Please use this identifier to cite or link to this item:
https://hdl.handle.net/11681/11607
Title: | Critical comparison of moving average and cumulative summation control charts for trace analysis data |
Authors: | University of New Hampshire. Department of Chemistry U.S. Army Toxic and Hazardous Materials Agency McGee, Irene E. Grant, Clarence L. |
Keywords: | Chemical analysis Quality control Control procedures Trace analysis Hazardous wastes |
Publisher: | Cold Regions Research and Engineering Laboratory (U.S.) Engineer Research and Development Center (U.S.) |
Series/Report no.: | Special report (Cold Regions Research and Engineering Laboratory (U.S.)) ; 87-21. |
Description: | Special Report Abstract: Percentage recovery estimates have been obtained for 15 analytes or surrogates of environmental concern by four commercial laboratories over a two-year period. These quality control analyses were performed using standardized methods on a control soil matrix. Over 100 lots of results were available for many of these analytes. This massive amount of data afforded an opportunity to compare the sensitivity of different quality control protocols for detecting "out-of-control" situations and also to compare the performance of the four laboratories. Recoveries averaged 90-100% for 11 of 15 analytes. Reproducibility of recovery estimates was surprisingly consistent from lab-to-lab, which permitted calculation of performance (capability) limits based on the mean recovery ±3o. From a comparison of moving average control charts (n = 2 and n = 3) with cumulative summation charts, the n = 3 moving average charts were considered most suitable for routine lot-to-lot control by contractors. The cumulative summation charts are very useful for situations requiring critical diagnostic analysis of problems. Where duplicate recoveries were obtained with each lot, lot-to-lot variability was similar in magnitude to within-lot variability. To avoid an excessive number of out-of-control responses, control limits should be based on total variability rather than within-lot variability. |
Rights: | Approved for public release; distribution is unlimited. |
URI: | http://hdl.handle.net/11681/11607 |
Appears in Collections: | Special Report |
Files in This Item:
File | Description | Size | Format | |
---|---|---|---|---|
SR-87-21.pdf | 2.44 MB | Adobe PDF | View/Open |