as the chief determiners of the course of one's life. Later generations have turned their attention to the tendencies to action found within man himself, which we call variously "capacities," "abilities," "traits of character," "habits," "dispositions," and the like. Such tendencies in an individual we can learn to understand through close association with him and observation of his actions under varying conditions; and we use such an understanding of him as the basis for judging how he may be expected to act in the future. But not often is this prolonged, intimate observation of an individual possible. We are constantly compelled to make judgments about people's abilities without having opportunity for long acquaintance. Especially is this true in selecting workers for employment, but it is no less true and no less serious a problem in many other fields of human activity Many methods are in daily use and many others have been advocated for judging abilities and predicting behavior with accuracy. In psychological tests we find one of the most recent and one of the most popular proposals for accomplishing these purposes. RECENCY OF PSYCHOLOGICAL TESTS It is now more than a third of a century since tests were first devised and used in psychological laboratories. But it is only 1 2 PSYCHOLOGICAL TESTS IN BUSINESS within still more recent years that they have begun to attract general attention. The work of the French psychologist, Alfred Binet, during the first and second decades of this century probably did as much as anything to stimulate the present wave of popular interest. The first really influential suggestion toward utilizing tests in industry came from Hugo Munsterberg in 1913. The use of psychological tests on an extensive scale dates from the work of American psychologists in the army camps during 1917-18. The testing movement is still in its infancy. But in spite of their newness, psychological tests are today enjoying unprecedented popularity. The term "mental test" has come to play an extraordinary role in popular speech. News- papers print '' memory tests ' ' daily. Monthly magazines contain frequent articles pertaining to tests and their use. Many people without psychological training, and almost everybody, it seems, with psychological training, are devising, using, or experimenting with tests of various sorts, or at least criticizing them. It is not surprising that the skeptic cries out, "Fad!" But the demonstrated success of tests in many directions shows that they are no mere fad. Today psychological tests are being success- fully used for practical purposes in a wide variety of fields. Schools, colleges, universities, vocational guidance bureaus, courts, reformatories, public-welfare agencies, physicians, the army, business men-these and others are using them, and new uses are being suggested almost daily. Naturally, the serious adoption of tests for so wide a variety of purposes is gratifying to the psychologist. It is a testimonial to the practical worth and the growing influence of his science. However, there is another side to it. Most psychologists are seriously concerned about the danger of overselling tests to the public. They know that the uncritical enthusiast, with his extravagant claims and unscientific practices, may be the worst enemy of the testing movement. Tests, carelessly prepared or hastily selected by incompetent testers, may lead to disappoint- THE NATURE OF PSYCHOLOGICAL TESTS 3 ing results and the inevitable reaction. An uncritical popular attitude, whether for or against tests, psychologists dread. They would have every business man take a genuinely scientific attitude toward tests-open-minded, yet critical. The business man has learned to intrust his other technical problems to experts who are competent by training and experience to handle the intricate issues involved. Psychologists would have him do the same with his problems of personnel selection and mainte- nance-problems not one whit less technical. WHAT IS A PSYCHOLOGICAL TEST ? But, after all, what is a psychological test ? What sort of a thing is this device, so popular and promising, and yet so liable to misuse? A brief and at the same time complete answer is impossible. A psychological test may assume any one of countless forms. Almost any kind of a task may be used for a test, so long as it can be used to give a reliable indication of an individual's ability or to predict with accuracy his future perform- ance. The test may require him to answer a series of questions or solve a number of problems, and his score will be the number he answers correctly. He may be required to sort, arrange, or mark a series of objects in specified fashion, and we may score him in terms of the time required to perform the task. Or a sample of his work-handwriting, composition, or manual construction-produced under prescribed conditions, may be graded as to its quality by comparison with a graded series of standard samples. Any of these--speed, accuracy, quality, or some combination of these-may serve as the basis on which we measure his performance, and provide the terms in which we express numerically his score. The principle of the psychological test is, then, to use a sample of one's behavior to indicate his abilities or other tenden- cies, and hence to predict his probable future behavior. This is indeed the basic principle of any sort of test, psychological or 4 PSYCHOLOGICAL TESTS IN BUSINESS otherwise-measuring a representative sample for the purpose of predicting what other samples are likely to be. The grain buyer does not need to examine each grain in a carload of wheat, nor does the structural engineer need to measure the strength of every beam that goes into a building. Having assured himself that the sample of grain or the single beam is typical of the rest, he grades or tests that sample and pins his faith to the uniformity and constancy of the conditions which produced those results. Similarly, when we give a typing test, a cancellation test, or an intelligence test to an individual, we assume that the internal conditions which determined his score on the sample are so constant that we can rely on his doing likewise under any similar conditions. The test then is given for purposes of prediction. WHERE ARE TESTS BEING USED ? i. Elementary education.-The grade schools were among the earliest and most active users of psychological tests. It was in their interest that Binet devised the first successful general intelligence scale. Tests of this and other types have multiplied rapidly in number and usefulness. With their aid pupils can be quickly classified into groups according to ability. Instruc- tion can thus be better adapted to children of different grades of ability, and individual progress at the most favorable rate encouraged. Various grades and types of mental defectives can be picked out and assigned to special classes, schools, or institutions, where they can have better care and training. Pupils who are "problems" can be diagnosed with the aid of different kinds of tests, and appropriate methods of treatment, either educational, disciplinary, social, or medical, prescribed. Specially gifted children can be discovered and given suitable opportunities for development. On the basis of test results pupils can be more intelligently advised how to choose studies and occupations. Recent years have seen a rapid increase in educational tests which measure achievement in arithmetic, THE NATURE OF PSYCHOLOGICAL TESTS 5 handwriting, reading, language, and other school subjects. These, which are really varieties of psychological tests, make pos- sible the more accurate estimation of the school progress of indi- viduals. This in turn makes possible the measurement of the com- parative efficiency of different teachers, schools, and school systems. 2. Higher education.-High schools, colleges, and universities are also actively using tests. By their aid students can be classified and assigned to appropriate classes and sections, causes of failure analyzed, and educational or vocational counsel given. Some higher institutions are even using them as a primary or supplementary method of deciding which applicants for admis- sion have sufficient ability to profit by college training. 3. Vocational guidance.-Bureaus, organized to guide young people in their choice of an occupation, have long been interested in the possibilities of tests as one means of discovering individual aptitudes, and thus of advising the youth in what occupations he has a reasonable chance of success, as well as what occupations he should not consider. In those bureaus which have made testing a part of their regular program, probably the chief value of tests thus far has been a negative one-to show pupils what occupations they should not choose. With the development of testing and of vocational information we may expect the uses of tests to become more positive and more specific, as well as more common. 4. Public-welfare agencies.-Many different agencies, public and private, are engaged in studying cases of poverty, delin- quency, and crime. Often they find in psychological tests useful instruments to aid in discovering the causes of social maladjustments, and thus in determining appropriate treatment. Some reformatories and prisons are accumulating in this way large bodies of information which will help us to find out how important a factor low intelligence or other measurable mental abnormalities are in causing crime, and how, therefore, our 6 PSYCHOLOGICAL TESTS IN BUSINESS treatment of individual criminals may be made more rational. Juvenile courts and other courts are using tests to determine how far individual offenders can be held responsible for their misdeeds, and what procedure in each case is wisest. Social case-workers find low mentality a common cause of poverty and misery, and are enabled to recommend corrective or relief measures which are better adapted to individual conditions. 5. Medicine.-The physician in an increasing degree looks to psychology to contribute methods of diagnosing disorders and defects of what he calls a functional character. Psychological tests and examinations, therefore, are playing an important role in the procedure of many specialists. 6. Army.-The successful testing of 1,750,000 men in the United States Army demonstrated the value of tests for such military purposes as picking out the mentally unfit, equalizing military units as to ability, and selecting men of superior general ability or of special trade proficiencies for certain tasks. 7. Business.-Large-scale industry, with its highly specialized occupations and functions, and its growing recognition of the importance of the human factor, is coming more and more to centralize its employment functions in the hands of those who themselves are not experts in the occupations for which they select workers, but who are specialists in problems of employment and labor management. This condition calls in increasing measure for standardized methods by which peculiar aptitudes or proficiencies can readily be discovered and directed into those channels which are most profitable for the employer and most promising for the worker. It is this field-the use of psycho- logical tests in business, and particularly in employment selec- tion-with which the present volume is concerned. In spite of the wide and rapid extension of the testing move- ment to the various fields we have named, we must not over- estimate the extent to which it has advanced. Even among schools, where the usefulness of tests is most clearly apparent, THE NATURE OF PSYCHOLOGICAL TESTS 7 there are undoubtedly more institutions where a standard test has never been seen, nor perhaps even heard of, than there are institutions which have ever thought of using them. In business, probably only a fraction of i per cent of American industrial and commercial establishments have ever seriously tried to use standardized tests, and in very few of these have they passed beyond the experimental stage. Of the thousands of different specialized occupations only a few have been analyzed from this standpoint, and in only a handful of these can we say with assurance that completely satisfactory tests have been found. Nevertheless, the outlook is emphatically favorable to tests. While many business men are skeptics on the question, and a great many more would admit that they don't know enough about tests even to be skeptics, many have been "sold" on the idea. TESTS AND EMPLOYMENT SELECTION As has been suggested, the use of psychological tests in business has been confined almost altogether to the selecting of applicants for employment or to assigning them to jobs. This is by no means the only possible use of tests in business. They may be of value for measuring the increase in proficiency of workers already employed, perhaps in connection with a training program. They may conceivably be used, either alone or in connection with other methods, for selecting workers for special purposes, such as promotion, transfer, or lay-off. But our discussion will center principally in the use of tests for selecting workers and for placing workers in positions for which they are, by native capacity, training, and temperament, best fitted to render service of maximum value. The use of tests for such purposes may be described as an effort to put " the right man in the right place." The desirability of such a condition needs no emphasis. Everyone knows how unsatisfactory are present conditions. No one in particular can be blamed for it; it is one of the by-products of modern life and 8 PSYCHOLOGICAL TESTS IN BUSINESS modern business. Probably the majority of adults feel that they are misfits in their life-work. How often do we hear the expression, "I could have done much better in some other occupation." Such a situation is a loss, both to employer and employee, and therefore to community and nation. A staff of workers ill fitted for the work they have to do means low quantity and poor quality of output. It means discontent, industrial unrest, wasted time, and friction. It means high cost of produc- tion and of labor replacement. For the employee, it means being tied down to tasks that are difficult and disagreeable. It means mediocre pay and slow or no progress. It means chronic dissatisfaction and all the harmful consequences, personal and social, that come from such a condition. To use physiological and psychological tests to avert such loss is not altogether a novel proposal. For a long time railroad and steamship companies have been testing engineers and navigation officers for color-blindness, since inability to dis- tinguish signal lights invites certain disaster. To hire an employee to operate a high-speed machine, who has not the nervous or motor capacity necessary to enable him ever to keep up the pace required, may not be socially so disastrous, but it is none the less unprofitable for the employer and unfair to the worker. The wise use of the proper tests in the beginning would prevent such mistakes, and the employee could be put at a task suited to his capacities where he would be both happier and more useful. The use of tests for vocational selection involves two funda- mental principles: First, the great variability of job require- ments; and second, the great variability of personal qualifications. The former has been on the increase ever since division and specialization of human labor began. A man may be eminently fitted for one job by the very personal make-up that totally JOB REQUIREMENTS AND INDIVIDUAL QUALIFICATIONS THE NATURE OF PSYCHOLOGICAL TESTS 9 disqualifies him for another. Hence, a first step in the selection of workers for any job, whether by tests or any other method, is to ascertain the abilities and traits which that job demands. This usually means some sort of a job analysis of that task. By a job analysis is meant a systematic investigation of the duties, operations, methods, conditions of work, qualifications, prospects, etc., of each job in an establishment. How detailed this must be depends on varying conditions, including the purpose for which analysis is made. Where the purpose is the devising or selection of tests, it may perhaps be very informal. But some familiarity with the job's demands must be gained, for we cannot decide whether a worker fits until we know what it is he has to fit. But of greater direct interest to us is the fact of individual variability. If people were all cast in the same mold, there would be no need of tests or other selective devices; knowing one, we should know all. But individuals are not all alike. From whatever point of view we compare and measure them they are found to vary widely. When a group is measured as to any one particular trait, a few individuals will be found to manifest that trait in an exceptionally high degree. A few others will be found almost or completely lacking in that trait. Most will be found to reveal it in some degree between these extremes. Such variability characterizes any trait we care to name, whether physical or mental. It is a commonplace that no two people look exactly alike. It is no less true that no two people are precisely alike in any other trait, hereditary or acquired, although the differences between them may range from infini- tesimal to great. Human traits are, moreover, practically countless in their number and diversity. The farther we carry our analysis of human personality the more numerous become the specific habits, abilities, and dispositions which we may enumerate, and which may contribute, positively or negatively, to vocational fitness. When we multiply by several hundred 10 PSYCHOLOGICAL TESTS IN BUSINESS or several thousand such highly specialized traits as "ability to add quickly long columns of three- or four-digit numbers," or " tendency to use caution in approaching a grade crossing when driving a motor truck," we begin to realize the tremendous com- plexity of the problem of testing all-or even the most impor- tant-traits which may be of vocational significance. One of the commonest mistakes in interpreting the results of tests arises from the failure to recognize adequately this rich diversity of human traits. It is easy to assume that because an individual ranks low on a general intelligence test he must therefore be of low ability in every direction. True, he may be. Psychologists do not take much stock in Emerson's Law of Compensation. But psychologists do recognize that inferior ability in one direction not infrequently goes hand in hand with superior ability in other directions. There are and must be many kinds of tests-far more than we have today-if we are to measure all aspects of human nature. Let us note, in passing, that measuring these traits by no means implies placing a corresponding moral or social valuation on them. Superior scores and inferior scores refer primarily to amount, not to value. Whether high scores are or are not desirable-and they may be either-depends on the purpose for which we would use them. It is most useful when attempting to measure any of these traits in a group of people to reduce them to quantitative terms and to show the distribution of the various degrees either in a distribution table or a distribution curve. As an illustration of these two ways we may show (p. n) the scores made by 367 pupils of the same school grade on an intelligence test. It is evident that to say of a pupil, "He is in such and such a school grade" tells little or nothing about his degree of intel- ligence. One pupil will be able to solve several times as many DISTRIBUTION TABLES AND CURVES THE NATURE OF PSYCHOLOGICAL TESTS 11 of the problems in an intelligence test as another in the same grade. Certainly, no one expects the same performance in school work of these two. If their relative ability remains constant, we need not expect the same performance of these two Test Score o to 9 Number of Individuals 2 IO to 19 14 20 tO 29 42 30 to 39 83 40 to 49 76 50 to 59 75 60 to 69 45 70 to 79 25 80 to 89 5 DISTRIBUTION TABLE DISTRIBUTION CURVE Number of Persons 90 80 70 60 S°- 40- 30- 20' 10 Score 0-9 10-19 20-29 30-39 40-49 50-59 60-69 70-79 80-89 when they enter business. One will be able to meet with com- parative ease difficult situations in which the other will be utterly at sea. A distribution table or curve for any other of the hundreds of psychological traits we might name would show similar form. Certain other facts about the distribution of individual abilities can be pointed out. When we measure the individual heights of a thousand men we find that their measurements do 12 PSYCHOLOGICAL TESTS IN BUSINESS not fall into sharply distinguishable classes-so many exactly 66 inches, so many exactly 67 inches, and so on. The measure- ments shade into one another by imperceptible degrees; they are continuous, not discrete. The same thing holds true for any trait, physical or mental, we want to measure. Ordinarily this is not a matter of serious concern, but if we try to classify people into a few distinct classes, we find it a very difficult task, since this continuous nature of individual differences com- pels us to put in one class people who are hardly perceptibly lower than those we put in the class above. Making distinctions under such conditions-particularly when dealing with traits of social importance-is often distasteful, though unavoidable. We can point out still another characteristic of these measure- ments. While they cover a wide range, they are not distributed uniformly from high to low, but tend to concentrate around the average. That is, most men are neither tall nor short, but of about average height. A smaller number are noticeably above or below the average. Still fewer are extremely tall or extremely short. Likewise with intelligence, or any other trait, as shown in the distribution curve just presented. The majority are of medium intelligence; a smaller number are bright or dull. A very few are exceptionally bright or exceptionally dull, includ- ing feeble-minded. This bell-shaped curve is the most common form of distribution of any trait. If we use large enough numbers of individuals and make our units of measurement fine enough, the distribution curve tends to become smoother and more truly bell-shaped in appearance, revealing clearly the continuous nature of the distribution. But for smaller numbers, of course, chance variations will be proportionately magnified and our curve will take a less regular form. INDIVIDUAL VARIABILITY AND VOCATIONAL SELECTION The importance of this fact of individual variability for employment selection should now be clear. Of the applicants THE NATURE OF PSYCHOLOGICAL TESTS 13 who present themselves for employment in a given job, not all will possess in adequate degree the various physical and mental characteristics essential for satisfactory work in that job. If it is an easy one, perhaps all but a few may possess the necessary degree of ability, and the problem will be to discover those few and to refuse to employ them for work in which they would probably fail. On the other hand, the work may be of such exacting character that only a few will possess the necessary traits in sufficient degree to make their success reasonably assured, and the problem will be to find these indi- viduals. Some of these traits are primarily products of heredity. Others have definitely been acquired through training or experi- ence. Most, if not all, involve something of both these influences. But to know their precise origin, while it would be interesting, is probably of less practical importance for our present purpose than it is to comprehend their nature, to be able to measure their degree, and to know how to utilize, control, and modify them. We can see now that the notion of vocational selection as "fitting the square pegs into square holes and the round pegs into round holes" is an inadequate statement of the problem. People are not like pegs of a few standard shapes. They do not fall into a few fixed types. If each person possesses a practically unlimited number of traits, and if each of these may exist in any degree from zero to maximum, the possible varieties of indi- viduals are seen to be infinite. There are as many types as there are individuals; or if you prefer, there is only the one type-the human-and all individuals are variants from that type. Like- wise, job demands are almost as varied as are people. Voca- tional selection can never hope to match man and job perfectly. Nor does it aim to. To change the figure, its object is rather to plant the individual in a soil where he can most effectively take root and grow. 14 PSYCHOLOGICAL TESTS IN BUSINESS TESTS AND OTHER METHODS OF JUDGING MEN IN BUSINESS A test of course is not the only device for discovering and measuring useful personal characteristics. Long and intimate acquaintance with a man is, after all, the surest way to tell what he is. But in employing men this is rarely practicable. As a rule, we have to depend on clues which are at best imperfect and partial indicators of character and ability. Different employers place reliance upon one or another of many supposed indications of personality. Among them may be mentioned letters of application, recommendations or references, school records, personal appearance, dress and manners, handwriting quality or handwriting peculiarities, answers to questions, facial characteristics, photographs, impressions made during interviews, and other devices. In so far as these can be proved to afford genuine indications of character and ability, they are, of course, valuable and should be utilized. Some of them afford a more or less direct expression of significant tendencies. But others bear only an indirect, remote, or chance relationship to the traits sought. School records often prove to be to some extent indicative of vocational fitness. The interview is com- monly regarded as indispensable, although it makes a difference what the interview comprises. Some interviewers lay stress on "first impressions," gained from the applicant's demeanor, appearance, dress, carriage, voice, mode of speech, and general responsiveness, while others depend more on information elicited through questioning. General letters of recommendation carry far less weight now than formerly, but confidential replies to direct inquiries are usually regarded as valuable-how valuable depends on how definite and objective the inquiries and answers are. Similarly, the value of formal, written applications depends on the definiteness and relevance of the information they yield. Some employers pin their faith to some system of character analysis, either their own or a commercialized system, based on peculiarities of handwriting or of facial characteristics. THE NATURE OF PSYCHOLOGICAL TESTS 15 But both experience and experiment show that all these methods are subject to error, and opinions based on them may be misleading. But so, indeed, are some of those based on tests. Why then should we use tests ? Because they show promise of being less subject to error and less misleading than the tradi- tional methods of judging applicants. The special strength of tests lies in the fact that they are standardized and more or less "fool-proof" instruments, the actual value of which in any given case can readily be discovered. We do not have to use them on faith. Nor do we have to try to prove their worth by the very questionable expedient so many system-makers use-citing a few favorable instances and ignoring all the cases that fail to support our pet theory. Since test scores are expressed in numerical terms, we can determine by computation just how accurately they indicate the abilities we want to measure. Thus we are in a position to gradually try out, revise, and improve our tests until we reach a satisfactory degree of reliability. We should not summarily reject other methods, but endeavor to submit them to the same sort of critical evaluation to which every good standardized test is submitted. Tests then are not to be regarded as an exclusive method of vocational selection, but potentially as a most valuable supplement to other employment methods. STANDARDIZED AND UNSTANDARDIZED TESTS A psychological test is by no means unique. It is a refinement of methods we are crudely using every day. When a layman is first introduced to a psychological test he may be surprised that it is so simple, and may think he could go ahead and devise one himself. Doubtless he could. In fact, a great many people with no psychological training have tried it. But whether the product will have any value is another matter. Anybody can take a stick and put marks along its length, but whether these marks mean inches or centimeters or nothing in particular is another question. Anybody can prepare a list of questions or 16 PSYCHOLOGICAL TESTS IN BUSINESS set down a number of problems of the sort he has seen in some test, but if these are not standardized nobody can tell what any given response to them means, except in terms of his own subjective opinions. For anyone else they are worthless. An ambitious but untrained amateur sets out to devise a test for a given occupation. He prepares a series of 100 questions and announces that 90 per cent answered correctly shall be regarded as passing. Then a technically trained tester happens along and begins to quiz him, asking such questions as: Are these questions worded absolutely unambiguously? How do you know? Are they all of equal-or accurately graded-difficulty, so that one means as much, and should mean as much, in the score as any other ? Is there a fixed time limit ? Why ? Or why not ? If so, is it made and kept absolutely uniform whenever the test is given ? Will a slow reader or a slow writer make a lower score than one who reads or writes rapidly ? If so, is it a trade test or a reading-writing test ? Will any subject be certain to get the same score, no matter who gives the test ? How do you know ? Will he be certain to get the same score, no matter who marks it? Is there any way in which a generous marker can score him higher than a hard marker could? If not, how has this personal element been eliminated ? Has the test itself been tested ? Why is 90 set as the passing mark ? Did you decide arbitrarily ? If so, what proof of the correctness of this figure have you ? Did you decide by taking the test yourself? If so, what evidence have you that your performance is the correct standard? Or did you try it on a group of workers ? How many workers have you tested with it ? Is this enough to con- stitute a reliable sample ? How did you pick out these workers? At random, or did you select cases favorable for your purpose ? Did the successful workers all score 90 or above? The poor workers 89 or below ? If not, do you know how reliable it is in separating these groups ? How did you determine which are "good" and which "poor" workers ? Was that a reliable method ? THE NATURE OF PSYCHOLOGICAL TESTS 17 Is this test available so that other people can try it out under precisely similar conditions and compare their results reliably with yours ? Are your results available for such comparison ? The well-meaning amateur has perhaps never thought of some of these things at all, or he cannot give answers that prove the test score to be a reliable, valid, and accurately interpretable sample measurement. True, even a standardized test may not rank A+ on all these points. But if the test-maker knows how close to reliability it is, that is worth a great deal. Such considerations distinguish the standardized psychological test from one which is nothing more than "somebody's guess." To work out and establish a sound technique, to obtain accurate norms for comparison and interpretation of test results, and to determine the validity and reliability of a test-these demand thorough psychological and statistical training on the part of the test-maker. His psychological training suggests to him, in the light of what he knows about human behavior, the sort of material and method of presenting it which will call forth the traits he wants to measure. It aids him in finding standards, against which to check up his beliefs about what the test measures. It makes him keenly alert to countless sources of error, unsuspected by the layman, which complicate psychological experimentation. It shows him how to guard against these errors. It helps him to understand the real significance of high or low test scores. His statistical training teaches him how to work out many details of technique; how to ascertain the validity and reliability of the test; how and where to get comparative standards or norms; and how to express these norms so as to permit accurate comparisons. CLASSES OF PSYCHOLOGICAL TESTS Perhaps it will help us to get a clearer notion of the nature of psychological tests, as well as to appreciate their infinite variety, if we try to classify them in various ways. Any classifi- 18 PSYCHOLOGICAL TESTS IN BUSINESS cation, we should remember, is no more than a convenience. It cannot be a rigidly mechanical affair, since there are literally hundreds of different standardized tests, all of which differ in some detail from every other, and consequently no two of which measure exactly the same thing. From whatever stand- point we choose to group them, there will be some tests whose nature is such that we would be equally justified in including them in any of two or more groups. With these precautions, let us proceed. The simplest way of classifying tests is into (i) individual tests, and (2) group tests. The earliest tests were mainly of the first sort, designed to be given to one person at a time. Group tests are more recent and involve a somewhat different technique. Indeed, the first widely used group intelli- gence test was that used in the United States Army in 1917-18. It is harder to maintain uniform controlled conditions with fifty subjects at a time than it is with one. Nevertheless, the great economy of being able to test from fifty to five hundred people in the same time it takes to test one has led to the rapid multipli- cation of group tests during the past few years. Another possible basis for classification is in terms of the form of response which the conditions of the test call for. These may be either (1) oral responses, (2) writing or marking, or (3) performance of any other sort. While individual tests may utilize any or all of these types of response, group tests usually necessitate writing or checking. Since people vary in their normal speed of writing, it is usually desirable to eliminate this complicating factor by having the subject respond by underlining or making a check mark opposite one of several alternative responses printed in the test. While all non-verbal tests may be classified as performance tests, it is hardly worth while to subdivide these, since, as will appear later, the kind of perform- ance called for varies with every test. We could, if we chose, classify tests as to their use, as (1) educational, (2) industrial, (3) clinical, (4) military, and THE NATURE OF PSYCHOLOGICAL TESTS 19 (5) social. But the value of such a classification disappears when we notice that many tests may be used equally well for several purposes. The army Alpha test, for example, has been widely used for every one of the previously mentioned purposes. CLASSES OF TEST ORGANIZATION Tests may be classified as to their form of organization, namely: (i) single tests, (2) test groups, and (3) scales. The second and third classes refer to groups of simpler tests, used either to get supplementary measures of the trait being tested, or to test a more complex trait, or perhaps to test a group of more or less closely related traits. Test groups differ from what are sometimes designated as "scales," in that the latter combine the scores of the separate tests into a single figure-a composite or one-dimensional score-while in the test group we record the scores of the separate tests of the group separately, thus affording a more analytical, multi-dimensional description of the subject's performance. Test groups (which, of course, means something quite other than group tests) are of diverse character, and a further word about the forms they take may not be amiss. Some of them are referred to as profile tests. Thus the Downey will-profile test measures twelve distinct volitional or temperamental tendencies, and the twelve separate scores are graphically recorded in such a way that the line connecting them depicts the individual's will-profile. Still another example of grouping tests is the battery of tests which the employment psychologist finds by experiment are measures of the traits necessary for a specific occupation. While one may prefer to combine, according to some formula, the separate scores of such a battery into a single figure, thus making a scale of it, he may, on the other hand, prefer to reveal, instead of to conceal, the scores on the separate tests and use the profile, test-group, or analytical method of describing a subject's performance. Seashore's series of tests 20 PSYCHOLOGICAL TESTS IN BUSINESS for the various elements entering into musical aptitude exempli- fies this latter procedure. While other possible classifications of tests might be suggested, we shall mention only one more, for many purposes the most suggestive. This is in terms of the kinds of ability or tendency the tests attempt to measure. FUNCTIONAL CLASSIFICATION OF TESTS i. Tests of proficiencies a) Educational tests ft) Trade tests 2. Tests of aptitudes «) General aptitude tests (i) General intelligence tests (2) Mechanical aptitude tests ft) Special aptitude tests (1) Physical tests (2) Motor tests (3) Sensory tests (4) Tests of other special mental functions 3. Tests of character and temperament traits With these various classes of tests and the traits they seek to measure, the rest of this chapter will be concerned. As has been pointed out, those characteristics which the tester measures are almost countless in number and variety. Human ability is not a single, unitary thing, but a vast complex of separate abilities. We cannot hope to get at all aspects of human ability by a single test. We can measure only one phase at a time. To get at every phase would necessitate more kinds of tests than have yet been devised. Some of these phases we designate as "proficiencies" of various kinds, others as "apti- tudes," still others, for want of a better term, we loosely call "traits of character," "temperament," and "personality." When we speak of proficiencies we are thinking of items of knowledge and skill in some well-organized field-a trade, or THE NATURE OF PSYCHOLOGICAL TESTS 21 a school subject-which have been acquired through definite training and experience. Ability to write a good business letter, to take and transcribe dictation, to operate a lathe or dynamo, to speak French, and to perform a thousand and one other skilled acts-these are definite proficiencies. Aptitudes refer to specific human capacities which have not been systematically developed toward definite ends, but the possession of which enables one to acquire with reasonable training the proficiency desired. They are potentialities of proficiency. Most, it is true, have had some casual development in the course of daily life, but not in the form in which we wish to use them. One who has never seen a typewriter may never- theless be proved, by proper tests, to have superior motor and mental aptitudes for this kind of activity, so that with proper training we know he can acquire a satisfactory typing proficiency. One's particular physical aptitudes, sensory aptitudes, motor aptitudes, and intellectual aptitudes, in their countless mutations, help to determine what special items of skill and knowledge one can profitably undertake to acquire. Still other phases of behavior-we refer to them variously as habits of character, emotional and temperamental tendencies, volitional qualities, traits of personality, etc.-can hardly be classified either as aptitudes or proficiencies, since they indicate not so much abilities to act, as natural or habitual levels of acting. A test for honesty, for example, would endeavor to measure not how honest a person could be if he tried, but how honest he habitually is. The importance of such tests for many purposes, we can readily see, is so great that any successful tests of these phases of behavior would be very useful in a great variety of ways. A functional classification must be even less rigid than those which have been given before. Often we cannot be sure which of several functions is being measured. Thus, when we think we are measuring general intelligence, we may really be in part 22 PSYCHOLOGICAL TESTS IN BUSINESS measuring the subject's proficiency in reading the instructions. A moral judgment test is likely to be more a test of intelligence than of morality. A low score on some intellectual test like color-naming or letter-canceling may prove, on further investiga- tion, to be really due to some sensory or motor defect. Some- times what we are testing varies with the way in which we conduct the test. If we ask a subject to name the colors of several objects, it may really be a test of his knowledge of color names instead of a test of his color sensitivity. Hence, we test color- blindness by handing him several skeins of colored yarns, asking him to select those which resemble in color certain stand- ard skeins, and we avoid mentioning color names. It follows, then, that the significance of any test must not be obscured by attaching the wrong name to it, such as color-blindness, instead of color-naming. Some even go so far as to advocate using no test names which imply particular mental processes. TESTS OF PROFICIENCIES Educational tests.-There are two classes of situations in which we most often demand of individuals the attainment of a certain body of organized knowledge and skill-the school and the trade. Proficiency tests have been in daily use in both these fields ever since teachers began examining their pupils and masters began examining their workmen. But it is only recently that these examinations have been standardized, so that we can be sure the tests are reasonably valid and results widely comparable. Today, for almost every elementary and secondary school subject there are available one or more standardized tests. An idea of their number and scope may be gained from the following list (p. 23), taken from a recent bibliography.1 While of course such tests are meant primarily for schools, it is readily apparent that many of them may have value for employment purposes, particularly for office jobs. 1 Bibliography of Tests for Use in Schools, World Book Co., 1922. THE NATURE OF PSYCHOLOGICAL TESTS 23 STANDARDIZED EDUCATIONAL TESTS Subject Number of Tests English .... 48 Composition, copying, grammar and punctuation, language, poetry, and spelling Mathematics •••• 45 General, arithmetic, algebra, and geometry Reading .... 40 Handwriting .... 24 Science .... 24 General, biology, chemistry, geography, and physics Foreign languages .... 17 General, French, German, Latin, and Spanish History .... 14 General, American, and ancient General school-survey tests .... 7 Commercial subjects .... 5 Drawing .... 4 Home economics '. .... 4 Vocational subjects and manual training .... 4 Music .... 3 Physical training .... 3 Citizenship .... 1 Journalism .... 1 Trade tests.-Trade tests have been worked out for special kinds of work in a few industries, and for a few standard occupa- tions, such as stenographers, typists, and general office clerks. These are usually standardized samples of the work done in the trade. We must distinguish clearly between these and the special aptitude tests which will be described later, and which are used for industrial selection. The latter are not trade tests, even though they relate to certain trades. The trade test is a test designed to measure not merely the aptitude for acquiring a trade, but the actually learned skill and knowledge involved in that trade. Trade tests can be of wide general use only when the specific items of knowledge and skill implied by the test are everywhere uniform, not limited to the practices of a single institution or even a single community. 24 PSYCHOLOGICAL TESTS IN BUSINESS The most conspicuous achievement in the trade test field is the group of trade tests devised for the United States Army. During 1918 tests for eighty-three trades were devised and standardized by the trade test division and used in various army camps for selecting those men who had the necessary trade ability-knowledge and skill-necessary in a great military organization. As examples of trades for which such tests were devised might be mentioned auto repairer (general test, and four sub trade tests), baker and cook, office clerk, electrician (five tests), stable boss, machinist and mechanic (twelve tests), printing trades (four tests), sheet metal worker (two tests), telegraph and wireless operator, and many others. These tests were standardized in industrial plants and were designed to be administered and scored by officers who themselves were wholly unfamiliar with the trade in question. They were arranged to reveal four grades of trade ability-novice, apprentice, journeyman, and expert-and were brief enough to meet the demand for economy of time. Three general types of trade tests were used: (a) the oral- question test, (b) the picture-question test, and (c) the perform- ance test. The first type, which was the most common, com- prised questions pertaining to the trade which could be, as a rule, correctly answered in only one way, and by a single word or brief phrase. This permitted scoring each answer unequivocally as right or wrong-a very important technical detail. As an example, among the fifteen questions in the "Bricklayer-General " test occur the following: Q. What are headers used for in a brick wall ? A. To bind (bond) (tie) Score 4 Q. What do you call a course of brick laid lengthwise of a wall? A. Stretchers Score 4 Q. In setting a sill course how much pitch do you give it ? A. One-quarter to three-quarters of an inch Score 4 THE NATURE OF PSYCHOLOGICAL TESTS 25 As the maximum possible score is 60, a score of 53 or above gives an expert rating; 44 to 52, a journeyman rating; 13 to 43, an apprentice rating. In the picture tests, used for cobblers, general carpenters, storage-battery electricians, etc., pictures of tools, processes, objects, or parts were shown and the questions referred specifically to these. In the performance tests-for example, truck driver, inside wireman, steamfitter, etc.