Two members of the CYFS Statistics and Research Methodology Unit have contributed a chapter to the recently published “Handbook of Structural Equation Modeling.”
James Bovaird, director of the SRM Unit, and Natalie Koziol, doctoral student affiliate in educational psychology and statistics, addressed advanced applications of the increasingly prevalent structural equation modeling technique. Their chapter outlines the modeling of ordinal data associated with latent variables, which are not directly observable but can be estimated from the values of other related variables.
Using a practical example, Bovaird and Koziol illustrated the similarities and differences among three estimation methods: one that relies on limited information to make indirect conclusions, and two that use more extensive information to directly model the categorical nature of ordinal data. Based on these comparisons, the chapter makes recommendations regarding which methods to employ under given conditions.
The 756-page handbook, edited by Duke University’s Rick Hoyle, is available through The Guilford Press.