Gage Run Chart

Gage Run Chart

The gage run chart shows every single measurement of a gage study: grouped into one panel per part, separated by colour and symbol per operator, against a common horizontal reference line. It computes nothing — it makes visible what a summary statistic compresses away.

In Minitab it lives under Stat → Quality Tools → Gage Study → Gage Run Chart. It is not the run chart — that one shows a single series over time with a median line and four runs tests, and has nothing to do with gages.

Panel per part: Each panel holds every measurement taken on one part. Inside the panel the operators sit in contiguous blocks, their repeat measurements connected by a line.

Reference line: The mean of all measurements by default. A custom value can be entered — the known value of a standard, say, or the centre of the tolerance band.

Data & operation

• Long/tidy layout with two required columns: parts and measurement data. Each row is one individual measurement.
• Optional operators column. Without it a single series per part is drawn — useful for single-operator studies or automated measuring systems.
• Row order in the worksheet defines the repeat. There is deliberately no separate trial column.
• Parts are ordered numerically when every label is a number (3, 4, 10, 15), alphabetically otherwise.
• Operators keep the order of their first appearance in the worksheet, so a deliberately chosen order survives.

With many parts the display wraps into several rows. All rows share one y-scale and one reference line and are therefore directly comparable. "Parts per row" controls the wrap.

The five free-text fields under "Gage information" document the study (gage, date, reporter, tolerance, misc) and appear above the chart. They are stored with the project.

Interpretation

• One operator sits consistently high or low versus the others: operator bias — different reading habits, fixturing, or reading of the inspection instruction.
• One operator scatters visibly wider: poor repeatability for that operator. The lines connecting repeat measurements show this directly.
• Large vertical jumps inside one panel: repeat measurements on the same part disagree — the measuring system is not reproducible there.
• Panels differ a lot, within-panel scatter is small: the normal picture of a usable gage study — part-to-part variation dominates.
• All panels sit close together: the parts do not span the process range. The study cannot meaningfully assess the measuring system that way.

The chart yields no numbers. For %GRR, repeatability and reproducibility run a gage study alongside — type 2 for variable, type 5 for attribute characteristics. The gage run chart shows why a number came out the way it did.

Pitfalls

• Name confusion. German Minitab calls the gage run chart "Messverlaufsdiagramm" and the run chart "Verlaufsdiagramm". If you are after the runs tests, use the run chart module.
• Unbalanced design. A part missing for one operator is reported as a warning. The chart stays readable, but the panels are no longer directly comparable.
• Too many parts per row. Beyond roughly ten panels per row the points become indistinguishable. Reduce the wrap rather than widening the chart.
• Attribute data. With 0/1 results almost all points sit at two heights — which is exactly what makes the few deviants stand out. The mean as a reference line, however, is then barely interpretable.
• Worksheet order. Because row order defines the repeat, sorting the worksheet changes the connecting lines.

Examples

This module ships with the following example datasets — load any of them in the app with a single click.

Available in the following cycles

• DMAIC: Measure
• DMADV: Measure
• 8D: D2 — Describe the Problem