Sampling When the Client Controls the Sample
You are auditing incoming inspection at a manufacturer of wellhead components. The client's inspector offers to pull the receiving records for you. You ask for all receiving records for the last six months for a critical forging supplier; the inspector returns twelve records, all passing. You notice the records are neatly sequential and all from the last eight weeks. When you ask for the earlier months, the inspector says the file room was reorganized. You walk to the receiving area and see a stack of receiving tickets on a clipboard, including two marked 'hold - dimensional.' The quality manager joins and says those were 'administrative holds, not rejections.'
1. How do you handle the sample and the discrepancy?
Read the worked answer
The audit trail is now unreliable, so the sample must be re-drawn by the auditor, not the auditee.
- Note the discrepancy: requested period (six months) versus produced records (eight weeks), and the stated reason. Record it as an observation until you can determine whether records exist.
- Go to the source. Take the receiving tickets on the clipboard directly, and ask to see the hold log and the nonconforming-material area. Auditee-selected samples are a classic bias risk; auditor-selected samples from the point of generation are stronger evidence.
- Test the 'administrative hold' claim against the hold records and disposition. If the holds were dimensional failures, they are nonconformities and should be in the NC system; if they were not, the records should show why.
- If records for the earlier period cannot be produced, that is a potential record-control nonconformity in its own right, separate from whatever the tickets show.
2. What is the risk if you accept the twelve passing records as your sample?
Read the worked answer
You would be auditing the auditee's curated view rather than the process. Twelve passing records from an eight-week window, selected by the person responsible for the process, tell you almost nothing about six months of performance and nothing about the two holds you can see with your own eyes.
The conclusion 'incoming inspection is effective' would be unsupported. The correct conclusion is either 'conformity demonstrated for the sampled period, with a limitation noted' or a nonconformity for records, depending on what the source documents show. The lead auditor must be able to defend the sample's independence.