The Western Electric Company, the Atlantic Refining Company, and many other firms have extended into the supervisory and executive levels their investigations of the usefulness of psychological tests. Considerable effort was devoted to the search for dependable aids in predicting ability to sell, both in retail stores and on the road. But the widest use of tests has been in the selection of typists, stenographers, file clerks, comptometer operators, and other office workers, following the pioneer work of Thorndike for the Metropolitan Life, and of Scott for Cheney Brothers.

The expectations of the uninformed that recourse to psychological tests could somehow relieve the employment manager of the necessity of using also the more familiar ways of sifting and placing applicants, were early dispelled. It was also recognized that vocational adjustment is a continuing process. It begins in the schools. Initial placement and replacement are incidents along the road of self-discovery and advancement. The industrial psychologist's interest, then, reaches back into the period of early vocational guidance, and continues throughout the worker's occupational career. If, historically, this interest seemed to find a locus first in the employment office, it almost immediately reached out into the plant. The processes of training on the job, bristling with problems essentially psychological, early engaged the attention of pioneers like Link, who had first entered industry to improve the procedures of hiring. The mere necessity of knowing the nature of the various jobs, and of getting dependable measures of the later occupational success of people hired, was enough to draw the psychologist out of his laboratory into the works. Miss Pond, of the Scovill Company, and Frazier, of Dennison's, for example, are typical of many who have begun by investigating employment tests and found themselves plunged almost at once into a consideration of supervisory relationships and problems of organization. Johnson O'Connor, of the General Electric Company, who for eight years has consistently held himself to research on employment tests, previously had had a background of industrial engineering and plant experience. The industrial psychologist cannot arbitrarily isolate his problems and concentrate his efforts on a single phase of the task of occupational adjustment without reference to the total setting.

Intelligence and Length of Service. The methods and results of employment psychology may be illustrated in somewhat greater detail by a few typical studies, the purpose of which has been to find the relationship between length of service on the job and intelligence test score, amount of schooling, and other variables obtainable at time of hiring.

Just what, for instance, is the relationship between length of service in routine clerical work and the intelligence of girls employed for such work? Can it be said that the brighter they are, the more stable they will be? Or does the opposite generalization hold, that the duller they are, the longer they stick to this kind of routine work? As a matter of fact, neither of these hypotheses is correct. Over and over again it has been found that the relationship between the two variables mentioned is not rectilinear. Scatter diagrams of intelligence test performance plotted against length of service show that routine clerical workers who are at once competent and contented, are more often neither very dull nor very bright. Above an upper critical point on the intelligence scale, as well as below a lower critical score, the proportion of girls who disappear from the payroll is greater than it is within the middle zone of mental alertness. Many other variables, such as wage rate, also are associated with permanency of employment, so that this characteristic relationship between brightness and stability on a relatively routine job is far from close, but it is striking enough to be significant. Since this fact was first definitely established by Yoakum and Bills some ten years ago, it has been kept in mind by all who use measures of mental alertness as one of the aids to selection of office workers. The hypothesis was not new in 1920, but its verification was possible only after reliable ways of measuring mental alertness had been developed.

A clean-cut demonstration of an upper critical score for success in a selling occupation was made at Carnegie Institute of Technology in 1919. Three years previously a group of twenty-seven companies maintaining national sales organizations established the Bureau of Salesmanship Research, later the Bureau of Personnel Research. This bureau was to pool the experience of the cooperating members, to evaluate their current procedures, and to experiment with new ways of selecting and developing salesmen. The first year's work, under Walter Dill Scott, resulted in a manual of"Aids in Selecting Salesmen" containing an improved personal history record or application form, a model letter of reference to former employers, a guide to interviewing which helped the interviewer to focus his attention on essential traits and to record his judgments quantitatively, and a set of five psychological tests with directions for administering them. Among these tests was a group intelligence examination, a forerunner of Army Alpha. It was given to various groups of salesmen and sales applicants, and their scores were then compared with their actual success as measured by amount of sales. Among the men so examined were forty salesmen for a food products company. To the dismay of the investigators, when the test scores were compared with the men's sales production records, the correlation was almost zero. This seemed to be a severe indictment of the test as a measure of intelligence. Then came the war, and with it a vast experience in personnel classification and intelligence examining. The psychological tests proved their worth as indicators of mental alertness. So, when Major Yoakum, with his background of Army experience, assumed direction of the Bureau of Personnel Research in 1919, he knew that the intelligence test methods were valid, and he sought a fresh explanation of the riddle in the findings of 1916. Using the same data, he computed the correlation between test performance and length of experience with the company. The correlation was not zero. It was negative: —0.40. In other words, the brighter the salesman, the sooner he tended to leave the employ of that concern. Yoakum repeated the experiment with seventy-six salesmen of the same company, using the best available adult intelligence examination. The correlation between test scores and length of experience was —0.46. A job analysis showed that the work required of these men was largely routine order-taking. The pay was meager. Chances of promotion were slight. Only plodders were content to remain long enough to get necessary experience and build up a creditable sales record. Examining the intelligence scores again, it was apparent that there was an upper limit as well as a lower limit within which the chances were large that an applicant for a position with this concern would make good. Below this zone he lacked the necessary mental ability. Above it the probabilities were that he would not be content to remain long enough to learn his work thoroughly. The earlier form of psychological test had, after all, been a reliable measure of mental alertness. The need had been for precise determination of the relationship between test scores and occupational success.