Anomalous results in G-factor models: Explanations and alternatives.

Abstract

G-factor models such as the bifactor model and the hierarchical G-factor model are increasingly applied in psychology. Many applications of these models have produced anomalous and unexpected results that are often not in line with the theoretical assumptions on which these applications are based. Examples of such anomalous results are vanishing specific factors and irregular loading patterns. In this article, the authors show that from the perspective of stochastic measurement theory anomalous results have to be expected when G-factor models are applied to a single-level (rather than a 2-level) sampling process. The authors argue that the application of the bifactor model and related models require a 2-level sampling process that is usually not present in empirical studies. We demonstrate how alternative models with a G-factor and specific factors can be derived that are more well-defined for the actual single-level sampling design that underlies most empirical studies. It is shown in detail how 2 alternative models, the bifactor-(S - 1) model and the bifactor-(S·I - 1) model, can be defined. The properties of these models are described and illustrated with an empirical example. Finally, further alternatives for analyzing multidimensional models are discussed. (PsycINFO Database Record

Invariance of ADHD Symptoms Across Sex and Age: a Latent Analysis of ADHD and Impairment Ratings from Early Childhood into Adolescence

“…The current model provides an acceptable, parsimonious description of the data, and is consistent with the extensive literature that supports a model with two correlated IN and HI factors to describe the structure of ADHD symptoms (Willcutt et al, 2012). However, bi-factor models of ADHD (e.g., Toplak et al, 2012) have also grown in popularity and been shown to demonstrate comparable or superior fit to two-factor models (Willoughby, Blanton, & Family Life Project Investigators, 2015), although recent work questions the validity and interpretability of these models (Eid, Geiser, Koch, & Heene, 2017). These models, in which a general factor is created from all 18 symptoms and IN- and HI-specific factors are created from residual variance in their respective symptoms, have demonstrated, among other things, that although the IN and HI factors are dissociable, their respective symptoms have more shared than unique variance.…”

The cross-national structure of mental disorders: results from the World Mental Health Surveys

“…To investigate this, bifactor models can be used that incorporate both a general factor and domain-specific group factors (e.g., Lahey et al, 2017). Although it has been pointed out that bifactor models can become complex and very hard to interpret (Eid et al, in press; Koch et al, in press), their use in structural research of mental disorders is an interesting topic for further investigation.…”

Internalizing Symptoms and Personality Traits Color Parental Reports of Child Temperament

“…This is not inherently problematic from an interpretive standpoint, especially as it is expected that parental agreement will be lower for more internalizing behaviors (Funder, ), and the Surgency dimension of the CBQ tends to have the least robust factor structure and internal coherence (Clark et al, ; Rothbart et al, ). However, Eid, Geiser, Koch, and Heene () have shown that negligible loadings in bifactor models, especially on the general factor, might be attributable to the use of “fixed” as opposed to “random” indicators (i.e., indicators from theoretically limited vs. unlimited pools). This concern about the bifactor model could be relevant for the primary results, given that each indicator represented a noninterchangeable and substantively coherent facet‐level scale.…”