ABSTRACT

Several authors have recently pointed out the usefulness of multiple indicators. Curtis and Jackson [3] argue that the use of each indicator separately has certain advantages over combining them into an index. It increases the number of predictions made by a particular model, enables the careful researcher to determine the existence of an unknown spurious cause, increases confidence in the validity of the indicators, and guides conceptual reformulation. Other authors, such as Siegel and Hodge [4], Costner [2], and Blalock [1] have illustrated the use of multiple indicators to determine the existence and nature of measurement error. This chapter addresses itself to the former use of multiple indicators, extends the implications of Curtis and Jackson to a more general model, and introduces a method that deals with the problems arising as a result of this extension.