Six Sigma Study Notes by Robert Perrine Release 1.1 October 2007

Overview This document is in an early stage of development. At this time, this document serves primarily as my study notes on Six Sigma. For this initial release the contents are notes taken while reading books or listening to audio recordings. The coverage is uneven. If I am already familiar with the content in a section, then I am taking few notes. If the material do not congeal in my mind, then I left it spares as I will come back to it at a later time. The order of these notes does not reflect the order in which I studied the material. Rather it represents the relevance I felt this material had to my immediate purpose. To a large extent, this first collection of notes is a glossary or a list of abbreviations with insufficient explanation. I plan to add explanatory material at a later date.

Thomas Pyzdek; The Six Sigma Handbook; McGraw-Hill; 2003; ISBN 0-07-141015-5.

INTRODUCTION ASQ=American Society Quality TQM=total Quality Movement MBO=management By Objectives DMAIC=Define, Measure, Analyze, Improve, Control DFSS=Design For Six Sigma DMADV=Define,Measure,Analyze,Design,Verify RR=Repeatability and Reproducibility SPC=Statistical Process Control DOE=Design Of Experiments

CHAPTER 1 WASTE=Potential Quality minus Actual Quality QUALITY=”value added by a productive endeavor” COPQ=Cost Of Poor Quality Note that 3 or 4 SIGMA QUALITY cost 25 - 40% METHOD=   Observe   Hypothesize   Predict   Test   Refine 6 SIGMA=Assumes 1.5 shift so it is actually 4.5 sigma PYZDEK’S LAW=”most of what you know is wrong” CONSTITUENCIES=   Customers   Shareholder   Employee All data must have a direct tie to a shareholder MOTIVATORS=   Be#1   Competitive   Current(tech)   skills(training)   Mandated(gov)   Demand(cust) CHANGE ROLES=   Formal Change Agent   Sponsor   Advocate   Informal Change Agent CHANGE GOALS=   Thinking   Norms   Processes CHANGE TOOLS=   Training   Education   SWOT   Time How to implement   1) Train senior executives in Six Sigma philosophy   2) Build customer feedback loops   3) Skills training for everyone   4) Metrics to feed continuous improvement   5) Choose strategic business process improvement   6) Execute the projects AHP=Analytic Hierarchy Process PELT=Process Excellence Leadership Team SCM=Supply Chain Management

CHAPTER 2 JIONER=   Customer Satisfaction   Cycle Time   Quality on first pass ROSE=   Category   >Goal   >Indicator+Definition   >Measure   >Parameter   >Means   >Notational   >Specific Metrics BALANCED SCORECARD=   Customer   Finance   Process   Learning CUSTOMER=   Quality   Timeliness   Service   Value OLAP=On-Line Analytic Processing STUDIES=Retrospective or Prospective FMEA=Failure Mode and Effects Analysis DASHBOARDS=   Trendline   + Histogram   + Process Contol   + Explainatory DRILL DOWN=Disaggregating dashboard metric DATA MINING   Goal definition   Data selection,   Data preparation   Data exploration,   Pattern discovery   Pattern deployment,   Pattern presentation   Business intelligence,   Data scoring and labeling,   Decision support systems   Alarm monitoring,   Pattern validity monitoring CAMP on BENCHMARKING   Identify need, plan and collect the data   Analysis   Communicate and set goals   Improve   Continuous improvement

CHAPTER 3 EDOSOMWAN=   Customer focused organization   Well defined market   Excellent quality   Competitive product   Satisfying customer needs and wants RESPONSES=   Open ended   Fill in the blank   Yes/No   Ranking   Rating   Guttman (higher alternatives include the lower)   Likert (intensity scale)   Semantic (unlabeled scale) CIT=Critical Incident Technique (Flanagan) QUESTIONS=   Relevant   Concise   Unambiguous   Focused   No double negatives KANO=   Basic   Expected and Exciting Quality Pyzdek notes that Six Sigma addresses Basic and Expected Quality. It takes innovation and creativity to get to Exciting Quality. QFD=Quality Function Deployment QFD PHASES=   Organization   Descriptive   Breakthrough   Implementation QFD DIAGRAMS=   Requirements (features) matrix   “house of quality” TRAGEDY OF THE COMMONS=though seeking the common good is obvious, each will seek their own good even when it detracts from the common good. For example, each fisherman seeks the maximum even though the end result is to overfish the common resource. DECISION RIGHTS MATRIX=ARCI for a process PELT=Process Excellence Leadership Teams PEX=Process EXcellence, like a PMO for six sigma BPE=Business Process Executive STRATEGY DEPLOYMENT MATRIX   1. Matrix to list strategies vs metrics   2. Assess relationship between each   3. Assign a weight to each

