Nice comments Shan. Yes - "low SES" masks a lot of heterogeneity! One thing that has been remarkable to me in my research is how consistently a very heterogeneous construct (education) predicts cognitive outcomes (I work on Alzheimer's disease and cognitive aging). Most of the social measures we use are both remarkably consistent and robust predictors of health and also explain a fairly small fraction of individual variability. Important point when thinking about how to improve population health and reduce disparities.
Maria