Statistical Analysis Technical Committee

GAS composition

Chairman: M. Claude Saby (Total)

The Statistical Analysis Technical Committee deals with inter-laboratoratory research carried out:

• in the Vehicle Industry
• in the Oil and Chemical Industry, Additive Industry
• and among Users.

The GAS participates in SDG, the CEC Statistical Development Group (ex SAAC)
.
The aim of the GAS within the GFC is to turn
"STATISTICS into a tool aiding decision-making".

Other relevant tasks are test methods validation, accuracy criteria,
new multi-varied analysis methods …

Two major achievements stand out:

A direct help to the Working Groups of the various GFC Technical Committees.

For the purpose of fine tuning test methods and establishing the repeatability and reproducibility thereof, the GAS :

• proposes programmes for the evaluation of test method fidelity parameters,
• ensures statistical analyses of round robin test results,
• helps the working groups in interpreting data correctly and fully.

A more prospective statistical thinking.

The statistical array does not always provide the tools suited to GFC more specific needs, in which case the GAS will convene a working group to research the problem, once the topic has been deemed of interest.

Here are a few significant examples among others:

• Terminology aimed at GFC members.
  It is essential to use a common language.
• Discrimination and discriminatory power.
  Basic concepts and quality standards for a test method, which will be useful only if it shows it can discriminate between two products - it is even better if the two products have different field performances. This gave rise to the opportunity of developing normality, comparison and variance tests...
• Outliers.
  This well-known topic is seen here in a new light: a set of methods reveals to the user data from an inter-laboratory test (in the form of a graph) before subjecting them to a relevant statistical analysis.
• Methodology for test method improvement.
  A statistical study must be part of the working group research. The statistical methodology for carrying out tests must meet the working group’s objectives and at the same time deal with the constraints.
  A methodology is then established, based on programmes, which will enable the investigation group to progress its research in the most rapid and efficient fashion.
• Workshops offer GFC members an opportunity to approach concrete statistical notions adapted to their concerns:
  Workshops:
  on 21 October 1988: Statistics – a tool to aid decision-making
  on 30 January 1992: La M.E.S.U.R.E. – the practical use of a test method repeatability and reproducibility values
  on 17 November 1995: Quality of a test method - Fidelity and discrimination.

This brief description aims at:

• stressing, if need be, how interesting and useful statistics are,
• showing how diversified, operational and prospective GAS research is.

In conclusion, the Statistical Analysis Technical Committee provides the necessary tools to support test methods devised in the other Technical Committees.