Statistical Considerations in the Interpretation of Size-Resolved Particulate Mass Data.

Random measurement error can account for a substantial fraction of the observed variability in coarse particle mass concentrations. This noise decreases the spatial correlation of coarse particle concentrations, tending to obscure any regional coherence in their variation. It also works to obscure associations between coarse particles and the effects to which they contribute. Multiple regressions can misattribute coarse effects to fine particles if the fine measurement is sufficiently more precise.