Quality Management Fundamentals
品質管理基礎
Production Management Operations Grade 3 生産管理オペレーション (BC検定) · ビジネス・キャリア検定 生産管理オペレーション3級 · Written by Tokutei Pass based on the official study texts; not copied verbatim.
1. 品質の概念(ねらいの品質とできばえの品質、真の特性と代用特性)と4M・PDCA
1. Quality Concepts (Quality of Design vs. Conformance, True vs. Substitute Characteristics) and 4M / PDCA
製品の品質は2つの段階で捉えます。市場ニーズや目標コストから企画・設計段階で狙い定める目標品質を「ねらいの品質(設計品質)」と呼び、設計図面どおりに実際に製造された結果の品質を「できばえの品質(製造品質・適合品質)」と呼びます。
Product quality is understood at two stages. The target quality aimed for at the planning and design stage based on market needs and target cost is called "Quality of Design (target quality / Nerai no hinshitsu)," whereas the actual resulting quality manufactured according to design drawings is called "Quality of Conformance (manufacturing quality / Dekibae no hinshitsu)."
また、消費者が本来求める「使いやすい・壊れにくい・書きやすい」といった要求品質を「真の特性」と呼びますが、そのままでは測定・管理しにくいため、寸法・硬さ・引張強さなど数値で測れる「代用特性(品質特性)」に置き換えて管理します。
Furthermore, the inherent quality requirements consumers truly desire—such as "easy to use, durable, smooth to write with"—are called "True Characteristics," but because they cannot be measured directly, they are translated into quantifiable, measurable values like dimensions, hardness, or tensile strength called "Substitute Characteristics (quality characteristics)."
製造現場の品質管理では、「品質は検査ではなく工程で作り込む」「後工程はお客様(不良品を受け取らない・作らない・流さない)」を鉄則とし、作業者(Man)・機械(Machine)・材料(Material)・方法(Method)の「4M」のばらつきを抑えます。
In shop-floor quality control, the ironclad rules are "Quality must be built into the process, not inspected in" and "The next process is the customer (do not accept, do not make, and do not pass on defects)," controlling variation across the "4Ms": Man (operators), Machine (equipment), Material, and Method.
そして日常は作業標準に従う「SDCA(標準S→実施D→確認C→処置A)」で品質を維持し、さらなる向上や問題解決には「PDCA(計画P→実施D→確認C→処置A)」を回して改善をスパイラルアップさせます。
Daily operations maintain quality by running the "SDCA cycle (Standardize → Do → Check → Act)" following work standards, while further improvements and problem solving run the "PDCA cycle (Plan → Do → Check → Act)" to spiral quality upward.
Key points
- ねらいの品質=企画・設計で定める「設計品質」/できばえの品質=設計どおりに作られたかの「製造品質(適合品質)」。
Quality of Design = target quality set during planning and design / Quality of Conformance = manufacturing quality reflecting how faithfully the product matches the design. - 真の特性=顧客が求める本来の要求(壊れにくい等)/代用特性=真の特性を保証するため数値化した測定値(引張強さ・寸法等)。
True Characteristic = the customer's inherent requirement (e.g., durability) / Substitute Characteristic = quantifiable measured values (tensile strength, dimensions) set to assure the true characteristic. - 4M(人・機械・材料・方法)のばらつきを管理し、維持の「SDCA」と改善の「PDCA」を回して工程で品質を作り込む。
Control variation in the 4Ms (Man, Machine, Material, Method) and rotate "SDCA" for maintenance and "PDCA" for improvement to build quality into the process.
| Word | Meaning |
|---|---|
| ねらいの品質(設計品質)とできばえの品質(製造品質) | Quality of Design (target quality) and Quality of Conformance (manufacturing quality) |
| 真の特性と代用特性 | True Characteristics and Substitute Characteristics |
| 後工程はお客様 | the next process is the customer |
| PDCAサイクルとSDCAサイクル | PDCA cycle and SDCA cycle |
2. データの種類(計量値・計数値)と基本統計量(平均・中央値・範囲・分散・標準偏差)
2. Types of Data (Continuous / Discrete) and Basic Statistics (Mean, Median, Range, Variance, Standard Deviation)
品質データは2種類に分かれます。長さ(mm)・重さ(g)・温度(℃)・時間(秒)など連続する値を測定器で測るデータを「計量値」と呼び、その分布は主に「正規分布」に従います。
Quality data falls into two types. Continuous values measured with instruments—such as length (mm), weight (g), temperature (℃), or time (seconds)—are called "Continuous Data (Variables / Keiryouchi)," and their distribution primarily follows the "Normal Distribution."