-a representative, though usually brief, task was set, a standard set of tools, materials, and instructions was provided, and performance was scored in terms of the quality of the finished product and the time required to complete it.1 TESTS OF APTITUDES The second general division of our functional classification is tests of aptitudes. Some of these aptitudes are general; others are more specialized. While tests of acquired proficiencies are important, they are far from being the only tests the employer needs to use. A large part of the work done in offices, stores, and factories is not of the standardized skilled-trade type at all. There is often no opportunity to acquire the precise knowledge and skill necessary to do certain kinds of work except within the business establishment itself, either in a vestibule school, by means of some apprenticeship system, or by working under the eye of a fellow-employee or foreman until the work is learned. In such cases, instead of proficiency tests, what is needed are aptitude tests-tests which will help to pick out those applicants who have the physical, sensory, motor, or intellectual aptitudes which will enable them to master the essential operations with a minimum of time and a maximum of efficiency. Sometimes these aptitudes are rather highly specialized. The abilities necessary for the routine work of an inspector of certain small parts may be accurate and quick co-ordination of par- 'J. C. Chapman, Trade Tests (Henry Holt & Co., 1921) gives a detailed de- scription of many of these tests, as used in the army. 26 PSYCHOLOGICAL TESTS IN BUSINESS ticular arm, finger, and eye movements, sufficiently good visual and touch discrimination to detect flaws, and enough brains to understand and remember the simple instructions. But in some other cases, the work is not so simple and mechanical. New problems are constantly arising, and an employee, to be competent, must be able to "use his head." In such cases what is most wanted may be a test of general intelligence, by which it will be possible to judge his ability to adjust himself to new situations. GENERAL APTITUDE TESTS Tests of general intelligence.-Tests of general intelligence are also referred to as general ability tests, mental alertness tests, or learning ability tests. For our purposes we may treat these terms as synonymous. They constitute, without doubt, the most conspicuous class of tests at present. So conspicuous are they, and so important, that not infrequently the term "general intelligence test" is thought of as synonymous with psychological test. But we have said enough to show that they constitute only one class of psychological tests; and we should not forget that, being only one class, their range of usefulness is definitely limited. Nevertheless, for many purposes it is so im- portant to know the average level of one's intellectual abilities that we are warranted in dwelling at some length on them. The earliest successful intelligence tests were devised by Alfred Binet and are called by his name, although various revi- sions of the original Binet scale have been produced. To pick out those children in the Paris schools whose general mental development was so delayed as to necessitate their being placed in special schools, he devised the plan of confronting them with a series of simple tasks of graded difficulty. By long experience he found which tasks were within the power of the average six-year old, the average seven-year old, and so on. These tasks included recognizing and naming simple objects, THE NATURE OF PSYCHOLOGICAL TESTS 27 stating similarities and differences between things, performing three simple commissions given simultaneously, pointing out absurdities in certain statements or omissions in pictures, rear- ranging disarranged sentences, telling the proper action in given emergencies, and the like. The Stanford Revision,1 the most widely used American adaptation of the Binet method, provides six tests for each year from three to ten, eight for twelve-year, six for fourteen-year, six for average adult, and six for superior adult intelligence. In the language of the Binet scale the child who can pass the tests equivalent to normal eight-year-old performance is said to have a mental age of eight. If his actual age is eight, he is said to be of normal intelligence, and his intelligence quotient or his I.Q.-mental age divided by chronological age, decimal point being omitted-is zoo. If, however, he is only six, but with eight-year-old intelligence, he is described variously as two years mentally advanced, of superior intelligence, bright, or of an I.Q. of 133. If, on the other hand, he has attained this eight-year intelligence level only at ten years of age, he is termed dull, retarded, or of an I.Q. of 80. Such terms are rapidly passing into everyday speech. In dealing with adults, we find the mental age and the I.Q. concepts of very limited usefulness; and if those adults are normal-i.e., not feeble-minded or border-line cases-we may well dispense with these terms. In the first place, the tests them- selves are inadequate in number and variety for the higher stages -above twelve years-of development. In the second place, these upper-age tests have not been so satisfactorily standardized as those for earlier years, since it is harder to round up and test a really average group in order to get reliable age standards after they have passed beyond the compulsory school age. In testing those of high-school age and beyond-to a large extent in the elementary grades as well-other methods have *L. M. Terman, The Measurement of Intelligence, Houghton Mifflin Co., 1916. 28 PSYCHOLOGICAL TESTS IN BUSINESS been evolved. These may be briefly characterized as point- scale group tests. Group intelligence tests.-Group tests may possibly be less accurate than individual tests, where the examiner can watch each individual and adapt his manner to the person's emotional attitude. But the greater economy of group tests has led to their wide adoption in schools, employment offices, and other places where it is desirable to be able to test several people in a short time. The Army Alpha is the prototype of many general intelli- gence scales which have been devised in the past few years. The bibliography previously mentioned lists no less than sixty-five intelligence scales. Not all of these serve the same purpose. Some are for primary-school children, others for high-school or college students, and still others primarily for employment pur- poses. A few are individual, but most are group scales. The National, the Illinois, the various Otis tests, the Thurstone, the Thorndike (Columbia college-entrance test), the Terman, the Haggerty, and many other equally well-known tests have largely replaced the Alpha scale, although having the same general type of material and organization. The Chicago Group Intelligence Test,1 which is here repro- duced in full (see pp. 30-36), has been extensively used with high-school students, and is typical both in material and form of the other scales mentioned. One modification worth noting is the "self-administering" or "cycle" form of organization, of which the Thurstone and some of the Otis tests are examples. In these the different kinds of tests are not separately timed, but are "scrambled," reappearing again and again throughout the test. This makes it easy to administer without error, particularly in the hands of those who 1 Thanks are due to Professors F. N. Freeman and H. 0. Rugg for permission to reproduce this scale. The time allowances on the five parts of the test are one, three, two, two, and one minutes, respectively. THE NATURE OF PSYCHOLOGICAL TESTS 29 are not trained testers, since no verbal instructions have to be given and the examiner needs to keep track only of the total time-thirty minutes being the usual time allowance. The inter- mingling of different kinds of material practically prevents any analytical study of relative performance on different kinds of material within the scale. Tests for illiterates.-The problem of testing the intelligence of illiterates is met by using one or more of several kinds of performance tests. Some of these are individual tests, while others can be used as group tests. Among these may be men- tioned the following: form boards of various sorts, in which variously shaped blocks are to be fitted into a frame or placed in similarly shaped holes in a board; dissected picture puzzles of various types; arranging a series of pictures in such order that they tell a connected story; completing imperfect pictures, as for example a face lacking a nose; and mazes, through which the subject has to find and trace the shortest correct path. Many other tests requiring no ability to read or write have been devised. The Beta group test, used in the army for testing illiterates, in- cluded seven such tests arranged in pictorial or other non-verbal form, printed in a booklet, the directions being given in panto- mime and the responses made by marking or checking, or at most by copying single letters or figures. Of this same type are numerous pictorial group tests for primary-school pupils who have not yet mastered reading skill. Some psychologists assert that we should distinguish at least three broad classes of general intelligence-the abstract or verbal, the mechanical, and the social types-with many subvarieties under each; and that the kind of general ability most of the present general intelligence tests measure is only one class-ability to deal with verbal material and abstract thinking. There can be no doubt of the great importance of this kind of intelligence; but it would be a mistake to think that it is the only kind that counts. Many people who would 30 PSYCHOLOGICAL TESTS IN BUSINESS The Chicago Group Intelligence Test F. N. FREEMAN H. O. RUGG University of Chicago Form B GENERAL INFORMATION Your Name Date Are you a boy or girl When is your birthday How old are you ..years. Name of your city Name of your school Name of teacher of the class in which you are taking this test DO NOT WRITE BELOW THIS LINE SUMMARY OF RESULTS Test No. No. of correct examples No. of incorrect examples Score 1 9 3 4 5 TOTAL SCORE Percentile rank (P. R.) .. (1. e. the pupil's rank in a typical group of 100 pupils: Example: A pupil with P. R. of say. 83, is exceeded in mental ability by. 17 per cent of the pupils in the same school grade.) GENERAL DIRECTIONS Teacher and Class should read these General Directions together. Be sure each pupil understands what is *ta be done. In this booklet there are five exercises which help to tell how well you can do certain things. You will work one exercise at a time and the teacher will tell you. when to begin each one and when to stop. When the teacher says "Begin" turn the page and begin working them exactly in the order in which they are numbered. Work as rapidly as you can but remember that it is impor- tant to get the examples right. Try each one but do not spend longer on any one example if you are sure you cannot do it. Do not write anything after you are told to "stop." 'As soon as you stop working one exercise, the directions will be given for the next one. Do Not Turn a Page Until You Are Told to Do So. Directions for Exercise No. 1-"Opposites" One of the three words which is joined to each of the words in the list in Exercis No. 1 expresses a meaning opposite to that of the word at the left of the brackei For examnle: small low short 7 "Short" expresses the opposite of the word "long." The other two do not. With each set of words check the word which expresses the opposite meaning. Now turn to Exercise 1. Long THE NATURE OF PSYCHOLOGICAL TESTS 31 EXERCISE No.. 1-"OPPOSITES" sloping down level to harm to destroy to curse 1. up 11, to bless 2. good mean bad weak 12. forcible considerate gentle peaceful late on time earlier ignorant illiterate stupid 3. early 13. 'intelligent 4. above 'below beside inside 14. tender rough cruel inconsiderate sell keep give away quiet self controlled sad 5. buy 15.. animated 6 cruel kind charitable loving 16. silly learned wise sensible smooth steep soft [careful safe [cautious 7. rough i 17. reckless 8. stale [clean new [fresh 18. enthusiastic uninterested apathetic unemotional confident certain secure careless superficial vacillating 9. doubtful 19. thorough 10. belief doubt denial hesitancy 20. ignorant [well informed wise [unskillful (Work No. 11 next-above) No. attempts u No. wrong No. right Score This is an example which explains the next exercise. In each row of numbers below, each number follows the one before it according to a rule. In this series, 7 8 9 10 11 12 13 14 each number is obtained from the one before it by adding 1 to it. Suppose the num- bers read 7 8 9 11 12 ... 14 Two numbers are missing. These can be supplied by adding 1 to 9, and 1 to 12,'giving 7 8 9 10 11 12 13 14 In the same way the missing numbers in this row,. 3 6 9 18 21 33 are 12, 15, and 24, 27, 30. On the next sheet there are rows of numbers like these from which some numbers are missing. You are to fill in the missing numbers. Remember speed and accuracy are both important. 32 PSYCHOLOGICAL TESTS IN BUSINESS EXERCISE No. .2-"MISSING STEPS IN THE SERIES" (A) 1 2 3 4 5 .... 7 8 9 .... 11 12 13 (B) 29- 24 .... 14 9 4 (C) 26 22 .... 14 10 .... 2 (D) 64 32 16 1 (E) 1 5 25 .... 625 3125 (F) 486 162 .... 18 6 .... (G) 7 16 19 28 31 40 .. .. (H) 3 18 33 53 (I) 72 37 2 (J) 84 4 54 30 6 No. attempts No. wrong No. rights..™- Weighted score Directions for exercise No. 3-"Proverbs" On the next sheet is given a list of proverbs or sayings arranged in pairs. Some of the pairs have the same or nearly the same meaning. In others, the meaning of one proverb is the opposite of the meaning of the other. In still others the meanings of the two proverbs are not related at all. Write an S before the pair which means the same or nearly"the same; an 0 before those which have an opposite^meaning, and an N before all other pairs. EXAMPLE: s (a) One swallow does not make a summer. (b) One tree does, not make a forest. THE NATURE OF PSYCHOLOGICAL TESTS 33 EXERCISE No. 3-"PROVERBS" 1 ((a) Shallow brooks are noisy. ((b) Little dogs bark the most. 2 ? (a) Every cloud has a silver lining') (b) One tree does not make a forest. 3 {(a) Married in haste, repent at leisure. ((b) Quick loving a woman means quick not loving a woman. i(a) A man of words and not of deeds is like a garden full of weeds. (b) Actions speak louder than words; 5 ((a) Absence makes the heart grow fonder. ( (b) Out of sight, out of mind. 6. ....... $ Don't put all your eggs in one basket. ••-"J (b) The rat that has but one hole is soon caught. 7 ((a) Make a virtue of necessity. {(b) Necessity knows no law. g ((a) Silence is golden. ((b) Silence gives consent. g ((a) What you would have well done, do yourself. ((b) Two heads are better than one. IQ ((a) A penny saved is a penny earned. I (b) A stitch in time saves nine. No. attempts ......... No. wrong..... ™ No, right™™™. Score.™.™.,., Weighted score Directions for Exercise No. 4-"Analogies" In each of the lines on the next sheet, the first two words stand in a certain relation to each other. Notice that relation and draw a line under the one word of the five in the parenthesis which has the same relation to the third word as the second has to the first. above-top: below-(?) (above, bottom, hang, over, sea) sky-blue: grass-(?) (grow, dead, green, cut, burned) cold-heat: ice-(?) (frost, refrigerator, steam, grass, cream) SAMPLES 34 PSYCHOLOGICAL TESTS IN BUSINESS EXERCISE No. 4-"ANALOGIES" 1. dress-cloth: hat-(?) (pretty, straw, woman, big, vote) 2. winter-cold: summer-(?) (short, dry, hot, long, sea-shore) 3. above-sky: below (?) (blue, heavy, level, earth, under) 4. food-eat: water-(?) (bucket, desire, drink, pour, sea) 5. sweet-sugar: sour-(?) " (bread, man, sweet, vinegar, miserly) 6. skirts-girl: trousers-(?) (hat, coat, boy, vest, advertisement) 7. uncle-nephew: aunt-(?) (brother, cousin, sister, niece, father) 8. sheep-wolf: mouse-(?) (fur, cat, kitten, dog, bite) 9. simple-complex: easy (?) (brittle, money, hard, work, compound) 10. anger-violence: love-(?) (woman, gentleness, child,'kiss, sweet) 11. eat-hungry: drink-(?) (fish, whisky, water, wine, thirsty) 12. truth-gentleman: lie- (?) (live, give, rascal, falsehood, dishonesty) 13. spoon-soup: fork-(?) (plate, cup, knife, meat, crackers) 14. pursue-bold: flee-(?) (run, afraid, fox, child, bear> 15. cannon-rifle: big-(?) (army, small, gun, bullet, bomb) 16. breeze-cyclone: spring-(?) (winter, shower,.water, bath, flood) 17. corn-horse: bread-(?) ' (flour, daily, butter, man, meal) 18. eat-mouth: breathe-(?) 7 (life, air, strangle, nose, foul) 19. pages-book: twigs-(?) (tree, wood, branch, leaves, bark) 20. rafters-house: skeleton-(?) (skull, body, bones, grace, frame- work) No. attempts. - No. wrong...™™™ No. right Directions for Test No. 5-"Selecting Best Reasons" On the ngxt sheet make a cross before the best answer to each question. THE NATURE OF PSYCHOLOGICAL TESTS 35 EXERCISE No. 5-"SELECTING BEST REASON: 1. Why are beans good food ? Because they 1. are raised in many states. .2. are nourishing and economical. 3. are eaten by many peoples. 4. look well when served. 2. If the health department says that the drinking water of the city is not pure, what is the thing to do? 1. drink coffee.? 2. drink«.only at meal times. 3. boil the water before using it. 4. eat plenty of salt. 3. Why should every house have a fire, extinguisher? Because 1. they are ornamental. 2. they cost only a little. 3. they are easily used. 4. they easily prevent a serious fire. 4. Why should food be chewed before swallowing? Because 1. it is better for one's health. 2. it is good for the teeth. 3. it is better manners. 4. solid foods are wholesome. 5. If you fall asleep on the train and ride past your station, should you 1. pull.the bell rope and stop the train instantly? 2. tell the conductor to take you back? 3. get off at the next stop and take another train back to your station ? 4. jump off and walk back? 6. Why is gold more costly than lead? Because 1. it is of finer appearance. 2. it is more scarce. 3. it is used more for jewelry. 4. it is yellow. 7. Why judge a man by what he does rather than by what he says? Because 1. it is wrong to tell a lie. 2. a deaf man cannot hear what is said. 3. it is wrong to judge anybody. 4. what a man does shows what he really is. 8. If your opinion were asked of a man whom you didn't know, what would be your reply? 1. I think he must be a fine man. 2. I will look him up and get acquainted. 3. I don't know and can't say. 4. I doubt his ability. 9. When you have a hard task to perform, should you ? 1. leave'it until tomorrow and you may not have to do it? 2. leave it to the other fellow? 3. get to work on it at once ? 4. think out the best plan; then go to work? 10. If you should sprain your ankle while on a hiking trip,.what should you do? 1. lie down under a nearby tree? 2. shout to attract attention? 3. take a drink from your water bottle ? 4. ask someone to send an ambulance? No. attempts No. wrong No. right. Weighted score 36 PSYCHOLOGICAL TESTS IN BUSINESS score low on the Alpha show a high degree of ability to deal quickly and competently with problems involving tools, or mechanical or manual operations, even problems entirely novel. Tests of general mechanical aptitude.-The Stenquist mechan- ical aptitude and assembling tests are good examples of tests for getting at this type of general ability. The assembling test consists of a series of small mechanical objects taken apart-a bicycle bell, a paper-clip, a cheap lock, a clip clothespin, a mouse- trap, a cupboard catch, etc.-which the individual is instructed to put together correctly as rapidly as possible. His perform- ance is scored both in terms of accuracy and time. The mechanical aptitude test is printed so that it can be given as a group test. Pictures of mechanical devices, tools, and the like are shown, and the subject indicates by letters and other marks the objects or parts which belong together. While all these tests involve more or less experience and training, and hence are definitely tests of acquired ability, that ability is of a general mechanical sort rather than ability pertain- ing to any specific occupation, and is to a large extent such as one of a mechanical turn of mind is likely to pick up for himself outside a trade school. They are, therefore, tests for aptitude and not trade tests. TESTS OF SPECIAL APTITUDES In contrast to the general aptitude tests we have described, tests of the next class, those of special aptitudes, are so varied that it is impossible to reduce them to elements or classes. Hence we can do little more than enumerate and describe some of those in most common use. The best-known reference book containing material on such tests is G. M. Whipple's Manual of Mental and Physical Tests, although far from all of the useful tests are described therein. We have suggested that THE NATURE OF PSYCHOLOGICAL TESTS 37 these tests may be crudely classified as physical, motor, sensory, and intellectual-mental in the narrower sense-and this order will serve to orient us in our discussion. Physical tests are not, of course, strictly speaking, the psychologist's business. But anyone who has worked with tests knows how impossible it is to draw fine distinctions between physical and mental tests. A test of endurance of hand-grip may test will as much as muscle. Leaving aside such hair- splitting issues, the psychologist does often have to know a great deal about traits which are definitely physical in order to supple- ment and interpret his other tests. Height, weight, skull- diameter and capacity, breathing capacity, chest expansion, strength of grip, endurance of grip, and a dozen other body measurements are a necessary part of any complete series of tests, whether for educational, clinical, or vocational purposes, and every tester is likely to have to make them. To some degree this is also true of the more strictly medical tests. Tests of motor aptitudes play an important role in many industries. Tests for quickness and accuracy of movement, and of muscular co-ordination and control, appear in the test literature in a variety of forms. Quickness of movement may be tested, for example, by tapping with a metal stylus on a metal plate, both arranged in an electric circuit with an electric counter which counts the number of taps in a given period. One kind of bodily control is measured by standing the blindfolded subject beneath a horizontal, inverted, smoked-paper surface upon which a vertical pointer, fastened to his head, traces his involun- tary movements. Other motor tests have been specially devised to duplicate as closely as possible the conditions under which specific manual work is to be done. Sensory tests are of many sorts, and most are well known. They include tests for optical defects-errors of refraction, convergence, etc.-as made by the optometrist with an ophthal- 38 PSYCHOLOGICAL TESTS IN BUSINESS moscope or vision chart and lenses, and which are necessary where perfect eyesight or perfectly fitted glasses are essential; tests of color-blindness for occupations like railroading or naviga- tion; watch-tick or whisper tests for acuity of hearing; tests for sense of pitch with tuning forks of slightly different vibra- tion rates, or with special phonograph records; tests of other musical capacities, such as those devised by Seashore,1 some of which are available on phonograph records, others requiring special apparatus. Tests for the sense of equilibrium and the perception of slight movements, by whirling or tilting the subject and observing his verbal or pointing responses, have been tried in aviation. Tests of the simpler mental functions are among the oldest we have. They vary from the simpler attention and perception tests, up through the memory, association, and imagination tests, to the tests of judgment and higher mental processes. Strictly speaking, none of these are tests of single or even of simple processes; their simplicity is at most relative. Probably it is of very little practical importance for the tester to know just how to classify or name them. It is more important to know something of their scope and nature, and to find, by experiment, how closely ability in each of them correlates with the abilities we want to measure. As an example of tests which have found a place in industry, we may mention the following: The cancellation test, in which the subject is required to cross out each "A" or other letter or letters in a page of printed capitals in the shortest possible time. CANCELLATION TEST KDFLHAZLPSHUEBDARSBENXRDBAPVKEAXZAF LBPRENFABDXLSERAZUABKEHMRSZBALPTHDE AMKFZCEBUASPORDPHRKBNVEASBAUTDMXHES HNRSPEZKDNBFAGLCEKBSDTRNEOKGBVADABR 1 C. E. Seashore, Psychology of Musical Talent. Silver Burdett & Co., 1919. THE NATURE OF PSYCHOLOGICAL TESTS 39 Sorting tests of various kinds may involve sorting cards, though they may use instead the special materials with which the job deals. Memory-span tests measure the number of letters, digits, or unrelated words one can remember and repeat after hearing them once. These have been suggested as useful for telephone operators and for stenographers. Substitution tests, in which a series of digits are to be substituted for corresponding letters according to a prescribed code, have been used by some investigators for testing typing proficiency. Many of the separate tests which appear as parts of various general intelli- gence tests may be used separately for testing more limited mental functions. There are literally scores of other special mental-ability tests which we cannot even enumerate here, but which get at one or another of a variety of traits of vocational importance.1 A number of tests have been devised by various investigators for some particular occupation. Such, among others, is the apparatus devised by Miinsterberg and described in his Psy- chology and Industrial Efficiency, for testing street-car motormen, or Link's modification of the Bogardus fatigue apparatus for testing machine operators. These cannot be classified as mental tests in the strictest sense, since they involve sensory and motor as well as intellectual aptitudes. But they are special rather than general because each was devised to fit a single occupation. TESTS OF CHARACTER, TEMPERAMENT, AND VOLITIONAL TRAITS Among the newer psychological tests are those for the many distinguishable volitional, temperamental, emotional, and character traits. These have not yet passed beyond the experi- mental stage. One of the best known is the Downey will- 1H. C. Link, in the appendix to his Employment Psychology (Macmillan Co., 1919), describes a number of these tests which he found useful. 40 PSYCHOLOGICAL TESTS IN BUSINESS temperament test, available both in individual and group form. It uses chiefly modifications in the speed and form of handwriting as its material. It is designed to measure several temperamental or volitional traits, to which the following names have been given: "speed of movement," "freedom from load" (tendency to work continuously at maximum speed), "flexibility," "speed of deci- sion," "motor impulsion," "reaction to contradiction," "resist- ance to opposition," "finality of judgment," "motor inhibition," "interest in detail," "co-ordination of impulses," and "volitional perseveration." Recent activities of many other workers in devising tests of such traits indicate the interest in this inadequately analyzed aspect of personality. Tests of such instinctive and emotional traits as susceptibility to anger or fear, or of such volitional traits as aggressiveness and perseverance, have been described in psychological periodicals, but they have not been sufficiently standardized to warrant description. Several tests of "moral judgment" have been proposed, such as a series of hypothetical acts which the subject is required to rank in order of relative merit or demerit. These have been used to some extent among delinquents and defectives and reveal the frequent lack of ability on the part of these unfortunates to estimate correctly even the more obvious moral values. But "circumstances alter cases," and, moreover, even the ablest judges will not always agree on which of two sins is the more heinous, or on which of two acts is the more praiseworthy. Hence such tests on any advanced scale have very doubtful value. But what is more important, moral judgment is only one factor in morality. Most people know better than they do. An intelligent criminal could doubtless make a perfect score in a test, telling what should or should not be done in certain situations, providing he thought he had anything to gain by passing such a THE NATURE OF PSYCHOLOGICAL TESTS 41 test. To know what, according to common standards, ought to be done, is no guaranty that one will have the habit of doing so or the desire to do so. The Voelker tests of habits of moral trustworthiness used in studying moral education among Boy Scouts have shown that there is a possibility of character tests. These tests set up critical situations to find whether the boy succeeds or fails in conforming his conduct to the accepted standards of honesty, punctuality, faithfulness, and the like. The success of such tests rests wholly upon the subject's not knowing that he is being tested. This condition it is rarely possible to achieve. There is another phase of social personality, however, more easily examined, and that is the range of one's interests. A common method is to provide a series of questions, each with alternative answers to be checked, or else statements some of which are true and some false, to be correctly indicated, and per- taining to a wide range of human interests. This method assumes, probably with justification, that the range and character of a man's knowledge is an index of the range and character of his interests. If one answers correctly every question about base- ball, fishing, and poker, and misses most of those about religion and literature, it indicates at least something about him. One test of this type asks fifty questions about such diverse recrea- tional interests as popular songs, sports and games, underworld slang, etc., and is said to be useful for selecting salesmen for work where wide and ready social adaptability to men of various classes is necessary. It will be evident by now that the field of psychological tests is one capable of indefinite-almost infinite-expansion; for only a comparatively few representative samples have here been mentioned. Some of these have proved their worth for occupational selection in various lines; others may yet do so. One who endeavors to select tests for a given occupation finds at 42 PSYCHOLOGICAL TESTS IN BUSINESS hand many ready-made standard tests, and his principal task may be to find out whether these tests do really serve to distin- guish good from poor workers. If he cannot find ready-made tests to meet his needs, his first task is to devise and standardize tests which he thinks will be worth trying out. But, in any case, his problem is not solved until such tests have been actually tried out on workers of known ability and the value of the tests for the specific purpose in mind scientifically determined. CHAPTER II SCIENTIFIC METHOD IN CONSTRUCTING PSYCHOLOGICAL TESTS FOR BUSINESS The distinctive characteristic of science is its method. This is as true of science in the service of industrial selection of workers as of science in the physics laboratory or science in weather prediction. The most important single fact about a psychological test is that its value has been scientificaUy determined. Stand- ardized psychological tests differ from ordinary employment methods not so much in their form and contents as in the process of scientific proof to which they have been subjected. The tests may contain some of the same questions that a foreman asks his applicants. Or the test devices may look like meaningless puzzles or amusing parlor games. The question, however, is not which devices look as though they ought to work best. It is simply: What scientific evidence is there that this test does work ? The insistence is upon scientifically proved methods of selection, regardless of whether tests are in question, or some other pro- cedure. Any of the definite items of information utilized in the acceptance and rejection of applicants can be subjected to pre- cisely the same scientific check and evaluation as can tests. It is not a whit less scientific, for example, to utilize a knowledge of the applicant's age, years of schooling, years of experience, and the like, than it is to use his scores on a series of psychological tests, provided the former items have been subjected to the same painstaking statistical check as have the latter. To obtain the point of view of scientific research in employ- ment procedure is more important than to decide the pros and 43 44 PSYCHOLOGICAL TESTS IN BUSINESS cons of particular tests or even of tests in general. If all the psychological tests in existence were destroyed tomorrow, there would remain the valuable influence they have exerted in the direction of increased application of scientific method in personnel practice. While, then, our discussion here will deal definitely with the scientific method of constructing and verifying psycho- logical tests, it should be clear as we proceed that the methods as such are of general applicability. THE TESTING OF PSYCHOLOGICAL TESTS Not all psychological tests are good. Some of them are worse than useless. The mere fact that tests are used in a given situation does not mean that results of demonstrated value are being achieved. Anyone can make a test, and anyone can use tests. The process is simplicity itself. One merely asks some questions, or brings together some puzzles, or instructs the per- son tested to perform some acts-acts that are foolish or sensible, depending upon the temperament and ingenuity of the test- builder. But with the statement that it is easy to construct tests, and that it is easy to give tests, let us always link the even more significant statement that there are bad psychological tests as well as good psychological tests. Constructing tests and constructing effective tests are not identical. Using tests on faith is a very different process from using tests on the basis of their demonstrated value. Here, pre- cisely, is the significance of the entire technique of test building and test standardization. The problem is in part the devising or choosing of tests that appear promising, but the decidedly more important part of the task consists of "testing the tests," that is, of ascertaining just how well each test works. If it were possible to tell by looking at tests which ones would prove satis- factory, and which would not, there would obviously be no need of careful methods for determining their relative merits. But experience has demonstrated that the description of a test reveals SCIENTIFIC METHOD 45 little concerning its value. The one safe method by which to judge the excellence of a test is to test it. A priori opinions regarding tests have been repeatedly contradicted by later results. So true is this, in fact, that scientific workers in the field fre- quently begin their search for a useful test by trying a rather large number of different tests in the confident expectation that many of those tried will prove unfit and be rejected. But these weak links cannot be detected and eliminated in advance. Re- course must be had to experiment. It follows quite reasonably that if one is to form sound judg- ments about psychological tests, it is more important to under- stand the technique of evaluating tests than to be familiar with particular sorts of tests or to know what these are alleged to be measuring. The real question about a psychological test is: Does it work? That question implies that there is a definite purpose for which the test is desired, and that the success with which that purpose is satisfied is, in the final analysis, the only basis on which the value of the test can be estimated. Ordinarily, the purpose is clear-the test is to aid in selecting people with respect to their ability for some particular line of work or course of training. The excellence of the test in fulfilling that purpose is the thing to be determined. Does the test enable us to make predictions as to people's fitness for the job in an appreciably better, cheaper, or more expeditious manner than can be made without the test? If it does, it is a successful test; if not, it is unsuc- cessful. The most straightforward procedure for ascertaining the "success" of a test can be very briefly stated: The test is given to a number of individuals whose ability in the job is already known or can later be determined. If test scores agree sufficiently closely with the known ability of the persons tested, the test is satisfactory. It can then be used to predict the ability of unknown persons of the same general group as those upon whom it has been tried. Clearly, however, this bare statement needs 46 PSYCHOLOGICAL TESTS IN BUSINESS a great deal of amplifying and qualification. Whether or not a test "works" is to be answered only after careful experimental investigation along the lines just indicated. Before we proceed to discuss the actual steps in the construc- tion and standardizing of tests, it is to be noted that there are some few tests the value of which has already been definitely established through many experimental tryouts. Where an occupation is highly standardized, and uniform from one business establishment to another, it is often possible to infer that a test which operated successfully in the one will, without necessity of further proof, prove satisfactory in the other. Thus a test for typists, once it has been carefully prepared and verified in several companies, may be widely adopted by other companies, provided reasonable precautions are observed in administering it and interpreting the scores. There are comparatively few of these ready-to-use tests avail- able, however. Even with fool-proof intelligence tests, and tests of trade proficiency, the purposes for which these are used and the conditions of their administration are likely to differ so markedly from plant to plant that experimental tryouts are indis- pensable. It has been found, for example, that the level of men- tal alertness of office boys and clerks in one office may require the use of very different standards from those in use in some other office. It has been found that tests which were successfully used in selecting shell and cartridge inspectors were almost worth- less in selecting workers for the allied occupation of gauging shells. It has been found that intelligence tests which proved worthless when used with certain groups of factory workers were decidedly valuable in another factory. Trade tests which gave good results in the army have failed to work in certain industrial situations. All such results that are available point to the same conclusion. The same test will give the same results only in case the conditions and purpose governing its use are closely similar. And it is usually impossible to know that the situations are sufficiently SCIENTIFIC METHOD 47 similar except by trying out the test anew to see whether it does give useful results. The fact remains, nevertheless, that some few tests for cer- tain widely standardized occupations have been tried in so many places and under so many different conditions that it is now pos- sible to use them as generally applicable tests for those occupa- tions. This is true, for example, of occupations such as typist and stenographer, and a number of the manual trades. It is also true of a wide variety of unstandardized occupations where " gen- eral intelligence" or mental alertness and abstract thinking abil- ity are important requirements. Thus general intelligence tests have been used to indicate the alertness of clerical workers, sales- men, office boys, and so on, just as they were used for testing men in the army and are used for testing children in the schools. Wherever these standardized tests are put into immediate use without special tryout, however, it is a wise policy to keep careful records of results, and to check back, as soon as this can conven- iently be done, to ascertain precisely how valuable the results are in this new situation. FUNDAMENTALS OF TEST CONSTRUCTION AND STANDARDIZATION i. Know the requirements of the job.-Some knowledge of the work for which people are to be selected is essential. If one is to devise a test for drill-press operators or typesetters or shipping clerks, he needs to know with considerable intimacy what is done by individuals in each of these occupations. The tests which are to be experimented with can be chosen only on the basis of some more or less plausible relationship between particular tests and the sort of duties performed in the job to be filled. 2. Choose or devise tests to be tried out.-After the test techni- cian has become sufficiently acquainted with the work for which men are to be selected, his next task is to decide upon tests that appear to be promising for the specific purpose at hand. Fre- 48 PSYCHOLOGICAL TESTS IN BUSINESS quently there are tests already available which have been used elsewhere for similar occupations or which appear for some other reason to be promising candidates for selection as successful tests. At other times, the psychologist finds no tests already in use which are quite the sort he desires, and in that case he may devise new ones. In this matter of choosing and devising tests, we can readily appreciate the importance both of a wide knowledge of tests and test results, and of ingenuity in planning new devices or variations of old ones. As was earlier mentioned, the expert in seeking tests for a job is usually not able, with any assurance, to state in advance which test, or even which kind of test, will prove successful, though this is not true of certain of the occupa- tions where tests have been widely standardized. The typical procedure, accordingly, is to begin the experiment with a number of tests. Some will early be eliminated; others are likely to give results sufficiently valuable to justify their retention and use. Clearly the expense and the success or failure of building tests depends largely upon the wise initial selection of tests to be tried. The office manager who tried out, as a test for bookkeepers, prob- lems in extracting cube root and solving algebraic equations was pretty definitely foredoomed to failure. Even clever guesses as to good tests, however, are worthless unless combined with a thoroughgoing experimental and statistical check to prove the wisdom of the choice. 3. Obtain the scores made on the selected tests by individuals whose ability in the job is ascertainable.-Once the tests to be tried are decided upon, the way to find whether they "work" is to try them out and see what happens. The way in which the tests are administered and all other conditions affecting test results must duplicate as nearly as possible the conditions under which it is proposed to use the tests. Hence, typically, the best procedure is to try the tests on all the applicants who are considered for the given job, just as they would be used in regular employment procedure. A less desirable method, but SCIENTIFIC METHOD 49 the one which for practical reasons is most used, is to try the tests on workers already employed in the occupation. The important thing is to have the conditions of the tryout and the kind of people who are tested as similar as possible to the conditions and people that will be involved in the final use of the tests. The number of persons for the experimental tryout must be sufficiently great to insure their being truly representative of the general run of applicants for that occupation, or of the general run of workers in the occupation, if the test is being tried out on men already employed. 4. Obtain some measure of the ability on the job of the indi- viduals who were tested.-In order to judge how well the tests indicate the abilities of workers, it is necessary to secure definite information as to the amount of ability in the given kind of work possessed by each of these men. There are three principal sources of this information which is to constitute the criterion against which test scores will be checked. In the first place, production records are available for many lines of work, which show the exact output of each individual worker. These may frequently be used directly as the measure of individual efficiency on the job. For very many occupations, however, individual production records are not obtainable, or are not an adequate measure of competence. In these instances, a second sort of criterion is used, namely, estimates of the worker's ability at his job by the executive under whom he works. These personal opinions have many shortcomings and grave inaccuracies, but when properly safeguarded, and obtained from several executives independently, they prove at least tolerably satisfactory. Another criterion of an entirely different kind is sometimes added to production records and ratings by superior executives. It is the record of the length of time the individual employees remain on the job. Clearly, it is important to ascertain whether men with high or low test scores show a higher rate of turnover than do those with less high or low scores. This criterion is 50 PSYCHOLOGICAL TESTS IN BUSINESS almost always to be used in conjunction with the measure of ability supplied in supervisors' ratings or production records. 5. Determine the agreement existing between test scores and ability on the job.-The two preceding steps are significant only as they are utilized for purposes of careful statistical comparison. The logic of the situation is simple. If the men with high test scores prove to be the men with high production records or supervisors' ratings, and the men with low test scores prove to be the workers who fail to make good, then to the degree that this agreement holds true, the test is successful. The inference is, of course, that if the test shows successfully the ability of the individuals in the group upon whom it is used experimentally, it will be similarly useful in predicting the ability of any unknown applicant. The usual statistical rhethod for ascertaining the closeness of the relationship is called "correlation," and the numerical index of the relationship is a "correlation coefficient." If there is considerable agreement, the correlation coefficient is high; if very little agreement, the correlation coefficient is low. The excellence of a particular test may, then, roughly be said to be indicated by the size of the coefficient. In everyday language, this is nothing more than the statement that the value of a test is shown by seeing how well it picks out good workers and poor workers. Often it is possible to make such comparisons more effectively by simple statements such as: Of all workers scoring below 138 on this test, 92 per cent failed to make good on the job, while above that score only 20 per cent failed at the work. Correlation coefficients and percentage comparisons do not automatically tell whether the test is a good one to use. That decision must depend upon other facts as well, one of which is the degree of success of alternative methods of selection for this job. If, for example, the years of schooling of the applicants show the same relationship to their ability at the work as do the test scores, it may not be worth while using the tests. It will SCIENTIFIC METHOD 51 not be worth while, that is, unless the tests and years of schooling combined work appreciably better as a basis for prediction than does the amount of schooling taken alone. 