Placing individuals in a multileveled context

Table of Contents

Placing individuals in a multileveled context
Initial notes
Annotations on common readings
Annotated additions by students
Idea: Different or even contradictory associations can be detected at different levels of aggregation (e.g., individual, region, nation), but not all influences can be assigned to properties of the individual—Membership in a larger aggregation can influence outcomes even after conditioning on the attributes of the individuals.

Initial notes

Prepare for class as follows:
Ecological & atomistic fallacies:

Neighborhood effects:

From neighborhoods to networks:

Income inequality, stature, and obesity:

Mini-lecture
Notes and annotations from 2007 course, 2009
Common readings and cases: Freedman 2001 (Ecological and atomistic fallacies), Diez-Roux 2002a, b (Neighborhood effects), Marcelli 2006 (Obesity)
Supplementary Reading: Balfour 2002, Coulton 1999, Dunn 2007, Korbin 2000, Lawlor 2005, Marcelli 2005, Oakes 2004


Annotations on common readings

Ana V. Diez Roux
Invited Commentary: Places, People, and Health
American Journal of Epidemiology Vol. 155, No. 6

In epidemiology, the triads of “agent, host, and environment” and “person, time, and
place” have been traditionally emphasized as important interactions in understanding the causes of ill health. Diez Roux argues that modern epidemiology’s disproportionate interest in personhood (individual characteristics) suggests that the effects of “places” can be ignored once individual-level risk factors have been accounted for. Well-established is the fact that health tends to exhibit gradients across geographic locations. The purveying putative explanation for this phenomenon being: the cumulative characteristics of people within a geographic locality giving rise to a distinct neighborhood health profile. Diez Roux notes a resurgent interest, among social scientists, to determine the ways in which neighborhood contexts may affect individual-level health outcomes.

Contextual and multilevel analyses have been the vehicles employed for investigating area effects on health. Diez Roux posits that area itself is consequential to the health of its residents. Previous studies document an independent effect of neighborhood socioeconomic characteristics on health, which persists after controlling for individual-level factors. Diez Roux concedes that the magnitude of neighborhood effects and their relative importance did vary. Empirically demonstrating that neighborhood environments are causally related to health is slated to have implications for disease prevention and health policy. However, Diez Roux underscores that determination of whether processes mediated by neighborhood environments are causally related to heath is often impeded by complexities. The principal quandary is establishing whether the derived associations truly reflect causal processes. Thus, they are not simply the artifices of model misspecification at the individual level or residual confounding by individual-level variables. Compounding this matter, analytic approaches commonly used in epidemiology are proposed to be ill-suited for examining complex causal models. In this commentary, Diez Roux critically reviews Balfour and Kaplan’s paper, ‘Neighborhood Environment and Loss of Physical Function in Older Adults: Evidence from the Alameda County Study;’ subsequently endorsing their analytical approaches to addressing several complex issues involved in assessing neighborhood effects.

“Neighborhood differences are not “naturally” determined but rather result from social and economic processes influenced by specific policies. As such, they are eminently modifiable and susceptible to intervention. In addition, the improvement of neighborhood environments is likely to have a multitude of benefits for people and society as a whole (Diez Roux).” (SY)


Annotated additions by students

(In alphabetical order by author's name with contributor's initials and date at the end.)