これに対し、1個、2個と数える不連続なデータを「計数値」と呼びます。計数値のうち、不適合品(不良品)の個数や不適合品率は「二項分布」に従い、布の傷や塗装ブツなどの「欠点数(不適合数)」は「ポアソン分布」に従います。
In contrast, discrete data counted in whole units (1 piece, 2 pieces) is called "Discrete Data (Attributes / Keisuuchi)." Within discrete data, the number or proportion of nonconforming (defective) units follows the "Binomial Distribution," whereas the "number of defects (nonconformities)" such as cloth scratches or paint specks follows the "Poisson Distribution."
データの中心位置(代表値)を表す統計量には、全データの合計をデータ数nで割った「平均値(xバー)」、データを小さい順に並べたときちょうど真ん中にくる「中央値(メディアン:データ数が偶数なら中央2つの平均)」、最も出現回数が多い「最頻値(モード)」があります。
Statistics representing the central location (representative value) of data include the "Mean (Average, x-bar)," found by dividing the sum of all data by sample size n; the "Median," the middle value when data is arranged in ascending order (or average of the two middle values if n is even); and the "Mode," the most frequently occurring value.
一方、データのばらつきの大きさを表す統計量には、「範囲(R=最大値-最小値)」のほか、各データから平均値を引いて2乗した和である「偏差平方和(S)」、Sを自由度(n-1)で割った「不偏分散(V=S÷(n-1))」、分散の正の平方根をとった「標本標準偏差(s=√V)」があります。
Meanwhile, statistics representing data dispersion (spread) include the "Range (R = Maximum - Minimum)"; the "Sum of Squares of Deviations (S)," found by subtracting the mean from each data point, squaring the differences, and summing them; the "Unbiased Variance (V = S ÷ (n - 1))," dividing S by degrees of freedom (n - 1); and the "Sample Standard Deviation (s = √V)," the positive square root of variance.
Key points
- 計量値(長さ・重さ・温度など連続量)→正規分布/計数値(個数を数える離散量)→不良品数は二項分布、欠点数はポアソン分布。
Continuous Data (length, weight, temperature) → Normal Distribution / Discrete Data (counted quantities) → Number of defective units follows Binomial Distribution; number of defects follows Poisson Distribution. - 代表値:平均値=総和÷データ数n/中央値(メディアン)=小さい順に並べた中央の値/最頻値(モード)=最も度数が多い値。
Central tendency: Mean = Sum ÷ n / Median = middle value when sorted in order / Mode = most frequent value. - ばらつきの尺度:範囲R=最大値-最小値/不偏分散V=偏差平方和S÷(n-1)/標準偏差s=√V。
Dispersion measures: Range R = Max - Min / Unbiased Variance V = Sum of Squares S ÷ (n - 1) / Standard Deviation s = √V. - 【例】5個の測定値(7, 9, 11, 11, 12mm) → 平均値=(7+9+11+11+12)÷5=10mm、中央値(メジアン)=11mm、最頻値(モード)=11mm、範囲R=12-7=5mm
Example: 5 measurements (7, 9, 11, 11, 12 mm) → Mean = (7 + 9 + 11 + 11 + 12) ÷ 5 = 10 mm, Median = 11 mm, Mode = 11 mm, Range R = 12 − 7 = 5 mm - 【例】同じ5個のデータ(平均10mm) → 偏差平方和S=(7-10)²+(9-10)²+(11-10)²+(11-10)²+(12-10)²=16、不偏分散V=16÷(5-1)=4、標本標準偏差s=√4=2mm
Example: Same 5 data points (mean 10 mm) → Sum of squares S = (7 − 10)² + (9 − 10)² + (11 − 10)² + (11 − 10)² + (12 − 10)² = 16, Unbiased variance V = 16 ÷ (5 − 1) = 4, Sample standard deviation s = √4 = 2 mm
| Word | Meaning |
|---|---|
| 計量値(正規分布)と計数値(二項・ポアソン分布) | continuous data (Normal distribution) and discrete data (Binomial / Poisson distribution) |
| 平均値・中央値(メディアン)・最頻値(モード) | Mean, Median, and Mode |
| 範囲(R)と偏差平方和(S) | Range (R) and Sum of Squares of Deviations (S) |
| 不偏分散(V)と標準偏差(s) | Unbiased Variance (V) and Standard Deviation (s) |
3. 統計図表(グラフ)の特徴と使い分け(レーダーチャート・Zチャート等)
3. Characteristics and Selection of Statistical Charts (Radar Chart, Z-Chart, etc.)
数値データを視覚的にわかりやすく表現するため、目的に応じて適切な統計図表(グラフ)を使い分けます。数量の「大小比較」には「棒グラフ」、時間の経過に伴う「増減変化(推移)」には「折れ線グラフ」を用います。
To present numerical data visually and clearly, statistical charts are selected according to purpose. "Bar Charts" are used for comparing magnitude across categories, while "Line Charts" are used for showing upward or downward trends over time.
