Multiplicative Factor Analysis with Latent Mixed Model for Exposure Assessment of Airborne Particulate Matter

• Activity Number: 420
• Type: Contributed
• Date/Time: Tuesday, August 2, 2011 : 2:00 PM to 3:50 PM
• Sponsor: Section on Statistics and the Environment
• Author(s): Margaret Claire Nikolov*+ and Brent Coull
• Companies: U.S. Naval Academy and Harvard School of Public Health
• Keywords: source apportionment ; latent variable ; factor analysis ; multiplicative error ; mixed model

Abstract

A major goal of air pollution research is to relate airborne particulate matter (PM) to speci?c sources which are often unmeasurable. Source apportionment and multivariate receptor modeling use standard factor analytic techniques to estimate source-speci?c contributions from a set of measured chemical species. We propose a multiplicative factor analysis with a mixed model structure on the latent source contributions. A factor analysis with multiplicative errors maintains the non-negativity of measured chemical concentrations, while the mixed model provides for systematic and random e?ects on the unobserved sources. We show that when applied to samples of ambient Boston aerosol, the multiplicative model provides better model fit over the standard additive model. Using this framework, we explore the effects of meteorological factors, such as wind direction and wind speed, on source-specific PM. Preliminary analysis of the Boston data indicates increased power plant PM associated with wind trajectories from the west/southwest, increased oil combustion PM associated with wind trajectories from the northwest, and elevated motor vehicle particles during stagnant air mass.