Professor Paul Switzer’s Conclusions
Unexplained heterogeneity of PM health effect estimates. Unresolved discrepant PM effect estimates point to probable inadequacies in the modeling approach such as an incorrect treatment of confounding variables or an incorrect characterization of the exposure-response relationship.
Enforced linearity of exposure-response. When non-proportional effects (e.g., a threshold) are allowed in the effect estimation model, the estimated ambient PM-effect relation often departs from proportionality, as can be seen for many cities in multi-city studies, such as Daniels et al. [2000] and Dominici et al. [2002].