全体に対する各項目の「内訳・構成比(割合)」を示すには「円グラフ」や「帯グラフ」が適しており、特に帯グラフを縦に複数並べると、年次ごとや工場ごとの構成比の推移・比較が一目でわかります。
To show the "breakdown or composition ratio (%)" of each item relative to the whole, "Pie Charts" and "Band (Stacked Bar) Charts" are suitable; stacking multiple band charts vertically makes year-by-year or plant-by-plant composition changes immediately clear.
また、複数の評価項目(例:P・Q・C・D・S・Mの6項目や作業者の多能工スキル)を中心から放射状の軸にとり、全体のバランスや弱点をクモの巣状の多角形で示すグラフを「レーダーチャート(クモの巣グラフ)」と呼びます。
A chart that plots multiple evaluation items (e.g., the 6 items P, Q, C, D, S, M or worker multi-skill levels) on axes radiating from a center to display overall balance and weaknesses as a web-like polygon is called a "Radar Chart (Spiderweb Chart)."
さらに、売上高や生産高について①月次(単月)高、②月次累計高、③移動年計(直近12か月分の合計)の3本の折れ線を1つの図に描き、アルファベットの「Z」の形で短期変動と長期傾向を同時に捉える図を「Zチャート」と呼びます。
Furthermore, a chart that plots three line graphs of sales or production on a single chart—① Monthly total, ② Cumulative monthly total, and ③ Moving annual total (sum of the most recent 12 months)—forming the letter "Z" to capture short-term fluctuations and long-term trends simultaneously is called a "Z-Chart."
Key points
- 棒グラフ=大小の比較/折れ線グラフ=時系列の推移・変化/円・帯グラフ=内訳・構成比(割合)の把握。
Bar Chart = comparing sizes / Line Chart = time-series trends and changes / Pie & Band Charts = grasping breakdown and composition ratios (%). - レーダーチャート(クモの巣グラフ)=複数の評価項目の全体バランス・長所と短所を比較するのに適する。
Radar Chart (Spiderweb Chart) = ideal for comparing overall balance, strengths, and weaknesses across multiple evaluation criteria. - Zチャート=「月次・累計・移動年計」の3本の折れ線で、季節変動を除いた長期トレンドと目標達成度を把握する。
Z-Chart = uses 3 lines ("Monthly, Cumulative, Moving Annual Total") to grasp long-term trends free of seasonal fluctuations as well as target achievement.
| Word | Meaning |
|---|---|
| 棒グラフ・折れ線グラフ・円グラフ・帯グラフ | bar chart, line chart, pie chart, and band chart |
| レーダーチャート(クモの巣グラフ) | Radar Chart (spiderweb chart) |
| Zチャート(月次・累計・移動年計) | Z-Chart (monthly, cumulative, and moving annual total) |
4. QC七つ道具①(パレート図・特性要因図・チェックシート・層別)
4. The QC 7 Tools ① (Pareto Chart, Cause-and-Effect Diagram, Check Sheet, Stratification)
現場の数値データを整理・分析する基本手法が「QC七つ道具」です。1つ目の「パレート図」は、不良項目や原因別に件数・損失金額を大きい順に並べた棒グラフと、その累積比率(%)を示す折れ線グラフを組み合わせ、「どの重点問題から手をつけるべきか」を明確にする図です。
The foundational toolkit for organizing and analyzing shop-floor numerical data is the "QC 7 Tools." First, the "Pareto Chart" combines a bar graph sorting defect items or causes in descending order of frequency or loss amount with a line graph showing the cumulative percentage (%), clarifying "which vital priority problem should be tackled first."
2つ目の「特性要因図」は、結果である「特性(不良など)」を背骨の右端に置き、そこに影響する「要因(4Mなど)」を大骨・中骨・小骨へと体系的に整理した図で、形が魚の骨に似ていることから「フィッシュボーンチャート(魚の骨図)」とも呼ばれます。
Second, the "Cause-and-Effect Diagram" places the resulting "characteristic (such as a defect)" at the right end of a central spine and systematically branches the contributing "factors (such as the 4Ms)" into large, medium, and small bones; because it resembles a fish skeleton, it is also called a "Fishbone Chart (Ishikawa Diagram)."
3つ目の「チェックシート」は、データ収集を簡単にし集計や点検の忘れを防ぐため、あらかじめ項目を印刷して記録・チェックできるようにした表(記録用と点検確認用)です。
Third, a "Check Sheet" is a pre-formatted table printed with check items so operators can easily record data or verify inspection points without omission (divided into recording check sheets and confirmation/inspection check sheets).
