Quality Practices and Measurement: Trustworthy Data and Controlled Variation
Measurement you can trust
Precision vs accuracy. Precision is repeatability - how tightly repeated readings cluster. Accuracy is closeness to the true value. A micrometer that reads 1.000 in. five times on a part that is truly 1.010 is precise but not accurate; it needs calibration. Questions swap these words constantly.
Tolerance and inspection. A tolerance is the permitted range around a nominal dimension. Go/no-go gauges check limits quickly; variable instruments (calipers, micrometers) give a number and demand proper technique - zeroing, temperature, cleanliness - plus periodic calibration against traceable standards.
SPC. Control charts separate common-cause (inherent) variation from special-cause variation. Control limits are calculated from process data; specification limits come from the customer. Plotting spec limits on a control chart is a classic trap. Never adjust a process showing only common-cause variation - that is tampering, and it increases variation. Capability indices such as Cpk are only meaningful once the process is in statistical control; an out-of-control process has no stable capability to report.
Problem solving. Use a Pareto chart to find the few defect categories causing most of the loss, a cause-and-effect (fishbone) diagram to brainstorm across people, machine, material, method, measurement, and environment, and 5 Whys to push past symptoms to root cause. Then verify the fix with data, and account for the cost of poor quality - scrap, rework, warranty, inspection.
Rule of thumb: if the chart is stable, ask whether the spread fits inside the tolerance; if it is unstable, find the special cause before computing anything.