Abstract
[Nonmetrical Statistical Techniques when Processing Questionnaire-Data, with An Application on Opinion-Research)
Questionnaires are frequently used in psychology. To deal with their data, classical synthetic statistics, such as factor analysis, cluster analysis, etc. are mostly used. But by using those techniques, one is working on a higher measurement level than those questionnaire-data usually allow. In fact, it would be better to use nonmetric statistics when processing those data. This article shows some basic elements of some nonmetric techniques, especially of the multidimensional scaling method MINISSA-I (Lingoes, 1973) which is based on the SSA-I as a variant of the Smallest Space Analysis (SSA) (Lingoes. 1965: Guttman, 1968). It treats ranks of similarity measures between the empirical data (ex. gamma-coefficients) to create a spatial configuration of the data in y dimensions. Some notions are given concerning the alienation-coefficient which measures the goodness of fit of our theoretical model, Some aspects of a study on the opinions of labourers and employees about the psychologist are presented. MINISSA-I is used to deal with our data. Seventy-seven items of the questionnaire are represented in an 8-dimensional space (values of the goodness of fit and the distance-scores on the 8 dimensions are shown in Tab. 1 and Tab. 2). From a nonmetric factor analysis NONFACT (Lingoes, 1973), based on the SSA-III (Lingoes, 1966; Lingoes & Guttman, 1967), seven clusters of items are withheld. Labourers and employees of the sample have a clinical image (mental health and psycho-social) of the psychologist, although industrial psychologists are not unknown There was almost no rejection of the psychologist in the sample.