CHAPTER 4 KSA=knowledge, skills and abilities Training needs assessment   Process audit   Assessment of knowledge, skills and abilities   Assessment of employee attitudes Audiences   Leadership=VISION, communication, conflict, ethics   BB=Methodology, especially statistical analysis   GB=Vocabulary+introduction to tools   Change Agents=Coach,mentor,negotiate,conflict   Facilitator=Communication including presentations JIT=Just in time JITT=Just in time training Aim for JIT Training for higher knowledge retention Training Evaluation   Reaction   Learning   Behavior   Results   Return on Investment

CHAPTER 5 Process Improvement=Kaizen or Breakthrough Reasons Quality Circles Failed   Could not enlist cooperation across departments   Management moved too frequently   Workers moved too frequently   (Team building takes time) Empowering   Bi-directional communications   Training and knowledge   Cooperative environment   Trust and acceptance   Feedback Conflict Resolution   Facts not opinions   Do not force a win-loose position   Persuasion not withdrawal   Consensus from a shared perspective on a point   Endorse diversity of perspectives overall   Beware of agreements reached in a vacumn   Avoid vote swapping and influence peddling   Envision the possible Team Formation (bruce W. Tuckerman, 1965)   Forming   Storming   Norming   Performing Team Problems   Floundering   Know-it-all   Dominance   Reluctance   Not fact focused   Rushing   Attribution (mind reading)   Ignoring   Wandering   Feuds Team Roles   Innovator   Researcher   Listener   Presenter   Perspective   Elaborator   Associative   Visionary   Judge   Motivator   Parlimentarian   Recorder

CHAPTER 6 Project Selection Criteria   excellent worksheets on pages 191 - 197 PPI=Pareto Priority Index=   (Benefits*Probability) / (Cost*Years) TOC=Theory of Constraints (Goldratt)   1 Identify the constraints   2 Exploit the constraint   3 Subordinate everything else   4 Resolve the constraint if #2 and #3 did not   5 Reassess CTQ=Critical to Quality CTC=Critical to COST CTS=Critical to SCHEDULE CTx=CTQ+CTC+CTS Synchronizer Resource   Key resources are required on multiple projects   Multitasking is inefficient   Therefore, single thread+buffer the key resource   TVM=Time Value of Money EBIT=Earnings Before Income Tax   EBIT = Q * (P - V) - F   Q=Quantity or number of units sold   P=Price per unit   V=Variable costs per unit   F=Fixed costs per unit FV=Future Value   FV = PV * (1 + I)**N   PV=Present Value   I=interest rate   N=number of terms OR, for continuous compounding   FV = PV * e**(I * N)   Where e is Euler’s constant of 2.71828 NPV=Net Present Value IRR=Internal Rate of Return Juran (1951) Goal is zero defects Cost of Poor Quality   Prevention costs   Appraisal costs   Failure costs - Internal and External   See table on pages 226 - 228 for details.

CHAPTER 7 Learning Models are process models that incorporate progressive elaboration. Examples include Plan-Do-Check-Act, DMAIC and DMADV. DMAIC details in chart form on pages 238-240 DMADV overview on page 241 SEL=Select-Experiment-Learn SEA=Select-Experiment-Adapt Deming actually preferred PDSA over PDCA Shewhart   A) data of experience where KNOWING begins   B) predict expected results from experiment   C) certainity assigned to the prediction   Then, running the experiment takes you back to A   Facts exist only within a framework of a theory   Specification->production->inspection->(circular) C.I.Lewis (not C.S.) Notes that all LOGIC is circular