4つ目の「層別」は、収集したデータをそのまま1つにまとめるのではなく、材料メーカー別・機械(号機)別・作業者別・昼夜勤別など共通の特徴(層)ごとにグループ分けして比較し、ばらつきの真の原因をあぶり出す考え方です(ヒストグラムや散布図と組み合わせて使います)。
Fourth, "Stratification" is the practice of dividing collected data into subgroups (strata) sharing common attributes—such as by material supplier, machine number, operator, or day/night shift—rather than lumping all data together, comparing them to uncover the true source of variation (used together with histograms and scatter diagrams).
Key points
- パレート図=項目別の棒グラフ(降順)+累積比率の折れ線グラフで、最優先の重点改善テーマを特定する。
Pareto Chart = descending bar chart by category + cumulative percentage line chart, used to identify the top-priority improvement theme. - 特性要因図(魚の骨図)=結果(特性)と原因(4Mなどの要因)の因果関係を体系的に整理する。
Cause-and-Effect Diagram (Fishbone Chart) = systematically organizes cause-and-effect relationships between a result (characteristic) and its causes (4M factors). - 層別=データを機械別・材料ロット別・作業者別などのグループ(層)に分けて比較し、違いの原因を見つける。
Stratification = separating data into groups (strata) by machine, material lot, or operator to compare and pinpoint the cause of differences. - 【例】不良総数100件(第1位:寸法不良50件、第2位:外観傷30件、その他20件) → パレート図の上位2項目の累積比率=(50+30)÷100×100=80%
Example: 100 total defects (1st: dimensional defect 50, 2nd: surface scratch 30, others: 20) → Cumulative percentage of top 2 items on Pareto chart = (50 + 30) ÷ 100 × 100 = 80%
| Word | Meaning |
|---|---|
| パレート図(重点指向・ABC分析) | Pareto Chart (priority orientation and ABC analysis) |
| 特性要因図(魚の骨図・フィッシュボーン) | Cause-and-Effect Diagram (Fishbone Chart) |
| チェックシート(記録用・点検用) | Check Sheet (for data recording and for inspection) |
| 層別(グループ分け比較) | Stratification (subgroup comparison) |
5. QC七つ道具②(ヒストグラムの形状・散布図・管理図)
5. The QC 7 Tools ② (Histogram Shapes, Scatter Diagram, Control Chart)
5つ目の「ヒストグラム(度数分布図)」は、計量値データの範囲をいくつかの区間(階級)に分け、各区間に入るデータ数(度数)を柱状グラフにして、中心位置・ばらつきの形・規格値との余裕を見る図です。
Fifth, a "Histogram (frequency distribution chart)" divides the range of continuous data into intervals (classes) and plots the number of data points (frequency) in each class as vertical bars to examine central location, spread shape, and margin relative to specification limits.
ヒストグラムの形状には、正常な左右対称の「一般型(正規型)」のほか、区間幅が測定目盛りの整数倍でないときに出る「櫛歯型」、規格外を選別除去したときに片側が切り立つ「絶壁型」があります。
Regarding histogram shapes, besides the normal symmetrical "Bell-Shaped (General/Normal) Type," there is the "Comb-Shaped (Multimodal Teeth) Type" caused when class width is not an integer multiple of the measurement scale, and the "Cliff-Like (Truncated) Type" sharply cut off on one side when nonconforming items have been screened out.
さらに、平均値が少しずつ異なる複数のロットが混ざった「高原型」、異質なデータが少量混入したり測定ミスがあったりした「離れ小島型」、平均値の異なる2台の機械や2社の材料が混ざった「二山型(双峰型)」があり、高原型や二山型は「層別」で原因を分けます。
Additionally, the "Plateau Type" occurs when multiple lots with slightly different means are mixed; the "Isolated Island Type" occurs when a small amount of abnormal data or measurement error is mixed in; and the "Twin-Peak (Bimodal) Type" occurs when outputs from two machines or two material suppliers with different means are mixed (Plateau and Twin-Peak types require "Stratification").
6つ目の「散布図」は、対になった2つの変数(例:焼成温度xと硬度y)を縦軸・横軸にとって打点し、両者の「相関関係(正の相関=右肩上がり、負の相関=右肩下がり、無相関)」を調べる図です。
Sixth, a "Scatter Diagram" plots paired variables (e.g., firing temperature x and hardness y) on horizontal and vertical axes to examine their "correlation (positive correlation = upward slope to the right, negative correlation = downward slope to the right, or no correlation)."