6. Set test standards for the acceptance, rejection, or classification of applicants.-If some of the tests tried out experimentally give sufficiently good results to warrant their further use, it is then necessary to set standards or "critical scores." For example, a score is specified at such a point that applicants scoring below this are not accepted for the job unless they have special compensating qualifications; or a score may be set at a point where it is possible to say that applicants scoring above this are to be given specially favorable consideration for employ- ment; or several ranges of scores may be indicated which will divide candidates into classes according to their ability. All these standards are set on the basis of the scores actually made by individuals of known ability. The foregoing account of the successive steps in scientific test-building needs to be greatly expanded. We turn, accord- ingly, to a more detailed discussion of the methods used in each of the six steps. JOB ANALYSIS AND TEST CONSTRUCTION The necessity for knowing something about the job for which tests are to be found is self-evident. But how is this information to be secured by the test psychologist ? How detailed must the knowledge be ? Can the information be obtained second hand, or must the tester himself observe the work, or must he even undertake to learn the operations himself in order to have the "feel" of the task? Is knowledge concerning the job more useful for test purposes if it is systematized and formulated, or are general impressions and "hunches" usually satisfactory ? We shall try to throw some light on these questions, though a complete answer is not possible. Some experts emphasize the importance of careful detailed analysis of the job before test 52 PSYCHOLOGICAL TESTS IN BUSINESS construction is undertaken, with an attempt systematically to describe the elementary psychophysical requirements of the work; others advocate only a rough acquaintance with the nature of the work, such as can be obtained by relatively short periods of casual observation. Despite these differences of opinion, however, we cannot lose sight of the fact that all agree as to the necessity of some familiarity with the work for which tests are to be built. The insistence upon rather detailed and systematized descrip- tions of job requirements as a preliminary to psychological testing has been markedly affected by the use of such job descrip- tions for other purposes. Job analysis as a generally useful method in personnel work has been greatly emphasized in recent years. And among the long lists of purposes which the detailed analyses are said to serve is frequently included the statement that they constitute a basis for the development of tests. With- out attempting to enter into a discussion of the nature and useful- ness of job analysis in general, we would point out merely that these studies of jobs rarely contain the sort of information that is likely to aid the test psychologist. After all, a job analysis includes those things which the analyzer by training and interest is prepared to describe. Typically, his purpose is not to find information of value for test construction and hence the ordinary analysis contains little material which the industrial psychologist would find of value for test purposes. But is it possible and desirable to have special job analyses made which will methodically gather exact information about the job specifically for purposes of test-building? As an ideal, this appears entirely praiseworthy. Nothing would be so satis- factory as to have a psychologist make a careful inventory of the elementary psychophysical requirements of a job and then apply certain of his tests to measure each of the required abilities. Perhaps some day we shall use this method. While up to the present time only a few attempts of the sort have been recorded, SCIENTIFIC METHOD 53 it is nevertheless possible to point to the kind of information that would be sought in the special analysis of a job for test- building purposes. It would include a detailed study of sensory capacities required, muscular co-ordinations and dexterity, strength of various body parts, emotional and temperamental characteristics, physical attributes, special intellectual aptitudes, etc. Each of these classes includes an immense number of sub- divisions. Detailed analyses of this sort are not a generally practicable method at the present time because of the great difficulty in ascertaining the qualities required and the equally formidable task of finding tests to get at these elementary abilities. Perhaps the most elaborate effort in this direction has been the analysis of the abilities required of a musician, with an attempt to find tests for these abilities. While some measure of success has been attained, it is important to note that the task has engaged the time of a group of research scientists over a period of years. Nevertheless, attempts such as this do point the way to possible future developments. We must distinguish between a complete systematized analy- sis of the qualifications for a job and the detection of a few outstanding requirements. While elaborate studies of the kind just mentioned are at present not practicable in industry, it is possible to analyze out a few of the abilities that are clearly needed. It is useful, for example, to recognize that a railroad engineer must have normal ability to see colors, or that factory workers in a particular job must have a certain strength of hand muscles or a certain speed of reaction. Simple specifications of abilities of this sort are not difficult to arrive at, and they do serve to suggest tests to be tried out. The special job analysis for test-building purposes is similar but it would include a great many of the detailed items and would aim at making an exhaus- tive catalogue of the qualities required. We return to the original statement that the test psycholo- gist must, as a matter of course, have some knowledge of the job, 54 PSYCHOLOGICAL TESTS IN BUSINESS adding now the opinion that at the present time it does not appear practicable to obtain the information by thoroughgoing, analyti- cal studies, however desirable these analyses may be for other purposes, and however justifiable they are theoretically. The alternative is that the expert should have sufficient familiarity with the fob to afford him a basis for wise predictions as to the sorts of tests that will work. Usually this means having a fair general understanding of the work and an insight into the essential features that determine success of workers in this job. This knowledge of the job is best gained in most cases by personal observation or even by performing the task. At times, doubtless, it is useful to set down methodically some outstanding require- ments of the work, though in most instances the necessity of even this is questionable. The emphasis we place upon informal familiarity with the work to be tested for and the "feel" of the job, as over against elaborate, long-worded descriptions, possibly seems a departure from scientific procedure. Further evidence in support of this common-sense approach, however, will appear in the ensuing discussion. Here we wish only to add the fact that in actual practice results have been obtained almost entirely by the non-analytic procedure. Even where investigators have made some effort to present careful analyses of the jobs, one usually comes away with the feeling that the analyses have about the same closeness of relationship to the actual test construction as the platform of a political party has to its subsequent activities. But this does not mean that a closer relationship would not be highly desirable. PRELIMINARY SELECTION OF TESTS The industrial psychologist is familiar with literally hundreds of different tests that have been used. In addition, he sees endless possibilities of new tests or modifications of old ones. When, now, he is confronted with the problem of deciding what particular tests to try experimentally as tests for a certain job. SCIENTIFIC METHOD 55 how is he to make the choice ? As a matter of fact, he depends largely on wise guesses. The ingenuity and insight manifested at this point will determine in no small measure his ultimate success in finding tests, the usefulness of which can be demon- strated. He may use either ready-made tests or he may feel that it is necessary to devise new ones. The decision as to what tests shall be tried out is based, of course, both upon the knowledge of the work for which tests are needed and upon a wide familiarity with tests of all sorts. The better equipped the test expert is in both directions, the better chance he has of getting results. In our present state of knowledge, an intimate and informal " feel " of the job is likely to be more richly suggestive of promising selective devices than is a formal analysis. And similarly, a close first-hand knowledge of testing methods and materials and the results obtained from these is more valuable in bringing fruitful guesses than is the attempt to use this test because it is called a test of "attention," and that test because it is called a test of "foresight," and still other tests because they are labeled "memory" or "reasoning" or "persistence." Usually, in other words, tests to be tried out can be better chosen through clever "hunches" based upon wide experience and keen insight than they can be through what appears superficially to be a more finished and precise scientific analysis. Lest it be thought that this "loose" procedure is a sign of weakness in test technique, we may recall that it is part of the general procedure of science, and that it is possible to cite from other branches of science illustrations of the same method. For, as a matter of fact, this process of forming wise guesses ("hypoth- eses" science calls them), often without antecedent detailed analysis, is a most typical step in sciences generally. Always, of course, the guesses are subjected to rigorous proof or disproof after they have occurred to the scientist. So are the guesses as to tests. The initial choice is only a first step. The indis- pensable sequel is the thoroughgoing testing of the test-or, in 56 PSYCHOLOGICAL TESTS IN BUSINESS more general terms, the verifying of the hypothesis. This further consideration, however, is to be noted: As a science develops and becomes more exact, dependence upon impressions as the starting-points for applications diminishes. There always remains, however, a considerable area at the frontier of a science where the accumulation of analytical knowledge is still insufficient basis for the solving of new problems. Industrial psychology may hope gradually to remove larger and larger parts of the test field from the realm where guesses must be made and then verified-just as, for example, the chemist or physicist has done. But at present the test expert has little established basis of facts from which to reason. He must accordingly guess, and then test his guesses. Consider an example or two of scientific "guesses" in other fields. Darwin formed his hypothesis of organic evolution when still a young man; the rest of his life was spent subjecting it to the test of meeting actual facts. Newton is said to have formed his initial notion of the law of universal gravitation when he saw an apple fall. The proof of that hypothesis involved years of astronomical observation and intricate mathematical calculation. Perhaps the gap seems too great between laws of evolution and gravitation and the construction of job tests. But the underlying scientific method is identical. Let us take one other example, somewhat less remote. An industrial chemist wishes to find the best metal composition to meet certain specified conditions. The product must withstand a variety of chemical actions, it must have certain strength, certain appearance, etc. The chemist has no answer to the problem ready to hand. He forms some wise guesses as to what sort of metal composition is needed. Probably his guesses are more sagacious and better founded than are those the test psychologist must make, simply because chemistry is farther advanced in its accumulation of exact knowledge. But essentially the methods are the same. The chemist proceeds to try out his best guess SCIENTIFIC METHOD 57 just as the test expert tries out his. The results of the experiment determine whether or not the guess was good enough. To return to our discussion of the choice of tests, it is well that we inquire in what ways they are related to the jobs for which they are intended. Tests for particular occupations may readily be classified according to the sort of relation the test bears to the job itself. Thus, the tests may be: i. Samples of the job. 2. Devices for measuring the separate abilities used in the job. 3. Devices for measuring the central or outstanding ability used in the job. 4. Imitative repetitions of the job. ("Replicas" or "miniature job tests.") 5. Devices bearing no apparent or obvious relation to the job. Of the tests which have actually been devised there are a number that fall in each of these five divisions. Different investigators have been especially partial to one or another of these classes in going about the choice of tests for experimental tryout. Without attempting to settle the question as to the advantages and disadvantages of tests of the several sorts, we shall give an example or two of each, and indicate some of the outstanding characteristics to be considered in adopting one or another variety. Tests which are samples of jobs have their principal place as proficiency tests-tests of trade knowledge and skill, or of educational attainment. The obvious way of testing an applicant claiming typing skill is to obtain, under standardized conditions, a measurable sample of his typewriting. Similarly with an individual's knowledge of French, his ability to spell, or his proficiency in operating a lathe. Army trade tests and almost all educational tests fall into this division of tests which are samples of the work in which ability is to be judged. The difficulty in the sampling method of test construction is to find samples that are sufficiently representative of the work as a whole. 58 PSYCHOLOGICAL TESTS IN BUSINESS Ordinarily this is reasonably practicable only where the job is fairly homogeneous, though trade tests have been constructed for rather complex trades. The sample method is apparently not useful in constructing aptitude tests because, in the nature of the case, samples or actual parts of the job performance cannot be a fair test for the individual who has not been trained in that line of work. The second class of tests-devices for measuring the separate abilities used in the job-has already been discussed in general terms. It has been pointed out that detailed analysis of the abilities required for a particular job has not thus far proved fruitful as a method of test-building. The theoretical desirability of this precise method has attracted certain test psychologists, however, and several tests have been worked out in this manner. Miinsterberg, for example, in one of the earliest vocational tests on record, attempted to judge the ability of telephone operators by testing separately " eight different psychophysical functions," including memory for digits, attention as indicated by the speed in crossing out "a's" on a newspaper page, intelligence as measured by a word-association test, exactitude of space perception, etc. An outstanding example of this method, earlier referred to, is Seashore's work in analyzing and testing musical aptitude. In spite of the difficulties in the use of this method, there is a great deal to be said in its favor. This is especially true if we include not only the formal and elaborate attempts at complete analysis but also the rather simple inquiries which aim to find a few of the more important abilities needed for the job. An informal analysis of the separate abilities used in the work is likely to suggest a number of tests to be tried out. Tests of simple sensory abilities and muscular movements are particularly apt to be hit upon in this way, although great uncertainty still attaches to the more complex processes. All the tests, no matter how clearly they seem to be measures of special abilities SCIENTIFIC METHOD 59 required for the job, must of course be subjected to tryout. The dangers in this analytical method of test selection are, first, the tendency to make the analyses more elaborate and detailed than is justified by the use made of them and, second, the tempta- tion to accept the results of the analysis without submitting these results to experimental verification. Wisely used, however, analytical inquiries as to the qualities needed in a job furnish a valuable basis for guesses of tests to be tried. The same statements can be made about the third class of tests listed-devices for measuring the central or crucial ability required for a job. To the extent that we can find such a central capacity, we are aided in the search for tests which show promise. Tests built upon the central capacity theory are well illustrated by another of Miinsterberg's earliest tests-that for street-car motormen. Miinsterberg decided the central ability required is "a particular complicated act of attention by which the mani- foldness of objects .... [is] continuously observed with refer- ence to their rapidity and direction in the quick changing pano- rama of the street." Accordingly, he devised an ingenious apparatus aimed to test this "particular complicated act of attention." His test for ship officers proceeded similarly with an attempt to tap the central or key mental process required. This method of test-building has not been much followed by later investigators. The last two varieties of tests in our list have been especially vigorously defended and attacked by different industrial psy- chologists. Imitative representations of jobs or miniature reproductions of the tasks have had a special appeal which is readily explainable. Somehow, it does seem more reasonable to test a prospective punch-press operator by having him push paper strips into a diminutive stamping apparatus than to have him draw pencil lines in a prescribed manner or press a telegraph key as rapidly as he can upon the occurrence of a set signal. It does seem more promising of results to test an applicant for 60 PSYCHOLOGICAL TESTS IN BUSINESS work as a file clerk by having her sort cards or alphabetize lists than to require her to take a mental arithmetic test. But actual results so frequently belie these confident anticipations that one begins to question how generally safe it is to rely upon apparent analogy between the job and the test. Too frequently the miniature operation only appears to be identical; the actual movements and, still more, the actual mental attitude may be very different indeed. It is probable that many tests will continue to be built as more or less close imitations or miniatures of the actual job. Doubtless not a few of these will prove valuable. But eternal caution is necessary lest we assume that because the test resembles the work, it must be a good test. Only by proceeding to test the test in the same rigorous experi- mental manner as is necessary in verifying less plausible looking tests can we be justified in the further use of the vocational miniature or replica. One important source of the popularity of this kind of test is the relative ease with which it may be "sold" to business men. It looks sensible. Many tests do not. And that, precisely, is why our warning is needed. A test cannot be accepted because of its plausible appearance or rejected because it seems remote or unrelated to the job. Only on the basis of experimental results can the test be judged. There is this to be said for the imitative test, however-that it frequently is the most promising sort of a test to choose for preliminary tryout, provided, of course, care is used to preserve the really important features of the actual job. One of its advantages-at times an important practical advantage-is that it will likely be better received by the indi- viduals to be tested than would be some apparently foolish or unrelated task. The same is obviously true of tests which are samples of the job. The final kind of test that may be used in the preliminary tryout we have called "devices bearing no apparent or obvious relation to the job." That is about all there is to say by way SCIENTIFIC METHOD 61 of description of such tests. They may be of any and every sort. The test expert tries a number of them, knowing that many will not work, but hoping that at least one or two of the array will fill the bill. This procedure, unscientific as it may sound at first, is perhaps the most commonly used, and has on occasion produced good results. As a matter of fact, the method is not nearly as haphazard as it is usually pictured to be. The test expert, in actual practice, does not choose his tests for tryout blindly or indiscriminately. When we speak of these tests as "bearing no apparent or obvious relation to the job," the words "apparent" and "obvious" need to be emphasized. The tests do bear a relationship to the job-and usually the tester can point out quite definitely, if he has to, what he conceives that relation- ship to be. Almost without exception it is possible to bring the test into one or another of our preceding classes. Since, however, this is not explicitly done, we are probably justified in continuing to use this fifth nondescript class. Several test studies are on record where the strictly empirical procedure implied by this sort of test has been used. An assortment of tests is tried out on a group of clerical employees, typists, factory workers, college students, or the like, and then from the results it is determined which tests were most successful and which ones least so. The method of experiment and statistical evaluation is, of course, identical with that necessary for any other of the varieties of tests we have been considering. We cannot too often repeat the thought that regardless of which of the preceding sorts of tests is chosen, experimental evaluation or tryout is indispensable. However good or however bad the test may look, the real value is to be ascertained only by testing the test. To the further steps in this process we accordingly give our attention. The distinction between the experimental tryout of the tests and their actual administrative use has already been emphasized. EXPERIMENTAL USE OF THE TESTS 62 PSYCHOLOGICAL TESTS IN BUSINESS During the experimental use of the tests, the purpose is solely to ascertain the value of the tests. The resulting scores are not to be utilized for making any decisions concerning the individuals tested. It is important that this clear differentiation in purpose be kept in mind. Three questions arise which we shall briefly deal with in the next few paragraphs. First, what people shall be tested for experimental purposes? Second, how many individuals shall be tested? Third, under what conditions shall these tests be conducted ? Let us take these questions one at a time. The tests may be tried out either with applicants for the job or with employees at present working at the job. The former is theoretically more desirable; the latter is usually more practi- cable. The advantage of using applicants whenever it can be done is that the conditions of the tryout are thus made closely similar to the conditions under which it is later planned to use the tests. The applicants constitute the actual group with whom the test is to be used in practice; hence, whatever relationship of tests to abilities is found in experiments with these applicants is pretty certain to be true later in the administrative use of the tests. There are, however, difficulties in the tryout of tests upon appli- cants--especially the delay involved. The number of applicants placed in a particular job within a relatively short period of time is usually not large. The tester must, therefore, wait over very long periods-a delay which is usually undesirable or entirely impracticable. There are likely, too, to be special inconveniences in the practical arrangements for the giving of the tests during the preliminary stages. The advantages of the tryout with appli- cants are such, however, as to make it thoroughly desirable to use this method as part, at least, of the test-evaluating procedure even where the main tryout must be with present employees. In the experimental stage, all applicants should be given the tests, but they should be interviewed and accepted or rejected wholly without reference to the test scores. The tests are being SCIENTIFIC METHOD 63 tested, not the applicants. If this process is continued for a time, it will then be possible to see whether the test scores, if they had been used, would have made more successful selection possible than was accomplished without them. It would be desirable during this experimental period to hire all applicants who showed even a minimum of possibilities for the job, so that the agreement between their test scores and their later efficiency on the job could be determined, as well as that of the better appli- cants. In practice this is often impossible. In fact, purely for practical reasons, tests are ordinarily not experimentally tried out with applicants at all, but rather are given to present employees. When present employees are used for the tryout, the testing can usually be done quite expeditiously. The most important requirement is that all workers at the job be tested or, if there are a great many, that a representative sampling of the whole group be tested. If only certain selected workers are tested-• for example, the extremely good ones and the extremely poor ones-the statistical interpretation of the results obtained will be almost worthless. For clearly in that case we cannot infer that the relations found in the artificially selected group will be true in the entire unselected group. Usually we are most interested in finding how well the tests pick out the poorest workers from the whole group of employees in that job. This can be definitely determined only in case we test the whole group (or samples of the whole group). Otherwise, we are only dis- tinguishing the worst workers from the best workers; the real problem is to distinguish the worst ones from the mediocre and fairly poor ones. In general, then, with either applicants or present employees, tests should be tried out with a typical group representative of the whole range of individuals upon whom the test is finally to be used. One further point is to be mentioned regarding the experi- mental testing of applicants and present employees. Any test, success in which depends upon special training or experience 64 PSYCHOLOGICAL TESTS IN BUSINESS acquired in the particular job tested for, clearly cannot be tried out on present employees and then used on applicants. For example, a trade test obviously is unfair if it contains questions the answers to which high-class tradesmen in this particular plant know, but tradesmen in other plants do not know. Such a test might show excellent results in sorting out good and poor workers among the tradesmen already at work in this one shop, and still prove worthless and misleading if used io judge men from other shops. This sort of pitfall is an ever present danger which even competent test psychologists do not always wholly avoid. Sometimes the influence of the present job upon men's ability to pass the test is very subtle and easily passes undetected. This consideration emphasizes the importance of trying out tests in part or wholly upon applicants. Another way of meeting the difficulty just discussed is to have the test tried out co-operatively in several different shops. That was the method followed in constructing army trade tests. Our second question asked how many individuals need to be tested. A dogmatic answer is not possible. It all depends. It depends upon the nature of the individuals tested, the nature of the tests, the conditions under which the tests are conducted, the excellence of agreement between test results and the ability to be predicted, etc. One safe and important generalization can be laid down: The number of individuals tested should be large enough that the statistical results would not be significantly changed if we were to include in our results a large number of additional test scores from the same general group as that from which the first individuals came. The test expert can, by proper statistical methods, determine in any given case approximately how many individuals must be tested to satisfy this requirement after a fair portion of these data are available as a guide. Actual test studies have used anywhere from five or ten cases, which is decidedly too few, up to several hundred. The usual number runs from twenty to one hundred. SCIENTIFIC METHOD 65 The third question-Under what conditions shall the tests be conducted?-may be passed over with a word. The out- standing thing to be said is: The experimental conditions should in all essentials be as nearly identical as possible with those conditions under which the tests are finally to be used. This means that the same directions are to be used; so far as possible, the same attitude is to be created in the individuals tested; the same scoring method is adopted; physical surround- ings, personality of the tester, and the like, should be not greatly different; and so on through the list of conditions which have an effect upon test scores. In order to determine the value of tests, predictions of ability from the test scores must be checked against some criterion. The tests are trying to tell what a man's ability is for a given line of work. To see how well the tests succeed, we must know what the man's ability actually is in the given work. There are in use two or three principal measures of that ability which we shall consider. The most obvious criterion for showing a man's working efficiency is the record of his output. Under favorable circum- stances, individual production records doubtless constitute the best indication of ability on the job. These records are, more- over, very often available in convenient form. Especially in repetitive factory operations, such as occur in the use of auto- matic machine tools, in highly specialized assembling operations, in pasting, sorting, folding, packing, inspecting, and the like, individual production records are often kept as the basis for piece-rate wage payments. Not a few office and miscellaneous occupations also afford opportunity for obtaining a measure of output in terms of pages typed, bills checked, orders taken, amount of insurance written, etc. Production records may show either quantity or quality of output, or both. Quantity records CRITERIA WITH WHICH TO COMPARE TESTS 66 PSYCHOLOGICAL TESTS IN BUSINESS are more used than are those of quality, though the two are frequently combined in arriving at an index of productive efficiency. As examples of quality records may be mentioned errors in typewriting, filing, computing work, and the like; and, in factory operations, the amount of waste material and the frequency with which work fails to pass inspection. Individual accident records have also been used as a criterion, as in the case of street-car motormen. A few of the dangers to be guarded against in the use of production records as measures of ability should be mentioned. In the first place, reasonable assurance is necessary that workers have sufficient incentive to produce. Otherwise the records show, not what the individuals can do, but what they happen to feel like doing. Again it is important to know whether the units of work turned out are equivalent. If they are not (for example, if a typist's ability is measured by the number of letters written, including letters of various lengths), then precautions must be taken to make allowance for the unequal units by averaging output over long periods or weighting the units in some way. It is necessary, further, to know that individual production is not markedly influenced by conditions beyond the worker's control, as would be true, for example, if workers are forced to wait for materials. Another danger is that the production records if taken over considerable periods, say week by week, may be thrown off by absences of the worker from his work. Hence hourly or daily averages are preferable. Precautions must also be taken to see that the records of workers involved are strictly comparable-that they are performing identical tasks under identical conditions, and that the effects of longer experience on the job are allowed for. These are a few of the more important difficulties that must be met. With proper care, this can frequently be done, and in that case the records of produc- tion are an excellent criterion against which to check the test findings. SCIENTIFIC METHOD 67 In many occupations, production records are out of the question. This is true wherever work is varied from individual to individual, and wherever the tasks of any one individual are not constant and repetitive. The most practicable measure of individual efficiency in these cases is the estimate of ability made by supervisors or superior executives. Estimates of this sort-"ratings" of ability, they are usually called-have been widely used, and have proved fairly satisfactory where sufficient care has been employed in obtaining them. Several kinds of rating scales-devices for recording the estimates or opinions- have been developed, but it is not necessary for us to consider these in any detail. The aim in all cases is to obtain a reliable and quantitative indication of the abilities of the individual workers. Experience has demonstrated that if reliable ratings are to be gotten, it is usually necessary to have three or more raters give independent estimates of each man's ability. All of the rating executives must, moreover, know intimately the indi- vidual rated. Many additional safeguards have been found desir- able in the use of rating methods, and with these the test expert needs to be thoroughly familiar. To describe the rating scales in use would, however, take us too far into technical detail. The simplest rating method, and probably the best, where it can be used, is the mere ranking of the individuals from highest to lowest with respect to their ability in the work they are doing. Then their ranking by test scores can be directly compared with this ranking by supervisors. In addition to production records and supervisors' estimates of ability, a criterion with which test results are occasionally compared is the length of service or period of employment of individuals. It is almost always supplementary to one of the other two criteria, rather than alternative to them. The purpose in finding the relation of test to these turnover figures is to see whether individuals with low, medium, or high scores are leaving the job especially rapidly or in especially large numbers. If the 68 PSYCHOLOGICAL TESTS IN BUSINESS turnover is great, for example, among individuals having low scores, and also those having high scores, while not great among those with medium scores, the conclusion might be justified that neither low-score nor high-score individuals should be employed for this particular job. There is, of course, danger of jumping to an unjustified conclusion in this way. The situation would call for further investigation. Perhaps the thing to do is not to begin rejecting the high-grade applicants, but to change the conditions of work or pay so as to hold these people once they are employed. The checking of tests against turnover figures is, in any case, likely to be a worth-while supplement to the compari- son of the scores with either production records or ratings by supervisors. ASCERTAINING THE RESULTS FROM THE EXPERIMENTAL USE OF THE TESTS The test scores and the measures of individual ability on the job constitute the concrete experimental data upon which conclusions regarding the value of the tests must be based. The process of arriving at conclusions is primarily a matter of statisti- cal procedure. The problem is simply to work out quantitatively the relation existing between the test scores made by the indi- viduals in the group and the corresponding efficiency ratings of those individuals. In other words, it is necessary to ascertain what degree of agreement there is between ability in the tests and ability on the job. Do individuals scoring high prove to be better workers than those scoring lower? Do those scoring very low turn out to be pretty uniformly the incompetent workers? Precisely how close is this relationship? There are two principal ways of working out this problem statistically. The one uses the correlation method; the other makes percentage comparisons between groups of poorer and better workers. The logic of the two methods is essentially the same. Only the SCIENTIFIC METHOD 69 barest indication can be given here of the nature of these statisti- cal procedures.1 First let us consider a simple example of the correlation of test scores with supervisors' ratings. Table I gives the data TABLE I Comparison of Mental Alertness Test Scores of Nineteen Office Employees with Ability Ratings Employee Test Score Ranking by Test Ranking in Ability by Two Super- visors A 67 2 B 63 2 3-5 C 61 3 5 D 59 4 6 E 54 5 7 F 52 6 G 50 7-5 14 H 50 7-5 9 I 48 9 3-5 J 47 IO 12 K 46 II 15 L 43 12 18 M 41 13 8 N 4° 14 16.5 0 37 15 IO P 36 16 13 Q 32 17 II R 23 18 16.5 S 15 19 19 obtained from the experimental testing of nineteen office em- ployees with a mental alertness (general aptitude) test and the ranking of these nineteen individuals by two supervisors. A rank of "1" is highest and "19" is lowest. In case of a tie, tA large part of the scientific literature on tests is devoted to statistical analyses of test results. Among the many books on statistical method which are valuable for workers in the field of tests, the following may be mentioned: H. O. Rugg, Statistical Methods Applied to Education; W. A. McCall, How to Measure in EAucation; G. U. Yule, Introduction to the Theory of Statistics; T. L. Kelly, Statis- tical Method; W. Brown and G. H. Thomson, Essentials of Mental Measurement; E. L. Thorndike, An Introduction to the Theory of Mental and Social Measurements. 70 PSYCHOLOGICAL TESTS IN BUSINESS both individuals receive the same rank. For example, employees G and H both scored 50 points and are tied for the seventh and eighth places. They are given their average rank, 7.5. The agreement in this table is seen to be fairly close between the ranking according to test scores and that made by the supervisors. The correlation coefficient, now, is merely a convenient number which can be calculated statistically to tell how close this agree- ment is. The coefficient is readily obtained from the rankings contained in the last two columns of the table. There are other methods of computing correlation coefficients that are especially valuable when we wish to use the actual test scores and actual measures of productive efficiency instead of the rankings of individuals. The correlation coefficient is zero when there is no agreement at all between the two columns compared; it is plus 1.00 if the two columns agree perfectly; it is minus 1.00 if the rankings of the two columns run in precisely opposite direc- tions, that is, if the individual ranked " 1 " in the tests is ranked "19" in the ratings, and so on. The correlation coefficients are always decimals, falling somewhere between minus 1.00 and plus 1.00. The coefficient actually found from the table was plus .76. This is somewhat higher than coefficients usually obtained between test scores and ability on the job, due to special condi- tions which need not concern us here. The significance of the correlation procedure for test purposes will be clearer if we interpret it in terms of prediction. After all, the whole purpose in obtaining test scores is to use them in predicting the ability of the individuals tested. And the closer the agreement between the scores and the actual ability of the individuals, the more accurate and reliable our predictions. Correlation coefficients measure this agreement and hence measure, also, the accuracy and reliability with which we can predict individual ability on the job from the test scores. The higher the coefficient, the better the predictions. Clearly, if the correlation coefficient is 1.00, the prediction is perfectly SCIENTIFIC METHOD 71 accurate, for we know that the individual who is highest in the test will invariably be highest in working efficiency. (Of course, coefficients of i.oo are never gotten.) If, on the other hand, the correlation is zero, then our predictions are no better than purely chance guesses. If the coefficient is between zero and i.oo, the predictions are better than pure guesses but are not unerring. The interpretation of a correlation coefficient of a given size is a difficult technical matter into which we need not enter. Corre- lation figures must be used with insight and conclusions based upon them must be cautiously drawn. Correlation is a con- venient tool for judging the predictive value of tests, but only when it is in the hands of an expert. One example only will be given to indicate the danger in an uncritical use of coefficients of correlation in test work. The size of the coefficient is tremendously influenced by the nature of the group involved. The same test may very well correlate with the ability of clerical workers in one office to the extent of .75, and give a correlation coefficient of perhaps .30 when used with a different group of clerks. The theoretical reasons for this need not detain us. The fact simply is that the coefficient is much larger where the people tested show great difference in ability among themselves, and is smaller where the individuals scatter over only a narrow range of ability. Our conclusion regarding the correlation procedure is this: It is decidedly useful in the hands of the expert who understands it, but it is very easy to misuse and to misinterpret. For the individual without statistical training, our warning is, "Distrust correlation coefficients! But trust the truly expert opinion based on them." Simple percentage comparisons frequently are more valuable in test evaluation than are elaborate correlations. Even where correlation coefficients are computed, percentage comparisons are likely to be decidedly valuable supplements. There is in a percentage statement, like the following, a direct and readily 72 PSYCHOLOGICAL TESTS IN BUSINESS understandable meaning not present in correlation figures: Of all the individuals scoring below 28 on this test, 85 per cent failed and 15 per cent succeeded, whereas among those scoring above 28, 74 per cent succeeded and only 26 per cent failed. A statement of this sort tells pretty clearly what results will be obtained from the use of the test. In practice, we are almost always concerned particularly either with the problem of how well we can detect the poor workers or with the problem of selecting the best workers, and are not so much interested in the excellence of differentiation at the other extreme of test scores. Percentage comparisons are thus especially useful in showing how well predictions can be made at some particular part of the total range of test scores. Mental Test Ratings Chart* showing a method for representing the results obtained in the tryout of psychological tests. Solid circles represent individuals who proved to be unsatis- factory; unshaded circles represent all other individuals. ♦Adapted from an article by L. L. Thurstone in the Journal of Educational Psychology, March, 1919. Graphic methods are especially revealing in making compari- sons of the sort just discussed. The results of a test tryout can be clearly pictured, for example, in a chart like that shown in the accompanying figure. This figure shows the results obtained in the tryout of a test group used to predict the ability of college students in their college work. Obviously the method would be identical in representing the value of tests for an SCIENTIFIC METHOD 73 industrial occupation. Each individual is represented by a circle placed above the particular test score he made. The test scores, as shown, range from i to 27. Students who have fallen below a certain standard of scholastic work are represented by solid black circles. All other students are represented by the unshaded circles. The chart shows that most of the unsatis- factory students are ones who made low test scores. Or, to use a more precise comparison, of the 15 individuals with a score below 9, not a single one made good; above that score 87 per cent made good (85 out of 98). DECIDING WHETHER TO USE THE TESTS TRIED OUT The decision regarding the choice of tests is based upon the statistical findings. But certain other considerations will play a part. Such matters as the ease with which a test can be given, the time it requires, the expense of giving it, the urgency of the need for some test, and many other incidental factors may be fairly important at times in reaching a practical decision, but it is not our intention to deal with these issues. Aside from such considerations, how good must the statistical results be to justify the adoption of a particular test ? Again the answer is, "It all depends." It depends on the care with which the tryout was conducted, the accuracy of the criterion with which the test scores have been compared, the extent to which the experimental conditions were identical with the conditions under which the test is to be used, the number of people tested, and so on. But more important than all these, it depends upon the competitors this test has. How well are other devices and methods of selection working without this test ? Does this test add anything useful to the information already obtained about an applicant as a basis for accepting or rejecting him? Are there any other tests that do as well as this test does? In a word, the decision as to the use of a test must depend upon the answer to the question: Is the proposed process of selecting 74 PSYCHOLOGICAL TESTS IN BUSINESS individuals for this job {including the use of this test) appreciably better, cheaper, or more expeditious than any reasonable alternative method that is available? The statistical evidence concerning the test does not settle the question. It is merely the most important single item of information that enters the final decision. But there is no automatic way of making sound judgments. SETTING TEST STANDARDS Before a test can be made part of the employment procedure, standards or norms must be established by which to interpret individual scores. The simplest method is to specify, on the basis of the experimental results, a "critical score" or "dead line," and then refuse to hire applicants who fall below the score. Provided this lower critical score or standard is wisely set, and provided above all that it is interpreted with good judgment, it has much to recommend it. An upper critical score or standard to denote applicants of special promise may similarly be used. Or, on occasion, an upper critical score may