The factor structure of the Hopkins Symptoms Checklist (HSCL-25) in a student population: A cautionary tale

Abstract

Aims: The Hopkins Symptoms Checklist (HSCL-25) is a widely used self-report measurement for mental health problems, but its factor structure is still uncertain, with divergent results in different social or cultural settings. We aimed to investigate the previously suggested factor structures of the HSCL-25, as well as a model including an explicit somatic factor among students in Norway. Methods: The study population is based on data from the Norwegian study of students’ health and wellbeing, SHoT (‘Studentenes Helse- og Trivselsundersøkelse’), and the present study comprises N = 13,525 participants. Using confirmatory factor analyses we investigated previously suggested factor structures, as well as a 3-factor structure, with specific subscales for anxiety, depression and somatic symptoms, suggested by the authors. After identification of the best-fitting model(s), measurement invariance across sexes, as well as associations with self-reported socioeconomic and social factors, use of medication and help-seeking behaviour were examined. Results: Based on the fit indices alone, bi-factor models fitted the data the best. However, upon further scrutiny when exploring the viability of the bi-factor models, we deemed the reliability of the specific subscales as extremely low and not viable as subscales. We therefore suggest that a uni-dimensional model was the most appropriate in our study. Conclusions: Based on considerations of fit indices, viability of subscales and associations with social and socioeconomic factors we suggest that a uni-dimensional model is most appropriate for HSCL-25 in a student population. Future investigations should examine how the revisions could improve the psychometric properties of the scale.