How this calculation works
Estimate center and spread from the first chosen baseline observations only. Keep those limits fixed while plotting and checking the whole time series, including later monitoring observations.
Sigma estimate = mean consecutive moving range / 1.128; I limits = baseline mean ± 3 sigma; MR UCL = 3.267 × mean MRA worked example
The first ten example observations average 10.2. The final value 16 is monitored against limits derived without using that final observation, avoiding a limit that shifts to accommodate the new point.
Good to know
Selected investigation signals only, not automatic causal diagnosis. Uses range length two and the stated constants. No other run rules or normality certification; zero-range baselines are degenerate.
Reference: NIST: individuals control charts (opens in a new tab) ↗
Keep the observation order
Moving ranges are absolute differences between consecutive measurements. Sorting the observations changes those ranges and destroys the time sequence being investigated.
The baseline count must not exceed the available observations. It selects a historical estimation interval, not a rolling recalculation that moves with every new point.
The process capability cp cpk pp ppk calculator adds compare Cp/Cpk from pooled within-subgroup variation with Pp/Ppk from overall sample spread, preserving off-center and negative capability estimates.
Freeze the reference before monitoring
Later observations are compared with the original baseline center and limits. A large new point must not widen its own limit by entering the baseline calculation automatically.
Both individual limit crossings and moving-range upper-limit crossings are shown. Their simultaneous appearance can reflect related evidence rather than two independent events.
Run rules add signals, not explanations
The optional rule flags the eighth and later consecutive points strictly above or below center. A point exactly on the center line breaks the run under this explicit convention.
Statistical control-chart interpretation still depends on the measurement process, dependence and baseline suitability. Flags suggest investigation; absence of flags does not prove stability or capability.
Use the result with the next measurement
The process capability Cp Cpk Pp Ppk calculator compares spread with specification limits. Control limits describe process behavior; specification limits describe a separate requirement.
Selected investigation signals only, not automatic causal diagnosis. Uses range length two and the stated constants. No other run rules or normality certification; zero-range baselines are degenerate.
A few questions, answered
Do later points update the limits?
No, the initial baseline remains frozen.
Does a center-line point continue a run?
No, it resets the same-side run.
Are trend rules included?
No, only the explicitly selected rules are applied.