be set for the purpose of keeping out of a particular job those individuals who score too high. A man can be too good for a job as well as not good enough. But typically, test standards are set for the elimination of low-score individuals or the special selection of those making high scores. Instead of having a single critical score, it is often desirable to have a series of standards or "norms" specified. Thus, for example, the total range of scores may be divided into five classes in such a way that an applicant may immediately be classified, as far as his test score goes, as "Far above average," "Above average," "Average," "Below average," or "Far below average"-or simply as in Class A, B, C, D, or E. Sometimes it is useful to make even finer differentiations, in which case one of the most effective methods is to give the individual a "per- centile score." The percentile score tells simply what percentage of all men hired for the job (or, perhaps, what percentage of present SCIENTIFIC METHOD 75 employees in this job) make a score lower than the individual under consideration. Thus, his "raw" test score may be 156, which, taken alone, means nothing; but his percentile score, we are told, is 88, and that means that 88 per cent of the men in this job make a score lower than 156. The individual is thus placed at once with respect to the general run of men in this work. Still other methods of expressing test scores are in use, but they are of less practical significance. Test standards cannot be set and then left forever to take care of themselves. The norms should be checked from time to time, especially during the early stages in the use of the test, and whenever necessary revisions should be made. The same con- sideration applies to the matter of verifying the test to see if it is operating successfully. Even though the test was carefully experimented with in a preliminary tryout and found reasonably successful, results obtained in actual practice should be checked now and then to see how well the test scores are actually indicat- ing the ability of men hired. Only after repeated checks can the value of the test be accepted as conclusively demonstrated, and the standards viewed as finally established. SCIENTIFIC CONSIDERATIONS IN USING TESTS After the selective test has been experimentally evaluated and approved, it is made part of the employment routine. The particular manner in which testing fits into the procedure of a given employment office is necessarily determined largely by local conditions. The tests should be administered by properly trained individuals, and should be conducted under as favorable conditions as possible, free from interference and interruption. Whether test results are to be interpreted and used by the tester himself or by employment interviewers, or whether all testing is to be done by the employment interviewer -these and similar questions must be decided in terms of the particular situation. Throughout the testing procedure there is 76 PSYCHOLOGICAL TESTS IN BUSINESS need for extraordinary tact, understanding, and regard for the atti- tudes and feelings of those tested. This is, of course, no less true during the experimental tryout of the tests than it is of their final use. All these matters we shall pass over with the mere mention of their very great practical importance. The combining of test scores with the results from other tests given the applicant, and with the other items of information accumulated in the employment interview, constitutes a final problem to be pointed out. It is perhaps the most important, and at the same time the knottiest, problem in the field of judging men's abilities. How are we to piece together our partial pictures of a man to get a view of the individual in his complete- ness ? How are we to weight the various elements in our size-up ? How are all the bits of knowledge about the man interrelated ? There is an elaborate statistical procedure ("partial correlation" and "multiple correlation") which attempts to throw light upon the problem. But even these methods break down when confronted with the actual complexity of the task. The best we can say is that the devices, tools, and methods of the expert help a little in the attack upon these infinitely complicated and stubborn problems. In practice, the best we can do is use whatever scientific methods are available, add to these what- ever of common sense and sound judgment we can muster, and be ready to revise our judgments whenever new evidence appears. CHAPTER III PSYCHOLOGICAL TESTS FOR OFFICE OCCUPATIONS In order to appreciate the development and the present status of psychological tests in business, it is necessary that we have in mind a picture of recent progress in employment methods generally. Employment tests are a product of practical indus- trial needs quite as definitely as they are a natural outgrowth of the psychological study of the abilities of individuals. These two sets of forces-that from within psychology and that of practical demands-have worked together in the stimulation and shaping of psychological tests for business purposes. Regarding the influences from within psychology, nothing more need be said at this point. The significant fact is that rapid historic developments in the study of individual differences and their measurement have made possible a scientific attack upon indus- trial selection problems. The other set of influences demands closer consideration. If we are to see employment tests in perspective and be able to place a just valuation upon them, it is indispensable that we view them in relation to all other employ- ment practices and devices. THE RECENT HISTORY OF VOCATIONAL SELECTION IN INDUSTRY Employment methods have been undergoing a profound change during the last decade, a change which is best summarized in the expression "functionalized and centralized employment." Scientific management, with its emphasis upon specialization and the separation of the various managerial functions, paved the way for the entrance of functionalized employment. The 77 78 PSYCHOLOGICAL TESTS IN BUSINESS hiring and firing activities which were formerly the prerogative of individual foremen have, in an increasing number of plants, been taken out of the hands of foremen and concentrated under a staff department which deals exclusively with personnel duties. Hiring is usually one of the most important of these duties. Centralized employment necessitates the development of two bodies of basic knowledge-a knowledge of the requirements of the tasks for which men are being selected and a knowledge of the equipment of the individual worker. In the days of foreman hiring, the need for written standardized job specifica- tions was not felt, nor was the need for technical devices for deter- mining men's abilities. Each foreman knew at first hand the jobs for which he was selecting men. Likewise, he could crudely ascertain the applicant's trade skill by asking questions or, as was more usual, by "trying him out." Since records of employ- ment costs were not kept and since labor was ordinarily plentiful, it mattered little how effective the selection procedure proved. All this changes when costs and records are kept, when one centralized department is responsible for efficient selection, and when the importance of having "the right man in the right place" has once been appreciated. Even where employment has not been functionalized or placed in a separate department, the problems of labor turnover and inefficient production have turned the attention of manage- ment sharply to the matter of improved selection methods. The obvious possibilities of adopting more careful interviewing methods and of utilizing various aids for interviewers, have led to marked progress, wholly aside from changes in the organi- zation for handling employment. The fullest use of scientific employment devices, however, has been pretty generally limited to centralized hiring departments. Among the devices and practices that have gained wide use as aids in sizing up men may be mentioned the application blank, the letter of recommendation, rating scales, psychological tests and trade tests, interest and TESTS FOR OFFICE OCCUPATIONS 79 personal-history questionnaires, and careful personal interviews seeking to get at a real understanding of the applicant and his nature. Tests are to be recognized as only one item in the list of devices calculated to introduce greater precision and efficiency into the judgment of men in vocational selection. In actual practice the tests are always to be thought of as useful in con- nection with other methods, and not in isolation. Wherever movement is in progress toward more adequate methods of hiring, tests become part of a general program and are to be evaluated in terms of the part they play in that general program. The scientific attack upon the problem of selection methods is scarcely more than a decade old. The pioneer book in the field was Miinsterberg's Psychology and Industrial Efficiency which appeared in 1913. The problems of vocational selection, it is true, had been prominent before this time among engineers in the field of scientific management, among vocational guidance experts, and within some few industrial plants. A small group of employment managers had begun meeting in Boston as early as 1910. But in none of these quarters had an effective effort been made to deal with the problems of vocational selection in any careful and scientific manner. Aside from Miinsterberg's book, one of the most important contributions to the development of a psychology of vocational selection that appeared before the war was Blackford and New- comb's The Job, the Man, and the Boss. In spite of the thoroughly unscientific and misleading character analysis methods included, this book proved distinctly valuable in drawing attention to the selection problem as it appears in its entirety to the employer, with healthy emphasis upon the human side of scientific manage- ment and upon individual differences and their utilization. During the war remarkable progress was made in meeting selection problems both in the armies and in industry, principally, though by no means exclusively, in the United States. More and more emphasis has come to be placed upon the use of quantitative 80 PSYCHOLOGICAL TESTS IN BUSINESS methods, upon the importance of the job-analysis side of the problem, and upon a broad consideration of employment policies and methods in relation to business management and labor problems. In the United States Army in particular, unparalleled advances were made (in no small measure by professional psy- chologists) all along the line of scientific selection-studies of occupational requirements, methods of interview, use of general intelligence tests, special vocational tests, trade proficiency tests, rating scales, and records of individual qualifications and progress.1 In the years immediately following the war, personnel work continued its rapid expansion and many psychologists entered industry as personnel workers, as consultants, or as students in pursuit of special researches. In part there has been an effort to bear to industry the methods used in the army, and in part the attempt has been made to develop new tests and other employment devices. Much of this work has supplemented in a valuable way the practical procedures that had been developing in industry. The net result of the combined research and applied activities has been a decidedly significant body of material in scientific employment methods. The more important features of these industrial selection methods, bringing together the experience of the army and of industry, are described in many of the recent books on employment and personnel and in numer- ous special articles.2 1 The direction of progress in the industrial use of psychological methods of selection may be seen by a comparison of Miinsterberg's Psychology and Industrial Efficiency with Hollingworth's Vocational Psychology and Link's more recent Employment Psychology. The development of selection methods is well reported in the committee reports published in the annual proceedings of the National Association of Corporation Schools. Valuable accounts of the achievements in the army are to be found in two official publications: Personnel System of the United States Army (Adjutant General's Department) and "Psychological Examin- ing in the United States Army" {Memoirs of the National Academy of Sciences, Vol. XV). More popular accounts of parts of the army work appear in Army Mental Tests, by Yoakum and Yerkes, and in Trade Tests, by Chapman. 2 Among the more significant books are the following: Tead and Metcalf, Personnel Administration; Shefferman, Employment Methods; Frankel and Fleisher, TESTS FOR OFFICE OCCUPATIONS 81 The fortunes of the newer employment methods, and particu- larly of psychological tests, have shown a marked tendency to rise and fall with changing business conditions. There can be little doubt, however, that the application of psychological methods to the selection of industrial workers has gained a permanent place in the increasingly important movement to regard the human element in business. A certain amount of "overselling" of the possibilities of psychological selection must be lived down; a considerable body of unscientific work must be scrapped; a more vital grasp of the employment problem in its industrial setting must be obtained; a host of painstaking and thorough researches must be pushed to completion. But the essentials of the movement are sound. The unfortunate quack- ery and charlatanism that always cling about a new movement must not blind us to the kernel of permanent values inherent in the development of scientific selection methods. PSYCHOLOGICAL TESTS AS ONE DIVISION OF SCIENTIFIC SELECTION METHODS Against this general background of employment develop- ments we are now to project a picture of accomplishments in the use of tests. Tests constitute the most distinctively psy- chological contribution to vocational selection. But, as we have emphasized, they are by no means the whole of scientific selection nor are they to be thought of as alternatives to such other methods as careful interviews and ratings, and detailed application blanks. Where employment tests have been standard- ized and scientifically evaluated they are decidedly valuable parts of the selection machinery; but they are not the whole of the machinery. The Human Factor in Industry; Kelly, Training Industrial Workers; Simons, Personnel Relations in Industry. Along somewhat similar lines but with greater psychological emphasis may be mentioned: Scott and Clothier, Personnel Manage- ment; Link, Employment Psychology; Scott and Hayes, Science and Common Sense in Working with Men; Reilly, The Selection and Placement of Employees. 82 PSYCHOLOGICAL TESTS IN BUSINESS Tests as selective devices have been especially popular since the war. Before 1917 vocational tests were a hope, an interesting possibility, with, it is true, a few scattered pieces of research as a basis. The war gave a radically changed status to test work and made practicable much of what had been merely possible. After the Armistice (and, in a few instances, before) tests and testers invaded industry. Progress has been made slowly. Selective tests adapted to industrial needs are still few; the move- ment is still immature; future possibilities are still uncertain. This is entirely healthy and normal if we recall the recency of the birth of scientific vocational selection. The danger-the aspect that is not healthy or promising-is the overselling of tests, the tendency to dwell upon advantages and successes and gloss over defects and failures, the tendency to speak of selective tests as an accomplished fact and not as a valuable field for experiment. The way to overcome these dangers is obviously through the accumulation of a body of knowledge based on facts concerning test results. Such facts are gradually being collected by test psychologists and employment experts in a number of progressive firms where tests are being put to the test. Many experimental tryouts of vocational tests have been reported in the literature of psychology and employment manage- ment. We shall bring together in the following pages the more important of these test studies. In the present chapter we shall deal only with tests for office occupations, leaving for later treat- ment a similar survey of test studies concerned with factory and sales occupations and others. TESTS FOR OFFICE OCCUPATIONS Greater progress has been made in the use of selective tests for office occupations than for any other class of vocations. The tests used have in the main been those of general intelligence and those of special acquired proficiencies, although a few attempts are also on record of tests to detect special aptitudes TESTS FOR OFFICE OCCUPATIONS 83 for particular lines of office work. We shall review some repre- sentative studies of office occupations and shall indicate the lines along which progress has been made, emerging finally with some conclusions as to the success attained in the selection of office workers by means of tests. For convenience, tests for office occupations will be divided into the following three classes: (i) tests for clerical occupations; (2) tests for typists and stenogra- phers; (3) tests for other office occupations. TESTS FOR CLERICAL WORK Tests for general clerical work have been used.in a large number of business establishments, apparently with considerable success. In some quarters they have been used mainly in experimental studies, but in a number of firms they have also become part of the routine method of judging applicants.1 Comparatively few reports are available, however, which give exact figures on the results obtained. But there is enough evidence to warrant a very favorable opinion as to the accomplish- ments and possibilities of tests in this field. The tests for non-specialized clerical work have consisted principally of general intelligence tests. In some instances adaptations of the usual intelligence tests have been made through the use of wrords, symbols, and examples from the clerical field. In addition to the general intelligence tests there 1 No complete tabulation of firms using different sorts of tests in a scientific fashion has been made. A few of the better-known companies which have tried out or used tests for clerical work are: American Rolling Mills Co., American Tobacco Co., Armstrong Cork Co., Atlantic Refining Co., Bell Telephone Co., Burroughs Adding Machine Co., Cheney Bros., Continental and Commercial National Bank, Curtis Publishing Co., Dennison Manufacturing Co., Eastman Kodak Co., Equitable Life Assurance Society, Marshall Field & Co., General Electric Co., Goodyear Tire and Rubber Co., Joseph & Feiss, R. H. Macy & Co., Metropolitan Life Insurance Co., National Cash Register Co., National Cloak and Suit Co., Phoenix Mutual Life Insurance Co., Sears, Roebuck & Co., Strawbridge & Clothier, Swift & Co., Western Electric Co., Western Union Telegraph Co., Westinghouse Electric and Manufacturing Co., Winchester Repeat- ing Arms Co. Many of these same firms have tried tests for non-office occupations. 84 PSYCHOLOGICAL TESTS IN BUSINESS has been considerable use of tests of educational achievement- tests of proficiency in arithmetic, spelling, handwriting, and grammar. Illustrative examples of some tests and test results in the selection of general clerical workers will now be presented. We shall find that clerical tests have been studied (a) in relation to the individual efficiency of the people tested, (6) in rela- tion to the grade of clerical work being performed by the individuals, and (c) in relation to the stability of the workers or the length of time they remain on the job. We shall call attention to examples of each of these three kinds of comparisons in the course of our consideration of the following reports on the tryout of clerical tests. i. Probably the most widely influential tests for clerical occupations are the "National Business Ability Tests."1 These tests are almost entirely tests of educational attainment and not of natural aptitude. The tests have been devised and tried out under the supervision of a national committee which was organized in 1913. In 1914, twenty simple tests were tried in the employment offices of six large commercial organizations and, according to Cody, they "met with the unanimous approval of the employment managers of the houses that co-operated in the experiment." The tests were later revised and experi- mentally administered in a large number of business establish- ments and schools, making possible the establishment of standards for scoring and for interpreting scores. Parallel series of equiva- lent tests have also been prepared from time to time. While no statistical evidence has been published to prove definitely the value of the tests, business firms which have used them seem to have found them decidedly worth while. In other words, we 1A description of the tests, together with tables of standards and a statement of the history and theory of the tests, is given in Commercial Tests and How to Use Them, by Sherwin Cody. The book and copies of the tests are published by the World Book Co., Yonkers, New York. TESTS FOR OFFICE OCCUPATIONS 85 have here some general impressions of close agreement between tests scores and individual efficiency. How far these impressions are based on facts we do not know. The "National Business Ability Tests" include the following parts: (i) The applicant must tabulate items according to certain somewhat complicated conditions given in the instruc- tions. This is an attempt to get at general intelligence or mental alertness. (2) The applicant must memorize and later reproduce some detailed instructions concerning the use of sales slips. The test aims to measure memory and willingness to work hard. (3) The applicant must fill in some sales slips in accordance with instructions and facts given. It is considered to be an intelli- gence test. (4) The individual is tested in the fundamental arithmetical operations. This and all the remaining tests are definitely ones of educational attainment. They may be merely indicated by name. (5) Business arithmetic (fractions and percentage). (6) Tests in English (spelling, grammar, and punctuation). (7) and (8) Tests in letter writing. (9) Steno- graphic tests. (10) Test on copying for the mimeograph, (n) Test on addressing envelopes with a pen, and on filing. It is not, of course, assumed that all these tests are to be given to each applicant. Only those that are appropriate are used. 2. More definite evidence of the value of tests given to clerical workers has been presented by other investigators. One psy- chologist,1 for example, has showed that the results from the clerical examination which he used gave a better prediction of the ability of office workers than did the age and amount of schooling of the individual tested. This test aims to measure general intelligence, though in so doing it makes use of the language and material of clerical work, thus gaining in its interest appeal and apparent reasonableness. There are eight parts: 1L. L. Thurstone, "Standardized Test for Office Clerks," Journal of Applied Psychology, III (1919), 248-51. The test materials are handled by the World Book Co. 86 PSYCHOLOGICAL TESTS IN BUSINESS (i) checking errors on a page of addition and subtraction prob- lems; (2) locating misspelled words in a printed passage; (3) canceling certain specified letters on a printed page; (4) learning a code; (5) alphabetizing a list of names in accordance with special instructions; (6) classifying insurance policies in a specified manner; (7) solving problems in arithmetic; (8) inter- preting the meaning of some proverbs. These tests have been standardized by giving them to about five thousand office clerks. The value of the tests was ascertained both by comparing test scores with the grade of clerical work being done, and by finding the degree of agreement of scores with individual efficiency. The first comparison was made by giving the test to one hundred employees in the office of an insurance company. The employees were rated in five classes according to the grade of office work they were doing. The test scores of the five different grades of employees were then com- pared and a considerable degree of agreement was found between the test rating of an employee and the level of work in which he was employed. This agreement was closer than was that between the grade of work and the combined items of age and amount of schooling. The second study of the value of these tests, comparing scores with individual efficiency ratings, is described in a later report.1 The test has been given to approximately twelve thousand clerks and has been put to a definite experimental check among five hundred clerks in a government office. The test scores of this last group were compared with their efficiency as clerks in the following manner: The clerks in each office division were ranked from best to poorest by their immediate superior. For each clerk there were available also a civil service efficiency rating, a record of education, and a record of salary. These four indicators of ability were combined to get a final 'Anonymous, "Testing Men Like Material," Factory (March, 1923), pp. 294-95. TESTS FOR OFFICE OCCUPATIONS 87 measure of each employee. The clerks who were above average in their standing were called good; those below average were called poor. When test scores were compared with these efficiency ratings, it was found that the people who made low scores in the test were usually the poor clerks, and those who made high scores were ordinarily the good clerks. In one group of twenty reviewing clerks, all but two of those above average in the test were good clerks; all but one below average in the test were poor clerks. The report states that, "What was true of the twenty reviewing clerks was found to hold approximately for the entire clerical force." A table showing the relation of test scores to success as clerks is given below. It is to be read Test Score Percentage Who Are Good Clerks 20-39 9 40-59 33 60-79 45 80-99 63 IOO-II9 87 120-139 100 TABLE II thus: Of all the clerks scoring between 20 and 40, only 9 per cent are good clerks (" good " defined as already explained); 33 per cent of those between 40 and 60 are good clerks; and so on. 3. The reports of three or four other series of tests given to clerical workers contain somewhat similar evidence bearing upon the relation of test scores to individual efficiency. One study1 reports the use of a mental alertness test (general intelligence test) given to several small groups of office employees, where comparisons were made between test scores of the individuals *A, W. Kornhauser, "Some Business Applications of a Mental Alertness Test," Journal of Personnel Research, I (1922), 103-21. This work is presented also in Scott and Clothier, Personnel Management. The part of the study which compares test scores and turnover figures is reported also by A. J. Snow, "Labor Turnover and Mental Alertness Test Scores," Journal of Applied Psychology, VII (1923), 285-90. 88 PSYCHOLOGICAL TESTS IN BUSINESS and the estimated ability of these workers by their supervisors. Rather close agreement was found-sufficiently close to satisfy the management as to the value of the tests. In one office group of nineteen workers only one person below average on the test was rated above average by the supervisors and only one above average on the test was considered poorer than average by the supervisors. In another group of fourteen employees, every individual below average in the test was rated as below average by the supervisors, and everyone above average in the test was ranked as above average in efficiency. Such results are, of course, somewhat more favorable than is typical. Scores of clerical workers on this same mental alertness test have been studied in relation to grade of work and in relation to length of service. The relation of average scores to different classes of office work is shown in the following table. Interesting TABLE III Occupation Average Score Men: Stenographers 50 Bookkeepers 5° Clerks 47 Office boys 33 Women: Stenographers 45 Typists 44 Clerks 36 differences between men and women office employees are inci- dentally to be noted. There is clearly a tendency here for people in the higher-grade occupations to have higher test scores. The relation between test scores and labor turnover was studied, in two groups of clerical workers. A definite relation was found in both instances. In one case, both low-score and high-score individuals were leaving in larger numbers during the TESTS FOR OFFICE OCCUPATIONS 89 first six months than were the employees with medium scores. In the other group, individuals with low scores and those with medium high scores were less permanent than those with medium scores or very high scores. Findings of this sort obviously require further investigation; they are the starting-points for detailed inquiries into conditions of work, wages, and other factors which cause employees to leave. It is equally clear that these results show something of the value of the tests. After all, success in a given job implies a fair degree of permanence in that work as well as a certain efficiency. Tests may be as important in giving predictions of the permanence of workers as in showing their efficiency. 4. Another tryout of tests for clerical work was conducted by Link1 at the Winchester Repeating Arms Co. A series of intelligence tests and also tests of "technique" were used, includ- ing such tasks as moving a metal pencil along a groove without touching the sides (test of steadiness), sorting cards, substituting letters according to a code, and solving arithmetic problems. In a preliminary experiment the results from a group of fifty-two office workers were compared with the opinions of the office head. According to the report, "The comparison showed a very marked agreement between the testimony of the tests and the rankings of the office manager." It was decided to give the test to all clerks coming into the organization and to get follow-up ratings on the efficiency of these newly hired workers. Nine hundred and thirty-five clerks were tested in this way. "The results of this follow-up showed very clearly that the tests were an aid in the selection of clerks." Statistical results are given for a group of 188 of these employees. The individuals were clerks recommended for employment on the basis of tests and then reported on by their supervisors each month for three months. The tabulation showing the percentage of those called good by their superiors is given in Table IV. XH. C. Link, Employment Psychology. New York: Macmillan Co., 1919. 90 PSYCHOLOGICAL TESTS IN BUSINESS The most important point here is that the office heads agreed more and more with the test findings as they came to know their workers better. TABLE IV Per Cent At the end of i month 75 At the end of 2 months 89 At the end of 3 months 92 Comparisons were also made between the average test scores for different occupational groups. The findings are given in Table V. These test differences correspond with the relative value and importance of the different grades of work. TABLE V Occupations Average Score in Tests for Technique Average Score in Tests for Intelligence Time study group 92 96 Ledger group 82 80 Statistical group 77 71 Computing machines and sorting group 79 69 5. There is another study1 in which evidence regarding the relation of test scores to grade of office work, to turnover among office employees, and to ratings of individual efficiency stands out very clearly. The test used was one of general intelligence, very similar to the one used in the army. The relation between test scores and grade of work is presented in the three tables which follow. In each case the classification begins with the least difficult jobs and ends with the most 1M. A. Bills, "Standardizing the Selection of Clerical Workers," Service Bulletin of the Bureau of Personnel Research, Carnegie Institute of Technology, Pittsburgh, Vol. V, No. 9 (May, 1923). Also, M. A. Bills, "Relation of Mental Alertness Test Score to Position and Permanency in Company," Journal of Applied Psychology, VII (1923), 154-56; and C. S. Yoakum and M. A. Bills, "Tests for Office Occupations," Annals of the American Academy, November, 1923. TESTS FOR OFFICE OCCUPATIONS 91 difficult. In Table VIII, the high average of Class C is explained as due to special efforts of the company to recruit more capable employees in the lower-grade jobs. A study of the relation between scores on this same test and length of service of clerical workers showed the following: Of TABLE VI Jobs in Office I Average Test Score A 102 B 109 C HQ D 126 E 150 TABLE VII Jobs in Office II Average Test Score A 85 B 77 C i°5 D 103 E 118 Jobs in Office III Average Test Score A 52 B 91 C Il8 D 102 E 107 F 113 TABLE VIII those who scored below 60, 61 per cent left within 30 months; of those who scored between 60 and no, 33 per cent left within 30 months; of those who scored above no, 83 per cent left within 30 months. In general, that is, clerical workers with medium scores were most permanent and those with high scores were least so. A more detailed analysis was made of the relation between scores and turnover for one group of 133 clerical workers. At 92 PSYCHOLOGICAL TESTS IN BUSINESS the end of two and one-half years, 63 of these people were still in the employ of the company. As a result of the sifting process that had taken place, there proved to be a clearer relation between test score and grade of work among the 63 workers than among the original 133. People with high scores had left the lower grade positions, and those with low scores had become less numerous in the higher grade work. The results indicate that mental alertness was a significant factor in the success of workers in these clerical positions. In another group a comparison was made of scores and indi- vidual efficiency ratings. Ninety office employees were rated by their superiors on a percentage scale. The relation between these ratings and average test scores is given in Table IX. It is clear that those rated high tend to have high scores in mental alertness and those rated low tend to have low scores. TABLE IX Percentage of Efficiency of Clerks as Rated by Their Superiors - Over 80 Average Score in Test 120 79-60 80 59-40 78 Below 39 20 6. We shall give one example for a somewhat more specialized clerical test than those we have already described-a test for file clerks.1 The test consists of six parts, requiring such per- formances as the alphabetizing of names, the classifying of numbers, the comparing of pairs of names and numbers to deter- mine whether they are identical, the classifying of items of information under topical headings, and the answering of specific questions based upon paragraphs that are read. The tests aim to reach the several classes of file clerks-sorters, numerical filers, alphabetical filers, and subject classifiers. One hundred and twelve file clerks in four companies were given the tests. 1 D. G. Paterson, "The Scott Company's File Clerk Test," Journal of Per- sonnel Research, I (1923), 547-61. TESTS FOR OFFICE OCCUPATIONS 93 The scores were compared with ratings of the clerks, based upon their ability and the class of work being done. The ratings were considered accurate in only one of the four companies, though even the inadequate ratings in two of the other companies are said to have agreed fairly well with the test scores. Forty-three clerks were rated in the company where the reliable estimates were obtained. Here, a very close relation was found between the ratings of ability and the test scores. The relation is indicated by the figures given in Table X. The general intelli- gence test did not work as well with these file clerks as did the special file clerk test, though even the former showed fairly good agreement. Rating of File Clerks Average Score Good 118 Medium 84 Poor 40 TABLE X 7. In this final section of our report of the results obtained with clerical tests, we shall mention several miscellaneous pieces of work done by different investigators. The Section of Psychology in the United States Army conducted a minor study,1 with a group of civil-service clerks. Seventy-three clerks were tested with the army intelligence test. The scores were compared with the civil-service ratings obtained by these clerks at the time of their examination for entrance into the service. A fair degree of correlation was found. Within the last year or two the United States Civil Service Commission has adopted clerical tests of a psychological sort.2 1 Memoirs of the National Academy of Sciences, XV, 837. Yoakum and Yerkes, Army Mental Tests, p. 201. 2 H. A. Filer and L. J. O'Rourke, "Progress in Civil Service Tests," Journal of Personnel Research, I, 484-520. In a series of special studies, L. L. Thurstone has recently published useful comparisons of clerical tests used by civil service commissions. Examples of test material are given and discussed. (L. L. Thurs- tone, "A Comparative Study of Clerical Tests," Public Personnel Studies, Vol. I, Nos. 2-6.) 94 PSYCHOLOGICAL TESTS IN BUSINESS Their new clerical tests include those in spelling, arithmetic, and penmanship, and those requiring the candidates to arrange names alphabetically, to check pairs of names to see if they are identical, to give the opposites of words, to classify items accord- ing to a code, and so on. The tests were tried on ninety clerks and the scores were compared with their efficiency ratings. "The examination scores were found to differentiate very satis- factorily between those rated high and those rated low in effi- ciency." The relation of the new tests to efficiency was closer than was that of the old civil-service examinations, and was also closer than were such items as age, schooling, and amount of experience. Mental alertness tests were tried on a group of general clerical workers in a large rubber factory by one investigator.1 The employees were rated in efficiency by the department heads under whom they worked. Here, too, a fair degree of correlation was found. In another study2 clerical workers in a New York depart- ment store were given a long series of tests. Some relation- ship was found between test scores and ability as estimated by supervisors. It also proved possible to differentiate fairly well on the basis of the tests between clerical workers and sales persons. A German investigator3 gave a series of clerical tests consisting essentially in the finding of names and words in directories and dictionaries. The author states that the test scores agreed well with the efficiency of the workers. 1 H. E. Burtt, "Employment Psychology in the Rubber Industry," Journal of Applied Psychology, IV (1920), 1-17. 2E. O. Bregman, "A Study in Industrial Psychology Tests for Special Abili- ties," Journal of Applied Psychology, V (1921), 127-51; "Studies in Industrial Psychology," Archives of Psychology, No. 59 (1922). 3 J. Duck, Die Berufseignung der Kanzleiangestellten. For a summary of this and many other articles in the field of tests, see B. Muscio, Vocational Guidance, Report No. 12 of the Industrial Fatigue Research Board, London. TESTS FOR OFFICE OCCUPATIONS 95 A number of other studies1 report results more or less similar to the ones we have considered. In almost all cases a fairly high degree of correlation is found between scores on the intelli- gence or mental alertness tests and measures of the working efficiency of the people tested. One investigator2 found no significant relation between scores made by a group of express clerks and their success at work. This study, however, stands alone in its negative results. TESTS FOR CLERICAL WORKERS-SUMMARY Our reason for mentioning all these studies should be clear. Results from any one investigation in the use of tests for voca- tional purposes are likely to be true only within the particular groups studied and under the particular set of circumstances. It is only when we find certain tests working successfully in many different groups and situations within an occupation that we can safely conclude that the tests are good tests for that occupation. In the case of tests for clerical work, results have rather consistently pointed to the value of general intelligence tests and adaptations of these. The same is true of appropriate tests for educational achievement. In addition to the actual evidence favorable to intelligence and educational tests as aids in selecting clerical workers, there is much that may be said on common-sense grounds in support of such tests. This is particularly true if we recall all the scientific evidence that is available concerning these tests in 1 Among others, there are the following studies: C. S. Carney, "Some Experi- ments with Mental Tests as Aids in the Selection and Placement of Clerical Workers in a Large Factory," Conference on Educational Measurements, Indiana University, N (1919), 60-74; F. C. Henderschott and F. E. Weakley, The Employment Depart- ment and Employee Relations. Chicago: LaSalle Extension University, 1918; M. S. Viteles, "Job Specifications and Diagnostic Tests of Job Competency," Psychological Clinic, XIV (1922), 83-105; W. D. Scott and R. C. Clothier, Personnel Management, p. 261. 2 J. K. Flanders, "Mental Tests of a Group of Employed Men," Journal of Applied Psychology, II (1918), 197-206. 96 PSYCHOLOGICAL TESTS IN BUSINESS non-business fields. Both intelligence tests and educational- achievement tests have been widely standardized and verified in the schools, as means for ascertaining the present advancement (educational tests) and potential capacities (intelligence tests) of the pupils in the work of the schoolroom. If these tests are thus useful in determining the arithmetical and spelling ability of students in the schools and in predicting their ability to learn and progress in such studies, the inference is not unreasonable that the tests will prove similarly useful in diagnosing the present and future abilities of clerical employees whose work is made up largely of just such school subjects. In other words, there is strong presumptive evidence in support of intelligence and educational tests as useful for diagnosing the abilities of clerical workers. If we combine all the available facts bearing upon tests for clerical employees, a very favorable conclusion is justified. We do not, of course, mean that tests are completely satisfactory in this field or that there is not need for a great deal of further research and improvement of test methods. Still less do we imply that tests are sufficiently accurate and reliable to warrant the abandonment of other methods of ascertaining the abilities of applicants, such as careful interviewing. What we do mean is that the evidence in support of intelligence and educational tests for clerical work is great enough to warrant the use of these tests as one device which, properly used, will aid in judging the merits of clerical workers. At the same time the use of the tests is to be viewed as experimental. That is to say, new standards must be established in each company where the tests are put into use and, as soon as feasible, the results from the tests must be checked and verified. Our general conclusion, then, regarding tests for clerical workers is that they are of considerable value. While it is wise to work out the specific degree of their value in each particular situation in which they are used, sufficient evidence is available TESTS FOR OFFICE OCCUPATIONS 97 to create a strong presumption in their favor and to warrant high hopes for their further perfection. We shall find that far less can be said in support of all other occupational tests. TESTS FOR TYPISTS AND STENOGRAPHERS A clear distinction is to be drawn between tests of present proficiency in typing and stenography and tests of aptitude or potential ability for becoming successful in these occupations. The examinations in proficiency are simply short standardized trials on the job or samples of what the individual can do. They are really "trade tests." They test the typist by asking her to type. Obviously, however, such a test is not useful in predicting which individuals can be trained to become good typists. To make this diagnosis of potential ability, aptitude tests are needed which do not depend upon any acquisitions in typing. Proficiency tests in typing and stenography may be very briefly dealt with. They are the most common-sense sort of tests. Actual trials at typing and shorthand transcribing have, in unstandardized form, been widely used in employment. The standardized tests consist simply of appropriate material to be copied or dictated under prescribed conditions. Scoring methods and standards for comparison of scores have been established so that an individual can be given a definite rating in proficiency upon completion of the test. Tests of achievement in typing and stenography have been constructed and used in a number of places. Cody1 gives a complete description of his tests in copying and in transcribing from dictation, including scoring methods and averages for comparison. The Army Trade Division constructed and standardized similar proficiency tests, one for typists and one for stenographers. Other standardized tests of this sort have been made by Thurstone, The Scott Co., Link, and others. More 1 Sherwin Cody, Commercial Tests and How to Use Them. 98 PSYCHOLOGICAL TESTS IN BUSINESS or less similar tests have been in use by many typewriting companies, schools for the training of typists and stenographers, and private business organizations. The importance of careful standardization and establishment of the value of the tests has, however, frequently been neglected. In addition to these simpler tests in typewriting and stenogra- phy, a carefully worked out series of standardized tests for proficiency in shorthand has been reported by one investigator.1 There is a test for ability to read shorthand, a test for speed in writing shorthand, a test for shorthand penmanship, and a test for knowledge of the shorthand system. All these tests have been tried out on large numbers of students of shorthand and results have been obtained to show that the tests are fairly satis- factory in revealing differences in proficiency among individuals. When we turn to aptitude tests for typists and stenographers, evidences of successful and consistent results are not plentiful. Some investigators have gotten quite promising results, but others using similar tests have failed to secure useful predictions of ability. There is thus a striking lack of the consistency which we emphasized as important in clerical test results. On the whole, however, the tests for typewriting have been far from discouraging. Sufficient progress has been made to justify the belief that further experimenting will bring accomplishments of proved value. Some of the more promising investigations will be sketchily presented in the following pages. 