Six Sigma Study Guide

📖 Core Concepts Six Sigma – Data‑driven methodology to reduce variation and eliminate defects; goal ≤ 3.4 DPMO (defects per million opportunities). Sigma Level – Number of standard deviations between process mean and the nearest specification limit; higher sigma = fewer defects. 1.5 Sigma Shift – Long‑term performance assumed to drift 1.5 σ worse than short‑term studies; a 4.5 σ short‑term process is labeled “Six Sigma” because of this shift. DMAIC – Core problem‑solving cycle for existing processes: Define → Measure → Analyze → Improve → Control. DMADV – Framework for designing new processes/products (Define‑Measure‑Analyze‑Design‑Verify). Lean Six Sigma – Fusion of Lean’s waste‑elimination with Six Sigma’s variation‑reduction; uses tools from both families. Process Capability Index (Cpk) – Quantifies how well a process fits within specification limits (higher Cpk = better capability). 📌 Must Remember Target defect level: ≤ 3.4 DPMO = Six Sigma. DPMO formula: $\text{DPMO} = \frac{\text{defects}}{\text{opportunities} \times \text{units}} \times 10^{6}$. 1.5 σ shift converts a 4.5 σ short‑term capability to a “Six Sigma” claim. DMAIC order is fixed; each phase supplies inputs for the next. Roles hierarchy: Executive → Champion → Master Black Belt → Black Belt → Green Belt. Key Six Sigma tools: 5 Whys, fishbone diagram, SIPOC, control charts, DOE, Pareto, CTQ tree. Lean vs. Six Sigma: Lean = waste elimination; Six Sigma = defect/variation elimination. 🔄 Key Processes DMAIC Workflow Define: Scope project, map SIPOC, capture CTQs, set goal (e.g., reduce DPMO by X%). Measure: Collect baseline data, compute current $C{pk}$, plot control/run charts. Analyze: Use cause‑and‑effect diagrams, 5 Whys, regression/ANOVA to pinpoint root causes. Improve: Conduct Design of Experiments (DOE) or Taguchi methods, implement mistake‑proofing (poka‑yoke), pilot changes. Control: Establish SPC charts, standard operating procedures, and hand‑off to process owner. DMADV Workflow Define: Align design goals with customer requirements and strategic objectives. Measure: Identify Critical‑to‑Quality (CTQ) characteristics, assess capability of proposed design. Analyze: Generate and evaluate alternative designs (DOE, simulation). Design: Select optimal alternative; detail specifications. Verify: Pilot the design, validate performance, transition to production. 🔍 Key Comparisons Lean vs. Six Sigma Goal: Lean → eliminate non‑value‑added waste; Six Sigma → reduce process variation/defects. Tool emphasis: Lean uses value‑stream mapping, 5S; Six Sigma relies on statistical analysis, control charts. DMAIC vs. DMADV When to use: DMAIC for improving an existing process; DMADV for creating a new process/product. Black Belt vs. Green Belt Commitment: Black Belt – full‑time project leader; Green Belt – applies tools while performing regular duties. ⚠️ Common Misunderstandings “Six Sigma = zero defects.” – It actually means ≤ 3.4 DPMO, not absolute perfection. Confusing the 1.5 σ shift with an extra safety margin; it’s a statistical adjustment for long‑term drift, not a design target. Assuming any statistical test is Six Sigma. Only tests that support process capability and root‑cause analysis (e.g., ANOVA, regression) are core. Believing Lean alone can achieve Six Sigma levels. Lean improves speed but does not guarantee the statistical rigor needed for σ‑level goals. 🧠 Mental Models / Intuition “Variation is the enemy of predictability.” Visualize a bell curve: the tighter (smaller σ), the fewer points fall outside specs. “Measure → Understand → Fix.” Data first, hypotheses second, experiments third – prevents “solution‑looking‑for‑a‑problem.” “Control is the safety net.” Even after improvement, continuous monitoring (SPC) ensures gains are not lost. 🚩 Exceptions & Edge Cases Short‑term vs. long‑term capability: A process may show a high short‑term $C{pk}$ but drop after the 1.5 σ shift; always adjust for the shift when claiming Six Sigma. Non‑normal data: If process data are skewed, standard sigma calculations may mislead; consider transformations or non‑parametric methods. Small sample sizes: Inferential statistics (p‑values, confidence intervals) become unreliable; ensure sufficient data before drawing conclusions. 📍 When to Use Which DMAIC → Existing process with measurable defects and a need for incremental improvement. DMADV → Designing a new product/process or when current capability cannot meet specifications even after improvement. Lean tools (5S, value‑stream mapping) → Early stage to identify and eliminate obvious waste before statistical analysis. Statistical tools (ANOVA, regression, DOE) → When root causes involve multiple interacting variables. 👀 Patterns to Recognize Pareto principle: 80 % of defects often stem from 20 % of causes – look for a few dominant factors in fishbone diagrams. Shifted control limits: Control charts that consistently drift toward a spec limit signal a hidden systematic change → trigger a re‑analysis. High $C{pk}$ but low $P{pk}$: Indicates the process is centered now but historically was off‑center; watch for recent changes. 🗂️ Exam Traps Choosing “Lean” over “Six Sigma” when a question asks for statistical reduction of defects – Lean alone lacks the required analysis. Ignoring the 1.5 σ shift and calculating DPMO directly from short‑term data → answer will be too optimistic. Mixing up roles: Selecting “Champion” as the person who performs data analysis (actually the Black Belt/ Master Black Belt does). Misreading “process capability” vs. “process performance”: $C{pk}$ compares to specs; $P{pk}$ reflects actual performance – exam may ask for the correct index. Assuming any control chart proves Six Sigma compliance. You must also demonstrate $C{pk} \ge 2.0$ (≈ 6 σ) after shift adjustment.