How to Build a Control Chart on Mac & Windows (I-MR & X-bar)

A control chart tells you whether a process is stable or whether something has changed. Here's how to build one on Mac & Windows and read the signals.

Statistical process control (SPC) plots a measurement over time against control limits computed from the data itself. Inside the limits, the process is behaving as usual; a point outside, or a non-random pattern, is a signal worth investigating.

Build it

Open your time-ordered data and choose Charts · Control Charts.
For individual measurements, pick an I-MR chart; for subgroups, an X-bar R or X-bar S chart.
Select the measurement column — Stratum computes the center line and the upper and lower control limits.

An I-MR control chart with control limits in Stratum — Center line and control limits, computed from your data.

Read the signals

Control limits are set at ±3 sigma. A point beyond them is the classic alarm, but subtle shifts show up as patterns too. Stratum applies the eight Nelson rules automatically — runs on one side of center, trends, points hugging the limits — and highlights any violations so you don't have to eyeball them.

Out-of-control points flagged by the Nelson rules in Stratum — Nelson-rule violations are flagged automatically.

Common vs special cause. Variation inside the limits is “common cause” — the normal noise of a stable process. A signal points to a “special cause” — something specific changed. The whole point is to react to signals and leave common-cause noise alone.

Beyond I-MR

For count data use p/np or c/u charts; to catch small, slow drifts use CUSUM or EWMA. We explain the ideas in What Is a Control Chart?, and the full set is in Lessons 19–21.

Download Stratum

Frequently asked questions

What control charts can I make on Mac & Windows?

With Stratum: I-MR and X-bar R/S for measurements, p/np and c/u for counts, and CUSUM and EWMA for small drifts — all native, no coding.

What are the Nelson rules?

Eight pattern tests (beyond 3 sigma, runs on one side, trends, alternating points, and more) that flag non-random behavior on a control chart. Stratum applies all eight automatically.

What's the difference between common and special cause variation?

Common cause is the normal random noise of a stable process (inside the control limits). Special cause is a specific, assignable change that produces an out-of-control signal.

How are control limits calculated?

From the data itself — typically the process center plus or minus three sigma, estimated from the within-subgroup or moving-range variation. They are not your spec limits.