UC Santa Barbara

In this thesis I present a guide to constructing a categorical measure to conform to the unique structure of a latent class nominal variable. A series of detailed steps are outlined from the starting point of construct definition through to the evaluation of models with construct validity as a common focal point at each stage of construct development and evaluation. The process of defining a construct to measure a categorical property and evaluate its validity is referred to in this thesis comprehensively as the practice of constructing measures. Currently, a pedagogical study which provides guidelines to applied education researchers for how to design a categorical construct, evaluate measurement, and build evidence for construct validity remains unaddressed in the literature. Furthermore, few applications of latent class analysis (LCA) exist which describe a process of developing a categorical construct from a measurement-oriented perspective. It is this author’s hope that by providing guidelines for constructing categorical measures researchers applying LCA methods will utilize these measurement practices and emphasize construct validity in their own research. Considering the construction of a categorical measure from the ground up requires a novel orientation towards measurement to be taken. Starting a discussion about the unique challenges of constructing a categorical measure provides an important contribution to the social sciences literature. This dissertation will provide a roadmap for defining and measuring a latent categorical construct. To demonstrate how construct measurement may be applied in practice a series of examples from education will be incorporated throughout the thesis. This paper is targeted toward applied education researchers with task-oriented recommendations intended to provide a starting point for researchers to engage in the practice of constructing measures.