From: Sam Devlin, QA Manager To: Plant leadership Subject: Q2 defect-rate review, read the sampling rule first
The raw defect counts below look alarming for Line C, but before anyone reacts: our inspection sampling is NOT uniform across lines or shifts. The defect rate you care about is defects per thousand inspected units, and how many we inspect depends on the sampling rule (separate card). Comparing raw defect counts across lines without normalizing by units inspected is the classic error.
What I need: the line and shift with the highest true defect rate once the sampling rule is applied, and whether the headline Line C is our worst line claim survives normalization.
Units produced and defects found, Q2. Each line runs two shifts (Day, Night). Columns: line; shift; units produced; defects found in the inspected sample.
Line A, Day: 40,000 units; 24 defects. Line A, Night: 40,000 units; 30 defects. Line B, Day: 30,000 units; 18 defects. Line B, Night: 30,000 units; 27 defects. Line C, Day: 50,000 units; 60 defects. Line C, Night: 50,000 units; 45 defects.
The QA team does not inspect every unit. The fraction of produced units inspected depends on the line's automation level and the shift.
Line A: Day 10 percent of units produced; Night 10 percent. Line B: Day 5 percent; Night 15 percent. Line C: Day 20 percent; Night 10 percent.
Defect rate is defined as defects found per 1,000 units inspected (not per 1,000 produced). Inspected units = produced units times sampling rate.
Based on all three Brightforge sources and the defect-rate definition, evaluate each statement, assuming each statement is otherwise accurate.
Five fresh questions every day, your progress tracked, every miss explained. Free with an account.