Statistical Process Control — SPC
Statistical Process Control (SPC) · Also known as: SPC, statistical quality control, process control charting, Shewhart control
Statistical Process Control (SPC) is a data-driven quality method that uses statistical techniques — primarily control charts — to monitor a manufacturing or service process over time. By distinguishing natural process variation (common cause) from unusual, actionable variation (special cause), SPC enables practitioners to maintain processes in a stable, predictable state and to detect problems early, before defective output reaches customers.
Read the full method
Sign in with a free account to read this section.
Method map
The neighbourhood of related methods — select a node to explore.
+40 more
When to use it
Use SPC when you need to monitor a repetitive process over time and distinguish random variation from real changes that require action. It is the standard tool for ongoing quality surveillance in manufacturing, healthcare, and service operations. SPC requires sequentially ordered, time-stamped measurement data; it is not suitable for cross-sectional datasets without a time dimension. Do not use SPC as a substitute for process improvement — it tells you when the process has changed, not how to make it fundamentally better. If the process is not yet sufficiently stable to estimate control limits reliably, short-run SPC or pre-control charts may be needed. Avoid using specification limits as control limits; doing so confuses product acceptance with process monitoring.
Strengths & limitations
- Provides real-time, data-driven signals that separate noise from genuine process shifts, reducing unnecessary operator interventions.
- Control charts are visually intuitive and can be used on the shop floor by operators without advanced statistical training.
- Early detection of process deterioration prevents defects from accumulating before they are discovered at final inspection.
- Establishes a documented, quantitative baseline of process behaviour useful for audits, customer requirements (e.g., IATF 16949), and process improvement projects.
- Applicable to both continuous measurements (X-bar, I-MR charts) and attribute data (p, np, c, u charts), covering most industrial measurement situations.
- Requires a stable baseline period of adequate length (20–30 subgroups) to compute reliable control limits; short or non-representative baseline data produce misleading limits.
- Control charts are retrospective monitors — they detect changes after they have occurred, not before. Proactive prevention requires designed experiments or failure analysis.
- Standard Shewhart charts assume approximate normality and independence of observations; autocorrelated data (common in continuous chemical processes) violate these assumptions and require specialised charts (e.g., EWMA, CUSUM).
- SPC addresses process stability but says nothing about whether the process is capable of meeting customer specifications — capability indices (Cp, Cpk) must be computed separately.
Frequently asked
What is the difference between control limits and specification limits?
Control limits are computed from actual process data (mean ± 3 standard deviations of the plotted statistic) and describe what the process naturally produces. Specification limits are set by the customer or engineer and describe what the product must achieve. A process can be in control but out of specification (stable but producing bad parts) or within specification but out of control (meeting requirements today but trending toward failure). Always keep them separate.
How do I choose the right control chart?
Start with the data type: if measurements are continuous and collected in subgroups of n ≥ 2, use X-bar and R (or X-bar and S) charts. If measurements are individual readings (n = 1), use an Individuals and Moving Range (I-MR) chart. For attribute data — defective items — use a p-chart (proportion defective) or np-chart (count defective) with constant or variable subgroup size. For defects per unit, use a c-chart (constant area) or u-chart (variable area).
My process data are autocorrelated — can I still use SPC?
Standard Shewhart charts assume independent observations. Autocorrelated data (e.g., temperature readings every minute in a continuous process) produce inflated false-alarm rates on Shewhart charts. Options include using EWMA or CUSUM charts (which are more sensitive to small shifts and handle autocorrelation better), fitting a time-series model to the residuals and charting those, or reducing sampling frequency to approach independence.
How many data points do I need before setting up control limits?
The conventional guideline is at least 20 to 30 rational subgroups collected under stable conditions. Fewer than 20 subgroups produce unreliable estimates of process variation, leading to control limits that are either too wide (missing real signals) or too narrow (generating excessive false alarms). If a stable baseline cannot be obtained, consider using short-run SPC techniques or provisional limits with planned recalculation.
What is the difference between SPC and Six Sigma?
SPC is a monitoring tool: it tells you whether your current process is stable and when it has changed. Six Sigma (particularly the DMAIC methodology) is an improvement framework: it diagnoses the root causes of chronic defects and redesigns the process to achieve a substantially lower defect rate. In practice, SPC is used in the Control phase of DMAIC to sustain gains after a Six Sigma project has improved the process.
Sources
- Shewhart, W. A. (1931). Economic Control of Quality of Manufactured Product. Van Nostrand. ISBN: 978-0873890762
- Montgomery, D. C. (2020). Introduction to Statistical Quality Control (8th ed.). Wiley. ISBN: 978-1119657118
How to cite this page
ScholarGate. (2026, June 3). Statistical Process Control (SPC). ScholarGate. https://scholargate.app/en/experimental-design/statistical-process-control
Which method?
Set this method beside its closest kin and read them side by side — the library lays the books on the table; the choice is yours.
- Control chartExperimental design↔ compare
- Design of experimentsExperimental design↔ compare
- Failure Mode and Effects AnalysisExperimental design↔ compare
- Process Capability AnalysisStatistics↔ compare
- Quality Function DeploymentExperimental design↔ compare
- Six Sigma DMAICQuality Management↔ compare