6.4 - Error, Confounding, Effect Modification in Ecological Studies

Sources of Error in Ecological Studies Section

  • Within group
    • Confounding
    • Bias (selection and information)
  • Between Groups (ecological fallacy):
    • Confounding-Background rate of disease in the unexposed population varies between groups
    • Effect modification-Rate difference between exposed and unexposed groups varies

Ways to Deal with Confounding and Effect Modification in Ecological Studies Section

Confounders

  • Confounders are not necessarily associated with exposure of the individual, as is the case for confounders in individual-level analyses
  • Methods to control for confounding:
    • Treat ecological measures of the confounders as covariates in the model
    • Standardize rates with confounders

Effect Modification

  • Interaction between ecological variables (X*Z) can be included if effect modification is to be considered
  • However, interaction (X*Z) of ecological variables is not equivalent to interaction term of individual-level variables (x*z)
  • If there is interaction at the individual level, inclusion of interaction in the ecological model leads to effect modification and ecological bias