1. Among the earliest studies of tests for typewriting aptitude was that of Lahy.2 Eleven experienced typists were tested with a series of special tests. The scores were compared with actual typing proficiency as demonstrated in a typewriting test. Four 1 E. R. Hoke, The Measurement of Achievement in Shorthand ("Johns Hopkins University Studies in Education," No. 6, 1922). 2 J. M. Lahy, "Les Conditions psycho-physiologiques de l'aptitude au travail dactylographique," Journal de Physiologic et de Paihologie general, XV (1913), 826-34. Summarized and statistically elaborated by Muscio in his Vocational Guidance. TESTS FOR OFFICE OCCUPATIONS 99 of the aptitude tests showed a high degree of relationship to ability in typing. These were tests involving immediate memory for concrete phrases, ability to grip equally strongly with the two hands, fineness of touch sensibility, and ability to judge slight differences in lifted weights. These four tests in combina- tion gave a very close prediction of typewriting efficiency among the eleven people tested. 2. Another investigator1 experimented with a group of students in a commercial high school, correlating test results with the abilities shown in school subjects. The test used was one in which the pupils had to substitute letters for one another in accordance with a code, the score being dependent upon the ability of the individual to improve in the rate of his performance. Fairly high correlations were found between the test and teachers' ratings of the pupils in typing, stenography, and business correspondence. There is some evidence, however, that the agreement is due to the teachers having estimated the general ability of the pupils rather than their specific ability in typing. 3. Several other studies have been more convincing than the foregoing. One experimenter2 used a series of association tests with a group of forty-five students of typing and stenography. The tests consisted of such tasks as giving the opposites of words, naming colors in rapid succession, following complicated direc- tions, substituting digits for geometrical forms according to a code, and so on. Measures of ability for the individuals tested were obtained as follows: In stenography a mid-year examination was given on the basis of which the students were graded as to their proficiency. In typewriting an achievement test was conducted once a month in which the individuals had to write from dictation under standardized conditions. Several of the 'J. E. Lough, "Experimental Psychology in Vocational Guidance," Pro- ceedings of National Conference on Vocational Guidance (1912), pp. 89-96. 2H. W. Rogers, "Some Empirical Tests in Vocational Selection," Archives of Psychology, No. 49 (1922). Also, "Psychological Tests for Stenographers and Typewriters," Journal of Applied Psychology, I (1917), 268-74. 100 PSYCHOLOGICAL TESTS IN BUSINESS aptitude tests showed fair agreement with the attainments of the students in typing and stenography. The tests which correlated best with typewriting were on the whole those which showed poor agreement with stenographic ability, and vice versa. Combinations of three or four tests were found which showed good correlations with stenographic ability and other combina- tions gave about equally close predictions of typing ability. The tests were also tried upon two groups of typists in a commercial establishment-one group of thirty-eight individuals who had worked in the same office division for at least ten months, and another group of sixty-five who had worked a shorter period. Records of output were obtained for typists in these groups in terms of the average number of sheets written per day. Test scores were then correlated with the production records. The agreement was a little less for these groups than for the student group, but results are well in accord with one another. 4. In another investigation,1 tests were tried out on a large number of typists and stenographers in an industrial plant and found to give useful results, though the statistical evidence has not been published. Aside from proficiency tests in actual typing and transcribing, the tests used consisted of the following: tests in spelling and grammar, a substitution or code test, and a test requiring the individual to supply omitted words in sentences. The individuals tested were twenty-two office typists, nineteen stenographers, and over four hundred candidates for positions as typists and stenographers. The author states that the tests have "proved themselves practical guides in the selection and grading of applicants." 5. In still another study2 an attempt was made to find tests which would be useful in predicting the abilities of applicants 1H. C. Link, Employment Psychology. 2 M. A. Bills, "Methods for the Selection of Comptometer Operators and Stenographers," Journal of Applied Psychology, N (1921), 275-83. Also "A Test for Use in the Selection of Stenographers," ibid., 373-77. TESTS FOR OFFICE OCCUPATIONS 101 for a training course in stenography. It was desired to eliminate failures and also to select individuals who would certainly be successful. Three tests were given to the students in the stenographic courses (eighty-four individuals). The tests com- prised one of general intelligence, one designed to show special aptitude in stenography and typewriting, and a "will- temperament test" which aims to get at certain qualities of character. Each of these tests is made up of several parts. Test scores were compared with the success of the students in their courses in typewriting and stenography. A person was considered a failure if her teachers did not recommend that she continue work in her course or enter an advanced course. Twenty-six of the eighty-four students were rated as failures. It was found that by combining scores on significant parts of the tests, failures and successes could be selected with considerable accuracy. To what extent the same combinations and standards would hold for other groups we do not know. Further experi- mentation is needed. As a matter of fact, some later work1 done by the same investi- gator did reveal the necessity for a change in the standards used. In order to test the results of the earlier study, three groups of stenographers in commercial organizations were tested. One group of twenty stenographers was rated by an executive as to their value to the firm. A high correlation was found between these ratings and the intelligence test. This, however, may have had little to do with any specific ability in typing or stenography. Similar results were obtained for another group of seventeen stenographers and eight secretaries. Decided differences in intelligence test scores occur between the secre- taries, stenographers rated as good, and stenographers who are just "getting by." On the whole, the results show a marked relation between mental alertness test scores and ability in stenographic work. 1 Reported in the second of the two articles cited. 102 PSYCHOLOGICAL TESTS IN BUSINESS 6. An English investigator1 has studied the relation of a series of tests to the ability of thirty typists employed in a govern- ment office. This series included tests of general intelligence, attainment in spelling and arithmetic, language ability and general information, and a number of proficiency tests in type- writing and shorthand-writing. Test scores were compared with rankings of the typists by their supervisor as to (a) their general office efficiency, (6) typewriting efficiency, and (c) efficiency in shorthand. The agreements between a number of the tests and ratings of efficiency are very close. It is to be noted, however, that the tests are largely ones of proficiency and not of aptitude. Correlations between the aptitude tests and the tests of actual achievement do not appear to be especially high. The author concludes, however, that "the results of the tests in the final form appear on the whole to be highly satisfactory." He doubtless refers more particularly to the proficiency tests than to those of aptitude. 7. Decidedly the most elaborate and thorough study of typing tests that has been made is the recent work of two British investigators.2 Especially worthy of note is their critical discussion of previous research in this field. The inconsistency of the results of different investigations is especially stressed, thus: The test of "attention" (cancellation), the test of acquisition of associa- tions (substitution), and the tests of "intelligence," have all yielded diver- gent results; one investigator obtaining a positive correspondence between typewriting efficiency and a given sort of test, and one or more investigators finding no such correspondence Indeed, the only tests with which a particular investigator obtained results so far uncontradicted are tests that have so far not been tried by other investigators. The authors aimed in their own investigation to find suitable groups in which the test findings would not be thrown off by 1 C. Burt, "Tests for Clerical Occupations," Journal of the National Institute for Industrial Psychology, I (1922), 23-27 and 79-81. 2B. Muscio and C. M. Sowton, "Vocational Tests and Typewriting," British Journal of Psychology, XIII (1923), 344-69. TESTS FOR OFFICE OCCUPATIONS 103 unusual conditions of age, experience, and the like, or by unreliable ratings of the efficiency of the individuals studied. Ten different groups were used consisting of typists in business offices and commercial schools. A great variety of tests were tried which are grouped as follows: i. Attention (rapidity of observation) Cancelling letters on a page, finding numbers in a table, noting the identity of names, etc. 2. Memorizing (acquiring associations) Substituting letters for forms according to a code, etc. 3. Immediate memory span Reproduction of sentences and numbers after looking at them. 4. Speed of association Giving words bearing a certain relation to suggested words. 5. Intelligence Several typical intelligence or mental alertness tests. 6. Tactile sensibility Discriminating by touch between cards having two, three, or four small holes punched in them. 7. Motor capacity Strength of grip in the right and left hands; tests of manual dexterity and speed of finger movement; etc. 8. Acquired associations (knowledge) Spelling and vocabulary tests. As a criterion of the actual typewriting efficiency of the indi- viduals tested, ratings by their supervisors and teachers were used. Special effort was made to obtain as reliable gradings as possible. The correlation between test scores and the gradings in typing efficiency are found to vary greatly. Some of the tests show close agreement with ability as typists, but the relation is not consistently shown for the different groups. We cannot do better than quote in this connection a paragraph from the origi- nal report. The most important feature of the results is the variation in the size of the correlation coefficients obtained from different groups with identical tests-identical not only in principle, but often also in material. This simply upsets the usual procedure in vocational investigations; that is, 104 PSYCHOLOGICAL TESTS IN BUSINESS the drawing of conclusions as to the vocational value of particular tests for specified occupations from coefficients obtained from a single group of industrial subjects. For there can be no doubt whatever that the size of a coefficient of a test with occupational efficiency is partly a function of the group to which the test is given. It is suggested by the authors of the report that the chief cause of the inconsistent results is the varying reliability of the efficiency gradings given in the different groups by different supervisors and teachers. This condition emphasizes the impor- tance of obtaining objective records of efficiency wherever possible, in place of personal estimates. Where objective records are not available, ratings may be increased in reliability by taking the average of estimates by several independent judges. The desirability of testing the test in more than one group is also clearly indicated. A second cause which is mentioned as an explanation of the variation in results is the fact of specialization within the occupa- tion of typists, and the requirements of different personal abilities in these different divisions of the occupation. This difficulty calls attention to the necessity of using, in the experimental tryout of tests, groups that are as nearly as possible identical with the group to whom the tests are to be applied in actual practice. The writers under consideration conclude, on the basis of their own investigation and all preceding work, that there are five tests which show some possible indication of being useful in diagnosing typewriting aptitude. These are: (i) a test for reproducing sentences immediately after they are read; (2) a test in carrying out certain directions; (3) a test requiring the finding of the products of two numbers in a table made up for that purpose; (4) a test in which appropriate words must be supplied in a sentence containing blanks; and (5) a spelling test. 8. Two still more recent studies of tests for typewriting aptitude have reported very favorable results-but with tests TESTS FOR OFFICE OCCUPATIONS 105 different from any of those recommended as promising in the conclusions of the foregoing report. The first of these investiga- tions1 made use of two specially devised tests for typing ability and of five well-known word association tests. One of the special tests was a "serial typewriting reaction test" in which the individual tested had to respond to a sequence of numbers by striking appropriate keys of the typewriter. As soon as one key was struck the next number was given. A continuous series of reactions was thus called for which required both speed and accuracy. The other special test was a "typewriting rhythm test" which measured the ability to strike series of keys with rhythmic regularity. All the tests were given to two classes in typewriting, numbering twenty and twenty-two respectively. The tests were used before the individuals began learning to typewrite. The test scores were later compared with several different measures of the typing proficiency attained by the individuals tested. Very close agreement was found between some of the tests and achievement in typewriting in the two classes tested. Two of the association tests gave fairly promising results and the serial reaction test was exceptionally successful. The two best association tests were "form substitution" in which numbers must be written below geometrical forms in accordance with a code that is given, and a "color-naming test" in which the person tested must call out as rapidly as possible the names of one hundred bits of color arranged on a sheet of paper. By combining these association tests with the serial reaction test, the investi- gator found results about as favorable as have been obtained for any other occupations. The rhythm test worked very well with one group, but not with the other. Further experimenting with these tests will be well worth while. XA. Brewington, "Prognostic Tests for Typewriting," American Shorthand Teacher, Vol. IV (1923), Nos. 1 and 2. A more complete account of this investiga- tion is contained in an unpublished thesis at the University of Chicago. 106 PSYCHOLOGICAL TESTS IN BUSINESS 9. The one other recent investigation1 which we shall report made use of a series of six tests with twenty students. Test scores were compared with actual typing proficiency. Three of the tests showed fairly good results. These were: (1) a test in speed of tapping, in which a typewriter key had to be struck over and over again in rapid succession; (2) a test in which the individual had to underline certain combinations of numbers and letters on a page containing rows of digits and letters; (3) a substitution test in which numbers were to be written beside symbols in accordance with a certain code. TESTS FOR TYPISTS AND STENOGRAPHERS-SUMMARY The selection of typists and stenographers on the basis of their present accomplishments is thoroughly feasible. Brief standardized tests in typing and shorthand-writing are readily constructed and used. There are a number of them on the market.2 These tests are a common-sense, and at the same time scientific, method of getting a sample of the applicant's actual efficiency. A considerable number of business offices are making use of such tests as a regular part of their employment procedure. The selection of individuals with respect to their future possibilities in typing and stenographic work is a problem of vastly greater difficulty. We have reported the more important attempts to solve this problem. While these studies have made a certain amount of progress, we are still unable to point with any great confidence to any test or set of tests which may be recommended for practical use in determining specific aptitude for typewriting and shorthand. No tests have been demon- strated to be consistently valuable. Certain tests show some relation, however, and these may well be tried experimentally 'W. W. Tuttle, "The Determination of Ability for Learning Typewriting," Journal of Educational Psychology, XIV (1923), 177-81. 2 See, for example, the World Book Co.'s Bibliography of Tests for Use in Schools, and the tests handled by the Gregg Publishing Co. TESTS FOR OFFICE OCCUPATIONS 107 in different places with a fair promise that they will prove of value. Intelligence tests, moreover, are useful among typists and stenographers, as among other workers, for indicating general adaptiveness and alertness aside from special abilities peculiar to the occupations. As to the more specific aptitude tests, we can say in conclusion only this: Enough work has been done to open up a promising field for research. We would advocate the use of tests for typing and stenographic aptitude-but only in case the emphasis is placed upon testing the tests and not upon using the results for the present as measures of the applicants' abilities. TESTS FOR OTHER OFFICE OCCUPATIONS Only a few studies of tests have been made for office occupa- tions other than those we have considered. Intelligence tests have occasionally been tried on executives and on office boys but few significant results have been published. There is good reason for assuming that with office boys these tests are of considerable value, arguing from what we know of mental testing of school children and the success of the tests in picking out the bright and dull. One or two studies of office boys and messengers and of executives may be briefly reported. The mental-alertness-test scores of messengers and office boys have been analyzed principally in relation to labor turnover.1 In one group which was studied, turnover was found to be high among the individuals making neither very high nor very low scores. Stability was greatest, in this company, among the boys with the highest scores. In another company, the scores of the boys who were discharged were compared with the average of those who were promoted. The former were decidedly lower. A more detailed report is given of the test results of in office boys whose records were studied about two years after the test scores had been obtained. The average score of the boys remaining with the company was higher than that of the boys 1 Scott and Clothier, op. cit; A. J. Snow, op. cit. 108 PSYCHOLOGICAL TESTS IN BUSINESS who had left, but the average of those who left to accept better positions was higher even than that of the group still employed. The discharged group averaged very low. The boys with high- est scores were receiving considerably higher salaries than were those with the lowest scores. It was found possible to set a "passing mark" in the test, which would differentiate fairly well between the boys who were successful and permanent as contrasted with those who were unsuccessful and likely to leave. Some agreement was also found between the test scores of the boys and the ratings of their abilities by executives. Material bearing on the results of tests for executives is meager. One experiment1 reports the tryout of a mental alert- ness test with 28 minor executives in a clothing manufacturing establishment. The 28 men were rated by six superior execu- tives. Exceptionally close agreement was found between the test scores and the ratings given the men by their superiors. In a recent study,2 102 business executives attending a con- ference in financial statistics were given an intelligence test. Experience records containing personal information about the individual^ were obtained from 73 of these men. These records were used as a criterion with which to compare the test scores. No agreement whatever was found. Regarding these test studies of executives, it is important to recognize that there is a great difference between distinguishing good executives from less successful ones, and selecting executives from the general run of men. Tests may be of very little value in diagnosing the precise degree of executive ability an individual possesses, and still be highly useful in differentiating between the whole class of men who may become executives and those who probably are not sufficiently intelligent. The specific results thus far secured offer no sound basis for assuming that tests 1 Scott and Clothier, op. tit. 2W. V. Bingham, "Intelligence Test Scores and Business Success," Psycho- logical Bulletin, XXI (1924), 103-5. TESTS FOR OFFICE OCCUPATIONS 109 can do more than give this very crude indication of executive ability. Some psychologists are of the opinion that there is no hope of developing more accurate tests in this direction.1 The only other office occupations to be mentioned are special- ized operations on office machines. Several investigations of this sort will be sketched. 1. In one study2 a series of tests was given to three groups of over 140 operators of computing machines (comptometers and Burroughs adding machines), to more than 120 candidates for positions in this work, and to 80 pupils entering a class for instruction in the operation of computing machines. The tests found most useful, aside from a proficiency test in machine com- puting, were a mental arithmetic test and a test requiring the substitution of numbers according to a code. Intelligence tests did not show any marked correlation with ability as operators. On the basis of the tryout the several tests mentioned were deemed of sufficient value to be adopted for use in employment. Definite report of results is made for only one selected group of twelve individuals, in which group agreement between test scores and executives' opinions was close. 2. A second investigation,3 also dealing with comptometer operators, used tests of intelligence, a series of tests for aptitude in typing, and a test for volitional qualities.4 Thirty-five comptometer operators in an evening class were tested, of whom fourteen were rated by their teachers as failures. A combination of the tests was found which picked out ten of these fourteen failures. Whether the same tests and standards would show equal success in other groups is not, of course, known. 1 See, for example, the discussion by H. C. Link in his Employment Psychology, chap. xvi. 2 H. C. Link, op. cit. sM. A. Bills, "Methods for the Selection of Comptometer Operators and Stenographers," Journal of Applied Psychology, V (1921), 275-83. 4 This study was made in connection with the fifth one reported in our dis- cussion of tests for typists and stenographers (p. 100). The same tests were used. 110 PSYCHOLOGICAL TESTS IN BUSINESS 3. In another study1 several tests were tried out with a group of twenty-seven Elliott-Fisher bookkeeping-machine opera- tors. The work is a specialized sort of operation involving simple typing and computing. Fair agreement was found between two or three of the tests and the actual ability of the individuals in their work. No final conclusions can be drawn, however, from an investigation as limited as this. One interesting point in this study is that the number of years of schooling of the indi- viduals agreed as well as did test scores with their actual efficiency. How true this would be of other office groups is not known. 4. In one experimental investigation2 the intelligence test used in the United States Army was given to 106 graphotype operators in the Treasury Department in Washington. The tests showed no agreement at all with either the speed or ac- curacy of these workers. 5. The final piece of work we shall mention is a study3 of selective tests for Hollerith machine-card punchers.4 About 130 experienced operatives were tested with a series of nine tests-mainly ones that are usually included as parts of a general intelligence test. Ratings of actual efficiency on the job were obtained for the individuals tested. The agreement be- tween test scores and the efficiency ratings is slight. It is never- theless true that a combination of three of the tests proved more valuable for predicting ability than did the civil service examination that had been in use. 1 A. W. Kornhauser, "A Statistical Study of a Specialized Group of Office Workers," Journal of Personnel Research, II (1923), 103-23. 2 "Psychological Examining in the United States Army," Memoirs of the National Academy of Sciences, XV (1921), 837. sL. Marcus, "Vocational Selection for Specialized Tasks," Journal of Applied Psychology, IV (1920), 186-201. * The Hollerith machine is a statistical device for sorting and counting cards carrying statistical data. The cards must be perforated or " punched " in a machine more or less like a simplified typewriter. TESTS FOR OFFICE OCCUPATIONS 111 These last studies of office-machine operations have added little to our positive knowledge of tests for office occupations. They are of interest primarily as indications of lines along which further more fruitful work may be expected. CONCLUSIONS REGARDING TESTS FOR OFFICE WORKERS Our general summary of what has been accomplished in the field of tests for office occupations can be very briefly stated. Most important is the method that has been developed for use in further test research. The significance of this permanent contribution of technique is not to be lost sight of, whatever be the fate of the more concrete test achievements. As to the specific occupational tests we have considered in the foregoing pages, there is especially strong evidence in support of (a) general intelligence tests and tests of achievement in spelling, arithmetic, etc., as aids in selecting clerical workers; and (6) proficiency tests (trade tests) in typewriting and stenogra- phy. Tests of aptitude in these last occupations and also in computing-machine work show promise, but have not progressed far enough to warrant a positive indorsement. On the basis of research in non-business fields as well as in business, a favorable view of intelligence tests is justified as a device for indicating the alertness and future possibilities of various classes of office workers, ranging from office boys to secretaries and perhaps to minor executives. As research is continued and expanded we may hope gradually to see these promising beginnings grow into rich bodies of concrete accomplishment. CHAPTER IV PSYCHOLOGICAL TESTS FOR NON-OFFICE OCCUPATIONS While the most successful applications of psychological tests have been in the selection of office workers, valuable investi- gations have also been conducted in a wide variety of other occupations. The fields in which studies of vocational tests have been made range from lathe operation to fife-insurance salesmanship and from type-setting to aviation. Some twenty- five to fifty different lines of work, aside from office occupations, have been the objects of specific pieces of test research. This is in addition to the occupations for which standardized trade tests have been constructed, numbering well over one hundred. The variety of vocations that have been worked with indicates something of the vigor and promise of the test movement. In the process of reaching out into many fields of work, however, selective tests have failed to develop to their fullest in any one field. It is necessary continually to remind ourselves that employment tests are very young and that only a few scientifically trained psychologists have been active in this work. If we consider the comparatively small amount of research that has been carried on, and then think of this research spread over several dozen occupations, we cease to expect great accom- plishments in specific lines. Psychological test studies have, on the whole, been exploring possibilities rather than intensively developing any one field. Many pioneer investigations have been made, but only in a few instances, thus far, have the promis- ing trails been followed farther. The meaning of all this is simply that vocational tests are still in an early experimental stage. Decidedly encouraging 112 TESTS FOR NON-OFFICE OCCUPATIONS 113 accomplishments are on record, but scarcely any of these can be called final and complete. The results have served mainly to show attractive possibilities in a large number of vocations and to develop a technique for further research. It remains for the future to follow up these leads and to apply the test methods in building actual test accomplishments that are more impressive than those we now have. Too conservative an interpretation is not to be placed upon the foregoing remarks. A number of test studies have reported actual accomplishments of considerable value. These we shall describe in the sections that follow. Even in these cases, however, results are somewhat tentative since in most instances only a single group of workers has been experimented with. Only after we have used tests on a number of different groups, or on one or two very large groups in an occupation, will we be able to say, "Here is a set of standardized tests for this occupa- tion," or even, "Here is a set of standardized tests for this partic- ular job in this particular plant." Nevertheless, positive accomplishments in a single restricted experiment are valuable. They merely need to be added to. But as far as they go they are useful and significant contributions.1 In our inquiry into test accomplishments, we need to begin by recalling the distinction between trade tests, or proficiency TRADE TESTS 1 Several summaries of the work done with tests in business and industry- have appeared during the past two or three years. Among these are the following: B. Muscio, "Vocational Guidance (A Review of the Literature)," Industrial Fatigue Research Board, Report No. 12 (London, 1921). A. W. Komhauser, "The Psychology of Vocational Selection," Psychological Bulletin, XIX (1922), 192-229. E. Claparede, "Problems and Methods of Vocational Guidance," International Labour Office Studies and Reports, Series J, No. 1 (Geneva, 1922). C. H. Crennan and F. A. Kingsbury, Editors, "Psychology in Business," Annals of the American Academy of Political and Social Science, Vol. CIX (November, 1923). R. Pintner, Intelligence Testing, chap. xix. New York: Henry Holt & Co., 1923. 114 PSYCHOLOGICAL TESTS IN BUSINESS tests, and aptitude tests. Trade tests attempt to measure the individual's present knowledge and skill in some particular line of work; aptitude tests try to determine the individual's potential ability, his capacity for achieving success in a particular direction, his possibilities of acquiring the requisite knowledge and skill. Trade tests, accordingly, proceed by asking questions about the job or by having the individual perform some repre- sentative sample piece of work. Aptitude tests must, on the other hand, avoid the use of items of knowledge or skill that are acquired in the work being tested for; they assume that the man tested is unacquainted with the particular occupation. Trade tests are evidently very simple. There is nothing mysterious about testing a typist by having her type a sample letter. Common sense would lead us to ask an applicant for a machinist's job some questions about his trade. But true trade tests are not mere haphazard questions or chance jobs given the applicant to "try him out." They are standardized. The questions and the sample pieces of work have been experi- mentally tested, verified, and refined. They are fixed and constant in the things they include, in the way they are adminis- tered, and in the way they are graded or scored. Nothing is left to the opinion of the person using the tests. Standards have been set showing precisely how well men of known ability do on the tests. These standardized trade tests are very different instruments from the usual trade interviews or trials on the job. Standardized trade tests were developed almost exclusively in connection with the work of the Committee on Classification of Personnel in the United States Army.1 Trade tests were 'Valuable descriptions of army trade-test methods are contained in the following: J. C. Chapman, Trade Tests. New York: Henry Holt & Co., 1921. Personnel System of the U.S. Army. Washington: Adjutant General's Depart- ment, 1919. P. J. Reilly, "The Selection and Placement of Employees," Federal Board for Vocational Education, Bulletin No. 49 (1919). B. Rumi, "Extension of Selective Tests to Industry," Annals of American Academy of Political and Social Science, LXXXI (1919), 38-46. J. C. Chapman, "Tests for Trade Proficiency," ibid., CIX (1923), 45-59. TESTS FOR NON-OFFICE OCCUPATIONS 115 used in the army to determine whether men were tradesmen in the lines in which they claimed skill, and approximately their grade of proficiency. Sixty-nine oral tests and twenty-three picture tests were constructed for a wide range of the common occupations. These tests consisted of carefully sifted and standardized sets of questions upon which the candidate could be rated. In some few trades, performance tests were used (twenty performance trade tests were completed), which required the candidate to perform some definitely specified task that was typical of his trade, a rating being assigned according to the excellence of the completed product and the time required. All the tests were tried out on groups of tradesmen in industry and were evaluated by comparison with the known ability of these men. These tryouts, in general, produced evidence of the value of the tests. No adequate studies have been made, however, of the actual results obtained through the use of the army trade tests. Not much has been added to the trade test work of the army. There has been considerable talk of the possibilities of trade tests in industry, but little actual progress has been reported in the use of standardized tests for non-office occupations. Adap- tations of the army trade tests were tried experimentally in some public employment bureaus,1 but for the practical needs of the situation unstandardized 11 selective trade interviews" were later adopted. A trade test in written form that could be used with groups was developed and tried out in one post- war experiment.2 The statistical evidence is favorable and indi- cates that trade tests of this kind may be feasible. Probably the chief work on trade tests, since the war, is some unpublished research of The Scott Company. Attempts were made to adapt army trade tests to industrial needs in several trades and to build new tests in several other trades. Some 1 J. C. Chapman, Trade Tests. Henry Holt & Co., 1921. ' J. C. Chapman and H. A. Toops, "A Written Trade Test," Journal of Applied Psychology, HI (1919), 358-65. 116 PSYCHOLOGICAL TESTS IN BUSINESS degree of success was achieved, but perhaps the most definite result was the conviction that trade tests can be most success- fully developed only through the co-operative efforts of a number of companies. It also appeared that for many industrial situa- tions unstandardized or partially standardized trade interviews may be more useful than formal tests. Adaptations of trade tests to the selection of department-store salespeople, according to their knowledge of merchandise, were also developed. These met with considerable success. All the work that has been done with trade tests, both in the army and in industry, has impressed investigators with the great possibilities of progress in this direction. Many optimistic discussions have appeared which emphasize the desirability of extending to industry the army trade test methods. It is pointed out that the tests are valuable not only in selection but also in training,1 and in the classification of tradesmen by employers and even by trade unions. The fact is, however, that actual accomplishments are still slight. The difficulties in constructing and standardizing trade tests are formidable, and in many cases unstandardized methods work sufficiently well. Unstandardized trade tests are, of course, widely used in industry. They cannot, however, be considered special scientific test methods as they use only ordinary common-sense procedure. To what extent the optimism will be justified regarding standardized trade tests remains at present an open question. APTITUDE TESTS The more important of the studies in tests of vocational aptitude in the several occupation fields will be summarized in the following pages. We shall divide the investigations into three classes: (1) tests for occupations in factory and workshop; 'See, for example, A. W. Kornhauser, "A Plan of Apprentice Training," Journal of Personnel Research, I (1922), 215-30. TESTS FOR NON-OFFICE OCCUPATIONS 117 (2) tests for salespeople; and (3) tests for a miscellaneous group of other occupations. TESTS FOR OCCUPATIONS IN FACTORY AND WORKSHOP Among the best studies of aptitude tests for factory occupa- tions are the researches of H. C. Link in the Winchester Repeating Arms Company.1 He conducted a series of careful, scientific test investigations. Special tests were constructed and stand- ardized for five different occupational groups in the shop. In each instance a thorough study was first made of the requirements of the work; then a series of tests was devised and tried out; then comparisons were made between test scores and measures of working efficiency-production records where these were available, and estimates by supervisors in other cases. These several test studies may be briefly viewed in turn. i. Shell and cartridge inspecting.-Fifty-two girls engaged in inspecting brass shells constituted the experimental group in the first study. Their work consisted in taking up handfuls of shells and carefully inspecting them for dents, scratches, stains, and other minute defects. Defective shells had to be extracted from the pile and thrown into one of several appropriate scrap boxes. Sixteen tests were tried out in a preliminary experiment, and eight of these were retained for more detailed study with the group of fifty-two inspectors. The test scores were compared with individual efficiency records showing the average hourly output over a period of four weeks. Results showed that three of the eight tests were fairly closely related to the ability of the inspectors. These successful tests were as follows: (a) a card- sorting test, in which the individual had to sort some cards into two piles, one pile containing cards with the letter "o" on them and the other containing the remainder; (6) a cancellation test, in which every " 7 " had to be crossed out on a page of mixed- 1H. C. Link, Employment Psychology. New York: The Macmillan Co., 1919. 118 PSYCHOLOGICAL TESTS IN BUSINESS up numbers; (c) a number-group-checking test, which required the individual to mark certain sets of numbers on a page. These same three tests were tried on small groups of workers in closely allied jobs-twenty-eight cartridge inspectors, ten bullet inspectors, and thirty paper shot-shell inspectors. With the first two of these groups the tests gave results almost as good as they did with the shell inspectors. Almost no agreement was shown, however, with the efficiency records of the paper shot- shell inspectors. This may have been due to the fact that there were no individuals whose production records were distinctly high or low, and hence there was little possibility for the tests to discriminate good and poor workers. On the whole, the conclusion based on the experimental work with these tests is very favorable. The investigator reports, "The results in their entirety were such as to justify the use of the three most significant tests together with the eye-sight test in the employment office." Further evidence regarding the tests was obtained from their actual use. Twenty-nine hundred applicants were tested in the employment office. Several small groups from this total number are reported on. Some individual case reports are given to show the superiority of the tests as selective devices as contrasted with the personal estimates by foremen. Follow- up reports on ninety-four inspectors are compared with their test standing and with their length of service. This comparison shows the following relation: TABLE XI Rated by Foreman as: Average Length of Service Satisfactory Unsatisfac- tory No Opinion Above the test standard 34 I 32 Weeks 9-56 Below the test standard I 17 9 1.05 TESTS FOR NON-OFFICE OCCUPATIONS 119 Another comparison was made for a group of forty-three inspec- tors who had been at work two months or more. Their actual production records were compared with their test scores on two of the tests. (The card-sorting test had not been given.) The agreement is not quite as close as in the original experiment, but is still far from unsatisfactory. 2. Shell gauging.-The same tests as those used with shell inspectors were tried out with a group of twenty-one girls engaged in gauging the head thickness of shells. The two jobs are very different in their requirements. The gauging is done in a very mechanical fashion which does not even require the use of the eyes. The tests, as might be expected, showed wholly dis- similar results with the two groups. Only one test proved useful in indicating the ability of the gaugers-a 11 tapping test" in which the individual had to press a telegraph key as rapidly as possible. The tests which gave good results with inspectors were of no value in this new kind of work. In both lines of work intelligence tests showed almost no agreement with productive efficiency. 3. Assembling of gun parts.-Tests were devised and tried out on several groups-one group of women and three groups of men. The work consists of placing the gun parts in position and then hammering, screwing, pushing, or springing them together. The tests used were two ''form-board tests" which require the individual tested to place blocks of different sizes and shapes into appropriate holes, and a strength of grip test. These tests were given to eighteen women applicants, twelve of whom were hired. After one or two months these girls were ranked in efficiency by their foreman and their section boss, and these ratings were then compared with the test scores. Estimates were obtained for only ten girls. Within this small group the agreement was fairly close between efficiency and test scores, but the number of individuals is too small to establish the value of the tests. 120 PSYCHOLOGICAL TESTS IN BUSINESS The groups of men tested consisted of thirty-one action- assemblers, fifty-one finishing-assemblers, and another group of twenty-six action-assemblers. The work was a little higher grade than that performed by the women assemblers. Tests similar to those given to the women were used. The Stenquist assembling test was added, a test which consists of a number of unassembled familiar objects to be assembled, such as a mouse trap, a bicycle bell, and a simple monkey wrench. As in the case of the women, it proved impossible to obtain objective production records as a criterion against which to check the tests. Accord- ingly each foreman ranked his men in efficiency and these esti- mates were compared with the test scores. The results can be summarized in the investigator's own words, "On the whole, the results of this experiment were not such as to justify using any of the tests tried as a basis for choosing new assemblers. Nevertheless, the correlations found justify continued research in this field." This same quotation might well conclude a score of other test investigations. 4. Dial-machine operating.-A specially devised test was tried out with several groups of machine operators. The job studied involves feeding a machine by placing pieces of metals quickly in position on a rotating table. The test which was constructed is somewhat similar. It requires the operator to drop steel balls through a slot in a rotating disk. The test was given to ninety-six dial-machine operators in three different shops. Test scores were then compared with production records. The agreement is fairly good. The test was considered useful enough to be adopted in the employment office. 5. Tool-making and machine-shop work.-Only a few men were tested in this final investigation. First, twelve men with from three to five weeks' training were tested with five tests. The test scores were compared with the foreman's ranking. Three tests showed exceptionally good results: a " form board" test similar to the ones tried with assemblers, the Stenquist TESTS FOR NON-OFFICE OCCUPATIONS 121 test, and a "cube-construction test" in which one must assemble twenty-seven small cubes into a large cube in such a manner that only the painted surfaces of the small cubes show. Two other groups consisting of eleven and twelve individuals respec- tively were tested with these same tests. The results were not as favorable as in the first group, but were still very good. Investigations by a number of other industrial psychologists have produced results of the same sort as those just described for a variety of occupations. A few of the more important studies follow: 6. Apprentice work in a machine-shop.-A study of tests for machine-shop apprentices1 has recently been reported which is similar to Link's experiments with apprentices in tool-making and machine-shop work. The tests used were about the same in this later work as those described under section 5 above. Twenty-five apprentices were tested and the test scores were compared with estimates of the ability of the men by their foremen and instructors. Some agreement was found, though it is scarcely close enough to prove that the tests have great value. In some measure, however, the results serve to support those that Link obtained in his earlier work. 7. Occupations in a rubber factory.-One series of test studies was carried on in the plant of a Canadian rubber company.2 A large number of tests, some thirty-two in all, were tried out in connection with several different jobs. Fairly successful sets of tests were found for tire-finishers and treaders, for workers who supply stock to the operatives, and for a miscellaneous group of moderately skilled factory workers. The tests for tire-finishers and treaders that were selected on the basis of the tryouts con- sisted of the following: (a) underlining in a large assortment of 1W. S. Berry, "Testing Tests," Scovill Bulletin, IX (1923), 5-7. This experi- ment is also reported in an article by H. C. Link in the Annals of the American Academy (November, 1923). 