Attribute sampling is based on the confidence level and quality level.  For example, to be 95% confident of a minimum 95% yield, the minimum sample size would be 58.4 units, which is rounded up to 59 units.

Variable sampling is based on the confidence level, quality level and ‘k’ factor of the process.  For example, to be 95% confident of a minimum 95% yield with a ‘k’ factor for this process of 3.33, the sample size would be 8 units.  As it is highly unlikely that 8 samples will produce a ‘k’ factor of exactly 3.33, it is recommended to slightly increase the sample size.  For additional information on sample size determination, refer to “Introduction to Statistical Quality Control” 5 .

Additionally, the confidence and yield requirements generally increase when associated with process parameters that are linked to failure modes that are associated with increased risk.