Is mental health problems measurement equivalent across disability, gender, age, and cohort in children? Findings from a Swedish national survey
Artikelförfattare: Magnus Ivarsson, Lina Homman, Henrik Danielsson
Abstract:
Aims: To investigate whether mental health problems can be described as a latent factor indicated by nine items in the Swedish Survey of Children’s Living Conditions (Barn-ULF), and to explore whether this structure varies across disability, age, gender and cohorts.
Methods: Cross-sectional survey data collected yearly from 2013 to 2019 regarding 3676 children (aged 10–18 years) was retrieved from two linked national registers: Barn-ULF (child-reported mental health problems data) and ULF/SILC (parent-reported data on child characteristics). The structural validity of the one-factor model was assessed using confirmatory factor analysis (CFA) and threshold invariance concerning disability, gender, age and cohort was tested using multigroup CFAs. The practical significance of invariance was assessed by comparing weighted scores from the singlegroup model and the multigroup models.
Results: A substantially modified one-factor model demonstrated adequate fit ( χ2 /df = 1.94, comparative fit index = 1.00, Tucker–Lewis index = 0.99, root mean square error of approximation = 0.02). Significant measurement non-invariance was found across all groups; however, its impact on the latent variable mean was minimal (<0.03 in Cohen’s d).
Conclusions: Barn-ULF might be a structurally valid tool for monitoring Swedish children’s mental health problems. However, differences in model performance across child characteristics and time – even if small – indicate that group comparisons must be conducted cautiously.
Forskningsfinansiärer: Svenska vetenskapsrådet