7つ目の「管理図」は、工程が安定状態にあるか異常原因が働いたかを時系列で判定するため、「中心線(CL)」の上下に標準偏差の3倍(±3σ)の幅で「上方管理限界線(UCL)」と「下方管理限界線(LCL)」を引いた図です。
Seventh, a "Control Chart" evaluates over time whether a process is in a stable state or affected by assignable (special) causes by drawing an "Upper Control Limit (UCL)" and "Lower Control Limit (LCL)" at plus and minus 3 standard deviations (±3σ) above and below the "Center Line (CL)."
管理図では、打点が管理限界線の外に出た場合だけでなく、限界線の内側にあっても中心線の片側に連続して並ぶ「連(7〜9点以上)」や、連続上昇・下降する「傾向(トレンド)」、「周期性」が見られる場合も工程異常と判定します。
On a control chart, a process is judged abnormal not only when a point falls outside the control limits, but also when points inside the limits exhibit a "Run (7 to 9 or more consecutive points on one side of the center line)," a continuous upward or downward "Trend," or "Periodicity (cycles)."

Key points
- ヒストグラムの形状:絶壁型=規格外の選別除去/二山型=平均の異なる2つの機械・材料の混在(層別が必要)/離れ小島型=異質データの少数混入・測定ミス。
Histogram shapes: Cliff-like = 100% screening removal of out-of-spec items / Twin-peak = mixture of 2 machines or materials with different means (needs stratification) / Isolated island = minor inclusion of abnormal data or measurement error. - 散布図=2変数の相関を見る図(xが増えるとyも増える右肩上がり=正の相関、yが減る右肩下がり=負の相関)。
Scatter Diagram = examines correlation between 2 variables (y increases as x increases = positive correlation; y decreases as x increases = negative correlation). - 管理図=CL(中心線)±3σにUCL・LCLを引き、管理限界線外の点や、限界内でも「連(片側連続)・傾向(連続上昇/下降)」があれば工程異常と判定する。
Control Chart = draws UCL and LCL at CL ± 3σ; judges the process abnormal if points fall outside limits or show a "Run" or "Trend" even within limits.
| Word | Meaning |
|---|---|
| ヒストグラム(絶壁型・二山型・離れ小島型・櫛歯型) | Histogram (cliff-like, twin-peak, isolated island, comb-shaped) |
| 散布図(正の相関・負の相関) | Scatter Diagram (positive correlation and negative correlation) |
| 管理図(UCL・CL・LCL:3シグマ法) | Control Chart (UCL, CL, LCL: 3-sigma method) |
| 管理図の異常判定(限界線外・連・傾向) | control chart abnormality rules (out-of-limits, Run, Trend) |
6. 検査の分類と抜取検査(OC曲線・生産者危険α・消費者危険β)
6. Classification of Inspections and Sampling Inspection (OC Curve, Producer's Risk α, Consumer's Risk β)
検査とは、品物を測定・試験した結果を判定基準と比較して、個々の品物の「良品・不良品」またはロットの「合格・不合格」を判定することです。段階別には受入(購入)検査・工程間(中間)検査・最終(出荷)検査があり、方法別には破壊検査・非破壊検査・人間の五感で判定する「官能検査」があります。
Inspection is the act of measuring or testing items and comparing the results against acceptance criteria to judge whether individual items are "conforming (good) or nonconforming (defective)" or whether a lot is "accepted or rejected." By stage, there are receiving, in-process, and final inspections; by method, destructive, non-destructive, and "Sensory Inspection" using the human five senses.
また対象数により「全数検査」と「抜取検査」に分かれます。不良品が1個でも混ざると人命に関わる重要保安部品や高価な単品製品は「全数検査」を行いますが、引張破壊試験などの「破壊検査」や大量のネジ、連続流体・粉体などは全数検査が不可能なため必然的に「抜取検査」となります。
By sample coverage, inspection is divided into "100% (Total) Inspection" and "Sampling Inspection." Critical safety parts where even one defect threatens human life or expensive one-off products require "100% Inspection," whereas "Destructive Testing" (like tensile breaking tests), high-volume cheap screws, and continuous fluids or powders cannot undergo 100% inspection and necessitate "Sampling Inspection."
抜取検査では、ロットからランダムに抜き取ったサンプルのみを調べるため、確率的に2種類の判定ミスが避けられません。ロットの真の不良率とロットが合格する確率の関係を示した曲線を「OC曲線(検査特性曲線)」と呼びます。
Because sampling inspection examines only random samples drawn from a lot, two types of probabilistic judgment errors are unavoidable. The curve plotting the relationship between a lot's true fraction defective and its probability of acceptance is called the "OC Curve (Operating Characteristic Curve)."