CHAPTER 8 Tools   Process Maps   Flow Charts   Check Sheets      Process      Stratified Defect      Defect Location      Cause and Effect Diagram   Pareto Analysis   Cause and Effect Diagrams Other valuable tools   Nominal group technique (brainstorm+delphi)   Force-field analysis (dynamics/free body diagram) 7M Tools (Management or 7MP management planning)   Affinity diagrams   Tree diagrams (WBS)   Process decision program charts (fault tree)   Matrix diagrams   Interrelationship diagrams (mindmap)   Prioritization matrices      (replacing matrix data analysis charts)      Full analytic criteria method      (AHP=analytic hierarchy process)      Combination ID/matrix method      Consensus criteria mathod   Activity network diagram

CHAPTER 9 Measurement scales (Harrington, 1992)   Nominal (categories) - attributes   Ordinal (ranking) - groupings   Interval (linear) - arbitrary y=( a*x + b )   like C vs F for temperature, 0C != 0F   Ratio (known zero) - physical y=a*x   Like kg vs lb for weight, zero is still zero Suitability of the measurement system   Bias - fixed by calibration   Repeatibility - precision by same tester + tool   Reproducibility - different people, same tool   Stability - same test at different times   Linearity - variation in bias over range of the tool Enumerative=study of what has happened   Deductive, average, mean, etc Analytic=study of the process to predict outcome   Inductive, control chart over time Inference   Deriving generalizations from facts   Domains=facts and conclusions   Approach=design-based or model-based   Design=Plan so the samples are representative   Model=statistical proof of reasonablness Permutations=n!, eg 4*3*2*1 Combinations=n!/(r! * (n - r)! ), eg 3!/(2!*1!) Resistance=suitable dispite outliers Robust=suitable even with deviation from assumptn Ogive=cumulative graph of frequency   For example the line drawn above a Pareto Mean=Sum(X) / n   X=value from this measurement   N=number of measurements Variance=Sum(X - Mean) / (n - 1) Standard Deviation=SqRt( Variance ) Standard Error=(StdDev) / SqRt(n) Binomial=Combinations() * p**x * (1-p)**(n-x)   Combinations=n!/(r! * (n - r)! )   N=number of measurements   X=number of arrangements/outcomes   P=probability of this outcome on a single item   Binomdist in Excel Poisson=u**x * e**(-u) / x!   U=number of defects in average sample   X=desired number of defects in this sample   E=Eulers constant   Poisson in Excel Hypergeometric=Hypgeomdist in Excel Normal=Normdist in Excel Exponential=Expondist in Excel Chi-square=Chidist in Excel Point estimators like mean, median, etc Confidence interval=estimator with a range   Confidence in Excel Significance level=percent of samples outside limits Hypothesis testing   Type 1 error=true hypothesis rejected   See significance level   Type 2 error=false hypothesis is accepted Distribution properties   Location=placement, eg mean   Spread=width of the distribution, eg std dev   Shape=range of the population   Central Limit Theorem (Shewhart)   Samples will tend towards normal disribution   Regardless of the shape of population SPC=statistical process control   Workflow on page 320   Process(p) provides performance information(pi)   PI is used to identify Actions on the Process(ap)   AP actions produce continuous improvement on P   PI is used to identify Actions on the Output(ao)   AO was the traditional approach via inspection Control (Shewhart)   Variation is within predictable limits   Common cause=random noise   Special cause=requires action

CHAPTER 10 Discrimination=resolution R Chart=Range S Chart=Standard deviation X-bar Chart=Averages Operational definition must include how top measure

CHAPTER 11 RUN CHARTS Prone to Type 2 error - special cause not detected Run length=number of consecutive points on the same side of the median. See table on page 364 for maximun run length guidelines. Number of runs=Table on page 366 for guidelines Trends=consecutive upward or downward runs. See table on page 367 for guidelines. EDA=Exploratory data analysis (patterns) (Tukey,1977) Stem and Leaf Plots=Histogram shape with data Box Plot=quartiles

