Art of Lean
Back to Reference
Quality

SPC (Statistical Process Control)

The use of statistical methods — particularly control charts — to monitor process stability, distinguish between common-cause and special-cause variation, and maintain processes in a state of statistical control so that quality is built into the process rather than inspected after the fact.

Japanese

統計的工程管理

tōkeiteki kōtei kanri

statistical process management

Also known as

Statistical Process Control, SPC, Process Control

Definition

Statistical Process Control (SPC) is the application of statistical methods to monitor, control, and improve processes. The core tool is the control chart, which plots process measurements over time against statistically calculated control limits. The fundamental distinction SPC provides is between common-cause variation (inherent to the process and stable over time) and special-cause variation (arising from assignable factors that change the process). A process operating with only common-cause variation is said to be “in statistical control” — its output is predictable within known limits.

Japanese Origin

Tōkeiteki kōtei kanri (統計的工程管理) directly translates the English concept: 統計的 (tōkeiteki, statistical), 工程 (kōtei, process), 管理 (kanri, management/control). Statistical methods were introduced to Japanese industry during the postwar period through multiple channels: the lectures of W. Edwards Deming to Japanese executives and engineers beginning in 1950, the Japanese Union of Scientists and Engineers (JUSE) quality education programs, and the broader adoption of American quality control methods during the reconstruction period.

History

SPC originated with Walter Shewhart at Bell Telephone Laboratories in the 1920s. Shewhart developed the control chart in 1924 and published his foundational work Economic Control of Quality of Manufactured Product in 1931. His key insight was that variation is inherent in every process, and that the practical question is not whether variation exists but whether the variation is stable (common-cause) or unstable (special-cause).

In Japan, JUSE began teaching statistical quality control methods in the late 1940s and early 1950s. Toyota adopted statistical methods as part of its quality management system during this period. The 7 QC Tools — of which the control chart is one — became the standard problem-solving toolkit taught to QC circles and line workers throughout Japanese industry.

At Toyota, SPC is part of the broader quality system but is not the central quality philosophy. Toyota’s primary quality approach is jidoka — building quality into the process through error-proofing (poka-yoke) and automatic detection of abnormalities. SPC complements jidoka by providing a method for monitoring process stability over time, detecting trends and shifts before they produce defects, and diagnosing whether a quality problem is systemic (common-cause) or event-driven (special-cause).

How It Works

The control chart:

  1. Collect data from the process at regular intervals (measurements, counts, proportions)
  2. Calculate the process average and control limits (typically at 3 standard deviations from the mean)
  3. Plot each data point on the chart as it is collected
  4. Evaluate the chart for signals of special-cause variation

Interpreting the chart:

  • In control: All points fall within the control limits with no non-random patterns. The process is stable and predictable. Only common-cause variation is present.
  • Out of control: A point falls outside the control limits, or a non-random pattern appears (trends, runs, cycles). Special-cause variation is present and should be investigated.

The two fundamental actions:

  • When special-cause variation is detected, find and remove the assignable cause. Something changed — a new batch of material, a worn tool, a different operator — and the change must be identified and addressed.
  • When only common-cause variation is present but the process is not capable of meeting specifications, change the process itself. Common-cause variation can only be reduced by fundamentally changing how the process works — different equipment, different materials, different methods.

Key rule: Do not tamper with a process that is in statistical control. Adjusting a stable process in response to individual data points (common-cause variation) actually increases variation rather than reducing it. This is Shewhart’s most counterintuitive but most important contribution.

Common Mistakes

Reacting to every data point. When a process is in control, individual points above or below the average are normal variation — not signals to adjust. Adjusting the process in response to common-cause variation (called “tampering”) makes the process worse, not better.

Setting control limits at specification limits. Control limits are calculated from the process data — they reflect what the process actually does. Specification limits define what the customer requires. These are different things. A process can be in statistical control but not capable of meeting specifications, or vice versa.

Using SPC without understanding the underlying concepts. Plotting points on a chart is not SPC. Understanding the distinction between common-cause and special-cause variation, knowing when to investigate and when to leave the process alone, and knowing how to reduce common-cause variation through process change — this is SPC.

Applying SPC to unstable processes and expecting it to fix them. SPC monitors stability; it does not create it. If a process has fundamental problems (broken equipment, untrained operators, inconsistent materials), those problems must be fixed before SPC can provide useful information.