第1の誤りは、本来は良いロット(合格させるべき品質水準AQLのロット)なのに、たまたま不良品が多く抜き取られて不合格にしてしまう確率で、「生産者危険(α:第1種の誤り・あわてものの誤り、通常約5%)」と呼びます。
The Type I error is the probability that a good lot (at Acceptable Quality Level, AQL) happens to yield many defective samples and is mistakenly rejected; this is called "Producer's Risk (α: Type I error / 'hasty error,' typically set around 5%)."
第2の誤りは、本来は悪いロット(不合格にすべき限界品質LQのロット)なのに、たまたま良品ばかりが抜き取られて合格にしてしまう確率で、「消費者危険(β:第2種の誤り・ぼんやりものの誤り、通常約10%)」と呼びます。
The Type II error is the probability that a bad lot (at Limiting Quality, LQ, which should be rejected) happens to yield mostly good samples and is mistakenly accepted; this is called "Consumer's Risk (β: Type II error / 'careless error,' typically set around 10%)."
Key points
- 破壊検査や大量低単価品・液体粉体は「抜取検査」しかできない/人命に関わる重要部品は「全数検査」。
Destructive tests, high-volume low-cost items, and liquids/powders can only use "Sampling Inspection" / Life-critical safety components require "100% Inspection." - 生産者危険(α:あわてものの誤り)=良いロットを誤って「不合格」にしてしまい、生産者が損をする確率。
Producer's Risk (α: hasty error) = probability of mistakenly "rejecting" a good lot, harming the producer. - 消費者危険(β:ぼんやりものの誤り)=悪いロットを誤って「合格」にしてしまい、消費者が損をする確率。
Consumer's Risk (β: careless error) = probability of mistakenly "accepting" a bad lot, harming the consumer.
| Word | Meaning |
|---|---|
| 全数検査と抜取検査(破壊検査・官能検査) | 100% inspection and sampling inspection (destructive and sensory inspection) |
| OC曲線(検査特性曲線) | OC Curve (Operating Characteristic Curve) |
| 生産者危険(α:あわてものの誤り) | Producer's Risk (α: Type I / hasty error) |
| 消費者危険(β:ぼんやりものの誤り) | Consumer's Risk (β: Type II / careless error) |
7. QCストーリーと品質保証・製造物責任法(PL法)
7. The QC Story, Quality Assurance, and the Product Liability Act (PL Law)
品質改善や問題解決を事実データに基づいて進める標準手順を「QCストーリー」と呼びます。①テーマの選定→②現状の把握と目標の設定→③活動計画の作成→④要因の解析(真因の特定)→⑤対策の検討と実施→⑥効果の確認→⑦標準化と管理の定着(歯止め)→⑧反省と今後の計画の順に進めます。
The standard step-by-step procedure for conducting quality improvement and problem solving based on factual data is called the "QC Story." It proceeds in order: ① Theme selection → ② Grasping current status & setting targets → ③ Activity plan → ④ Analysis of causes (identifying root causes) → ⑤ Studying & implementing countermeasures → ⑥ Confirming effects → ⑦ Standardization & establishing control (preventing recurrence) → ⑧ Reflection & future plans.
また、消費者が安心・満足して製品を長期間使用できるよう品質を請け負う体系的活動が「品質保証」です。万一、製品の欠陥によって消費者の生命・身体または財産(その製品自体の損害のみの場合は除く)に被害が生じた場合に被害者を救済する法律が「製造物責任法(PL法)」です。
Systematic activity guaranteeing quality so consumers can use products safely and satisfactorily over a long period is "Quality Assurance." If a product defect causes damage to a consumer's life, body, or property (excluding damage solely to the product itself), the law providing victim relief is the "Product Liability Act (PL Law)."
民法の原則ではメーカーの「故意・過失」を証明する必要がありましたが、PL法ではメーカーの過失の有無を問わず、製造物に「欠陥(通常有すべき安全性を欠いていること)」と損害との因果関係さえあれば賠償責任を負う「無過失責任(欠陥責任)」を採用しています。
Whereas traditional Civil Code liability required proving the manufacturer's "intent or negligence (fault)," the PL Law adopts "Strict Liability (no-fault / defect liability)"—holding the manufacturer liable for damages regardless of fault as long as the product had a "Defect (lack of safety it normally should provide)" that caused the injury or damage.
PL法の対象となる「製造物」は「製造または加工された動産(不動産・未加工農産物・電気・ソフトウェア単体は対象外)」であり、欠陥は①設計上の欠陥、②製造上の欠陥、③指示・警告上の欠陥(取扱説明書や注意ラベルの不備)の3つに分類されます。
"Products" covered by the PL Law are "manufactured or processed movables (real estate, unprocessed agricultural goods, electricity, and standalone software are excluded)," and defects are classified into three types: ① Design Defects, ② Manufacturing Defects, and ③ Instruction/Warning Defects (inadequate manuals or warning labels).