CHAPTER 12 SPC=Statistiical process control   Goal is improved decision making   Use simple, but valid methods   Should have more than 10 values   Fewer than 20% of the values should repeat Chart selection criteria on page 419 R Charts=pages 393 - 398, variable, 2<n<9 S Charts=pages 398 - 401, variable, n>9 Run Charts=pages 401 - 405, variable, n=1,non-norm X-bar Charts=pages 401 - 405, variable, n=1, normal P Charts=pages 406 - 409, count groups, n varies NP Charts=pages 409 - 411, count groups, n fixed U Charts=pages 411 - 416, count units, n varies C Charts=pages 416 - 419, counts units, n fixed Attribute Control Charts   Stabilized attribute control charts=p445-449   Demerit control charts=pages 449 - 452 EWMA=exponentially weighted moving averages

CHAPTER 13 PCA=Process Capability Analysis This chapter has detailed instructions

CHAPTER 14 Excellent summation of mathematical theory Least Squares B=sum((X-Xbar)*(Y-Ybar)) / sum(X - Xbar)**2 A=Ybar - b * Xbar

CHAPTER 15 Summation of projrct management Change cycle   Denial   Anger   Negotiation   Depression + Acceptance   Or Decision + Action

CHAPTER 16 - I took no notes on this chapter

CHAPTER 17 - I took no notes on this chapter

CHAPTER 18 Juran quality trilogy   Planning   Control   Improvement Standardization   Approved processes list   Approved parts list

CHAPTER 19 DFSS=Design for Six Sigma Based on DMADV=   Define   Measure   Analyze   Design   Verify CTQ is minimum   Need to get beyond expected to excited Affinity Diagram to map the CTQ criteria Customer demands translated into design criteria Assign importance weighting to categories and subcategory items Subcategory item weight   *category weight   =global importance weight Simulation and modeling can predict process improvements

CHAPTER 20 Muda=waste   Anything not contributing to the finished product   Defects   Overproduction   Inventories   Unnecessary processing,   Unnecessary movement of people   Unnecessary transport   Waiting   Or a design that does not meet customer requirements Kaizen is a lifestyle, not just a business tool

Minitab; Meet Minitab; Minitab; 2004; ISBN 0-925636-48-7

Subir Chowdhury; The Power of Six Sigma; 2001; Dearborn Trade Publishing; ISBN 0-7931-4434-5 A well written introduction to Six Sigma told as a conversation between two friends.

Howard S. Gitlow and David M. Levine, Six Sigma for Green Belts and Champions, Prentice Hall; 0-13-117262-X

PART 1 DMAIC=   Define   Measure   Analyze   Improve   Control ROLES=   Executive   Executive Committee   Champion   Master Black Belt   Black Belt   Green Belt PDSA=   Plan   Do   Study(check)   Act SDSA=   Standardize   Do   Study   Act UNIT=Area of study for this DMAIC cycle DPU=Defects per Unit DPO=Defects per Opportunity DPMO=Defects per Million Opportunities PROCESS SIGMA=see pg 32 YIELD=ratio good per total RTY=Rolled Throughput Yield=   Probabilty of good yield through ALL of the steps in the process

PART 2 DASHBOARD=   Mission   Key Objectives   Key Indicators   Tasks & Projects OBJECTIVES=   Financial   Process   Customer Satisfaction   Employee Growth INDICATORS=   Attribute   Measure   Binary   List   Gantt FLAG INDICATOR=rollup of dept indicators to corp TASKS=Zero, Increase or Decrease COST REDUCTION=bottom line COST AVOIDANCE=not spent but not bottom line, eg reducing the hours for salaried employee only hits the bottom line if there is other work to do KANO QUESTIONARE=survey with response range KANO CATEGORIES=   If done   If not done   Dollars KANO ANSWERS=   One-way   Must-be   Attractive   Indifferent   Questionable   Reverse COST=Uniform, Triangular, J-shaped 5W1H=Questions, 5 Why + 1 How PART-TO-PART VARIATION=   same operation but different results REPRODUCIBILITY=   same result from different operation   (different person or lab, etc) VARIATION DUE TO GAGES=   Repeat!ility   Calibration   Stability   Linearity BIAS OVER TIME=drift BIAS OVER DOMAIN=variability over range PERCENT R&R=repeatability + reproducibility RESOLUTION=precision R-CHART=range chart MEASUREMENT SYSTEM STUDIES=   MSA checklist   Test-retest study   Gage R&R study MSA=measurement system analysis CHECK SHEET=Attribute, Variable or Location NUISANCE VARIABLES=   Uncontrolled   Controllable   Noise (lurking) DESIGN OF EXPERIMENTS=   Screening   Full factorial   Fractional factorial   Two-factor factorial   Response surface methodology ANOVA=Analysis of variance DIFFUSION=Disseminating the program ACCEPTANCE=   Innovator   Early Adopter   Early Majority   Late Majority   Laggard

