The Discipline of Control: From Shewhart to Sakurako
経験に対抗するための道具
物語の中で、桜子は「Machineではない」と判断しました。
それは直感ではありません。
対立でもありません。
統計的手続きに従った結果です。
管理図は単なるグラフではありません。
それは、工程に手を加えてよいのか、
それとも触れてはいけないのかを示す境界線です。
品質問題の多くは、不良そのものよりも
「不安に基づく調整」によって悪化します。
本章では、管理図の原理とその歴史を整理します。
経験や権威に対抗するために、
なぜ統計が必要なのか。
その背景を確認します。
1. What Is a Control Chart?
In the story, Sakurako’s confidence that the issue was not primarily “Machine-related” came from her disciplined use of the Control Chart.
A Control Chart visualizes the behavior of a process over time by adding a Center Line (average) and statistically calculated Control Limits (UCL/LCL). Its primary purpose is to distinguish between Common Cause Variation and Special Cause Variation.
• Common Cause Variation: The inherent, unavoidable variation present in a stable system.
• Special Cause Variation: A statistical signal indicating the presence of a specific disturbance, such as raw material issues, operator error, equipment malfunction, or environmental changes.
The breaches of control limits and sustained upward trends Sakurako observed were statistical signals suggesting that the process was not in a state of control.
⸻
2. The History: Born at Bell Labs
The history of the Control Chart is closely tied to the origins of modern Quality Control.
The Revolution of Walter Shewhart (1920s)
In 1924, Walter Shewhart, a physicist and statistician at Bell Telephone Laboratories, proposed the concept of the Control Chart.
At the time, it was common practice to adjust machinery whenever a defect appeared. However, Shewhart recognized that unnecessary adjustments to a stable process introduce additional variation—a phenomenon later described as tampering or over-adjustment.
Shewhart formalized the concept of a “State of Statistical Control”: a condition in which only Common Cause Variation is present. In such a state, the process should not be adjusted unless statistical evidence indicates a Special Cause.
This principle marked a fundamental shift in industrial thinking.
Dr. Deming and the Legacy in Japan
As Sakurako wondered, “How would Dr. Deming interpret this?”, Dr. W. Edwards Deming was one of the key figures who promoted Shewhart’s ideas internationally.
After World War II, Deming was invited to Japan, where he trained engineers and executives in Statistical Quality Control (SQC). His teachings influenced the broader quality movement in Japan, which later intersected with production systems such as the Toyota Production System (TPS).
The emphasis on statistical thinking, system stability, and management responsibility became foundational elements of Japan’s postwar industrial development.
⸻
3. How the X̄–R Control Chart Works
The X̄–R (X-bar and R) Control Chart used by Sakurako remains one of the most widely applied tools in manufacturing environments.
• X̄ (Mean) Chart: Monitors whether the central tendency of the process has shifted.
• R (Range) Chart: Monitors the magnitude of variation within each subgroup.
By interpreting these two charts together, one can determine whether a shift in the process average has occurred, whether within-subgroup variability has increased, or whether the process has become unstable.
In Sakurako’s case, the analysis suggested that the process was being destabilized when material lots changed. The issue was not merely a simple dimensional shift; it involved a loss of statistical stability.
⸻
Summary
The Control Chart serves as a structured method for translating the “voice of the process” into statistical evidence.
It does not assign blame.
It does not rely on authority.
It distinguishes variation types.
In environments where decisions are often driven by experience or hierarchy, statistical tools such as the Control Chart can serve as one of the most effective means of breaking through assumption-based reasoning.
Sakurako’s breakthrough did not come from confrontation.
It came from disciplined interpretation.
境界を守るということ
管理図の核心は、異常を見つけることではありません。
境界を守ることです。
すべての変動が異常ではありません。
すべての逸脱が罪でもありません。
統計は人を責めません。
変動を分類するだけです。
シューハートは「過剰調整」の危険を示しました。
デミングは「システムの責任」を強調しました。
桜子が行ったことも同じです。
仮説を感情で選ばない。
順序を守る。
証拠が示す範囲を超えて断定しない。
Man。
Machine。
そして次は Materials。
分析はまだ終わっていません。
しかし一つだけ確かなことがあります。
統計は沈黙を守ります。
だからこそ、強いのです。
