PD CEN/TR 16797-2:2015

European foreword
0 Introduction
1 Scope
2 Terms, definitions, abbreviations and symbols
3 Context
4 Quality level of performance with respect to the
   release, emission or content of RDS
5 Methods for statistical assessment of declared
   values for RDS
6 Assessment by variables
7 Assessment by attributes
8 Statistical principles of the rules of application
9 Additional sampling requirements
10 Indirect tests
Annex A - Examples of the rules of application
Annex B - Distribution of test values
Annex C - Checklist for Technical Committees
Annex D - Model clauses for product standards
Annex E - Critical values for assessment by variables
Annex F - Gamma factor
Bibliography

Gives guidance on the statistical assessment of declared values with respect to the release, emission and/or content of dangerous substances.

This Technical Report provides guidance on the statistical assessment of declared values with respect to the release, emission and/or content of dangerous substances. This report provides statisticallybased criteria for type-testing (TT), further-testing (FT) and where a product has been shown to be consistent with measured values for the release, emission or content that are significantly below the declared values, the point where no-further-testing (NFT) is permitted.

A series of fundamental principles are defined in CEN/TR 16797-1 and two statistical approaches are defined. The first approach is to use assessment by variables and this approach requires the data to be normally or log-normally distributed. This approach is recommended as the default option. The alternative approach based on assessment by attributes is appropriate for data sets that are not normally or log-normally distributed. The downside to this form of assessment is that more test data are needed for the same level of reliability. CEN/TR 16797-1 introduces these assessment procedures and CEN/TR 16797-2 provides more detail and the statistical proof that they satisfy the principles defined in CEN/TR 16797-1. With both of these approaches the minimum frequency of testing is a function of the distance between the mean value and declared value and the variability of the data set, i.e. the sample standard deviation.

To reduce the costs of testing, production plants producing a similar product may share data, e.g. be grouping the product into clusters for statistical assessment of declared values. Rules for the use of clusters are given in this document.

This document also contains rules for identifying outliers within a data set and guidance on using tests other than the reference method for FT.

A list of tasks for product technical committees is given in this document as is a model clause for including in product standards and rules of applications that may be cited in the product standard or copied into product standards.