PART 3 DESCRIPTIVE STATISTICS=full population INFERENTIAL STATISTICS=sample ENUMERATIVE=point in time ANALYTIC=cause and effect, predictive FRAME=defined subset of the population GAP=underrepresentation of the pop by frame NONPROBABILITY SAMPLE=no frame, biased NONPROBABILITY SAMPLE TYPES=   Convenience   Judgement   Quota PROBABILITY SAMPLE=   from frame so representation is known PROBABILITY SAMPLE TYPES=   Simple random   Stratified   Systemic   Clustered PARETO, “other” catch all is always at far right SAMPLE->statistics FRAME->parameters Probability=   Classical=# desired outcomes / # outcomes   Empirucal=# times observed / # observations   Subjective=expert opinion Collectively exhaustive=entire set of outcomes Mutually exclusive=both cannot occur same time Joint probability=both occur Binomial distribution   Fixed number of data points   Mutually exclusive   Probability of an outcome is known and stable   Each outcome is independent of prior outcomes   Example - rolling dice Poisson distribution   Population is bounded and sample is a count   Probability of an outcome is known and stable   Each sample interval is independent of others   Shrinking boundary will cause count to go to zero   Example - customers queueing per minute Hypothesis testing   ART=Alpha error is Rejecting when True   BAF=Beta error is Accepting when False Pg 454 flow chart for choosing the right control chart

Pete Pande and Larry Holpp, What is Six Sigma, McGraw-Hill; 0-07-138185-6

CTQ=Critical To Quality DMAIC=Define, Measure, Analyze, Improve, Control STEPS=Selection, Formation, Charter, Doing, Transition CAUSES=Methods, Machines, Materials, Measures, Mother-Nature, People GAINS=Experience, Exposure, Excitement, Enlighten SIPOC=Supplier, Input, Process, Output, Customer CHARTS=X-causes,Y-results

Greg Brue; Six Sigma for Managers; 2002; McGraw-Hill; ISBN 0-07-138755-2

Capability Index=(USL - LSL)/6s OR lesser of (USL - mean)/3s or (mean - LSL)/3s Benchmakring Clearinghouse www.apqc.org/free/conduct.cfm

Pete S. Pande, Robert P. Neuman and Roldand R. Cavanagh, The Six Sigma Way, McGraw-Hill Audio; 0-9724889-5-2 (audio CD)

CPQ=Cost of Poor Quality

Jay Arthur; Six Sigma Simplifies - Green Belt Training Made Easy; 2004: LifeStar; ISBN 1-884180-19-1 Audio cassette training and motivational presentation

Customers desire   Better   Faster   Cheaper Just in time training because 80% of classroom learning is lost if not applied. Search for the “mother load” - don’t just pan for gold.   Goal is breakthrough projects, not just incremental. Use quick-win projects   achieve breakthroughs   showcase the value Differentiators   Innovation - like Intel   Satisfaction - like Nordstoms   Efficiency - like Walmart

AICPA; Accounting for the Costs of Computer Software Developed or Obtained for Internal Use (Statement of Position 98-1); American Institute of Certified Public Accountants (AICPA); 1998

PROJECT GOALS   Reduce costs,   Operate more efficiently   Improve internal controls   Service customers better   Gain competitive advantage

J. Mike Jacka and Paulette J. Keller; Business Process Mapping, Improving Customer Satisfaction; John Wiley and Sons; 2002; ISBN 0-471-07977-4 A useful reference. I did not take notes while reading this as it aligned well with existing practice.