Assessing footwear using statistical sampling techniques
Testing and inspection play a vital role in ensuring that footwear supplied is of consistent and acceptable quality.
The inspection and testing of products form a key quality control process for many manufacturers. Even those with quality management systems such as ISO 9001 often use inspection and testing at the raw material, in-process and finished product stages of production. For some companies without a formal quality management system, inspection may be the only means of confirming product quality, and is often a contractual obligation.
The main aim of any system is to confirm that the customer will receive the required quality, and the amount examined will depend on a number of factors. Absolute confirmation of product quality can only be achieved by 100 per cent inspection and testing, however, this is not always practical or desirable. For instance, some product characteristics may be verified only by destructive or damaging testing. At the other extreme, simply carrying out limited spot checks may not be sufficient to reveal the extent of any quality defects within the whole batch.
A practical compromise is to inspect and test product samples using a sampling programme based on mathematical probability. Although this has the disadvantage that not all products are examined (and so faulty items may be missed), the risks can be calculated and a plan chosen that satisfies both manufacturer and customer. In fact, the amount of inspection and the product characteristics to be checked are often agreed at the tender stage and form part of the contract.
There are no international rules that detail sampling requirements specifically for footwear. Testing once per season, per style or per colour is commonplace for some properties. A daily or batch-testing regime for certain key tests is better. A number of specifications require a set of test results per 1,000 pairs, while many factories employ 100 per cent visual inspection.
Even shoes that appear to be exactly alike will vary. As an example, one SATRA study tested 41 apparently identical shoes taken from bulk production. In the heel pull-off test (SATRA TM113:1996 – ‘Measurement of the strength of attachment of heels to footwear and the backpart rigidity of such footwear’), the nails pulled out in all cases, although the results varied considerably from 373N to 667N.
Checking the batch
Statistical sampling of a batch evaluates the overall performance of bulk delivery. The standardisation of military standard MIL-STD-105D sampling inspection tables led to the development of international standard ISO 2859-1:1999 and BS 6001-1:1999+A1:2011 (UK).
ISO 2859-1 specifies sampling procedures for inspection by attributes. Based on an agreed ‘acceptable quality level’ (AQL), the customer accepts batches of product, provided that the production non-conformities do not exceed a defined amount after a percentage of products have been inspected from a batch. As the sampling plan is based on mathematical probability, it is assumed that the remainder of the batch will exhibit similar characteristics.
‘Statistical sampling’ – The selection of a subset of individuals from within a statistical population to estimate characteristics of the whole population.
‘Non-statistical’ sampling – The selection of a test group based on the examiner's judgment rather than a formal statistical method. For example, an examiner could use his or her own judgment to determine the sample size or the items selected for the test group.
‘Convenience sampling’ (also known as ‘grab sampling’, ‘accidental sampling’ or ‘opportunity sampling’) – A type of non-probability sampling that involves the sample being drawn from that part of the population that is close to hand.
ISO 2859-1 identifies different inspection levels, within which normal, tightened and reduced inspection regimes can apply. The inspection level and batch size determine the number of samples to be inspected, with the AQL fixing whether the batch passes or fails overall. The AQL can be fixed so that any defect causes the whole batch to be rejected. More normally, however, it is agreed that a certain level of non-conformity is acceptable within the batch, although different characteristics may be categorised as ‘major’ or ‘minor’, with different reject levels. The onus is squarely on the manufacturer to ensure product quality meets the requirements or risk the whole batch being rejected, while the customer can be confident that any non-conformity within the batch is at or below a level considered acceptable for their type of business. ISO 2859-1 procedures are not limited to end-of-the-line inspection, but can also be applied at in-process stages with rejected quantities being reworked and inspected again.
The standard may be applied by the retailer prior to distribution. Table 1 shows the sample size required for certain batch sizes. It is common for inspection to be undertaken in accordance with ISO 2859-1, but it is also appropriate to also use the standard to select test samples and use AQLs to determine performance characteristics. Standards such as ISO 2859-1 enable the footwear supplier to demonstrate that the required performance level is present consistently throughout the batch, based on the balance of probability.
|Table 1: Single sampling plans for normal inspection – taken from table II-A of
|Batch size||Sample size||AQL
0 per cent
1 per cent
2.5 per cent
|Fault figures indicate the minimum number of defectives found to reject the batch.|
It is also to be noted that there are other parts of this standard available from which to choose, and these could be used in order to be specific in the requirements for sampling. An example of this is ISO 2859-4:2002 – ‘Sampling procedures for inspection by attributes – Part 4: Procedures for assessment of declared quality levels’.
Most batch inspections have three fault categories that can be assessed visually:
- a ‘critical’ fault – a potential safety problem that requires attention – for instance, a loose heel
- a ‘major’ fault – a potential problem, likely to lead to notable returns, that requires attention, such as a failed seam
- a ‘minor’ fault – a potential problem, likely to give rise to some returns, to which it is preferable to attend, including loose thread ends.
Each batch must, as far as practicable, consist of items that are manufactured under essentially the same conditions, at essentially the same time. The inspected batch should contain no critical faults. Major faults are often set at 1 per cent or 2.5 per cent acceptability. Minor faults are often set at 2.5 per cent or 4 per cent acceptability.
A sample is selected and any faults present are listed. Table 1 details that for 1,000 items at a 1 per cent AQL (1 in 100 with a major fault), a representative sample of 80 is taken and checked. Thus, the acceptability of the batch is determined. This inspection can be based upon visual identification of attributes and/or test data.
When using these procedures for inspection by attributes, it is vital to have attributes that are not subjective and are clearly specified. A fault category such as ‘excessive toe spring’ may be listed as a minor fault, but this is open to interpretation. Unambiguous fault classifications are required so that they can be consistently applied.
Suppliers need to select a level of testing that minimises the risk of unacceptable items reaching the customer. It is likely that the frequency and quantity of sampling will change over time, based on the manufacturer’s experience and specified sampling procedures. If, after an initial high level of inspection and testing, confidence in the overall production is high, then the level of inspection and testing can be reduced (subject to contractual arrangements).
The purpose of testing and inspection is not merely to identify unsatisfactory items, but to drive continuous improvement and stop further defects occurring. It is through an appropriate level of testing that suppliers can demonstrate that they have taken all reasonable steps to guarantee the safety and quality of their product. Use of ISO 2859-1 principles will demonstrate that consistent quality is present throughout each and every production batch.
ISO 2859-1 can be used as a standalone scheme or as part of an ISO 9001:2015 quality management system (QMS). As a globally-recognised quality standard, ISO 9001:2015 is an excellent way to demonstrate a commitment to overall quality and service, while also providing the framework to aid efficient manufacturing, minimise defects and review feedback and returns.
How can we help?
Please email firstname.lastname@example.org for further information on how SATRA can help you with ISO 2859-1 implementation and sampling, in addition to ISO 9001:2015 quality management systems.
This article was originally published on page 22 of the May 2020 issue of SATRA Bulletin.