|
Activity Number:
|
362
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Wednesday, August 6, 2003 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Section on Statistics & the Environment
|
| Abstract - #301447 |
|
Title:
|
Accuracy and Precision of PM 2.5 and Co-pollutant Samplers Used in the Steubenville Comprehensive Air Monitoring Program
|
|
Author(s):
|
Richard A. Bilonick*+ and Stephen E. Winter
|
|
Companies:
|
CONSOL Energy Inc. and CONSOL Energy Inc.
|
|
Address:
|
1800 Washington Rd., Pittsburgh, PA, 15241-1405,
|
|
Keywords:
|
PM 2.5 ; accuracy ; precision ; bias ; Grubbs estimators ; SCAMP
|
|
Abstract:
|
One of the goals of the Steubenville Comprehensive Air Monitoring Program (SCAMP) is to determine actual exposure to PM 2.5 and various co-pollutants using personal monitors. The precision and accuracy of these personal samplers relative to federal reference method samplers were determined using Grubbs estimators and related techniques. During summer and fall 2000, a set of collocated samplers collected a series of daily average concentrations. The results of the analysis are illustrated graphically and then compared to the incorrect use of regression analysis that is often reported by researchers trying to characterize bias and precision. Also discussed is the problem of assuming a constant bias when the bias changes as a function of the true concentration. To simplify the computation of the various estimates, a set of functions written in R was developed and will be made available as a contributed R package. These functions provide a simple way to compute the maximum likelihood estimates of the precision, corresponding approximate confidence intervals, and the parameter estimates for the constant bias model and for the nonconstant bias model.
|