Key points
- QCストーリーの手順:テーマ選定→現状把握・目標設定→要因解析→対策実施→効果確認→標準化と管理の定着(再発防止の歯止め)。
QC Story steps: Theme selection → Current status & Target → Cause analysis → Countermeasures → Effect confirmation → Standardization & Control (locking in to prevent recurrence). - PL法(製造物責任法)=メーカーの「過失」ではなく、製品の「欠陥」を要件とする「無過失責任(欠陥責任)」。
PL Law (Product Liability Act) = adopts "Strict Liability (Defect Liability)" based on product "Defect" rather than manufacturer "Fault/Negligence." - PL法の対象は「製造・加工された動産」であり、3つの欠陥は「設計上の欠陥・製造上の欠陥・指示警告上の欠陥」。
PL Law covers "manufactured or processed movables," and its 3 defect types are "Design Defects, Manufacturing Defects, and Instruction/Warning Defects."
| Word | Meaning |
|---|---|
| QCストーリー(問題解決の手順) | QC Story (problem-solving procedure) |
| 製造物責任法(PL法) | Product Liability Act (PL Law) |
| 無過失責任(欠陥責任) | Strict Liability (no-fault / defect liability) |
| 設計・製造・指示警告上の欠陥 | Design, Manufacturing, and Instruction/Warning Defects |
8. FMEA・FTAとISO 9000シリーズ・品質マネジメントの7原則
8. FMEA / FTA, the ISO 9000 Series, and the 7 Quality Management Principles
設計・工程段階で故障や不良を未然防止する信頼性解析手法に「FMEA」と「FTA」があります。「FMEA(故障モード影響解析)」は、構成部品のあらゆる故障モード(断線・摩耗など)から出発してシステム全体への影響を上向きに調べる「ボトムアップ手法」です。
Two reliability analysis techniques used to prevent failures and defects at the design and process stages are "FMEA" and "FTA." "FMEA (Failure Mode and Effects Analysis)" is a "Bottom-Up Method" that starts from potential failure modes of individual component parts (such as wire breakage or wear) and traces upward to evaluate their impact on the overall system.
これに対し「FTA(故障の木解析)」は、起きてはならない重大事故(頂上事象)から出発し、その発生原因を論理記号(ANDゲート・ORゲート)で木の根のように下向きに展開して追究する「トップダウン手法」です。
In contrast, "FTA (Fault Tree Analysis)" is a "Top-Down Method" that starts from an undesirable major accident (the Top Event) and unfolds its causal pathways downward like tree roots using logic symbols (AND gates, OR gates) to identify root causes.
一方、国際標準化機構(ISO)が定めた品質マネジメントシステム(QMS)の国際規格群が「ISO 9000シリーズ」です。基本概念と用語を定めた「ISO 9000」、第三者認証の基準となる要求事項を定めた「ISO 9001」、組織の持続的成功への指針「ISO 9004」、監査指針「ISO 19011」で構成されます。
Meanwhile, the international standard family for Quality Management Systems (QMS) established by the International Organization for Standardization (ISO) is the "ISO 9000 Series." It consists of "ISO 9000" (fundamentals and vocabulary), "ISO 9001" (requirements used for third-party certification), "ISO 9004" (guidelines for sustained success), and "ISO 19011" (auditing guidelines).
ISO 9000シリーズが掲げる「品質マネジメントの7原則」は、①顧客重視、②リーダーシップ、③人々の積極的参加、④プロセスアプローチ、⑤改善、⑥客観的事実(エビデンス)に基づく意思決定、⑦関係性管理です。
The "7 Quality Management Principles" set forth in the ISO 9000 series are: ① Customer focus, ② Leadership, ③ Engagement of people, ④ Process approach, ⑤ Improvement, ⑥ Evidence-based decision making, and ⑦ Relationship management.

Key points
- FMEA(故障モード影響解析)=部品の故障から全体への影響を積み上げる「ボトムアップ手法」。
FMEA (Failure Mode and Effects Analysis) = "Bottom-Up Method" building upward from component failure modes to system-wide effects. - FTA(故障の木解析)=頂上事象(重大トラブル)から論理記号で原因を掘り下げる「トップダウン手法」。
FTA (Fault Tree Analysis) = "Top-Down Method" drilling downward from a Top Event (major failure) into causes using logic gate symbols. - ISO 9001=品質マネジメントシステムの「要求事項(認証取得の対象規格)」であり、7原則(顧客重視・プロセスアプローチ等)に基づく。
ISO 9001 = specifies the "Requirements (standard for certification)" of a Quality Management System, grounded in the 7 Principles (Customer focus, Process approach, etc.).
