This commentary briefly summarises past work that has used multilevel analysis to investigate the multilevel determinants of health and outlines possible new directions in this area. Topics discussed include the need to (1) examine contexts other than neighbourhoods; (2) improve measurement of group-level constructs; (3) apply techniques more appropriate for causal inference from observational data; (4) analyse data from “natural experiments” involving exogenous variations in contextual characteristics; (5) examine dependencies between groups (such as spatial dependencies) more broadly and allow for reciprocal relations between individuals and contexts; and (6) contrast multilevel statistical models (or regression models generally) and complex systems models in the study of multilevel effects.
Statistics from Altmetric.com
Competing interests: None declared.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.