Paper cup visual inspection agreement matters when two inspectors make different decisions about the same rim, print mark or seam appearance. Changing the machine cannot resolve an unclear acceptance rule. A structured assessment shows whether the disagreement comes from the definition, the viewing method or inconsistent application of an agreed rule.
This guide addresses categorical visual decisions. It does not replace a dimensional measurement study or prove that a visually acceptable cup has a sound liquid barrier. Keep appearance requirements distinct from functional tests and customer-specific acceptance criteria.
Diagnostic decision table
| Pattern in the results | Possible issue | Next review |
|---|---|---|
| One person changes their answer | Within-inspector inconsistency | Viewing method and repeat assessment |
| Inspectors disagree consistently | Different rule interpretation | Reference examples and definitions |
| Everyone accepts a known defect | Shared blind spot | False-accept examples and training |
| Agreement is high on easy cups | Unrepresentative challenge set | Include realistic boundary cases |
Write an observable decision rule
Select one defect family first, such as a particular print blemish or a defined rim appearance. State the inspection zone, permitted handling, viewing orientation and decision categories. Replace vague instructions such as good appearance with the customer-approved description and suitable reference evidence. Do not invent an acceptance boundary during the assessment.
If multiple defects are present, record each relevant category rather than asking inspectors to guess which one the study organizer intended. Separate an uncertain reference classification from an inspector error. The technical or customer authority should resolve disputed reference samples before those samples are scored as having a known correct answer.
Build a representative physical sample set
Include acceptable examples, clear defects and realistic cases near the agreed boundary. Preserve sample identities using codes that do not reveal the intended verdict. Record material and production context separately. Inspect the samples for damage between rounds because handling can change a fragile rim or add a mark to a previously acceptable surface.
Plan the number of independent samples and assessment rounds with the quality specialist. Many repeated ratings of a few memorable cups do not provide the same coverage as a varied sample set. If photographs are used instead of physical cups, state that the result applies to that image-based task and may miss defects visible only when the cup is turned.
Run blind repeat assessments
Have each participating inspector evaluate the coded samples independently under the defined lighting and handling conditions. Change the presentation order between rounds and conceal earlier answers. Allow the normal inspection method without coaching during the scored rounds. Record the answer given, the sample code, inspector and round without correcting the record afterward.
Keep training separate from assessment. If an inspector asks for clarification because the rule is ambiguous, record the issue and resolve the definition through the responsible quality authority. When the rule changes materially, begin a new assessment version rather than combining results collected under incompatible instructions.
Separate consistency from correctness
Attribute agreement analysis distinguishes agreement within an appraiser, between appraisers and against an established reference. Review all three where the study permits. Inspect a disagreement table by defect type; an overall percentage can hide a recurring failure to identify a small but important class of unacceptable cups.
Count false accepts and false rejects separately against the approved reference. False accepts release an unacceptable sample, while false rejects discard an acceptable one. Have the analyst report the denominators and uncertainty, particularly when few examples represent a category. Avoid declaring the inspection system validated from one attractive agreement statistic or from consensus alone.
Correct the method and verify again
Use the disagreement pattern to revise reference photographs, viewing instructions or the decision definition. Target the actual confusing feature instead of repeating general training. Retain the old and revised method versions so the improvement can be traced. Protect reference samples from deterioration and establish who can approve their replacement.
Repeat the assessment using representative examples that are not simply memorized from the training session. Agree release criteria with the quality owner before scoring. Send HANNAI the resolved defect definition and representative images when requesting a process investigation; consistent identification makes machine-side evidence substantially easier to interpret.
Related equipment and next checks
Review the paper cup production line and the related guide to defect classification and containment. The Engineering Notes archive connects these checks with wider machine planning.
Technical reference
- Minitab: attribute agreement analysis — general reference; confirm applicability to the installed equipment.
Frequently asked questions
Does agreement prove the decision is correct?
No. Inspectors can agree with one another while disagreeing with the approved reference.
Can one overall percentage describe every defect?
It can conceal category-specific errors; review the disagreement details.
Should inspectors see their previous answers?
No. Conceal earlier responses during independent repeat assessments.
Can this replace leak testing?
No. A visual decision study addresses the defined appearance task only.
Review the evidence with HANNAI
Send your machine model, component identification and the observations described in this guide. Include the drawing or material reference and any before-and-after measurements so the required next check can be identified.