ABSTRACT
Attribute type data have only two values: pass or fail, go or no-go, and so forth. An example of attribute data can be collected during final visual inspection of a product for cleanliness, chips, cracks, and metal marks. Clear-cut cases of acceptable, rejectable, marginally acceptable, and marginally rejectable products should be identified and photographed by responsible engineers. This is the most critical step for attribute data taking, namely developing precise definitions of conforming and nonconforming product conditions. You want to take human judgment out of attribute data as much as possible. When you are creating sample products as attribute standards for final visual inspection, you, your customer, and/or your subcontractor should all be in agreement. If you are trying to correlate with your customer and/or with your subcontractor, you all should have copies of photographs for these attribute standard products for final visual inspection with descriptive explanations so that there is no controversy when a product lot is rejected.
