Factors Affecting Precision
Why aren’t repeat results identical? Why do repeat results differ most of the time?
Genetics, environment and nutrition, management and disease influence the properties of wool as a natural fibre. Wool is a variable commodity and wool testing is used to provide an estimate of these properties based on a sample taken from the bulk. As AWTA Ltd samples and tests this variable commodity, the components of that variation, and variation introduced by the adoption of the official IWTO Methods and Regulations, contribute to each result. For example, the Confidence Limits are calculated from components of variance, which include:
- between-core or between-grab variation;
- within-laboratory variation; and
- between-laboratory variation.
Between-Core/Between-Grab Variation
This source of variation is beyond the control of AWTA Ltd, and is largely beyond the control of the wool classer, since much of the variation occurs within individual fleeces. This variation differs for wools of different origins and extreme wool types. For example, core sampling schedules are based on taking sufficient number of cores to achieve a sampling precision of ± 0.7% Wool Base. More cores/bale must be taken from bales of South American wools than for Australian wools in order to obtain a similar precision, because the bales of South American wools are usually more variable for yield than Australian wools.
Within-Laboratory Variation
Within a laboratory, minor equipment and operator differences can exist. Stringent quality control, equipment maintenance and supervision minimises this source of operator variation. The procedure of measuring sub-samples/specimens on different instruments reduces the effects of variation between machines and operators. For example, the IWTO Airflow Test Method requires 2 Airflow instruments to be used and the results averaged for certification. Similarly, staples sampled for ATLAS Staple Length and Strength certification must be prepared by at least 4 operators to minimise bias of any individual operator who might consistently draw longer or shorter staples.
Between-Laboratory Variation
The subject of bias has already been discussed. Even with bias eliminated, a component of variation between laboratories exists. Each instrument will give some small difference in performance, and laboratory procedures may differ in their interpretation.
In addition, the amount of testing conducted has a direct relationship with the precision of the answer – the greater the amount of testing, the more repeatable is the test result. Testing multiple samples randomises some of the within-laboratory effects and increases the chance of identifying an error before the result is issued.
When developing test methods, the sources of variation are determined from international inter-laboratory trials. These data form the basis for the calculation of the precision limits of the new method. Precision of a test result is also dependent upon the amount of sampling and testing conducted. The greater the amount of sampling and testing, the better the precision of the test result.
The precision of an individual test result is usually expressed as 95% Confidence Limits, ie. the limits on either side of the “true” result within which you can expect 95 of 100 repeat measurements to lie. The 95% Confidence Limits for Wool Base, Vegetable Matter Base, Mean Fibre Diameter, Colour, Staple Length and Strength measurements as defined in the IWTO Test Methods and Regulations are shown in Column A of Table 1.
