In this paper we evaluate a number of models and approaches in psychometrics, which had influenced greatly over the development and evolution of the theory of psychological scaling. We also discuss their mutual interference and relationships. We consider and compare the two procedures of empirical data collection: pair comparison procedure (a respondent compares a pair of objects with respect to the certain feature/attribute) and the direct response procedure (a respondent answers to a question about a single object). Later on we compare the two mathematical models of psychological scaling that may be applied to the pair comparison data: the Thurstone model (and implied Bradly-Terry-Luce model) and Coombs' unfolding model. The applicability of these models to direct response data is further discussed (see papers by David Andrich).