next up previous index
Next: asa.stat.environ.05 Up: ASA Statistics and the Previous: asa.stat.environ.03

asa.stat.environ.04


Submitted by ASA-ENVR for competition

Session Slot: Sunday, 4:00-5:50

Estimated Audience Size: 50-60

AudioVisual Request: overhead and slide projectors


Panel Session Title: How Low Can We Measure: The Statistics of Detection and Quantitation

Theme Session: Yes

Applied Session: Yes


Session Organizer: Coleman, David Alcoa


Address: Applied Math & Computer Tech. (483), ATC-D-10 Alcoa Technical Center, 100 Technical Drive, Alcoa Center, PA 15069

Phone: 724-337-5913

Fax:

Email: david.coleman@alcoa.com


Session Timing: 110 minutes total (Sorry about format):

Opening Remarks by Chair - 5 minutes Panelists Comments/Discussion - 85 minutes Floor Discussion - 20 minutes

Abstract: A multitude of statistical issues surrounds the seemingly simple question: ``How low a concentration or level can we reliably measure or detect?'' This question is pertinent to environmental sciences and analytical chemistry, but also to medical sciences and national defense. In this session, four distinguished panelists with diverse background present and tackle some of these issues, then discuss and debate the merits of various solutions. One concern is how to balance detection sensitivity with false positives/false alarms. Also important is selecting a population subsample; estimating between-lab and within-lab sources of bias and variation; developing calibration methods that will work in the face of heteroscedasticity and the presence of outliers; describing the relation between measurement uncertainty and significant digits; and applying equivalence testing.

The first panelist will comment on the statistical and conceptual properties of various estimators for the quantification limit and address issues of bias, variability and robustness. The second panelist will describe the calculation of detection limits for ion-chromatographic data and discuss the strength and weaknesses of two approaches. The third panelist will comment on the application of statistical models to accommodate heteroscedasticity and traditional laboratory approaches for estimating detection limits. The fourth panelist will describe the newly adopted ASTM standard practice for detection and the proposed standard practice for quantification.


This session addresses the theme of statistics as a guide for policy. Everyone has or should have an interest in how low a concentration - such as lead in water, PCB's in soil, carbon monoxide in air - can reliably be measured. The requirement to protect human health and ecosystems has caused regulatory agencies to push for lower limits on pollutants, but the ideal 0% is unattainable; so policy must be established to determine acceptably low limits. That policy should be based on statistics, analytical chemistry, and practical considerations.

This session will be of interest to applied statisticians. Because all measurements are uncertain, statistics and analytical chemistry must be used jointly to estimate the lowest concentration that can be reliably detected and the lowest concentration at which one can reliably report a numerical measurement value. Issues at hand that will be discussed and are of interest to applied statisticians are selection of sample population, biases (e.g., between labs), heteroscedasticity, blind trials, false detections, power of detection, significant digits, and calibration.

This session will be of interest to members of other ASA sections including the Sections on Physical & Engineering Sciences, Government Statistics, Biometrics, and Health Policy. Statisticians, environmental scientists, and policy-makers in government regulatory agencies, USGS, DoD, and NIST will be interested in this topic.


Session Chair: Wendelberger, Joanne R. Los Alamos National Laboratory


Address: Los Alamos, NM 87545

Phone: 505-665-4840

Fax: 505-667-4470

Email: joanne@lanl.gov

Gibbons, Robert   University of Illinois


Address: Chicago, IL

Phone: 312-413-7755

Fax: 312-996-2113

Email: robert.gibbons@uic.edu

Vanatta, Lynn   Air Liquide


Address: Dallas, TX

Phone: 972-995-7541

Fax:

Email: ALAV%mimi@magic.itg.ti.com

Davis, Charles   Environmetrics & Statistics Limited


Address: Envirostat, Environmetrics and Statistics Ltd., 1853 Wellington,Henderson, NV 89014

Phone: 702-456-8994

Fax: 702-456-8114

Email: bromero@compuserve.com

Coleman, David   Alcoa


Address: Applied Math & Computer Tech. (483), ATC-D-10 Alcoa Technical Center, 100 Technical Drive, Alcoa Center, PA 15069

Phone: 724-337-5913

Fax: 724-337-4911

Email: david.coleman@alcoa.com

List of speakers who are nonmembers: 1 - Lynn Vanatta


next up previous index
Next: asa.stat.environ.05 Up: ASA Statistics and the Previous: asa.stat.environ.03
David Scott
6/1/1998