| Word | Meaning |
|---|---|
| FMEA(故障モード影響解析:ボトムアップ) | FMEA (Failure Mode and Effects Analysis: bottom-up) |
| FTA(故障の木解析:トップダウン) | FTA (Fault Tree Analysis: top-down) |
| ISO 9001(品質マネジメントシステム要求事項) | ISO 9001 (Quality Management System requirements) |
| 品質マネジメントの7原則 | 7 Quality Management Principles |
試験によく出るポイントと覚え方
High-Frequency Exam Points and Memory Aids
品質管理の計算では、データ(例:5個の測定値)から「平均値・中央値(メディアン)・範囲R(最大-最小)・不偏分散V(偏差平方和S÷(n-1))」を求める問題が定番です。分散はnではなく「n-1」で割る点に注意してください。
In Quality Management calculations, a staple question gives a small dataset (e.g., 5 measurements) and asks for the "Mean, Median, Range R (Max - Min), and Unbiased Variance V (Sum of Squares S ÷ (n - 1))." Be careful to divide by "n - 1" rather than n when finding unbiased variance.
データの分類では「計量値(長さ・重さ・温度・引張強さ=正規分布)」と「計数値(不良品数=二項分布、傷・欠点数=ポアソン分布)」の組み合わせが頻出です。
For data classification, exams frequently test the pairings: "Continuous Data (length, weight, temperature, tensile strength = Normal Distribution)" and "Discrete Data (number of defectives = Binomial Distribution; number of scratches/defects = Poisson Distribution)."
QC七つ道具では、ヒストグラムの「絶壁型(規格外の選別除去)」「二山型(平均の異なる2群の混入)」「櫛歯型(区間幅のとり方のミス)」と、管理図の「±3σ」「限界内でも連や傾向があれば異常」が狙われます。
In the QC 7 Tools, expect questions on Histogram shapes—"Cliff-Like (100% screening removal of out-of-spec items)," "Twin-Peak (mixing two groups with different means)," and "Comb-Shaped (improper class width)"—and Control Chart rules: "±3σ limits" and "even within control limits, a Run or Trend indicates an abnormality."
また抜取検査の「生産者危険α(あわてもの=良いロットを不合格)」と「消費者危険β(ぼんやりもの=悪いロットを合格)」、および「FMEA=ボトムアップ、FTA=トップダウン(木の上から下へ)」はセットで暗記しましょう。
Memorize these pairs together: "Producer's Risk α (hasty error = rejecting a good lot)" vs. "Consumer's Risk β (careless error = accepting a bad lot)," and "FMEA = Bottom-Up vs. FTA = Top-Down (from the top of the tree downward)."
Key points
- 計量値=正規分布/計数値=不良個数(二項分布)・欠点数(ポアソン分布)/不偏分散V=偏差平方和÷(n-1)。
Continuous Data = Normal Distribution / Discrete Data = Defectives (Binomial Distribution), Defects (Poisson Distribution) / Unbiased Variance V = Sum of Squares ÷ (n - 1). - ヒストグラム:絶壁型=規格外の全数選別/二山型=異なる2台の機械・材料の混在/離れ小島型=異質データの混入。
Histogram: Cliff-like = 100% screening of nonconforming units / Twin-peak = mixing 2 machines or materials / Isolated island = inclusion of abnormal data. - 生産者危険α(あわてもの)=良いロットを不合格/消費者危険β(ぼんやりもの)=悪いロットを合格/FMEA=ボトムアップ、FTA=トップダウン。
Producer's Risk α (hasty) = rejecting a good lot / Consumer's Risk β (careless) = accepting a bad lot / FMEA = Bottom-Up, FTA = Top-Down.
| Word | Meaning |
|---|---|
| 計量値と計数値の分布 | distributions of continuous data and discrete data |
| 不偏分散(S÷(n-1)) | Unbiased Variance (S ÷ (n - 1)) |
| 生産者危険αと消費者危険β | Producer's Risk α and Consumer's Risk β |
| FMEA(ボトムアップ)とFTA(トップダウン) | FMEA (Bottom-Up) and FTA (Top-Down) |
Quality Control (QC7, Inspection, ISO 9001)
Practice this chapter in the appOfficial references
- 中央職業能力開発協会(JAVADA) ビジネス・キャリア検定試験 3級 生産管理オペレーション 試験範囲 — https://www.javada.or.jp/jigyou/gino/business/subject.html
- 中央職業能力開発協会 編『ビジネス・キャリア検定試験 標準テキスト【専門知識】生産管理オペレーション 3級〔第4版〕』 — https://www.javada.or.jp/jigyou/gino/business/text.html