Book excerpt: This book is based on the assumption that the quantities social scientists want to measure, such as attitudes, abilities, price levels, and mobility, can best be represented as unknowns in a statistical model. This method includes both theoretical and practical problems. On the theoretical side, one examines the formulation and testing of an appropriate model and then employs the standard methods of estimation and prediction. On the practical side, one explores the difficulties of implementation deriving from various fields of application. This distinctly statistical approach to social measurement recognizes that in any measuring situation there are two sorts of variables: (1) manifest or indicator variables that one can directly observe, and (2) the latent variables that one cannot observe but wishes to measure. By regarding these all as random variables on common ground, Bartholomew reduces measurement problems to the familiar calculus of probability. Key Features * Argues that social measures should be derived by reference to a statistical (stochastic) model of the process being studied * By analyzing a variety of models, Bartholomew finds the common cords that bind them together in a single whole * Distinctive in its steady attention to the statistical aspects of social measurement and focus