3 H. E. Burtt, "Employment Psychology in the Rubber Industry," Journal of Applied Psychology, IV (1920), 1-17. 122 PSYCHOLOGICAL TESTS IN BUSINESS digits adjacent pairs of numbers whose sum is ten; (b) finding consecutive numbers scattered at random over a page; and (c) responding as quickly as possible to a signal shown by the experimenter. These tests agreed to some extent with the ratings of efficiency of the workers. Another set of three tests proved about equally diagnostic of ability among the individuals who hand out stock. Similar tests were tried for tire-builders, but no agreement was found, despite the fact that the work is very similar to that of the tire-finishers. A set of five tests of the sort usually included in general-intelligence tests proved somewhat useful in judging the ability of the miscellaneous group of factory workers. These experimental results are promising and worthy of further investigation. The work that has been described is, however, not at all conclusive, since the number of people tested is not known and since the statistical methods used are open to considerable question. 8. Lathe operating.-Another investigation1 of a wholly different character involved the tryout of tests for predicting ability in lathe operation. An assortment of tests was given to sixty engineering students in a college course in shop practice. Scores on the tests were correlated with the ability of the students as shown by the accuracy and quality of the work they produced in the course. Their products were carefully and objectively measured in order to secure a sound rating of efficiency. From the seventeen different test scores computed, five were retained as the ones of greatest value. The combined score from these five tests agreed fairly well with the measures of actual ability. Whether this would be true of other groups of lathe operators can be ascertained only through further research. The tests which proved useful are: (a) and {b} a specially devised co-ordination test, which represents in miniature certain elements *E. F. Patten, "An Experiment in Testing Engine Lathe Aptitude," Journal of Applied Psychology, VII (1923), 16-29. TESTS FOR NON-OFFICE OCCUPATIONS 123 of lathe operation; (c) a test in which the six parts of a lock must be assembled; (d) a test in which a "puzzle box" must be opened by manipulating levers according to directions; and (e) a test in which the individual must quickly mark the centers of a number of circles of different sizes. 9. Printing: hand-composing.-An exceptionally well con- ducted investigation was made in the printing industry by a British psychologist.1 The principal work was done with the job of hand-composing. The hand-composer or compositor assembles the lines of type to be used in printing. He must pick the letters from the appropriate compartments on a wooden tray which holds all the letters and other signs used in printing. The type must be quickly and correctly placed, the lines must be kept even, the pages of type must be properly locked with space for margins, and so on. The qualities required for the work are found to be: (a) good eyesight, (5) physical strength above average, (c) dexterity of right hand and arm, (d) rapid visual observation, (e) good "immediate memory" or ability to hold material in mind for a short interval, (/) ability to estimate size and form, (g) moderate degree of general intelligence. Six tests were tried out on groups of compositors in four printing works. The groups contained twenty-five, twenty-four, forty, and eleven individuals, respectively. In each instance the workers were ranked according to their test scores and also according to their efficiency as rated by the managers. These rankings were then compared in order to see how well the tests succeeded in picking out the good and the poor compositors. Three of the tests gave very good results. These were: (a) a "match-stick insertion test" which requires that match sticks be inserted in small holes in a board; (b) a "cancellation test" in which every e on a page has to be crossed off; (3) a "directions test" where the individual must carry out precisely certain 1 B. Muscio, "The Psycho-Physiological Capacities Required by the Hand Compositor," Industrial Fatigue Research Board, Report No. 16 (section A). 124 PSYCHOLOGICAL TESTS IN BUSINESS specific instructions. Combined scores on these three tests agreed closely with ratings of the ability of the compositors. Some evidence appeared that a fourth test has value-an immedi- ate memory test in which the individual must remember and write down sentences he has just read. 10. Other occupations in the printing industry.-Two earlier studies in the selection of workers for the printing industry had been published in Germany.1 These were war-time investiga- tions aiming to aid in the selection of women applicants to replace men. The first investigator used a series of five tests, and, on the basis of a tryout among sixty applicants, the tests were recommended for use and standards were set. Definite evidence in support of the tests was not published. The second investigator used substantially the same tests in the regular examination of applicants in a large printing industry. Follow- up reports indicated that there was considerable agreement between the test scores and the ability of the employees as rated by their supervisors. 11. Occupations in a feather factory.-In another investiga- tion,2 seventy-five women employees in a feather and fancy- ornament factory were tested. Some nineteen different tests were used, most of them well-known standardized tests of general intelligence and quickness of association. Seven specially devised tests were tried out. These consisted of several tasks analogous to the actual operations and of other tests which gave promise of tapping the abilities involved. The seventy-five workers tested were ranked in order according to their scores on each test, and these rankings were compared with an "effi- 10. Lipmann, Die Berufseignung der Schriftsetzer. Leipzig: Barth, 1918. D. Krais, Eignungsprufungen bei der Einfuhrung von weMiche Ersaizkrdften in das Stuttgarter Buchdruckgewerbe. Leipzig: Barth, 1918. These studies are summarized in Muscio's "Vocational Guidance (A Review of the Literature)," Industrial Fatigue Research Board, Report No. 12 (London). 2 Emily Burr, "Psychological Tests Applied to Factory Workers," Archives of Psychology, No. 55 (May, 1922). TESTS FOR NON-OFFICE OCCUPATIONS 125 ciency roster" compiled from ratings made independently by four members of the management. Remarkably high indexes of agreement are obtained (though the statistical methods are not clear) but these are of doubtful importance, due to the fact that the people tested composed a most heterogeneous group, ranging from skilled operatives in the "fancy department" to errand girls and routine feather-sorters. In actual practice the employment problem is rarely to select better and poorer workers from a group so varied. A separate study was made of twenty girls in the selecting department, all engaged in sorting feathers according to size and color. Three tests were found to work successfully with this group: a test in estimating the size of pieces of cardboard, a color discrimination test, and a test in sorting feathers (really a sample of the job itself). There is some question whether these are aptitude tests or whether they measure abilities that were in part acquired on the job. Three years after the original investigation, inquiry was made concerning the progress of the employees who had been tested. The investigator states that, "the findings of the tests are vindicated by a fairly close parallel between what the tests predicted and what actually developed in the operation of the factory." The evidence, however, is not unambiguous. 12. Operations in a camera factory.-Some test work at the Eastman Kodak Company1 has been reported, but the results are not given in a manner that makes possible any judgment as to their value. Some 3,000 women workers have been tested, with apparently very satisfactory results. Mechanical tests were used, consisting of boards with holes in them into which small metal pins were to be inserted. Experimental tryout of the tests is said to have demonstrated their usefulness in picking out good and poor operators. 1 A. L. Mann, "Placement of Operators through Tests," Personnel Administra- tion (June, 1922). 126 PSYCHOLOGICAL TESTS IN BUSINESS 13. Factory work in a silk mill.-Several other investigators have tested factory workers with general intelligence tests. The results are not in agreement. In one study1 an intelligence test was used with four hundred workers in a silk mill. No agreement at all was shown between the tests and production records. 14. Factory work in a clothing establishment.-An investiga- tion2 in a large clothing factory, however, obtained very favor- able results from the use of a general intelligence test. Test scores of 290 employees were compared with accurate produc- tion records extending over a four-week period, and fair agree- ment was found. This study is especially valuable and is almost unique in one particular. The investigator scientifically compared the results gotten by means of tests with the results obtained by other available methods in the actual employment situation. This pro- cedure seems the obvious thing to do, but scarcely any studies have done it. In the investigation just referred to it was found that there was no agreement between the size-up of applicants by two experienced interviewers and the later efficiency of these applicants. The tests, then, were considered valuable not because they were nearly perfect but because they were better than any reasonable alternative. 15. Ball-bearing inspecting.-One other study should be mentioned by reason of the great popularity it has enjoyed. It was an early piece of work conducted by Thompson, a scientific management engineer, and reported some years later by F. W. Taylor.3 A simple test was used in selecting girls engaged in inspecting ball bearings. It appears that the test consisted in timing certain elements of the work; that is, it was probably 'A. S. Otis, "The Selection of Mill Workers by Mental Tests," Journal of Applied Psychology, IV (1920), 339-41. 2 H. A. Wembridge, "Experiment and Statistics in the Selection of Employees," Journal of the American Statistical Association, XVIII (1923), 600-606. 3 F. W. Taylor, The Principles of Scientific Management. New York: Harper, 1911. TESTS FOR NON-OFFICE OCCUPATIONS 127 not a standardized test at all. It is stated that thirty-five of the selected workers accomplished the work that 120 girls had previously performed. The nature of the test and the results is very vague and the study probably is of little significance. Other changes were introduced simultaneously with the new selection methods, thus making it impossible to say definitely what part of the gain was due to the test.1 TESTS FOR FACTORY OCCUPATIONS-SUMMARY A consideration of the foregoing studies leads to an apprecia- tion of the possibilities of tests for special manual jobs, and at the same time it emphasizes the relatively slight achievements that have been made. In other words, the status of psychological tests for factory workers can be summed up in the phrase, "a good beginning." Tests of value have been found for a few specific occupations, but even in these cases, with two or three exceptions, the results have not been conclusive. Typically, the results have not been obtained consistently enough or obtained from large enough numbers to warrant any definite judgment that the tests are valuable. But, in almost all instances, the results have been encouraging and have seemed to justify further research. In the field of special aptitude tests for workshop and factory jobs, Link's work, especially with shell inspectors, and Muscio's study of hand-compositors may be pointed to as representative of the best that has been accomplished. These investigations have produced results which appear to be well established. A few other pieces of research have also obtained favorable results though of more doubtful validity. The value of general intelligence tests for factory occupations is not clear. A number of writers have emphasized the impor- tance of eliminating the feeble-minded and very dull applicants 1 For an excellent critical discussion of this case, see the Appendix to Muscio's "Vocational Guidance (A Review of the Literature)," Industrial Fatigue Research Board, Report No. 12 (London). 128 PSYCHOLOGICAL TESTS IN BUSINESS by means of intelligence tests, as well as the possibility of select- ing for promotion individuals of special alertness. Actual investigations demonstrating these uses of tests are lacking. Of the few studies that have been reported, some (Burtt, Burr, Wembridge) have found intelligence test scores of value, while others (Link, Otis, several unpublished studies) have obtained opposite results. These differences may well be due to the fact that the groups tested and the nature of the work were very different in the several cases. It appears reasonable to suppose that in certain situations intelligence tests for factory workers are decidedly worth while. As to trade tests, the possible usefulness is especially clear. But the actual accomplishments of standardized trade tests are almost nil, save, of course, in the army work. The reason for this lies chiefly in the fact that the standardization of tests is difficult and expensive, and unstandardized trade interviews and sample jobs work well enough. A few of the special ability tests that have appeared successful for factory occupations bor- der on being trade tests (for example, some of the tests used by Burr and by Patten). No one seems to doubt that trade tests are practicable and of value. Whether or not there is sufficient need to justify standardizing the tests in a given situa- tion is the question. So far the question has been answered almost everywhere in the negative. A few studies have been reported on the selection of salesmen by means of tests. We shall briefly summarize these results and then see what conclusions can be drawn. i. As far back as 1915, Scott1 published a little material on the selection of salesmen by tests. He reports the use of an intelligence test with ten traveling salesmen of a large tobacco TESTS FOR SALESPEOPLE 1W. D. Scott, "The Scientific Selection of Salesmen," Advertising and Selling (October-December, 1915). These articles are reprinted in Bloomfield's Employ- ment Management, pp. 222-37. TESTS FOR NON-OFFICE OCCUPATIONS 129 company. The early sales records of these men showed no agreement with their test scores, but some months later the salesmen who remained (only five men) stood in the same order in their selling records as they had in the tests. Scott and his associates later tested a number of other groups of salesmen, but the results have not been published. In many of the studies little relation was found between test scores and sales records. 2. Munsterberg1 also did a little experimenting with tests for salespeople. In one instance he compared the test scores of five of the best salesmen and five of the poorest salesmen of a large manufacturing concern. Two tests were found to differen- tiate especially well. Both of the tests involved selecting or rearranging letters to make them spell words or names. Similar tests were given to ten good and ten poor salesmen in a clothing establishment. No agreement was found between test scores and sales ability in this group. In another study tests were used with department-store salespeople-thirty-five efficient sales clerks and thirty-five inefficient ones. Several tests gave favorable results, especially the test involving the rearrangement of letters to form words (probably a test of intelligence). 3. In another investigation2 of department-store salespeople there are reported the results of a large number of tests (principally intelligence tests) given to groups of especially competent and especially incompetent sales clerks. About fifty individuals were tested in the main study. Tests were found which differ- entiated very well between selected groups of salespeople. The tests are shown also to afford a basis for distinguishing between capable salespeople and office clerical workers, again using the selected groups. It is not known how successfully the tests 'H. E. Burtt, "Professor Miinsterberg's Vocational Tests," Journal of Applied Psychology, I (1917), 201-13. 2E. O. Bregman, "Studies in Industrial Psychology," Archives of Psychology, No. 59, New York. Substantially the same material is reported in an earlier article: E. O. Bregman, "A Study in Industrial Psychology-Tests for Special Abilities," Journal of Applied Psychology, N (1921), 127-51. 130 PSYCHOLOGICAL TESTS IN BUSINESS could be used to indicate differences in ability in the ordinary run of sales clerks or applicants for sales positions. It is a point of considerable interest that in almost all the tests the good salespeople made lower scores than the poor salespeople. That is, the agreement between test scores and sales ability was in the reverse direction from what might have been expected. This is especially remarkable since other studies, including an earlier investigation by this same writer,1 found the opposite relation. 4. Another series of test studies2 reports the use of intelligence tests with several groups of shoe salesmen. One group of retail shoe salesmen was given an intelligence test, and the scores were compared with the manager's classification of the salesmen into a "good" and a "mediocre" group. On the average the good salesmen made decidedly higher scores than the poor salesmen. In another study, thirty-seven retail shoe salesmen and saleswomen were tested with an intelligence test. Again marked differences in average score appeared between the "above average" group, the "average" group, and the "below average" group. Similar results were obtained with another group of twenty-nine shoe sales people, but in this in- stance both an intelligence test and an "occupational shoe test" were used. In all these comparisons the groups overlap to a considerable extent, even though the averages are rather far apart. 5. A final study3 to be reported differs from the others in that the tests used were not intelligence tests. A set of tests attempting to get at character traits such as persistence and flexibility was given to two groups of life-insurance salesmen. The salesmen were classified as to selling ability on the basis 1 E. Oschrin, "Vocational Tests for Retail Saleswomen," Journal of Applied Psychology, II (1918), 148-55. 2 Daniel Starch, "The Use and Limitations of Psychological Tests," Harvard Business Review, I (1922), 71-80. 3 M. J. Ream, "A Group Will-Temperament Test," Journal of Educational Psychology, XIII (1922), 7-16. TESTS FOR NON-OFFICE OCCUPATIONS 131 of the insurance each sold over a period of eleven weeks, in the same territory and at the same time. The groups contained forty-seven and seventy-five men, respectively. Marked differ- ences were obtained in the average scores of the "successful," "doubtful," and "unsuccessful" salesmen. There was, of course, considerable overlapping of the groups despite the clear differ- ences between them on the average. _ The investigator concludes that "the tests are of positive value in predicting success in selling life insurance." He states also that "results from these tests are much more significant for sales work than intelligence test results." TESTS FOR SALESPEOPLE-SUMMARY Only a very small dent has been made in the problem of selecting salesmen by means of tests. Intelligence tests have been used in almost all investigations despite the common recogni- tion that selling ability is far different from abstract intelligence. Some studies have, however, reported favorable results from the use of intelligence tests. This is not surprising. The individuals tested had already been selected with some regard to their posses- sion of tact and ability to meet people, and it is to be expected that, these other qualities being equal, the more intelligent man will be the better salesman. In so far, then, as intelligence tests are used as supplements to other methods of choosing salesmen, they may prove useful. Their usefulness must be determined, however, in each specific situation where they are to be used. The last study we reported opens up a promising field for research with tests of character qualities and non-intellectual traits. Later work, following upon that presented, has not entirely agreed with these published results. Many psychologists are of the opinion, however, that tests of selling ability will be developed along these special lines rather than through the use of the better-known tests. The use of interest questionnaires 132 PSYCHOLOGICAL TESTS IN BUSINESS and items concerning the personal history and social life of individuals may prove even more valuable than any tests. The tests, however, are likely to supply useful supplementary informa- tion. Accomplishments at least justify that prediction. Tests have been tried for a number of occupations other than those we have discussed. The more important of these studies we shall summarize even more briefly than the foregoing ones. 1. Two or three test investigations have been made in teleg- raphy. One experimenter1 gave a series of tests to twenty-two boys studying telegraphy in a continuation school. A selected list of six tests was found, the scores on which agreed closely with the ranking of the boys in ability by their instructor. Another investigator2 tried out eight tests on a group of 165 men taking a war-time course in radio-telegraphy. Test scores were compared with records of the speed with which indi- viduals could receive messages. Fair agreement was found, especially for an original "rhythm test." A German investigator3 used a test which was a sample or miniature of actual telegraphic work. No results are published to show the success of the test. 2. Several studies have also been made of telephone-operating. One of Munsterberg's4 early sets of tests was for this occupation. He used a series of eight tests with a group of about thirty opera- tors in training. No statistical results are published, but the tests are stated to have been satisfactory. TESTS FOR SOME MISCELLANEOUS OCCUPATIONS 1E. S. Jones, "The Woolley Test Series Applied to the Detection of Ability in Telegraphy," Journal of Educational Psychology, VIII (1917), 27-34. 2L. L. Thurstone, "Mental Tests for Prospective Telegraphers," Journal of Applied Psychology, III (1919), 110-17. 3O. Lipmann, Die psychische Eignung der Funkentelegraphisten. Leipzig: Barth, 1919. <H. Miinsterberg, Psychology and Industrial Efficiency. Boston: Houghton Mifflin Co., 1913. TESTS FOR NON-OFFICE OCCUPATIONS 133 Another investigator1 tested nine operators with two tests, in one of which the individual tested had to operate a miniature switchboard. The tests proved fairly effective in picking out the good and the poor operators in this small group. The most important work with telephone operators is some research done in Geneva.2 A series of tests was given to twenty- seven operators and test scores were compared with the man- ager's ranking of the girls. A set of eight tests was found which gave results in close agreement with the efficiency ratings. 3. Tests for street-car motormen have been tried out in three or four investigations. Another of Miinsterberg's3 early studies was in this field. He used an ingenious test which it was hoped would represent the same sort of psychological situation as is present in the actual running of a street car. The test was given to a number of motormen-good, medium, and bad ones. The results are said to have shown "a far-reaching correspondence" between test performance and actual efficiency. Statistics are not given. A later investigation4 in Germany made use of a more or less similar test with six women applicants for positions as "motormen." Only three applicants remained at the work. This number is too small to warrant any conclusion concerning the test. t Another German investigator5 tested twenty motormen with a test involving elaborate apparatus-calculated to simulate 'H. C. McComas, "Some Tests for Efficiency in Telephone Operating," Journal of Philosophy, Psychology, and Scientific Method, XI (1914), 293-94. 3 J. Fontegne and E. Solari, "Le Travail de Telephoniste," Archives de Psychol- ogic, LXVI (1918), 81-136. sH. Miinsterberg, Psychology and Industrial Efficiency. 4W. Stern, Uber eine psychologische Eignungsprufung fur Strasscnbahn- fahrerinnen. Leipzig: Barth, 1918. s H. Sachs, Studien zur Eignungsprufung der Strassenbahnfuhrer. Leipzig: Barth, 1920. 134 PSYCHOLOGICAL TESTS IN BUSINESS the complex task of running a car along a busy street. The results indicate that the test is of value though the agreement of test scores and ability as motormen is far from perfect. Another brief report1 states that some psychological tests have been used for selecting motormen in Dallas. The tests are said to have shown good agreement with the men's records. 4.* There are also some reports of intelligence tests applied to policemen. One investigator2 tested thirty candidates for posi- tions as policemen and firemen. No direct evidence is given of the value of the tests. In another study3 several hundred policemen in Detroit were tested with the army intelligence test and one or two other tests. The investigator concludes that he is "not able to find anything strikingly diagnostic in any of the tests." This conclusion was based on a comparison of test scores with the detailed service records of the policemen. Police officers made lower test scores than the patrolmen. Moreover, the men with longer periods of service scored lower than those who had served a shorter time. The interpretation is probably that "the brightest men who enter the police service drop out from that occupation in favor of other occupations where their ability is better recognized." It is pointed out that the same tendency is present in some other occupations and in the police force of another city. 1 P. W. Gerhardt, "Psychological Tests for Workmen," Industrial Management, LI (1916), 605-6. 2L. M. Terman, "Trial of Mental and Pedagogical Tests in a Civil Service Examination of Policemen and Firemen," Journal of Applied Psychology, I (1917), 17-29. 'L. L. Thurstone, "The Intelligence of Policemen," Journal of Personnel Research, I (1922), 64-74. The same investigator has reported a later study of some unstandardized civil-service tests for policemen in Philadelphia. The men who received high test ratings almost all made good records in the Police Training School; those with low test ratings proved to include the good and the poor in about equal numbers. (L. L. Thurstone, "The Civil Service Tests for Patrolmen in Philadelphia," Public Personnel Studies, II [1924], 1-5). TESTS FOR NON-OFFICE OCCUPATIONS 135 5. An intelligence test study1 has been reported in which the scores of a group of restaurant waitresses were compared with the length of service of the individuals. The waitresses who had served four months or less made decidedly better scores on the average than did those with longer periods of service. This is the same sort of relation as that reported of policemen in the preceding study. 6. A set of ingenious tests for journalistic aptitude has been described by one investigator.2 He presents the results obtained with several groups of students of journalism and a few reporters. The group comparisons and the correlations of scores with instructors' ratings indicate that the test is effective. 7. An attempt has been made by a British investigator3 to find tests for dressmaking aptitude. Tests of speed and accuracy in a variety of simple movements were tried, and also some tests for observation, ability to follow directions, judgment of dis- tances, memory of form and color, and so on. The tests were given to dressmaking classes in London trade schools, and the results were compared with estimates of the girls' abilities by their teachers and by their companions. Scores from the com- bined tests agreed well with the ranking of the girls in actual ability, but some doubt concerning the tests is justified by the fact that a second set of test scores showed very poor agreement with the results earlier obtained for the same group of people. 8. During the world-war there was a great deal of psycho- logical study of military occupations, especially of aviation. These investigations can merely be touched upon here. Far more effort has gone into the study of aviation from the point of view of vocational selection than into any other single 1 Daniel Starch, "The Use and Limitations of Psychological Tests," Harvard Business Review, I (1922), 71-80. 2 M. Freyd, "Test Series for Journalistic Aptitude," Journal of Applied Psychol- ogy, v (1921), 45-56. 3 W. Spielman, "Vocational Tests for Dressmakers' Apprentices," Journal of the National Institute of Industrial Psychology, I (1923), 277-82. 136 PSYCHOLOGICAL TESTS IN BUSINESS occupation, military or civil.1 Elaborate physical examinations and painstaking interviewing methods had been used in selecting aviation candidates from the beginning, but the predictions of future success had proved sufficiently unreliable to warrant intensive research on the problem by psychologists as well as by physiologists and medical men. A certain measure of success was attained. General intelligence tests and personal informa- tion concerning the candidate were shown to be useful, and a number of psychophysical tests were evolved to supplement these and to get at the candidate's probable ability to fly and to endure such special conditions as partial asphyxiation. One of the later reports of research2 concludes that: "The tests, as a whole, and some of them singly, are to some extent diagnostic. Their precise value, however, can be known only after trial under more favorable conditions." Careful experimental studies were made of other military and naval tasks, one or two examples of which may briefly be noted. One investigator3 developed valuable methods for XA great number of articles on aviation and the selection of aviators are available in both English and foreign languages. We shall give only a few refer- ences in this field since it is somewhat aside from the main line of vocational tests. Some of the chief American research is represented in the following: G. M. Stratton, H. C. McComas, J. E. Coover, and E. Bagby, "Psychological Tests for Selecting Aviators," Journal of Experimental Psychology, III (1920), 405-23- V. A. C. Henmon, "Air Service Tests of Aptitude for Flying," Journal of Applied Psychology, III (1919), 103-9. K. Dunlap, "Psychological Research in Aviation," Science, XLIX (1919), 94-97- R. M. Yerkes, "Report of the Psychology Committee of the National Research Council," Psychological Review, XXVI (1919), 83-149. Certain prominent features of European investigations are summarized in: F. C. Dockeray and S. Isaacs, " Psychological Research in Aviation in Italy, France, England, and the American Expeditionary Forces," Journal of Comparative Psychology, I (1921), 115-48. a Stratton, McComas, Coover, and Bagby, op. cit. sR. E. Dodge. His work is described in the report by Yerkes, already referred to. TESTS FOR NON-OFFICE OCCUPATIONS 137 selecting gun-pointers in the United States Navy by means of specially constructed apparatus. Tests were also devised for selecting recruits for plotting-room service and for the school for listeners in the navy. Work along somewhat similar lines was conducted in the British navy for selecting candidates for training in hydrophone listening.1 Other investigators2 report some research on tests for men in the look-out and signal service in the navy. 9. Finally, there are a number of test studies that lie along the border-line between business uses of tests and educational uses. Most prominent of these are Seashore's investigations in methods of determining musical ability.3 He has developed an elaborate array of tests aiming to measure the constituent special abilities involved in music. Evidence showing just how valuable these tests are has not been published. Tests have been used to determine the ability of professional engineering students4 with considerable success. Tests were also used in connection with other devices in an investigation5 to determine the differential fitness of engineers for work as salesmen, designers, and executives. 1 C. S. Myers, "Psychology and Industry," British Journal of Psychology, X (1920), 177-82. aE. F. Ferree and G. Rand, "A Study of Ocular Functions with Special Reference to the Look-Out and Signal Service of the Navy," Psychological Bulletin, XVII (1920), 77-78. 3 C. E. Seashore, The Psychology of Musical Talent. Boston: Silver Burdett & Co., 1919. * Experiments of E. L. Thorndike reported in C. R. Mann, "A Study of Engineering Education," Bulletin No. 11, Carnegie Foundation for the Advancement of Teaching, New York, 1918; L. L. Thurstone, "Intelligence Tests for Engineering Students," Engineering Education, XIII (1923), 263-318; H. E. Burtt and F. W. Ives, "Vocational Tests for Agricultural Engineers," Journal of Applied Psy- chology, VII (1923). 178-87. 3 B. V. Moore, Personnel Selection of Graduate Engineers. Published at Carnegie Institute of Technology, Pittsburgh. 138 PSYCHOLOGICAL TESTS IN BUSINESS Some German investigators1 have also experimented with the use of tests among students in vocational schools, but no results are given to prove the value of the analyses made or of the tests used. A number of other test studies along similar lines have been made, but they would take us too far from the main line of our discussion. We shall leave the reports of particular investi- gations and proceed to an attempt to draw some conclusions from all the material presented. ACCOMPLISHMENTS OF PSYCHOLOGICAL TESTS FOR BUSINESS-SOME CONCLUSIONS We should be able now to reach a general statement as to the accomplishments of psychological tests for the different occupational groups. While any generalization must be pre- sented with some misgivings, we shall venture to summarize in a few short paragraphs the material we have been reviewing. With our conclusions we shall combine some warnings. Above everything else is to be emphasized the necessity for reaching a judgment concerning tests which recognizes both their possibilities and their shortcomings. It is easy to condemn or approve; it is difficult to suspend judgment. We are prone to say of new methods, such as tests, that they are good or that they are bad. Usually they are neither. We must avoid extreme and unqualified approval of tests just as we must avoid ungrounded criticism and disapproval. Psychological tests for business occupations have done something; but they have done far from everything. In this connection a second point is to be kept clearly in mind. Psychological tests must be judged in terms of the prac- tical problem to be solved and the actual alternatives available 1N. Braunshausen, Psychologische Personalbogen als Hilfsmittel filr Pddogogik und Berufsberatung. Leipzig: Barth, igig. O. Lipmann and O. Stolzenberg, Methoden sur Auslese hochwertiger Facharbeiter der Meiallindustrie. Leipzig: Barth, 1920. TESTS FOR NON-OFFICE OCCUPATIONS 139 rather than by comparison with a standard of perfection. Even the best of tests are a long way from the roo-per cent mark in their results. Can more be said for any other method of selecting men? The proper question is never, Do tests enable me to judge men for jobs with unerring accuracy ? Of course they do not. The real practical question is simply, Do tests aid me in arriving at appreciably better judgments than I can reach without them? The answer to this question is frequently a definite affirmative. Not, perhaps, that the tests are so good; rather that the ordinary methods are so poor. Definite comparisons of the results gotten by test methods with those obtained from non-test methods are rare. Research of this kind is especially needed. Only by knowing how well present methods of selection operate can we form sound judg- ments as to the usefulness of tests as a new method. In the absence of direct comparisons of this sort, the best we can do is to point to the serious mistakes and inaccuracies of the usual selection process and then to infer that wherever tests show even fairly good results they are a valuable addition to employment methods. In the case of office occupations, the chief accomplishments have been: (i) the development of successful tests for clerical workers, especially general intelligence tests and tests of achieve- ment in spelling, arithmetic, etc.; (2) the development of successful proficiency tests (trade tests) for typists and stenogra- phers; (3) promising pioneer work on aptitude tests for typing and stenography and for computing machine operating; and (4) collection of evidence as to the value of intelligence tests for indicating the alertness and future possibilities of various classes of office workers. Psychological tests for factory occupations have made less progress. Some good beginnings have been made, however, prominent among which are the following: (1) the army con- struction of trade tests for a large number of occupations; 140 PSYCHOLOGICAL TESTS IN BUSINESS (2) the tentative conclusion that intelligence tests are useful for selecting within some groups of factory workers but are worthless with other groups; (3) the development of a few successful aptitude tests for particular occupations (for example, Link's work and Muscio's); and (4) promising attempts at the construc- tion of aptitude tests for a number of other occupations. Little in the way of positive progress has been reported in the test selection of salespeople. The research that has been done shows that (1) intelligence tests sometimes are useful and some- times are not, and (2) results of possible value may be obtained through further investigation with tests of non-intellectual characteristics and interest questionnaires. Intelligence tests are probably valuable wherever other qualities-social and temperamental-can be assumed to be satisfactory. Other things being equal, greater intelligence is an asset for salesmanship. In several other occupations test research has made consider- able progress. Thus: (1) good pioneer work has been reported in special aptitude tests for telegraphers, telephone operators, street-car motormen, journalists, aviators and other military occupations, musicians, and others; (2) intelligence tests have been used with some measure of success among policemen, waitresses, professional-engineering students, and a number of other groups more or less definitely vocational; and (3) army trade tests have been constructed for a variety of occupations. Our survey of test accomplishments would be incomplete without mention once more of the method and technique that have been developed. All the specific test achievements are sec- ondary to the scientific procedure for test construction and for testing the tests, which have grown out of the efforts of applied psychologists in this field. CHAPTER V THE PLACE OF TESTS IN THE PERSONNEL PROGRAM In this and the final chapter we shall consider several aspects of psychological tests in business which have been mentioned in the preceding pages, but which have not been developed. In particular, the following matters deserve further discussion: (i) problems of test administration and the place of tests in the actual employment procedure; (2) uses of tests for other purposes than hiring employees; and (3) the future of tests in business, and the conditions necessary for progress in industrial test work. Up to this point the major emphasis has been placed on test research. We have described the methods and the chief accom- plishments in the research development of employment tests. But there are many problems on the administrative side of test work in business, as well as on the side of research. We shall suggest some of the technical problems that arise in the routine use of tests, and shall also indicate how tests can be fitted into the actual administrative procedure of the employment department. Emphasis has been placed also, up to this point, on the use of tests for hiring and placement. It has been suggested in passing that this is not their only possible use, but it is the use to which they have principally been put. In addition, however, tests are useful aids in making readjustments in the working force, and in connection with training programs, personnel surveys, investi- gations into the causes of labor turnover, vocational guidance procedures, and so on. Illustrations of these applications of tests will be given in this chapter. 141 142 PSYCHOLOGICAL TESTS IN BUSINESS Finally, in the concluding chapter, we shall point out certain of the broader relationships of the testing movement, particularly with reference to the outlook for the future development of vocational testing. Problems connected with the use of psychological tests in business fall into two distinct classes-those of test research and those of test administration. Most of what has been said in this book deals with test research. This is true partly because tests are still so young and so definitely in the experimental stage. In part, too, it is a recognition of the fact that the prob- lems of test administration are relatively much simpler and less technical than those of research. Test research includes such things as studying the particular requirements of the various occupations for which tests are desired, searching out or constructing tests which seem promis- ing, standardizing the technique of these, trying them out on workers of known efficiency, ascertaining the relative value of each test and combination of tests, deciding on those that are to be used, determining critical scores or other standards to be used in applying them to individuals, and, in general, settling the first and the basic problem-that of preparing a selective instrument of tried and proved value. Then, and only then, does the problem become one of test administration. Perhaps the time will come when occupational descriptions will be so well standardized, conditions of work so uniformly regulated and understood, and tests standardized for such a variety of tasks, that employment officers can refer to an index and determine at once what test methods have been found to be appropriate for each kind of work. But that time is far in the future. Indeed, so long as occupations and industries multiply, new or improved machines and methods are intro- duced, and work continues to become more and more specialized, TEST RESEARCH AND TEST ADMINISTRATION TESTS IN THE PERSONNEL PROGRAM 143 there is no hope for a ready-made standardization of selection methods for every kind of occupation. Hence, test research may be expected to continue to be an important and a growing phase of scientifically managed industry, and in practice insepar- able from test administration. There is to be expected a constantly growing demand for competent, trained research investigators, even after testing has reached the point where the actual giving of tests to applicants for employment can be turned over to routine workers. But even when testing has reached the administrative stage, we should not make the mistake of thinking of tests as being standardized once for all. Research on test results is likely to be even more valuable after the tests have been used in inducting hundreds of workers into a business, and after their success in selecting the suitable workers has been demonstrated, than was the research done in originally introducing the test program. PROBLEMS OF TEST ADMINISTRATION Test administration is at least on the surface a relatively simple task. Once a procedure has been worked out, adequate for the different jobs and occupations represented in the establish- ment, the giving and scoring of tests and the recording of test results of applicants for employment is not especially compli- cated. Nevertheless, test administration offers problems which by no means solve themselves. i. Provision must be made for suitable quarters and appara- tus, and for getting and maintaining those physical and mental conditions which enable the applicant to do himself justice. One example of precaution in this direction is the practice of starting each test with a "shock-absorber"-some easy and amusing problem, such as a picture puzzle-to instil self-confidence in the applicant and awaken his interest. 2. A routine procedure must be established and a daily sched- ule has to be prepared in order that tests can be given in the most 144 PSYCHOLOGICAL TESTS IN BUSINESS economical and effective way, and in order that wherever possible tests shall be given to groups of applicants simultaneously, to minimize expenditure of time and effort. 3. A thousand and one details of technique of test-giving and scoring, connected with timing, arrangement of materials, verbal instructions, encouragement or explanation properly to be given or withheld, economical methods of scoring large numbers of tests, have to be mastered and kept unchanged, to avoid the many possibilities of error which only training enables one to foresee. 4. Where different occupations in an industry require a variety of groupings and weightings of the many tests used, the formulas which have been worked out to achieve this end have to be applied and adequate records made. 5. The making and keeping of records of test results and of other facts that are considered in employee selection, in such form that they shall be readily accessible and easily interpretable, is an important and often a difficult problem. Unless records are of such form and nature as to facilitate future reference, either for checking up individual success or for statistical tabulation and analysis, half their value is lost. 6. Test scores of individual applicants have to be interpreted in their relation to other personal data entered on the application blank or revealed in the interview or references, appropriate decisions made, and action taken. 7. Test materials and supplies have to be kept in order and prevented from getting into unauthorized hands. It seems inevitable that much more attention shall be given to this detail as tests become more widely used and prospective applicants learn to prepare in advance for the tests. Every teacher makes it a point to keep secret his prospective examination questions, sometimes at the cost of elaborate care. The same is conspicu- ously true of state examinations for admission to practice law or medicine. But practically no concerted effort has been made TESTS IN THE PERSONNEL PROGRAM 145 or even suggested among psychologists or test-publishers to keep psychological tests from falling into unauthorized hands. Already complaints are heard that the value of the army Alpha test has been lessened by too much publicity. 