Using item response theory to investigate potential gender bias in hoarding symptoms

Abstract

Gender and/or sex differences in the prevalence of hoarding disorder have been well documented, yet there remains a striking lack of understanding as to potential reasons. Critically no study to date has examined whether gender differences may stem from underlying biases in hoarding assessment instruments. The current study used item response theory to evaluate the Saving Inventory-Revised (SI-R; Frost et al., 1994), to determine if there might be differential item functioning (DIF) between men and women across the three core features of hoarding. We first examined responses in a sample of individuals with clinically significant hoarding symptoms (N = 255), and then replicated our findings in a non-selected community sample (N = 719). There was evidence of modest DIF, which affected only a handful of items in each SI-R subscale. In the clinical sample, women scored significantly higher on the clutter and acquisition latent continuum, whereas there was no gender difference for difficulties discarding. In contrast, there were no gender differences in any of the subscales for the community sample. Findings indicate that while there are likely gender differences in the manifestation and associated features of hoarding, those gender differences are not the result of measurement artifacts in the SI-R.