spell, to write unusual words, to file letters and to find letters in a file. On the basis of this wider knowledge we construct a series of performance tests which take, say, half an hour for any stenographer to go through. We may find that our first test is entirely too complicated and that only three or four of the best individuals in a large typewriting office can pass it. But we are just as likely, on the other hand, to make a test which is too easy, so easy indeed that even the poorest members of the group can pass it. We then work at the method until the average good stenographer can pass the test. Naturally the test must be constructed with reference to the field of general stenographic work. This test would not be suitable for accountants, for news- paper reporters or for any other vocational field. Such work is being done constantly by many business houses in consultation with psychologists. Indeed, many of the larger business houses now employ psychological experts for devising such tests. The construction of the test is a research problem. The use of the test can possibly be left to individuals who are not psychologists, but who may have received special instructions in its use. A slightly different task it at hand when we have to ask the question: What are the general factors which enter into the make-up of a successful lawyer, statesman, newspaper corre- spondent, or aviator. We may say in advance that no very ser- viceable correlations throwing light upon such questions have yet been obtained. To make the problem specific we may take up the study of the make-up of a successful aviator: What factors in the previous life of the aviator make for success in flying? Has the amount and kind of academic education anything to do with it? The stratum of society from which he comes, age, pre- vious occupation and salary obtained in it, marriage, and athletic attainments? To answer the question the necessary data are obtained from successful and unsuccessful pilots. The material so obtained is then subjected to statistical treatment and the co-efficients of correlation obtained.3 ____________________ | 3 | For method see Thorndike Mental Measurements. For the extent to which such correlations have been worked out see Thorndike article "Science", 1919. | -47- |