Causal Diagrams

Statistical analyses of observational data may benefit from systematic approaches to determine the variables that should and should not be adjusted for. Tools such as directed acyclic graphs (DAGs) which represent the causal relationships between exposures, outcomes, and other covariates can be very useful in this regard (Greenland et al, 1999). DAGs are useful in identifying the appropriate statistical specifications of controlling and stratifying variables in order to achieve causal estimates of a particular exposure in relation to an outcome. Often there is an over-control for variables that are on the causal pathway between an exposure and a disease especially when dealing with such distal factors such as neighborhood environments. In some situations, adjustment can introduce rather than control for confounding (Fleischer and Diez Roux, 2008). DAGS can be used to explore different situations and help develop an analytic plan.

3.4.1 Randomization

There has been one attempt to study neighborhood effects in a randomized trial setting, the Moving to Opportunity Project (Goering and Feins, 2003). In this study, individuals in poor neighborhoods were randomized to either stay in that area or move to a non-poor area (based on a pre-specified set of criteria). The obvious advantage of this approach is that it randomizes individuals to living in different neighborhoods and therefore improves our ability to draw causal inferences. However, even in this randomized design, questions remain. For example, the extent to which persons randomized to move to a non-poor area actually experienced a significant improvement in neighborhood conditions has been questioned (Clark, 2008). A bigger issue is the fact that this was an experiment on the effect of moving and not the effect of improving neighborhood environments. Given the complexities of randomized experiments in this field, a number of researchers have called for better use of natural experiments and quasi-experimental designs to identify the health impact of changes in neighborhood conditions.

0 0

Post a comment