The Differential Association of Socioeconomic Vulnerabilities and Neglect-Related Child Protection Involvement across Geographies: Multilevel Structural Equation Modeling

Objectives: This paper explores the use of social geographic data and multilevel latent modeling to make initial predictions on geographic variation in child protection involvement for reasons of neglect, resulting in novel findings regarding the relationship between poverty and neglect in low-density geographies in the province of Quebec, Canada.

Method: This study used multilevel structural equation modeling, which combines both structural equation modeling and multilevel modeling, to test how a latent construct of socioeconomic vulnerability – across 10,650 small area geographies and 166 community health and social service regions in Quebec – is connected to child protection intervention for neglect in child population density quintiles across these geographies. Small area geographies and health and social service regions in this study are defined and discretely organized using full six digit postal codes and larger community regions provincially defined according to the public social and health services administered by the province. Full alphanumeric postal codes allow for more granular analysis than studies using only the first three digits. The rate of substantiated neglect cases is calculated per 1000 children aged 0 to 9 years, for the years 2006 to 2016, inclusively. The socioeconomic data for each of the small area geographies and health and social service regions was drawn from the 2011 Canadian National Household Survey and then matched to provincial administrative child protection data using full alphanumeric postal codes allowing for more granular analysis than studies using only the first three digits. Model fit evaluations (comparative fit index and Tucker-Lewis index) were conducted and results suggested acceptable model fit.

Results: Overall results suggest a consistent association between socioeconomic vulnerability and the increased likelihood of child protection involvement for reasons of neglect. However, latent socioeconomic factors were most associated with neglect cases in the lowest density small area geographies, suggesting that increased vulnerability to investigation for neglect is related to a sparse geographic spread of the child population.

Conclusions: The association between socioeconomic vulnerability and neglect is not surprising, but the fact that this relationship varies in strength according to child population density is a novel finding. The finding of greatest association between socioeconomic vulnerability and neglect in the lowest density small area geographies suggests that more research should be done to explore the ways that population density may relate to the likelihood of exceptional involvement by child protection authorities for neglect. Implications of this study for policymaking intended to prevent neglect, particularly related to chronic need, include the importance of taking a tailored approach to preventative service provision in small area geographies with lower child population densities, considering the challenges families face in more remote areas in accessing appropriate supports.