8. Many problems of test administration arise in individual cases, where there is a question of what tests to use, or of the interpretation of the scores. Even in cases where the employ- ment machinery is in smooth running order, the routine of testing applicants and "passing" or "failing" them is not wholly mechanical. The procedure varies from case to case with cir- cumstances. Sometimes the candidate applies for a specific job. More commonly he applies for "some" job, with certain very general restrictions, such as that it shall be office work or machine work. In the former case, he may be a skilled worker claiming certain proficiencies. Here proficiency tests, such as trade tests, enable the employer to determine whether he actually possesses the requisite degree of knowledge and skill. But if he is an unskilled worker, or applies for work of a type which has to be learned on the job, aptitude tests of various sorts permit the employer to determine whether or not he has the traits, physical, motor, sensory, intellectual, or temperamental, which previous investigation has shown to be necessary to insure success in that particular work. In times of labor shortage the procedure will be to eliminate only those definitely below the minimum critical score, the predetermined lowest level for prob- able success. If, on the other hand, the supply of applicants exceeds the need, the one or more rating highest will be selected for immediate employment and the others perhaps asked to leave their applications on file for possible future use. In view of all these problems, it is not to be expected that even the routine work of testing applicants for employment can safely be done without at least the supervision of trained special- ists. Most tests require at least a little special training and experience on the tester's part before he can administer them 146 PSYCHOLOGICAL TESTS IN BUSINESS easily and without liability to error. When an employment program involves a great variety of tests, this requirement is magnified in importance. One present tendency evident among test-makers is to make administration simpler and easier, as, for example, in the " self-administering " tests of Otis and Thur- stone. But even at its simplest, test administration is not some- thing which just any clerk can pick up and carry on without special instruction and supervision. When to the work of test administration we add the inseparable problems of test inter- pretation, test selection, and test standardization, it is evident that we have not only a new tool but a new technology-or profession. TESTS AND THE GENERAL EMPLOYMENT PROCEDURE It would be a great mistake to think of the administration of tests, or even of test research, as a particular kind of personnel work which stands apart from the other activities that confront the worker in this broad field. Neither in their technique, nor in their relations to the general employment function, are psy- chological tests wholly unique. Still less are they to be thought of as constituting the substance of the work of employment management. A course in psychological testing in a university-or a succes- sion of courses-will not alone equip a man or woman to become a competent employment specialist. While it is evident, from what has been said, that adequate preparation for genuinely scientific personnel research necessitates rigorous, systematic training in research methods, this kind of training is the ideal rather than the prevailing practice. Many more employment officers have come to their present position by way of some other department or position in the organization they serve than have come by way of specialized, technical training courses for employ- ment work. While such actual business experience alone is not an ideal preparation for scientific personnel work, it does at any TESTS IN THE PERSONNEL PROGRAM 147 rate furnish an indispensable element in such preparation-a first-hand knowledge of business in general and of that business in particular. And it does give one a perspective of the place of tests in the employment program that the university classroom or laboratory does not give. One of the things the employment manager soon learns is that the use of tests or other scientific devices in selecting workers is only one phase-and perhaps in practice a very minor phase- of his everyday work. He learns that he must have constant recourse to a great many other methods-some good, some poor-for helping to form judgments on the fitness of applicants and placing them wisely. He learns that his ideal programs are being constantly shattered by conditions of the labor market and by demands from other departments of his organization. He finds that a thousand and one details that fall-or are dumped by other departments-within his jurisdiction keep him from cultivating the leisurely, detached, scientific attitude which one associates with the term "research." These problems have to be faced and worked into his daily program. If he can use tests and test technique to effect a more satisfactory solution for any of them, so much the better. But he quickly learns to value the tests only as means to his ends, and not as ends in themselves. STEPS IN THE EMPLOYMENT PROCESS In a large establishment where the employment functions have been centralized in a single department, the steps in the procedure of hiring and placing employees are fairly definite and uniform. Assuming that jobs have been analyzed and job specifications (or some other form of occupational descriptions) have been made available for the use of the employment office, let us outline the steps in the employment sequence in order to see just where and how tests fit into the process. i. Contacts are first made with potential sources of labor supply. One important source is the present working force. 148 PSYCHOLOGICAL TESTS IN BUSINESS When a vacancy is to be filled, the first thought of the employ- ment manager should be, Is this an opportunity to promote or transfer some deserving present employee? Ordinarily psy- chological tests are less important in selecting from within the business than from without, but they have real possibilities of giving supplementary information about even well-known workers. The outside sources of labor supply are numerous and varied. Applicants may be secured by means of advertising; labor scouts; personal introduction by employees, customers, or stockholders; the influence of the firm's presence and reputa- tion in the community; and contacts with outside organizations such as firms, fraternal and social bodies, churches, public and private schools, business schools, high schools and colleges. All these may afford a possible, though ordinarily a very hetero- geneous, source of supply. Little preliminary weeding-out of undesirable material is possible, except at the point where, as sometimes happens, co-operative relations are established with certain schools which have facilities for selecting from among their students those whom they recommend. 2. Preliminary classification of applicants may be made at the door of the plant into such groups as office workers, sales- men, unskilled labor, and skilled tradesmen, in order to refer them, for greater convenience, to different clerks, desks, or rooms in the employment office. 3. Applications are received. Usually at this point the applicant fills out an application blank. While application blanks vary widely from one establishment to another and one kind of business to another, they ordinarily seek to elicit such information as seems likely to be of importance, either in inter- viewing, hiring, and placing the worker, or in maintaining future relations with him. Personal and family history, facts about race, personal habits, religion, education, health, work prefer- ences, experience, interests, reasons for applying, and almost anything else may be asked about. Probably the majority of TESTS IN THE PERSONNEL PROGRAM 149 application blanks ask questions which have no value whatever for ascertaining the applicant's fitness; and probably as many fail to call for certain information which might prove very sig- nificant. There is almost as great a need for a thoroughgoing, scientific investigation of the values of each of the different kinds of personal data as there is for investigating the value of various selective tests. 4. Inquiries are made from references named by the candi- date. These are either former employers or others who can speak of his character, training, and ability. Employers place no reliance on general letters of recommendation; but confidential statements from references, while usually considered an indis- pensable part of the procedure, are, like application blanks, also in need of more scientific standardization. 5. The applicant is interviewed. With the application blank in hand, the interviewer endeavors by questioning the applicant, and observing his appearance and behavior, to ascertain the type of work he desires and his probable fitness for that kind of work. If, as is frequently the case, the applicant has no clear idea of the kind of position he wants, a considerable part of the interview may consist in the interviewer's giving him information about different kinds of work to help him decide intelligently. As with application blanks and inquiry letters, there is a need for better standardization of the interviewing methods. There can be no doubt that many interviewers place reliance on home-made or ready-made systems of 11 character reading" which are utterly without scientific value. Yet this part of the employ- ment procedure is one of the most crucial. Hollingworth1 and Scott and Clothier2 show by striking illustrations how completely at variance with one another may be the judgments on the fitness of applicants which are made by professional interviewers, who 1H. L. Hollingworth, Judging Human Character, chap. v. D. Appleton & Co., 1922. 2 Scott and Clothier, Personnel Management, chap. ii. A. W. Shaw Co., 1923. 150 PSYCHOLOGICAL TESTS IN BUSINESS differ from one another in the sort of questions they ask, the phrasing of those questions, the weight they attach to different sorts of reply, and the details of behavior from which they estimate the applicant's habits and promise. Some of the most important facts which interviews and inquiries from references seek to elicit, it should be kept in mind, are not ability factors, but what may be called "character" and "personality" factors. Ability to do work is, after all, only one of the conditions which determine a worker's success in a position. His moral habits, his perseverance, his personal agree- ableness, appearance, voice, carriage, manner, tactfulness, the strength of his ambition, his interest in his work-any of these may affect adversely the efficiency of a worker who possesses the ability to do a piece of work; or they may go far toward com- pensating for an inferior degree of ability. No tests are success- ful as yet in getting at these traits. Hence, the utilization of any and all means by which these characteristics may be determined in advance, and the scientific determination of the actual value of such means, constitute indispensable parts of the personnel program. 6. In order to verify the applicant's statements about his qualifications, to determine with greater accuracy the extent of those qualifications, or to discover them if unknown, the next step logically is to test the applicant. This will ordinarily be done by a special clerk, trained in the technique of giving and scoring tests and recording results. What has been said about the diversified problems of test administration applies primarily to the work of this individual. The decision as to the hiring or rejection of the applicant, and the position to which he shall be assigned if employed, is, however, ordinarily referred to the interviewer or some other responsible individual, in whose hands the test scores and all other significant data are brought together. TESTS IN THE PERSONNEL PROGRAM 151 7. At this point come any remaining details of the general- employment procedure, such as reaching an agreement about compensation, physical examination, provision for giving surety bond if one is required, giving the new employee general informa- tion, instructions, or rules, and introducing him to his new foreman or manager. 8. Induction into the job is accomplished by a variety of methods, some systematic, some very haphazard. Often the new employee serves a "probationary period," and the final decision as to his retention is in theory deferred until the expira- tion of that period, unless he proves himself unfit before that time. Not infrequently this preliminary period is supplemented by training courses designed to fit the employee for advancement to more responsible positions. Of the possible uses of tests in such training programs, more will be said later. 9. Following up the worker in his work is an indispensable part of sound personnel procedure. This is important not only in order to ascertain whether the worker's value is increasing or decreasing, but to determine the actual success of the employ- ment procedure by means of which he was originally selected and placed. Follow-up methods are of many kinds. They include the maintenance and utilization of production records, measure- ment by standardized tests, interviews, supervisory methods, records of errors, of demerits, of attendance, of performance in training courses, and of other significant items, and periodical efficiency ratings. In some establishments, even where workers are carefully selected and placed, it is the usual thing for them then to be lost sight of, unless through some unusually good or unusually bad performance they force attention to themselves. Consequently, many good employees' abilities are wasted, or their services are lost to the concern, because they are not given opportunities for development commensurate with their abilities. Measurement of ability, whether by tests or otherwise, after 152 PSYCHOLOGICAL TESTS IN BUSINESS employment is hardly less important than measurement at the time of hiring. TESTS FOR PURPOSES OTHER THAN EMPLOYMENT We turn now to the second of the general considerations men- tioned at the beginning of this chapter-that of the use of tests for other purposes than the original selection of workers. It is this use-hiring and placement-to which most attention has been given, and which has, in actual practice, been much the most conspicuous. It is at this initial point that the need of tests is most clearly apparent, because here the question of personal suitability is uppermost, and there is least opportunity to use such other methods as will become available only after the candi- date is actually in the firm's employ. Some other directions in which tests are actually or potentially useful are, however, well worth considering. In the first place, psychological tests have a possible value in effecting transfers, promotions, and lay-offs, and in selecting employees for special duties. This is usually done on a basis of informal estimates or systematic ratings of workers, but the difficulty of getting reliable estimates for these and other pur- poses has led some personnel men to look to the more objective tests as a possible aid. In the training program, also, tests can be used to advantage. The measurement of aptitude for different types of training, the selection of individuals who are competent to advance more rapidly than others in training courses, the measurement of individual progress in training-these are tasks in which psychological tests give useful assistance. Likewise, the analysis of labor turnover has on several occasions been facili- tated and illuminated by studies of the relation of test scores to the duration of employment of workers in particular occupations. Work and fatigue studies, including investigations of the effect of such variable conditions as the relative distribution of work and rest periods, of hours of labor, and the like, have frequently TESTS IN THE PERSONNEL PROGRAM 153 utilized standardized tests as convenient measures of working efficiency under varying conditions. The comparisons of differ- ent work groups, based on sex, race, amount of schooling, age, and other distinguishing characteristics, or comparisons of groups of workers from different shops or departments, or of the workers in a given shop or department from year to year, are aided by studies of test scores. Finally, vocational guidance offers a field of great potential usefulness for tests as they increase in number and as our understanding of them becomes more perfect. Let us consider a few of these uses in a little further detail. TESTS IN TRAINING PROGRAMS Business establishments have developed a wide variety of methods for acquainting new employees with the details of the work to which they have been assigned, and of guiding them until they are able to work without constant supervision. Apprentice- ship systems, understudy methods, breaking-in by experienced employees, and vestibule schools or company schools of many kinds are designed to meet this need. Closely related to these introductory training courses are the various types of educational program, general or vocational, intramural or extramural, which most large and progressive business houses foster for the purpose of fitting their employees for promotion to more advanced work, or for the purpose of making good the educational deficiencies of those employees who have left school early. In both these classes of situations tests have made themselves useful. Preliminary tests (including general intelligence tests and other kinds of aptitude tests) are useful for selecting those who can profit by certain types of training and are an aid in placing them in courses best fitted to their capacities. Scott and Clothier1 report that "mental alertness" or general intelligence tests have been found to be better than amount of schooling as a basis for classifying employees in a factory training course, into 1 Personnel Management, chap. xxi. 154 PSYCHOLOGICAL TESTS IN BUSINESS sections based on ability to progress with the course. In schools and colleges this classification of students into sections or groups for the purpose of facilitating instructional processes is one of the most important uses to which tests have been put. One who is interested in this problem will find a wealth of educational literature in books, periodicals, and monographs dealing with this and related practices. Proficiency tests of various sorts, both educational tests and trade tests, are useful for measuring the progress made by the learner, and they help also to determine the relative efficacy of various training methods. Toops, in his Trade Tests in Educa- tion, points out three types of use of such tests: (i) to test the learner's progress and maintain his interest in trade instruc- tion; (2) to enable the educational director to make proper allowances for individual differences in the rate of acquiring a trade proficiency; (3) to aid in educational and vocational guidance and school-placement work. TESTS AND THE ANALYSIS OF LABOR TURNOVER One of the most studied problems of personnel maintenance is that of labor turnover, not only because of the enormous expense which a high percentage of labor mobility involves, but because of the difficulty in discovering its causes, and thus intro- ducing preventive measures. While the causes for the discharge of unsatisfactory employees are not uncommonly a matter of record, the reasons stated are usually superficial, and do not get at the root of the difficulty. Still more difficult is it to get at the causes of the voluntary separations, for even in the none-too- frequent instances where inquiry can be made, either because of the employee's unwillingness or his inability to state the real reasons, the answers are likely to be worth little. Studies made by The Scott Company1 in several firms have shown that "mental alertness" (i.e., general intelligence) is a not insignificant factor 1 Scott and Clothier, op. cit., chap, xxvi; A. J. Snow, "Turnover and the Mental Alertness Tests," Journal of Applied Psychology, VII (1923), 285-90. TESTS IN THE PERSONNEL PROGRAM 155 in this connection. Both test scores and age-grade status at time of leaving school have been used by them as indexes of relative brightness. Frequency of resignations, expression of desire to leave, and length of service have been used as indexes of stability. Characteristically, definite relationships have been found between the degree of mental ability and the stability of workers in various kinds of occupation. In one sort of work, repetitive and mo- notonous in nature, the curve which indicates instability is low for the less intelligent groups and rises rapidly as the degree of intelligence increases. Here the need of determining and using a "maximum critical score" in assigning workers to this type of work is suggested. In some other position the reverse relation- ship may be true. The less intelligent workers find the work too hard, and the more intelligent remain. In one case, stability was found high among women workers of low intelligence, falling rapidly as intelligence grades advanced, but rising again with the highest grades of intelligence. In this case, a more active promotional policy affecting the middle grade of workers was recommended, in order to prevent their leaving in too great numbers. In another group of clerks the greatest instability was found at two points: among the least intelligent and again among those of moderately high intelligence. An unsatisfactory wage condition was found to be the cause of the undesirable loss among a good grade of workers. Thus, while test scores may not in themselves tell directly the cause of the turnover, they may help to reveal the presence of certain causes of dissatisfac- tion, perhaps not clearly recognized by the workers. These and a great variety of other related problems point to the need of reliable test scores for determining general employ- ment policies, as well as in hiring individual workers. In most large business establishments it is a safe presumption that com- plete records of employees' test scores, as well as of their pro- ductive efficiency, special interests and abilities, and the like will prove of great importance both for research and for admin- istrative purposes. 156 PSYCHOLOGICAL TESTS IN BUSINESS TESTS IN VOCATIONAL GUIDANCE Closely related to the problems we have been discussing, and particularly to those of vocational selection and vocational training, is the matter of vocational guidance. While the pri- mary responsibility for advising youth as to the choice of a voca- tion does not often fall directly on the business man, the status and prospect of vocational guidance in community and nation are matters of direct concern to him. His future labor problems will be made easier, and his relations with the schools of his community to which he looks to supply him with youthful recruits will be made more satisfactory, if the rising generation enjoys competent vocational counsel and training, based on a thorough and accurate understanding of the needs of society, the requirements of various industries, and the potentialities of each boy and girl. Moreover, as has been pointed out, it often becomes necessary, in actual practice, for the one in charge of hiring to give voca- tional counsel to applicants who do not themselves know just what kind of work they want or are fitted to do. Here the employment officer's problem is much the same as the vocational counselor's-to give the applicant vocational information, to help him discover his abilities and interests, and to guide him into that position in which he gives greatest promise of progress and success. Professional vocational counselors, whether connected or not with schools, have hitherto made much less use of psychological tests than it seems likely they will in the future. They have given major attention to other phases of the problem. In part this is because of the lack of sufficient accurate information about the requirements of different occupations; in part it is because we have almost no tests which have proved themselves useful for vocational guidance. More use has been made of general intelligence tests than any other kind. They seem to afford a helpful guide in narrow- TESTS IN THE PERSONNEL PROGRAM 157 ing the range of occupations within which the youth should make his choice. It is safe to predict that a boy whose general intelligence ranking is in the lower part of his high-school class should not ordinarily look forward to a professional career. One who stands still lower should limit his range of choice still farther. But in any case, we have not reached the point where we can say to him, "You should choose this, and only 'this, career." Probably we never shall. There are too many other determining factors which psychological tests do not as yet enable us to measure. Some of these may ultimately be measur- able by tests; others will not. Assuming an adequate minimum of ability, one's vocational interest-provided it be an intelligently informed interest and not merely a transitory whim-is probably one of the most useful vocational guides; for interest and ability are ordinarily closely related. Some work has been done with interest tests and questionnaires, but their technique is still in the experimental state. One present vocational need is to assemble and make accessible more full, specific, and authoritative vocational infor- mation in order that interest may be intelligently formed. Other factors which must help determine one's choice include physical suitability, financial ability to acquire the necessary training, vocational opportunities or lack of opportunities in the com- munity, economic and other rewards. Inasmuch as vocational guidance and vocational training are inseparable phases of the problem of vocational suitability, we may expect that progress in standardizing trade tests will provide vocational schools and courses with another instrument which will help to select those of superior trade fitness, and thus to advise and assist them wisely. Test experimentation and the extension of job analyses are also helping us, little by little, in the same general direction. Tests of temperamental and voli- tional traits may sometime help still further to determine voca- tional fitness, as in all probability will many types of aptitude test. 158 PSYCHOLOGICAL TESTS IN BUSINESS But we are still far from having test information of wide applicability for vocational guidance purposes. We must expect the problem of guidance tests to be solved more slowly than that of employment tests, for there are greater inherent, difficulties. Much of our needed knowledge for vocational guidance will come, however, from experimentation in employment psychology in business organizations. But this knowledge will have to be based on extensive test work in many business establishments before we shall be able to say with assurance that certain tests do indicate, in a useful way, superior vocational ability. One or two of the special difficulties of vocational guidance tests may be indicated by contrasting them with tests for selec- tion purposes. In the first place, most successful employment tests are suitable primarily for rather specialized occupations and within a somewhat restricted range. Successful guidance tests would have to discover more general trade abilities, and ought also to predict probable success in the higher and less definite levels of occupations. Selection methods begin with a specific set of requirements to be satisfied. The results of selec- tions can be checked, and in that way poor methods are elimi- nated and good methods can be demonstrated to work effectively. In guidance, on the other hand, the individual is taken with all his complexity of interests and abilities, and an attempt is made to find the best line of work for him among all the thousands of occupations. The results can rarely be checked, for there is no way of knowing certainly whether or not the individual would have succeeded better in some other direction. In the face of these and other difficulties, it is not surprising that little has been accomplished in the development of special vocational guidance tests. OCCUPATIONAL INTELLIGENCE LEVELS But test scores are not wholly without significance for voca- tional guidance. The accompanying chart (p. 160), taken from Volume XV of the Memoirs of the National Academy of Sciences, TESTS IN THE PERSONNEL PROGRAM 159 entitled "Psychological Testing in the United States Army," is suggestive not only for vocational guidance, but also for the employer of labor. This chart shows the variation between the average intel- ligence levels revealed by representatives of many different occupations among the drafted men in the United States Army of 1917-18. The lowest degree of intelligence was found among the unskilled laborers and the trades requiring least skill and training. Above these are the semi-skilled trades. Next come the skilled trades and the office and commercial occupations, demanding, as a rule, either more general education or more prolonged specific training or both, than the semi-skilled and unskilled occupations. Highest of all are the semi-professional and professional occupations, requiring more of a general educa- tional background, more exercise of independent judgment, more ready adaptability to changing situations. The chart shows not only the average (median) intelligence level for each occupation listed, but also, by means of the hori- zontal line, the range of the middle half of the men in that occupa- tion. That is, the least intelligent one-quarter and the most intelligent one-quarter are omitted, giving us a fairly clear picture of the average intelligence level of several representative occupations. These represent, of course, only a few of the many occupations. Further testing will probably extend the list. This has already been undertaken by certain workers.1 1 Douglas Fryer, "Occupational Intelligence Standards," School and Society, XVI (1922), 276. G. A. Feingold, "The Relation between the Intelligence and Vocational Choices of High School Pupils," Journal of Applied Psychology, VII (1923), 143-53. See also discussions of this point by the following: H. C. Link, " Psychological Tests in Industry," Annals of the American Academy (November, 1923), pp. 32-44. Scott and Hayes, Science and Common Sense in Working with Men. Daniel Starch, "The Use and Limitations of Psychological Tests," Harvard Business Review, I (1922), 71-80. 160 PSYCHOLOGICAL TESTS IN BUSINESS Presumably the grades found to characterize the different occupations can be taken as representing the degree of intelligence required for reasonable success in these occupations. The data Laborer . Cen. miner Teamster . Barber . . C- Horseshoer ... -. Bricklayer . . - Cook ...... Baker ....... Painter Gen. blacksmith . . Gen. carpenter . . . Butcher . Gen. machinist . •• . Hand riveter . . . . Tel. & tel.,lineman Gen.pipefitter . . . Plumber ...... Tool and gauge make] Gunsmith .....' Gen. mechanio • . • Gen. auto repairman Auto engine mechanii Auto assembler . . Ship carpenter . . . Telephone operator , C Concrete const, foreman Stock-keeper Photographer Telegrapher ...... R.R. clerk . Filing clerk. ..... Gen. clerk Army nurse ...... Bookkeeper C+ Dental officer . . . Mechanical, draftsman Accountant . ... . Civil engineer . . . Medical officer • • B A Engineer officer TESTS IN THE PERSONNEL PROGRAM 161 cited give of course no indication of these workers' vocational success, other than that they reported these as being their previous occupations. But another equally important fact is revealed by the chart, that of the very great degree of overlapping between different occupations. For most of the skilled trades the average intelligence ratings, as well as the middle half of the scores, are almost identical. An individual's test score tells nothing about which of these trades offers him the greatest likelihood of success. Whether his particular aptitudes indicate that his best chances would be as a painter, a carpenter, a plumber, or a mechanic must be determined wholly by other means, such as special aptitude tests, studies of his specific interests, his training, his experience, his ambitions, and other factors not measured by intelligence tests. The most his test score can tell us is whether or not he has the necessary minimum of intelligence, and whether or not he has more than can profitably be utilized in these trades. Between occupations varying considerably in intelligence averages there is still perceptible overlapping, as, for example, between two such diverse occupations as those of dental officers and bakers or painters. Without more information than the chart gives us about the relative occupational success of the least intelligent one-third of the dental officers and of the most intel- ligent one-third of the bakers, we are not in a position to predict, on the basis of a test score which falls within this overlapping range, one's relative prospect of success in these two occupations. Therefore, while the chart affords definite evidence that intel- ligence is one important factor in the hierarchy of occupations, from unskilled labor up to the professions, it tells little or nothing more than that. Successful guidance demands the use of measures of other and more specific traits. There is one other consideration which it is important to bear in mind in interpreting this chart of occupational intelligence levels. It may be that the intelligence test scores made by the 162 PSYCHOLOGICAL TESTS IN BUSINESS people in any particular occupation are in part a result of the specially favorable or unfavorable influences of that occupation in preparing individuals for the kind of performance called for in intelligence tests. The scores, too, may be influenced by the amount and kind of education required for entrance into the different occupations. Clerks, for example, have much higher test scores than plumbers or tool makers. The ordinary inter- pretation of this is that the one group has more inherent intel- lectual ability than the other. But the difference may be due, in part, to the advantage the clerks have in practice with paper and pencil performances and with the sort of semi-clerical tasks included in the intelligence tests. It is also possible that they have had more formal schooling of a sort that would influence test scores. Similar possibilities obtain with respect to. many others of the occupations compared. How large a part these influences play in producing the results shown in the chart, it is difficult to say. They do at least suggest the need for caution in making interpretations. It may be, for example, that if all the plumbers and all the clerks reported in the test results had been tested when they were twelve years old, there would have been no difference between the groups. The difference may have come about wholly through the later influence of unlike training and the development of unlike interests. In that case, the chart would have no usefulness for vocational guidance purposes. In all probability this inter- pretation in such extreme form is not true. But there remains a question whether, in less extreme form, it may not contain considerable truth. SUMMARY From what has been said it should be clear, now, that there is more involved in the use of tests in business than is included in a description of the scientific methods employed in standardiz- ing and evaluating tests. There is more, too, than is added by a consideration of the results that have been secured through the TESTS IN THE PERSONNEL PROGRAM 163 application of test research methods. Research and research accomplishments do serve as the solid basis on which progress in employment tests must rest. But in actually putting tests to use, other problems and other applications appear. The tasks of planning and administering test procedure, of recording and interpreting test results, and of co-ordinating the testing program with other parts of the employment process- these matters call for careful study and intelligent handling. The value of tests is determined by the way they are used as well as by the merits of the tests themselves. Most of these problems that arise in connection with the use of tests are specific matters which must be solved in terms of the given situation. Hence we have made little attempt to offer general principles for effective test administration. But these adjustments in each specific case are none the less important and worthy of most careful consideration. The use of tests, we have also emphasized, is not limited to the initial hiring of workers. Tests are of value in many other personnel activities, both administrative and research. On the administrative side they have been used in such ways as selecting men for training and in measuring their progress during training, in selecting men for transfer and promotion, and in helping to make sound vocational counsel possible. On the research side they have shown their possibilities by assisting in detailed analyses of labor turnover, in comparisons of different working groups, in explaining inadequate individual output, and so on. New and unforeseen uses appear when least expected. It would certainly, at the present time, be ill-advised to think of tests as too narrowly restricted in their scope and possibilities. Poten- tially they may do much. Actually, however, let us recall that we have only a few promising beginnings in any of the directions that have been mentioned. CHAPTER VI THE OUTLOOK FOR TESTS IN BUSINESS Any just estimate of the prospects for psychological testing in business must take into consideration a number of factors. It must consider the present extent of their use, both in this country and abroad. It must take account of the attitude, present and prospective, of the various parties interested in the problems which tests endeavor to solve. It must point out the conditions which operate to affect the future use and develop- ment of tests, for good or ill. To these considerations the present chapter will be devoted. PRESENT EXTENT OF USE OF TESTS It is easy to overestimate the present status of the psy- chological testing movement in business, both the extent of use of tests and the degree of their perfection. So much has been said and written about their possibilities that the uninformed enthusiast is likely to think of testing as a well-established part of the employment practice in most large business establishments. The truth is quite the opposite. However many concerns there may be which have more or less seriously considered the adoption of such methods, certainly only a small percentage have made any attempts at all to use them, and only a fraction of these have undertaken test research in any really serious, persistent, scientific fashion. Attempts to enumerate those American concerns which are known (by published reference or personal inquiry) to have used tests to any considerable degree reveal not more than two score. The actual number is probably larger than this, but at most amounts to only a handful. It should, furthermore, be borne in mind that even in the majority of these 164 THE OUTLOOK FOR TESTS IN BUSINESS 165 the use is confined to a few tests for only a part of the positions and does not often amount to an organized and comprehensive test program.1 Before the war little more had been done than to talk about using tests. The experience of the American army with psycho- logical tests on a wholesale scale brought them into prominence, but most of the enthusiasm over employment tests immediately following the war subsided during the business contraction of 1920. A few concerns which had hastily and uncritically adopted tests of the intelligence-test type abandoned them on finding that they failed to prove valuable for uses to which they should never have been put. INDUSTRIAL TESTS IN EUROPE2 Nevertheless, testing has been taken up more extensively in American industry than it has in Great Britain and France, where hardly more than a half-dozen large firms have adopted any extensive testing program. In the main, British practices have followed those in this country. Their literature is full of citations of American test literature, although numerous British psychologists, including Burt, Spearman, Muscio, Watts, and others should be named among the more significant contributors to vocational psychology. But, as a rule, British industrial psychologists have concerned themselves chiefly with problems of optimal work conditions and methods, fatigue studies, and motion studies. The content of the phrase, "business psychol- ogy, " is quite different there from here. 1 The 1921 report of the National Association of Corporation Training listed thirty-two firms which reported using tests, in more or less-usually less-thorough- going fashion. 2 Brief discussions of European progress with tests in business may be found in the following: M. S. Viteles, "Psychology in Business-in England, France, and Germany," Annals of the American Academy (November, 1923), pp. 207-20; H. D. Kitson, "Industrial Psychology in Europe," Journal of Applied Psychology, V (1921), 287-90. 166 PSYCHOLOGICAL TESTS IN BUSINESS In France, tests have gained no foothold in the business world, owing to the conservatism, even skepticism, of the French busi- ness man toward scientific improvement of business methods. Nevertheless, one or two experimentalists (Lahy being the best known) have undertaken research problems in this direction. In Germany, on the other hand, vocational selection by tests has received much attention from both psychologists and busi- ness men. Post-war Germany has been alert to the necessity of using all available scientific methods for reducing waste and increasing efficiency in production if industrial prestige is to be regained. Street railways and state railways have used tests both for selecting workers and in connection with apprentice- ship training; and several lines of manufacturing-the metal trades and printing trades, especially-have been reported as adopting similar practices on quite an extensive scale. In the present political and economic crisis in Germany it is difficult to state facts or make predictions which will be true a few months later. We know that German industrial psychologists have been very active in devising and trying out tests in large numbers; but their published reports are very deficient in the statistical data necessary to enable us to evaluate their success. It is for this reason that earlier chapters have made comparatively little reference to them. There are indications,1 however, of a growing recognition of the need for a more scientific checking up of the actual results of tests which are in use. There is also evidence that some of the tests, such as are used in the post-office and in various workshops, have demonstrated their real value in selecting better workers. That the test idea has gained definite standing in Germany seems beyond dispute. Nor, while we are considering the present status of tests in business, must we overestimate the finality and perfection of what has been done even where based upon sound research 1 G. H. Miles, "The Berlin Conference in Applied Psychology," Journal of the National Institute of Industrial Psychology, I (1923), 190-92. THE OUTLOOK FOR TESTS IN BUSINESS 167 methods. Establishments which have done the most with tests are cautious, although in general optimistic, in expressing their conclusions on the subject. The truth about the present status of psychological testing may be summarized somewhat as follows. Industrial tests are still in the experimental stage. But they have proved worth experimenting with. A scientific technique has been well established. A few definite general principles and a considerable number of facts of more restricted scope about the selective value of tests have been proved. What we need now is to accumulate, with the aid of this technique, a much larger body of facts. We need negative as well as positive facts. We need to know also the relative value of selective methods other than tests. We need to know what tests and methods fail to select, as well as what ones succeed; and we need to know the relative degree of their success. We know something about selecting suitable employees for a considerable number of occupa- tions, and a great deal concerning a few. But about most occupa- tions we know little or nothing of scientific worth. UNDEVELOPED FIELDS We do not know, for example, what tests will select competent executives, although there is evidence that general intelligence has something to do with it. But many highly intelligent men are failures as executives, and some executives who would make mediocre intelligence test scores do get things done effectively. There are dynamic factors in personality, not describable in intelligence terms, and not yet adequately measurable by tests or any other device, which go far to determine success as an executive. We use such ill-defined and unanalyzed terms as "initiative," "aggressiveness," "leadership," "courage," and the like, to describe them. There are also important social factors-interests, "social intelligence" and the like-which are Likely to affect success in this as in other occupations. Just what 168 PSYCHOLOGICAL TESTS IN BUSINESS these "non-intellectual" determining factors are, and how to detect and measure them, is a task for the future. The selection of salesmen is in a somewhat similar state. We do know that the term "salesman" is only a blanket term for a very wide range of occupations, dealing with a vast diversity of commodities, calling for very different types of personal qualifications, and having, in short, only one common feature- that they all sell something. Here again, intelligence, or any single mental aptitude which can at present be tested, is usually less of a determining factor in success than are interests and social adaptability. None of the learned professions have been adequately surveyed from the vocational test angle. In engineering educa- tion and the profession of teaching, a start has been made, enough to reveal the complexities and difficulty of the problem. As in the case of executives, we know that intelligent teachers are often poor teachers, and rather unintelligent teachers often do well. When we turn to the fine arts and many of the practical arts, the extent of our knowledge is equally limited. The tests we have for these occupations merely scratch the surface. The able and scientific work done by the numerous competent workers mentioned in earlier chapters touches only a small fraction of the thousands of trades and occupations, and can be regarded as conclusive, even provisionally, in only a few of these. But trails have been blazed and we may hope that coming years will bring richer contributions to our understanding of these fields. FACTORS DETERMINING TEST PROGRESS The outlook for psychological tests in business cannot be predicted categorically. Their future depends on the attitude and actions of at least four groups of people, who have, potenti- ally, a direct interest in the problem of the scientific determina- tion of vocational fitness. These four are: THE OUTLOOK FOR TESTS IN BUSINESS 169 i. The psychologists and others interested in the technical side of test research and test administration. 