When: Monday, Oct 23, 2017 @ 10:00 AM – 11:30 AM

Where:  Norm Endler Seminar Room (BSB 164)

Speaker: Dave Flora, PhD, Department of Psychology, York University

Title: Higher order and hierarchical measurement models: Implications for subscale interpretation

Abstract: Many psychological scales are designed to measure a general, broad construct as well as narrower subcomponents of that construct, leading to the calculation of both total scale scores and separate subscale scores. Higher order and hierarchical measurement models have a long history of being used to represent such structures, but hierarchical models (particularly bifactor models) have enjoyed a recent surge in popularity. Higher order and bifactor models are very similar statistically (and occasionally equivalent), with both types of model featuring a general factor that influences all item response variables. But the additional factors included to capture multidimensionality have fundamentally different interpretations across these models. In recent literature, the specific factors of the bifactor model are often considered synonymous with (latent) subscales. But I argue that the traditional, more intuitive, subscale interpretation is consistent with the lower order factors of a higher order model. I will conclude with a discussion of implications for total score and subscale reliability assessment.