Spatial path models with multiple indicators and multiple causes: Mental health in US counties

This paper considers a structural model for the impact on area mental health outcomes (poor mental health, suicide) of spatially structured latent constructs: deprivation, social capital, social fragmentation and rurality. These constructs are measured by multiple observed effect indicators, with the constructs allowed to be correlated both between and within areas. However, in the scheme developed here, particular latent constructs may also be influenced by known variables, or, via path sequences, by other constructs, possibly nonlinearly. For example, area social capital may be measured by effect indicators (e.g. associational density, charitable activity), but influenced as causes by other constructs (e.g. area deprivation), and by observed features of the socio-ethnic structure of areas. A model incorporating these features is applied to suicide mortality and the prevalence of poor mental health in 3141 US counties, which are related to the latent spatial constructs and to observed variables (e.g. county ethnic mix).

► Considers suicide and mental health in US counties to in relation to unobserved ecological risks. ► Finds deprivation, fragmentation and rurality as positive area risk factors for suicide. ► Finds community social capital to reduce area rates for poor mental health. ► Finds social capital itself to be negatively related to deprivation, and to fragmentation. ► Finds lower social capital scores in the south eastern US.