2. The workers who are to be selected or guided by the aid of tests. 3. The public, who will be affected, directly and indirectly, by any method which promotes or hinders the more efficient utiliza- tion of human resources in social and industrial organization. 4. The employers, who are responsible for determining policies, initiating research programs, appointing technical experts, and adopting or rejecting programs and methods which the experts may propose for their consideration. Of the interest of the technical workers, nothing further need be said. They have proved that the problem is one which they can and are willing to attack. They are active and productive. Their numbers, while not in the aggregate large, are growing about as fast as the demand increases. Every university and most colleges offer one or more courses in psychological tests or personnel methods, and a great many men and women each year receive advanced degrees representing training in research methods which bear directly or indirectly on this kind of work. Some go into other business pursuits; some into college or uni- versity teaching; many with training in test research are to be found in educational administration. It can safely be assumed that as the demand becomes intensified, workers trained in applied psychology will come from the laboratories or from other pursuits to meet that demand. About the attitude of the second party-the worker-it is not so easy to speak definitely. It is clear that a favorable, co-operative attitude on the part of those tested is one of the pre-conditions of success with this method. But when we look for an answer, we are struck with the paucity of evidence, one way or another. There has been no such definite, conscious ATTITUDE OF LABOR TOWARD TESTS 170 PSYCHOLOGICAL TESTS IN BUSINESS formulation and expression of favor or hostility to tests as there has been with reference to certain other attempts to apply scien- tific methods to labor problems, such as "Scientific Manage- ment."1 Indeed, there is no single attitude. Those who have worked with tests know that individuals in a group vary greatly in their reactions. Most, probably, take it as a matter of course. Like filling out application blanks, giving references, and submit- ting to interview, it is something to be gone through with, and that is all. Others are frankly curious and interested. Still others express scorn or disgust, if not indignation. Link2 describes the favorable, often genuinely intelligent, comments of workers to whom he had explained the general purpose of the selective tests he was introducing: "That's not physicology," said one, " that's common sense." It need not be surprising that labor, particularly class- conscious, organized labor, sometimes views with a little suspicion the introduction of selective methods of this type. We can always expect the attitude of hostility and suspicion with which men who distrust the motives of their employers greet any innova- tion which seems to them to have for its purpose increasing profits at their expense, whether or not their suspicions have any real grounds. A part of this is, of course, the general attitude of hostility toward anything new and little understood. But occasionally this attitude of suspicion toward scientific innova- tion receives more definite expression. Hoxie, in this country, and others elsewhere have shown the prevailing attitude of hostility to scientific management to be due not to the principle involved, but to abuses in its practice. Workers have seen it used in diverse ways as devices for getting more work out of them without equivalent recompense. While there has been little conspicuous expression of such an attitude with reference to tests, it is not impossible that organ- 1 R. F. Hoxie, Scientific Management and Labor. D. Appleton & Co., 1915. 2 Link, Employment Psychology, chap. vi. Macmillan Co., 1919. THE OUTLOOK FOR TESTS IN BUSINESS 171 izations of skilled labor might be led to think of selective tests as an infringement on their peculiar prerogatives to say who should or should not enter certain lines of employment. They can, how- ever, easily be led to see tests as a useful instrument in recruiting their ranks, classifying their members, and raising their standards of competency. Unorganized and unskilled labor might con- ceivably be led to think of tests as an instrument for discriminat- ing against them, putting them out of their jobs, or making mere machines of them. They should, on the contrary, be helped to think of tests as a way of discovering and utilizing their own resources more adequately. To know that these suspicions are short-sighted and baseless is not enough. It is the obligation of every employer who introduces tests, and of every psychologist who works with them, to use care and tact, to regard them him- self, and to describe them to others, in terms of their larger social perspective. SIGNIFICANCE OF VOCATIONAL TESTING TO LABOR AND THE PUBLIC Viewed from the standpoint of their broadest social utility, selective tests are not only profitable to the employer, but they are also of vital importance to the worker and of real significance to society at large. This is equally true of other selective devices besides tests. Indeed, there are those who assert that the justi- fication for all improved personnel methods in industry is not the economic gain, but the enhancement of human welfare, and, specifically, the welfare of the worker. It should be recognized clearly that neither psychology nor any other science is inherently partisan. Pure science seeks to discover truth; applied science seeks to put that truth to the service of human ends. Whose ends it is used to further depends not on the science itself, but on who it is that cares to utilize the science. That tests and other selective devices have been utilized more commonly by employers than by labor does 172 PSYCHOLOGICAL TESTS IN BUSINESS not mean that they are of value only to the former; still less, that they are in any degree antagonistic to the interests of the latter. A worker-owned plant would find psychological tests as useful as do plants in the management of which workers have no voice. Whatever makes for the more adequate discovery of a worker's resources and their more effective utilization is of the utmost significance to him. Suppose that every man could know just that life-work in which his personal endowment would bring him the maximum of success and the minimum of failure and disappointment. His increased efficiency and higher productivity would mean increased earnings. But this would be the least of his gains. It would mean increased likelihood of advancement to higher levels in his occupation, with better remuneration, greater opportunities, and more certain tenure. It would mean more leisure, strength, and means for recreations. It would mean less of the discontent that economic maladjustment always brings. It would mean more of the happiness that comes from the sense of achievement, and more of that absorbing interest which is called forth only by those activities one can do best. But not alone do employer and employee profit by the advan- tages of successful programs of occupational selection and guidance. Society at large has as direct an interest in the utili- zation of human resources to their maximal efficiency. One of the major causes of high living costs is wastefulness in production. And there is no wastefulness so far-reaching as that of misdirected human activity. When we read repeatedly of the reduction of labor costs from 5 per cent to 50 per cent through the introduc- tion of improved working methods and the better selection and placement of workers, the possibilities of saving to society are seen to be enormous. Not only the monetary saving, moreover, but the higher standards of living and the consequent enrich- ment of life would be shared by a society organized on the prin- ciple of every man in the best possible place. THE OUTLOOK FOR TESTS IN BUSINESS 173 A democratic social organization, we know, does not mean equality in the sense of equality of ability and responsibility, for there is no such thing. Individual differences do exist and must be taken into account. So also must racial differences. Tests have helped to make clear the nature and extent of these differences. But a critical survey of all the facts seems to support neither of two extreme positions which we find vigor- ously debated today, with reference to the significance of tests for democracy. Such a survey hardly warrants the pessimistic claims of those who quote test results to prove the impending decline of American civilization. But no more does it support those who, opposing what they term an "undemocratic, educa- tional determinism," condemn testers for recognizing and mak- ing educational provision for individual differences. A genuinely democratic society is not one where everyone is leveled down to a mediocre plane, but one in which everyone has equal opportunity-that is, the utmost possible opportunity- to discover, develop, and utilize his own particular abilities to the best advantage, consistent with equal opportunities to others. Such an ideal can be realized only when society has both the disposition and the means by which to enable each one to dis- cover his own capacities and weaknesses, and to be guided into an occupation suitable thereto. Such considerations as these bear directly on the employee's attitude toward tests. The clear recognition of these facts will go far to facilitate the introduction of scientific selection methods on a large scale and with a minimum of friction. The methods by which one can insure that potential opposition from employees will be turned into cordial co-operation will vary with circum- stances. In many establishments, shop committees or other representatives of labor share in determining working conditions. In such places the logical procedure, it would seem, is for the administration of any program of vocational selection to be carried out under joint control. In other establishments, where 174 PSYCHOLOGICAL TESTS IN BUSINESS no such joint control is practiced, the employer can doubt- less do much to decrease potential opposition by a straight- forward, judicious presentation of the whole plan. He should lay all his cards on the table, and make it clear that the employees have as much to gain by it as the employer. THE FUTURE OF PSYCHOLOGICAL TESTING IN BUSINESS It is clear that the testing movement is not an enterprise that can safely be left to advance by its own momentum. It may have a future that is scientifically sound and economically profitable. Or it may bring itself into disrepute. Which of these it shall be depends less, perhaps, on the psychologists than on the attitude of those who are ultimately responsible for busi- ness policies-business men themselves. If employers are con- tent to turn over their employment problems to any irresponsible, self-styled psychologist or character-analyst who happens along, or to minor executives or clerks, without adequate scientific train- ing, scientific personnel research will be set back indefinitely. If, on the other hand, employers realize that continued test research is absolutely essential, and that only intelligent and technically competent specialists can be safely intrusted to do that research and to oversee the introduction and administra- tion of scientific employment programs, the testing movement may be expected to go on steadily to greater success and profit. Furthermore, the business man can do a great deal to effect the success or failure of even the best-trained investigator's researches. Test research, to be successful, has to be carried out under the most favorable conditions, and only the business man himself can say whether these conditions shall be provided. CONDITIONS OF SUCCESS IN BUSINESS RESEARCH Important among these conditions may be mentioned the following. To be successful, test research must be conducted: i. Under laboratory conditions in business establishments 2. Continuously THE OUTLOOK FOR TESTS IN BUSINESS 175 3- Co-operatively 4. With adequate resources in the way of personnel and production records 5. With intelligent and hearty co-operation of executives and employers 6. As an integral part of the personnel selection and main- tenance program. Let us consider these six conditions of success in a little more detail. 1. Test research to be successful must be conducted under actual laboratory conditions in business establishments. This may seem contradictory, but it is both possible and necessary. It has been found indispensable in other lines of industrial research. Some of the finest and best-conducted chemical, electrical, and mechanical laboratories are today to be found in manufacturing plants. Similar facilities must be provided for personnel research. On the technical side there must be competent research workers, with adequate physical facilities and equipment, which in psychological testing work are ordinarily much simpler than in chemical or electrical research. There must be freedom from external, economic pressure to "show results." Valuable discoveries cannot be made to order. Often weeks or months may be spent in the fruitless search for an elusive device or fact which may turn up at a time and place when least expected. Such conditions, facilities, and opportuni- ties are provided by university research laboratories, and this has made the modern university so prolific a center of discovery and advancement. In practical fields, business enterprises which seek to profit by scientific methods must provide similar opportunities. Effective research on practical personnel problems can be done only in close contact with the actual work situation, and under working conditions. The university laboratory, with all its resources, still lacks the essential materials for personnel 176 PSYCHOLOGICAL TESTS IN BUSINESS study-employees at their work. Tests are of real value only when they measure abilities as they are actually exercised on the job. This means that the investigator must have constant access to the men he needs for his purposes, and that he must have the cordial co-operation of foremen and executives-in practice by no means an easy condition to realize. The intelli- gent and tactful psychologist will, needless to say, make an effort to fit his own plans, so far as it is at all possible, into those of the executives with whom he deals. But just in so far as he is limited by their failure to co-operate, just in so far will his find- ings fail to represent accurately the real working conditions. 2. Test research, to be successful, must be continuous. The problem of finding the best selective methods cannot be settled once for all. Work conditions and distribution of duties are constantly changing in every large establishment. Job titles are not accurately indicative of the character of the work done. So it is necessary constantly to check up and revise the methods of selection, not only to improve them but to keep them workable under changing conditions. It has been pointed out, moreover, that while a set of tests may be standardized on a group of old employees and then used for selecting new employees, those tests are much more likely to be of maximum usefulness if the test scores made by new employees are checked against their own later records. This, of course, means a continuous, and not sporadic, research program. CO-OPERATIVE RESEARCH 3. The varied and intricate problems of personnel research which concern any business organization can never be solved completely by any one individual, no matter how competent he may be. They demand the combined efforts, experiences, and judgment of many investigators, each working in his own corner of the field, but sharing his findings and views with others. It has been repeatedly pointed out that selective methods which THE OUTLOOK FOR TESTS IN BUSINESS 177 have proved successful in one establishment may or may not prove successful in another. Prolonged, co-operative inquiry into the exact effect upon these methods of varying environ- mental, administrative, technical, and social conditions is indis- pensable if we are ever to formulate them as general principles of practice. The rapid and successful development of the army psy- chological testing program in 1917 and 1918 illustrated strikingly what co-operative effort can accomplish. There is every reason to believe that co-operative research can become a normal basis of procedure in industrial psychology as it is coming to be in other lines. Men trained in our universities-the largest source, by far, of trained personnel workers-are almost wholly free from the traditional petty jealousies and short-sighted self-interest which have hampered progress in competitive society, but which are now rapidly decreasing. Scientists do not hesitate to make public their findings and to share in the results of others' work. In one form or another, co-operative research in the scientific world is an accepted principle. As the business world sees more clearly the mutual advantages of co-operative research, and that effective personnel administration is more than a mere matter of increased profits and decreased costs, the extension of the principle into business research may be expected to follow. Co-operative research can be carried on in many ways. A first way, which is co-operative in effect if not in form, is the wide sharing of research findings in individual investigations through the medium of various journals and their further trial, criticism, revision, and improvement in other situations. Among the many fruitful sources of suggestion in this field are the Journal of Personnel Research, the Journal of Applied Psychology, Industrial Management, Management and Administration (and its two predecessors, Administration and Management Engineer- ing}, the Annals of the American Academy, the Journal of the National Institute for Industrial Psychology (British), Praktische 178 PSYCHOLOGICAL TESTS IN BUSINESS Psychologic (German), and many others, to which footnote references have been made in earlier chapters. Reports of pro- ceedings of the Personnel Research Federation, and of other conferences of psychologists, personnel workers, labor leaders, employment managers, and others serve a like function. Another type of semi-co-operative research utilizes the research facilities and skill provided by departments of psy- chology in our universities. Although very few doctors of philosophy in psychology are devoting their major attention to industrial research, a much larger number have a secondary interest in it, frequently supplemented by practical experience. The Psychological Corporation,1 organized in 1921, has as one of its major objects the bringing about of a closer relationship between the competent psychologists of the nation and business firms with psychological problems. While this is, of course, co-operative research only in a limited sense, it is a definite step in that direction. The private, independent research organization, of which there are a number in the field of personnel, affords another type of medium for sharing and testing findings. One example of such organizations is The Scott Company. The men in this group undertook numerous important research investigations in various cities in the years following the war and added measur- ably to our understanding of methods and devices for dealing more effectively with workers as individuals. It is significant, however, that their experiences sharply brought out the necessity of having, inside each business organization, as a permanent part of its executive staff, one or more trained personnel workers to facilitate and to carry on the work initiated by outside investi- gators. No personnel program can be carried on wholly from the outside, nor can it ever be self-operative. An example of what co-operative research can do is afforded by the Division of Applied Psychology and the Bureau of 1 J. McK. Cattell, "The Psychological Corporation," Annals of the American Academy (November, 1923), pp. 165-71. THE OUTLOOK FOR TESTS IN BUSINESS 179 Personnel Research at the Carnegie Institute of Technology,1 active for several years in carrying on research with business organizations in Pittsburgh and other cities, in inspiring the establishment of like enterprises elsewhere, and in training pro- fessional personnel research workers. Other institutes and bureaus with similar aims are to be found in this and other countries. Sometimes these are affiliated with universities and sometimes they are independent. Under various forms of organization they afford an arrangement whereby business houses having similar problems unite to maintain a co-operative bureau employing trained workers. Such a program involves many difficult and delicate problems of maintenance, but it has definitely proved its worth, both by way of actual achievement and in promise of larger usefulness. Other extensive co-operative organizations and enterprises, in a variety of forms, have found their inspiration, their methods, and their workers in the Carnegie Institute, the National Research Council,2 or other research organizations. Among them may be mentioned the Life Insurance Sales Research Bureau,3 and the Bureau of Public Personnel Administration,4 which are carrying on investigations of mutual interest and far- reaching significance. THE NEED FOR BETTER PERSONNEL RECORDS 4. Test research, to be successful, must have at its command better criteria for ascertaining scientifically the comparative value of different tests and other selective devices. While this 1W. V. Bingham, "Co-operative Business Research," Annals of the American Academy (November, 1923), pp. 179-89. 2R. M. Yerkes, "Psychological Work of the National Research Council," Annals of the American Academy (November, 1923), pp. 172-78. s C. F. Hansen, "Psychology in the Service of the Life Insurance Business," Annals of the American Academy (November, 1923), pp. 190-93. 4L. L. Thurstone, "Psychology in the Civil Service," Annals of the American Academy (November, 1923), pp. 194-99. 180 PSYCHOLOGICAL TESTS IN BUSINESS is, of course, primarily a matter of technical procedure, it does concern the employer. It demands that he shall provide more extensive, more accurate, and better-kept records than most establishments keep, not only of many details of the personal history of each worker and his employment, but also of his pro- motions, transfers, and salary increases, his errors, his output, his efficiency ratings, and any other possible measure of his personal value to the organization. A selective test can be no more valid than the standard against which its validity is checked. 5. Test research, to be successful, must have the intelligent and hearty co-operation of employers, minor executives, and employees. To this end, the investigator should be a recognized part of the regular staff of the organization, with opportunities for wide contacts and far-reaching observation. If he is an outsider, he can produce results only if he has the constant assistance of a competent, technically trained man who is a permanent part of the organization, endowed with authority to command the facilities and resources that have been described. He should have the support of the administration in his efforts to allay the suspicious attitude on the part of workers and fore- men, so frequently noticed when an ill-understood innovation is introduced into the routine of work. But perhaps of still greater importance is an intelligent and sympathetic attitude on the part of his superiors, and particu- larly those who are responsible for policies. A program of per- sonnel research can be successfully undertaken and carried through only when the responsible official has a genuinely scientific attitude toward the project. This does not, however, mean that he must be technically trained. Perhaps it is better if he has not even a smattering of psychological jargon. But it is essential that he shall display certain qualities which are indispensable in the successful pursuit of scientific research. EMPLOYERS MUST HAVE A SCIENTIFIC ATTITUDE THE OUTLOOK FOR TESTS IN BUSINESS 181 a) He must have a real appreciation of the meaning of special- ized technical training. The sound judgment of a successful business man is often revealed in his recognition of his own limitations, his willingness to refer technical matters to competent specialists and abide by their recommendations, and his ability to find and retain such competent technologists. b) He must be scientifically impartial. Whatever his own opinions may be, he must be willing to submit them to the test of experiment, and abide by the results. He must have mastered the fine art of maintaining an attitude of suspended judgment about questions on which there is not yet sufficient evidence to make a decision. c) He must be critical yet open-minded; critical enough to demand adequate proof, but sufficiently intelligent to recognize proof when he sees it, and open-minded enough to accept it, even at the expense of discarding his own preformed opinions. d) Finally, he must be patient. As has been mentioned, scientific research is often a long-drawn-out game, one that taxes the patience not only of the investigator but even more the patience of the employer who is eager for practical results. Probably nothing has hampered scientific research in business so much as the impatience of employers. Too often they under- take a research project in much the spirit that a child seizes a new toy which promises him much pleasure, but of which he quickly tires as the novelty wears off. Such an attitude is fatal to successful research. No executive has any right to undertake a scientific research project in his establishment if he is not willing to wait a long time, if need be, for results, and to give the investi- gator carte-blanche to whatever time and whatever facilities he needs to make it a success. To undertake it with any other spirit is to waste his own money and the scientist's time, and to bring scientific research into disrepute. With the administra- tion of devices which research has provided, the case is different. 182 PSYCHOLOGICAL TESTS IN BUSINESS Here he has a right, within the bounds of reason, to demand results and set dates. But research can never be forced. TESTS ONLY A PART OF PERSONNEL RESEARCH 6. Finally, test research, to be successful, must be made and treated as a part of the entire program of personnel selection and maintenance. On the one hand it ties up intimately with job analyses and job specifications. Tests can be made of the widest usefulness only when occupations are more thoroughly stand- ardized, the country through, than is now the case. A few attempts have been made in this direction,1 but the task is a staggering one under present conditions. But within the indi- vidual business establishment, the two undertakings are insepar- able. On the other hand, testing is, as has been repeatedly pointed out, only one method of vocational selection, one part of a wide series of closely interrelated investigations that need to be made if a vocational selection program is to be thoroughly intelligent and successful. Test research must be supplemented by research into the prognostic value of such personal data as age, sex, marital status, amount of schooling, previous experience, references and recommendations, methods of character judgment and of interviewing, value of photographs and of personal appear- ance as criteria of character, and a host of other possible indexes of character and ability. While none of these may be as valu- able, in some occupations, as well-chosen tests, there is always the possibility that they may have some value, and that as supplementary to tests they may serve to make selection and placement even more successful. In still another direction, testing relates itself to the vexing problem of measuring the efficiency of workers already in service, by means of records of production, success records, error records, and, very commonly, rating-scale systems. No one of these 1 See, for example, the Descriptions of Occupations, prepared for the United States Employment Service by the Bureau of Labor Statistics, 1919. THE OUTLOOK FOR TESTS IN BUSINESS 183 related problems can be settled independently of the test problem, and tests themselves cannot be selected successfully without due regard to the accuracy of alternative or supplementary methods of judging personal fitness. But enthusiasm for research, whether with tests or other personnel methods, should never be permitted to obscure the fact that coefficients of correlation and tables of averages are not ends in themselves. They are only means to the one all- important end, that of arriving at a better understanding of the individual worker. The reason that scientific personnel methods are less commonly used in small than in large establishments is because they are less needed. In the small shop the worker is known to his employer as an actual human being. In the large establishment he is only a unit; as a person he is lost in the crowd. The function of scientific personnel devices is to find him again, to come to understand him, his abilities and his motives; and to make his personality, so far as possible, a matter of record, in order that he shall not be lost sight of again. THE BROAD MOVEMENT TO STUDY MEN Psychological tests, then, are seen to be only one small part of the broad problem of understanding the human factor. As the nineteenth century was the era of unprecedented progress in the natural sciences and the technologies they made possible, so the indications are that the twentieth century is to be marked by great advances in our understanding of men and our control of human energies. Some far-sighted thinkers are proclaiming their conviction that our increasing mastery of physical forces is a potential source of disaster to civilization, unless man's increasing control of gigantic agencies of power and of destruc- tion is paralleled by an equally great development of his under- standing and control of himself. It is not only the business world that is interested in finding better methods of under- standing people. The same interest is reflected in the publishers' 184 PSYCHOLOGICAL TESTS IN BUSINESS catalogues, the library call-lists, the columns of newspapers and magazines, on the stage, the screen, and the pages of the popular novel, as well as in the growth of the university departments of physiology, psychology, anthropology, and sociology. It is in the air about us. The business world is increasingly awake to the various prob- lems which the human factor presents-the problems of analyz- ing, measuring, and understanding men, both workers and con- sumers, their character, their motives, and their abilities; the problems of changing men, of modifying their motives and inter- ests, and training their abilities; the problems of providing such conditions, physical, psychological, and social, as shall enable them to accomplish their ends with the greatest success and the least pains. Any devices and methods will be valued which can prove helpful in this drive to substitute intelligent under- standing and control for guesswork and prejudice in the realm of human action. So far as tests can aid, we may be optimistic as to their future. APPENDIX SUGGESTIONS FOR FURTHER READING The following list is not intended to serve as a complete bibliography. Its purpose is merely to suggest to the interested reader some of the books which will give him further information about psychological tests and their use in business. Footnotes throughout this book offer additional sugges- tions along the same line. A still more extensive body of literature deals with the uses that have been made of psychological tests in schools. Although no detailed refer- ence to that literature is made here, it contains much that is more or less directly relevant to the use of tests in business. J. C. Chapman, Trade Tests, Henry Holt & Co., 1921. Describes the construction and use of trade tests in the United States Army during 1917-18, and reproduces some of the tests in full. Sherwin Cody, Commercial Tests and How to Use Them, World Book Co. 1920. Describes the National Business Ability Tests-mainly educational proficiency tests-and their use in commercial education and selection of office workers. C. H. Crennan and F. A. Kingsbury, Editors, "Psychology in Business," Annals of the American Academy of Political and Social Science, Novem- ber, 1923. A series of summary articles by American psychologists on various aspects of the application of psychology to business, including several articles on psychological tests and other selective methods. H. L. Hollingworth, Judging Human Character, D. Appleton & Co., 1922. A survey of various methods used in judging traits of character, such as photographs, letters of application, recommendations, personal inter- views, psychological tests, and rating-scale systems. The value of each method is discussed in so far as scientific studies have thrown light on the matter. H. L. Hollingworth, Vocational Psychology, D. Appleton & Co., 1916. One of the earlier books in this field. Brief discussions of vocational tests, character analysis methods, self-estimates, and ratings by associates. 185 186 PSYCHOLOGICAL TESTS IN BUSINESS H. C. Link, Employment Psychology, The Macmillan Co., 1919. Probably the best description of sound scientific methods in construct- ing and standardizing selective tests for use in an industrial establishment. Discusses briefly certain other employment methods and problems. Hugo Munsterberg, Psychology and Industrial Efficiency, Houghton Mifflin Co., 1913. The pioneer work in this field. Good statement of the problem of vocational selection, and a clear picture of the possibilities in the field. On the whole, less scientific than some of the later books. "The Personnel System in the United States Army"; Vol. I, History of the Personnel System; Vol. II, The Personnel Manual; U.S. War Depart- ment, 1919. Describes the various devices and methods used during the recent war for ascertaining, recording, and utilizing more effectively the qualifications of men in the United States Army. Rudolf Pintner, Intelligence Testing, Methods and Results, Henry Holt & Co., 1923. Presents a good bird's-eye view of the intelligence testing movement, including a historical introduction, analytic description of the more common individual and group scales, and chapters on the use of tests with school children, college students, delinquent, dependent, and defective classes, army, racial groups, and employees. Helpful bibliographies. W. D. Scott and R. C. Clothier, Personnel Management, A. W. Shaw Co., 1923. A comprehensive discussion of the problems of personnel administra- tion, from a standpoint scientific yet practical. Embodies results of numer- ous investigations conducted by members of The Scott Company in a variety of large business enterprises. Includes chapters on tests. W. D. Scott and M. H. S. Hayes, Science and Common Sense in Working with Men, Ronald Press Co., 1921. A brief and popular treatment of several outstanding problems of per- sonnel management, with some discussion of tests in business. L. M. Terman, The Measurement of Intelligence, Houghton Mifflin Co., 1916. Describes in detail the methods of using and interpreting the "Stan- ford Revision" (the most widely used American form) of the Binet-Simon general intelligence test. Some discussion of the relation of intelligence to school progress. SUGGESTIONS FOR FURTHER READING 187 H. A. Toops, Trade Tests in Education, Teachers College, 1921. Shows how trade tests can be used to advantage in connection with industrial training courses. G. M. Whipple, Manual of Mental and Physical Tests, 2 volumes, Warwick & York, 1919. A useful manual of tests of special abilities. Good bibliography of the earlier literature on psychological tests. R. M. Yerkes, Editor, "Psychological Testing in the United States Army," Vol. XV of Memoirs of the National Academy of Sciences, Government Printing Office, 1921. Detailed statistical analysis of psychological test results in the army during 1917-1918. C. S. Yoakum and R. M. Yerkes, Army Mental Tests, Henry Holt & Co., 1920. Describes in detail the various general intelligence tests used in the United States Army during the recent war. Illustrative results are also given. Bibliography of Tests for Use in Schools, World Book Co., 1923. A classified list of several hundred psychological and educational tests and scales, with names of publishers. Journal of Applied Psychology, Indiana University Press, quarterly. Publishes numerous articles on problems of industrial psychology. Journal of Educational Psychology, Warwick & York, monthly. Publishes frequent articles on tests and mental measurements, prin- cipally of educational, but occasionally of industrial, interest. Journal of the National Institute of Industrial Psychology (London), quarterly. Contains general articles and reports on research studies in industrial psychology, including some on tests. Journal of Personnel Research, Williams & Wilkins Co., monthly. Each number contains one or more articles describing scientific personnel researches, as well as reviews of books and abstracts of recent periodical articles on personnel topics. Public Personnel Studies, Johns Hopkins Press, bi-monthly. Presents reports of comparative studies and original investigations undertaken by the Bureau of Public Personnel Administration in the field of selective and administrative methods in public employment. INDEX INDEX Alertness, mental. See Intelligence Alpha test. See Army psychological tests Application blanks, 14, 148 Aptitude tests, 21, 25-39, 97 ff-, 114, n 6 ff., 145. Ste Intelligence, tests of. Army psychological tests, 6, 80, no, 158-60; Alpha intelligence test, 18,19, 28; Beta intelligence test, 29; trade tests, 24-25, 64, 97, 114-16 Army, United States, personnel system of, 80, 186 Assemblers, tests for, 11^-20 Assembling tests, 36, 120 Aviators, tests for, 135-36 Bagby, E., 136 Ball-bearing inspectors, tests for, 126-27 Berry, W. S., 121 Beta tests. See Army psychological tests Bibliography, 185-87 Bills, M. A., 90, 100, 109 Binet, A., 2, 26-27 Bingham, W. V., 108 Blackford, K. M. H., 79 Bookkeepers, tests for, 88 Bookkeeping machine operators, tests for, no Braunshausen, N., 138 Bregman, E. O., 94, 129 Brewington, A., 105 Brown, W., 69 Burr, Emily, 124 Burt, C., 102, 165 Burtt, H. E., 94, 121, 129, 137 Business: use of tests in, 6; uses of tests for purposes other than selection, 7,151-63 Camera factory workers, tests for, 125 Carnegie Institute of Technology, 179 Carney, C. S., 95 Cattell, J. McK., 178 Chapman, J. C., 25, 80, 114, 115, 185 Character traits: tests of, 21, 39-41, 130-32 Chicago Group Intelligence Test, 28, 30-35 Civil Service, 86-87, 93, 134 Claparede, E., 113 Classes of tests, 17 ff. Clerical workers, tests for, 47, 83-97 Clothier, R. C., 81, 87, 95, 107,108,149, 153, T54, 186 Clothing-factory operators, tests for, 126 Clothing-store salesmen, tests for, 129 Cody, Sherwin, 84, 97, 185 Color blindness, test of, 22 Compositors, tests for, 123-24 Comptometer operators. See Comput- ing-machine operators Computing-machine operators, tests for, 90, 109 Co-operative research, 64, 176-79 Coover, J. E., 136 Correlation, 50, 69-72 Crennan, C. H., 113, 185 Critical score. See Tests, standards and critical scores Democracy and tests, 173 Department-store salespeople, tests for, 94, 129-30 Distribution of human traits, 10-12 Dockeray, F. C., 136 Dodge, R. E., 136 Downey, J. E., 19, 39-40 Dressmaking aptitude, tests for, 135 Duck, J., 94 Dunlap, K., 136 191 192 PSYCHOLOGICAL TESTS IN BUSINESS Education, use of tests in, 4-5 Educational tests, 4, 22-23, 84-85, 95 Employers. See Labor Employment selection. See Selection Engineering students, tests for, 137 Europe, industrial testing in, 165-66 Executives, tests for, 107, 167-68 Factory occupations, tests for, 116-28, 139-40 Feather-factory workers, tests for, 124 Feingold, G. A., 159 Ferree, E. F., 137 File clerks, tests for, 92 Filer, H. A., 93 Firemen, tests for, 134 Flanders, J. K., 95 Fleisher, A., 80 Following-up employees, 49-50, 65-68, 151 Fontegne, J., 133 France, industrial testing in, 165 Frankel, L. K., 80 Freeman, F. N., 28 Frequency tables and curves, n Freyd, M., 135 Fryer, D., 159 Gaugers, shell tests for, 119 General intelligence. See Intelligence Gerhardt, P. W., 134 Germany, industrial testing in, 166 Graphic methods, 72 Graphotype operators, tests for, no Great Britain, industrial testing in, 165 Group intelligence tests, 18, 28-36 Gun-pointers, tests for, 137 Hayes, M. H. S., 81, 159, 186 Henderschott, F. C., 95 Henmon, V. A. C., 136 Hoke, E. R., 98 Hollerith machine operators, tests for, no Hollingworth, H. L., 80, 149, 185 Hoxie, R. F., 170 Illiterates, tests for, 29 Individual variability, 9-13 Industrial occupations. See Factory occupations Inspectors, shell, cartridge, and bullet, tests for, 117-19 Insurance salesmen, tests for, 130-31 Intelligence: classes of, 29 s.; occupa- tional levels, 158-62; quotient, 27; tests of, 26-36, 47, 69-70, 83, 85 ff., 126 ff., 155 Interests, tests of, 41, 131-32, 157 Interviews, employment, 14, 149-50 Ives, F. W., 137 Job analysis and specifications, 9, 47, 51-54, 78, 182 Job requirements and individual qualifi- cations, 8, 47, 51-54 Jones, E. S., 132 Journalistic aptitude, tests for, 135 Kelly, R. W., 81 Kelly, T. L., 69 Kingsbury, F. A., 113, 185 Kitson, H. D., 165 Komhauser, A. W., 87, no, 113, 116 Krais, D., 124 Labor: attitude of, toward tests, 169- 71; significance of tests for, 171-74; sources of supply of, 147-48. See Turnover, labor Lahy, J. M., 98 Lathe operators, tests for, 122 Layoffs, use of tests for, 7, 152 Ledger clerks, tests for, 90 Life Insurance Sales Research Bureau, i79 Link, H. C., 39, 80, 81, 89, 97, 100, 109, 117, 121, 127, 159, 170, 186 Lipmann, O., 124, 132, 138 Lough, J. E., 99 Machine operators, tests for, 120 Machine-shop work, tests for, 120-21 Mann, A. L., 125 Mann, C. R., 137 INDEX 193 Manufacturing operations. See Fac- tory operations Marcus, L., no McCall, W. A., 69 McComas, H. C., 133, 136 Mechanical aptitude tests, 36 Medicine, use of tests in, 6 Mental age, 27 Mental alertness. See Intelligence Messengers, tests for, 107 Metcalf, H. C., 80 Miles, G. H., 166 Moore, B. V., 137 Moral judgment tests, 22, 40 Motor aptitude tests, 38 Motormen, tests for, 133-34 Miinsterberg, H., 2, 39, 58, 59, 79, 80, 129, 132, 133, 186 Muscio, B., 94, 98, 102, 113, 123, 124, 127, 165 Musical ability, tests for, 19, 20, 38, 58, 137 Myers, C. S., 137 National Association of Corporation Schools, 80 National Association of Corporation Training, 165 National Business Ability Tests, 84 National Research Council, 179 Navy, tests in, 137 Newcomb, A., 79 Office boys, tests for, 47, 88, 107 Office occupations, tests for, 69, 77, 82- in, 139. See Clerical workers, tests for O'Rourke, L. J., 93 Oschrin, E., 130 Otis, A. S., 28, 126 Paterson, D. G., 92 Patten, E. F., 122 Personnel management and tests, 77-81, 141-63 Personnel Research, Bureau of, 178-79 Personnel Research Federation, 178 Physical tests, 37 Pintner, R., 113, 186 Policemen, tests for, 134 Prediction, tests and, 3-4, 70-71 Printing industry, tests for, 123-24 Production records, 49, 65-66 Professions, tests for, 168 Proficiency tests, 20, 22-25, 57~58, 97, 113, 145. See Trade tests, Educa- tional tests Promotions, use of tests for, 7, 152 Psychological Corporation, The, 178 Psychological tests. See Tests Public Personnel Administration, Bu- reau of, 179 Public-welfare agencies, use of tests in, 5 Radio-telegraph operators, tests for, 132 Rand, G., 137 Ratings of efficiency, 49, 67 Ream, M. J., 130 References, inquiries from, 14, 149 Reilly, P. J., 81, 114 Research, test, 43-76, 141-42, 174-83 Rogers, H. W., 99 Rubber-factory workers, tests for, 121-22 Rugg, H. O., 28, 69 Rumi, B., 114 Sachs, H., 133 Salespeople, testsfor,47,128-32,140,168 Scales, 19 School subjects, tests of. See Educa- tional tests Scientific attitude toward testing. 180-82 Scientific management, 170 Scientific method in test construction, 43-76 Scott, W. D., 81, 87, 95, 107, 108, 128- 29, 149, 153, 154, 159, 186 Scott Company, The, 92,97,115,154,178 Seashore, C. E., 19, 38, 58, 137 Selection, employment or vocational: development of, 77-81; individual variability and, 12-13; methods other than tests, 14-15, 78-79, 131-32; tests and, 7, 81-82, 146-52 194 PSYCHOLOGICAL TESTS IN BUSINESS Selling. See Salespeople Sensory tests, 38-39 Shefferman, N. W., 80 Shoe salesmen, tests for, 130 Shorthand. See Stenographers Silk-mill operators, tests for, 126 Simons, A. M., 81 Snow, A. J., 87, 107, 154 Solari, E., 133 Sorters, tests for, 90 Sowton, C. M., 102 Spearman, C., 165 Spielman, W., 135 Standardization. See Tests, standardi- zation of Stanford revision of Binet test, 27 Starch, D., 130, 135, 159 Statistical clerks, tests for, 90 Statistical methods in test-construction, 50, 64, 68-73 Stenographers, tests for, 47, 88, 97-107 Stenquist tests. See Mechanical apti- tude tests, and Assembling tests Stern, W., 133 Stolzenberg, O., 138 Stratton, G. M., 136 Taylor, F. W., 126 Tead, O., 80 Telegraph operators, tests for, 132 Telephone operators, tests for, 58, 132- 33 Temperament tests. See Character traits, tests of Terman, L. M., 27, 134, 186 Tests, psychological: administration of, 75-76,141-46; classification of, 17 ff., 57-61; criteria for checking validity of, 49-50, 65-68; defined, 3-4; democracy and, 173; experimental use of, 61-65; extent of use of, 6, 83, 164-66; future of, in business, 164- 84; for illiterates, 29; and move- ment to study men, 183-84; organi- zation of, 19; place in employment procedure, 150-52; recency of, 1; research with, 43-76, 141-42, 174-83; scientific method in constructing, 43- 76; selection of, 54-74; self-admin- istering, 28, 146; social significance of, 171-74; standardization of, 15-17, 47-76, 113; standards and critical scores, 51, 74-75; testing of, 44 ff.; uses of, 4-7; uses of in business for purposes other than employment, 7, 152-63; value of, 15 Thomson, G. H., 69 Thorndike, E. L., 69, 137 Thurstone, L. L., 28, 72, 85, 93, 97, 132, 134, 137 Time-study clerks, tests for, 90 Tool-makers, tests for, 120-21 Toops, H. A., 115, 154, 187 Trade tests, 23-25, 113 ff., 128, 154 Training workers, use of tests in, 7, 152, 153-54 Transfers, use of tests for, 7, 152 Turnover, labor: as criterion of test validity, 67-68; use of tests in analyzing, 88-89, 152, 154 Tuttle, W. W., 106 Types, human, 13 Typists, tests for, 47, 88, 97-107 Viteles, M. S., 95, 165 Vocational guidance, 5, 156-62 Vocational selection. See Selection Voelker, P. F., 40-41 Waitresses, tests for, 135 Watts, F., 165 Weakley, F. E., 95 Wembridge, H. A., 126 Whipple, G. M., 36, 187 Will-temperament test, 19, 39-40 Work and fatigue studies, tests and, 152-53 Yerkes, R. M., 80, 93, 136, 187 Yoakum, C. S., 80, 90, 93, 187 Yule, G. U., 69 PRINTED IN THE U.S.A.