Rationale of Norms on an Adult Personality Test, the 16 P.F. - For American Women

ALTHOUGH THE validity of a test is more im portant than its standardization, sincevaluable work can be done in research without an accurate standardization, yet there comes a time when the more extended use of a well constructed test re quires attention to the finer points of representative standardization. The Sixteen Personality Factor Questionnaire, which is the only factored personality questionnaire based on fully attained oblique simple structure, and an array of behavior representative of the total personality sphere, was released as a research in strument just ten years ago. Now that its factor structure has been checked in two independent re searches (2, 4), and the unitary nature of its fac tors has been witnessed by clinical and general ev idence (5, 6, 7), there is increasing demand that its temporary standardization shall be replaced by one having as much care in its calculation as went into the original obtaining of the factor valid items themselves. So far, the standardization may be considered adequate in regard to University students, but prob ably not adequate in regard to the general pop ulation. The present report describes the proce dures in what we regard as a more adequate stan dardization, and brings out certain dangers in com mon procedures on which we and others have hith erto depended. Up to the time of the present article, the gen eral adult standardization (as distinct from the undergraduate student standardization) was de rived from data obtained in the course of testing adults gathered in a variety of occupations, princi pally for the purpose of establishing, for vocational guidance and selection practice, the differences in personality profiles which exist among persons stably settled in various occupations. The principle followed in deriving general popu lation norms from data for occupational groups re quired that we 1) classify the occupations according to the ten major categories in the U.S. Census Re port, with the help givenbyShartle's work on clas sification of occupations (9), and 2) that we multiply the number of individuals in all but the largest cat egory by the ratio necessary to bring the numbers in our samples from the main occupational groups to the same proportions as those existing in the country as a whole. This principle of stratified sampling according to main occupational categories'' is admittedly a rough one. It does not insure that the age distribu tion correspond to that in the general population, nor can one be sure that c er tain narrowly specific occupations that one happens to be able to test are a random selection from the broader occupational which they are allowed to represent. For example, we used airmen to represent the military group, though they may be somewhat differently selected from army and navy personnel. However, the literal man on the street is the hardest possible subject to get and extremely few psychological measures are really based on him. A scrutiny of such surveys of mental tests as Buro's Mental Measurement's Year Book will show that all but a minority of tests for adults actually describe standardizations only for students, and that still fewer gave an standardization based on a proper sampling of the population in occupa tion, age, and region of the country. Although we believe that our previous principle of stratified sampling according to occupational category is one of the few available ways of getting at the general population, it admittedly still misses the un employed and persons in uncommon occupations, and, as we found in the standardization of the Cat tell intelligence test, twenty years ago (1), it turns out that the higher social status occupational cate gories are almost invariably more reliably sampled than the lower. In the case of women, a majority of whom are in the home, the sampling by occupations, how ever, becomes inappropriate and open to a substan tial systematic bias. Our results below will show the female population sampled in this way dif fers significantly from one sampled by better meth ods. This article is concerned with improved tech niques and with resulting new 16 P. F. norms for