ASTM D3244 Standard Practice for Utilization of Test Data to Determine Conformance with Specifications
8. Obtaining the Assigned Test Value (ATV)
8.1 The following procedure will produce an ATV with precision control based on the reproducibility of the test method.
8.2 The receiver and supplier should obtain independent test results, XR and XS, respectively.
NOTE 2 - The supplier's result must be on the test sample (see Section 5) and not a reported value by the supplier. In many cases, a reported value by the supplier is obtained on a different sample, for example, at point of manufacture, and may be the average of several determinations.
8.3 ATV Procedure:
8.3.1 If the absolute value of Δ = XR - XS is less than or equal to R, the reproducibility of the test method, average the two results to obtain the following in accordance with 6.3.1:
ATV = (XR + X S)/2
8.3.2 If the absolute value of Δ exceeds R, reject both results and retest on portions of the retain sample to obtain XR', XS'.
8.3.3 If the absolute value of Δ' = XR' - XS' is less than or equal to R, average the two results to obtain the following in accordance with 6.3.2:
ATV = (XR' + XS')/2
8.3.4 If the absolute value of Δ' exceeds R, obtain a new test value XRL from a referee laboratory (6.5).
8.3.5 If Δ3 = Xmax - Xmin is less than or equal to 1.2 R, obtain the following:
ATV = (XR' + XS' + XRL)/3
8.3.6 If Δ3 exceeds 1.2 R, obtain ATV as the average of the closer pair.
NOTE 3 - This last step for obtaining an ATV does not comply rigidly to statistical concepts. It is done in this manner because in most cases the test sample (see Section 5) is depleted.
8.4 The above procedure will always yield an ATV. If the supplier's and receiver's laboratories have little or no bias relative to each other, then the procedure will end at 8.3.1 about 95 % of the time, and some 95 % of the remaining 5 %, at 8.3.3.
8.5 If any particular supplier and receiver pair find they frequently must go as far as calling for a reference laboratory test, they should carefully check their running of the test and, if possible, calibrate their results with other laboratories.
8.6 This procedure for obtaining an ATV is designed for the test of samples obtained according to Section 5.
8.6.1 If more extensive testing is needed for special situations, comparable procedures can be developed. A statistician or quality control expert should be consulted to do this.
9. Product Quality Conformance
9.1 A product should be considered as conforming to the specifications if the ATV of each property meets the AL value.
9.2 The supplier should ship product only if there is confidence that each property meets specification values.
9.3 When the receiver has obtained a single result, the product quality should be considered suspect if the test result fails the AL value (see A3.1.5).
9.4 A dispute between supplier and receiver may arise whenever a receiver's result fails the AL value.
9.5 The dispute should be resolved by obtaining an assigned test value (ATV) for the product as an estimate of the "true value" and comparing this to the acceptance limit (AL) as determined in 7.3.
