Accession Number : AD0016708


Title :   TWO NEW APPROACHES TO FACTOR ANALYSIS


Descriptive Note : Annual technical rept.


Corporate Author : INTERNATIONAL PUBLIC OPINION RESEARCH INC NEW YORK


Full Text : https://apps.dtic.mil/dtic/tr/fulltext/u2/016708.pdf


Report Date : Jun 1953


Pagination or Media Count : 50


Abstract : Mental ability tests, personality schedules and ratings, and attitude questionnaires were studied in a new approach to the problem of factor analysis called radex (a radical expansion of complexity). Two distinct notions are involved: a difference in kind between tests of the same degree of complexity (circumplex) and a difference in degree between tests of the same kind (simplex). The hypothesis of linearity with its attendant monotonicity and single-valuedness is rejected in favor of polytonicity and multivaluedness. Indefinitely many common-factors are assumed to operate in a test battery; these give rise (under certain circumstances) to orderly arrangements among test intercorrelations without any specific theory as to the nature of common-factors and without any calculations aimed at locating them. Two new computing techniques were devised: quantiles and latent vectors. The latter resulted in the derivation of new techniques for computing with large-order matrices. Radex theory provides a new picture of the functional nature of the interrelationships among different abilities and traits; the method is believed to be capable of better predictions with less tests than former techniques. The procedure which is used to test whether a given set of data forms a simplex is illustrated by an example, and lists of approximate simplexes and circumplexes are included. Considerations are discussed on results relative to a study of multivalued and polytone attitudinal interrelations.


Descriptors :   *FACTOR ANALYSIS , THEORY , LINEARITY , EXPANSION , HYPOTHESES , INTELLIGENCE TESTS , PICTURES , QUESTIONNAIRES , FORMS(PAPER) , ATTITUDE(INCLINATION) , DATA MANAGEMENT , MENTAL ABILITY


Subject Categories : Statistics and Probability


Distribution Statement : APPROVED FOR PUBLIC RELEASE