Please comment on how laboratories can discard erroneous test results. Also, please comment on the example of a test result which is clearly an outlier, but without a discernible cause.
Suspect test results should not be discarded, but should be treated according to the laboratory's quality assurance plan covering the test method. At a minimum, the plan should include steps to document that a result is suspect, the reason the result is considered suspect, the results for the sample that are not considered suspect, the steps that were taken to obtain the non-suspect result, and the steps that were taken to assure quality of future analyses. Reporting requirements (to EPA) only include submission of the final result. However, documentation of erroneous results should be kept for auditing purposes.
An outlier which can be clearly identified through statistical means, may be discarded when allowed by the laboratory's quality assurance plan. These steps must be clearly documented as explained above. EPA will not accept retesting on a regular basis as a means to achieve favorable test results through test to test variation. Also, it is not considered appropriate to only retest unusually high (or unfavorable in the case of properties like oxygen content) test results. An outlier in the favorable direction should also be investigated. (7/1/94)
This question and answer was posted at List of Reformulated Gasoline and Anti-Dumping Questions and Answers: July 1, 1994 through November 10, 1997 (pdf)