Examining mental health in adults with intellectual disability: The benefits of multilevel modelling

Background:Adults with intellectual disabilities (ID) commonly live in specialised residential settings. Data collected from these adults are often nested. The aim of this study is to examine the impact of this data structure. Method:Referent caregivers completed the Reiss Screen for Maladaptive Behaviour for each of the 571 adults with ID who participated in the study. Single-level and two-level multilevel modelling was compared in the context of confirmatory factor analysis and group difference analysis. Results:For both types of analysis, the two-level solution showed a better fit. Furthermore, all group differences highlighted using the one-level model disappeared, i.e., became non-significant, in the two-level solution. Conclusions:It is important to control for nested data when analysing mental health in adults with ID. Otherwise, the risk of underestimating the standard errors and obtaining biased results in significance tests appears greater, as was the case in the present study.